{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.26,
  "eval_steps": 500,
  "global_step": 26000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1e-05,
      "grad_norm": 1.6317715205474108,
      "learning_rate": 3e-06,
      "loss": 10.867,
      "step": 1
    },
    {
      "epoch": 2e-05,
      "grad_norm": 1.6173147870740345,
      "learning_rate": 6e-06,
      "loss": 10.8657,
      "step": 2
    },
    {
      "epoch": 3e-05,
      "grad_norm": 1.6387509359885835,
      "learning_rate": 9e-06,
      "loss": 10.8658,
      "step": 3
    },
    {
      "epoch": 4e-05,
      "grad_norm": 1.597552357815991,
      "learning_rate": 1.2e-05,
      "loss": 10.865,
      "step": 4
    },
    {
      "epoch": 5e-05,
      "grad_norm": 1.6454471252189307,
      "learning_rate": 1.5e-05,
      "loss": 10.8617,
      "step": 5
    },
    {
      "epoch": 6e-05,
      "grad_norm": 1.6407925722175996,
      "learning_rate": 1.8e-05,
      "loss": 10.8593,
      "step": 6
    },
    {
      "epoch": 7e-05,
      "grad_norm": 1.6096088910322361,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 10.8456,
      "step": 7
    },
    {
      "epoch": 8e-05,
      "grad_norm": 1.4682698515009915,
      "learning_rate": 2.4e-05,
      "loss": 10.8184,
      "step": 8
    },
    {
      "epoch": 9e-05,
      "grad_norm": 1.3934246427009196,
      "learning_rate": 2.7e-05,
      "loss": 10.8113,
      "step": 9
    },
    {
      "epoch": 0.0001,
      "grad_norm": 1.3326087040550991,
      "learning_rate": 3e-05,
      "loss": 10.7969,
      "step": 10
    },
    {
      "epoch": 0.00011,
      "grad_norm": 1.2173090118888668,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 10.7801,
      "step": 11
    },
    {
      "epoch": 0.00012,
      "grad_norm": 1.176457724285593,
      "learning_rate": 3.6e-05,
      "loss": 10.7688,
      "step": 12
    },
    {
      "epoch": 0.00013,
      "grad_norm": 1.1304424318539597,
      "learning_rate": 3.9e-05,
      "loss": 10.7498,
      "step": 13
    },
    {
      "epoch": 0.00014,
      "grad_norm": 1.1158244568462428,
      "learning_rate": 4.2000000000000004e-05,
      "loss": 10.739,
      "step": 14
    },
    {
      "epoch": 0.00015,
      "grad_norm": 1.10654756415174,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 10.7299,
      "step": 15
    },
    {
      "epoch": 0.00016,
      "grad_norm": 1.0751770608444569,
      "learning_rate": 4.8e-05,
      "loss": 10.7126,
      "step": 16
    },
    {
      "epoch": 0.00017,
      "grad_norm": 1.046143788290158,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 10.6968,
      "step": 17
    },
    {
      "epoch": 0.00018,
      "grad_norm": 1.0230391412556632,
      "learning_rate": 5.4e-05,
      "loss": 10.6806,
      "step": 18
    },
    {
      "epoch": 0.00019,
      "grad_norm": 0.9869982733638543,
      "learning_rate": 5.7e-05,
      "loss": 10.6649,
      "step": 19
    },
    {
      "epoch": 0.0002,
      "grad_norm": 0.9728818553338922,
      "learning_rate": 6e-05,
      "loss": 10.6526,
      "step": 20
    },
    {
      "epoch": 0.00021,
      "grad_norm": 0.9439994749998407,
      "learning_rate": 6.3e-05,
      "loss": 10.639,
      "step": 21
    },
    {
      "epoch": 0.00022,
      "grad_norm": 0.9279528910342688,
      "learning_rate": 6.599999999999999e-05,
      "loss": 10.6244,
      "step": 22
    },
    {
      "epoch": 0.00023,
      "grad_norm": 0.9208663519602571,
      "learning_rate": 6.9e-05,
      "loss": 10.6103,
      "step": 23
    },
    {
      "epoch": 0.00024,
      "grad_norm": 0.9178551557561957,
      "learning_rate": 7.2e-05,
      "loss": 10.5993,
      "step": 24
    },
    {
      "epoch": 0.00025,
      "grad_norm": 0.9193923250060233,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.5847,
      "step": 25
    },
    {
      "epoch": 0.00026,
      "grad_norm": 0.9190901609677985,
      "learning_rate": 7.8e-05,
      "loss": 10.5717,
      "step": 26
    },
    {
      "epoch": 0.00027,
      "grad_norm": 0.913753327244254,
      "learning_rate": 8.1e-05,
      "loss": 10.5597,
      "step": 27
    },
    {
      "epoch": 0.00028,
      "grad_norm": 0.9119625217070774,
      "learning_rate": 8.400000000000001e-05,
      "loss": 10.5467,
      "step": 28
    },
    {
      "epoch": 0.00029,
      "grad_norm": 0.9131038863398008,
      "learning_rate": 8.7e-05,
      "loss": 10.5323,
      "step": 29
    },
    {
      "epoch": 0.0003,
      "grad_norm": 0.9186172829723749,
      "learning_rate": 8.999999999999999e-05,
      "loss": 10.517,
      "step": 30
    },
    {
      "epoch": 0.00031,
      "grad_norm": 0.9155453723962563,
      "learning_rate": 9.3e-05,
      "loss": 10.5024,
      "step": 31
    },
    {
      "epoch": 0.00032,
      "grad_norm": 0.909575589137279,
      "learning_rate": 9.6e-05,
      "loss": 10.4882,
      "step": 32
    },
    {
      "epoch": 0.00033,
      "grad_norm": 0.9102175928717151,
      "learning_rate": 9.900000000000001e-05,
      "loss": 10.4717,
      "step": 33
    },
    {
      "epoch": 0.00034,
      "grad_norm": 0.9103991171564593,
      "learning_rate": 0.00010200000000000001,
      "loss": 10.4554,
      "step": 34
    },
    {
      "epoch": 0.00035,
      "grad_norm": 0.9164468459870767,
      "learning_rate": 0.00010500000000000002,
      "loss": 10.4388,
      "step": 35
    },
    {
      "epoch": 0.00036,
      "grad_norm": 0.9129850089149896,
      "learning_rate": 0.000108,
      "loss": 10.4213,
      "step": 36
    },
    {
      "epoch": 0.00037,
      "grad_norm": 0.9029808327462479,
      "learning_rate": 0.000111,
      "loss": 10.4047,
      "step": 37
    },
    {
      "epoch": 0.00038,
      "grad_norm": 0.9074583652458479,
      "learning_rate": 0.000114,
      "loss": 10.3859,
      "step": 38
    },
    {
      "epoch": 0.00039,
      "grad_norm": 0.9101621604618185,
      "learning_rate": 0.000117,
      "loss": 10.3665,
      "step": 39
    },
    {
      "epoch": 0.0004,
      "grad_norm": 0.9144345472354501,
      "learning_rate": 0.00012,
      "loss": 10.3443,
      "step": 40
    },
    {
      "epoch": 0.00041,
      "grad_norm": 0.9045315909874942,
      "learning_rate": 0.000123,
      "loss": 10.3264,
      "step": 41
    },
    {
      "epoch": 0.00042,
      "grad_norm": 0.9101518170592343,
      "learning_rate": 0.000126,
      "loss": 10.3059,
      "step": 42
    },
    {
      "epoch": 0.00043,
      "grad_norm": 0.9147491310031046,
      "learning_rate": 0.000129,
      "loss": 10.2831,
      "step": 43
    },
    {
      "epoch": 0.00044,
      "grad_norm": 0.9138166723808987,
      "learning_rate": 0.00013199999999999998,
      "loss": 10.2617,
      "step": 44
    },
    {
      "epoch": 0.00045,
      "grad_norm": 0.9166634386783579,
      "learning_rate": 0.000135,
      "loss": 10.2388,
      "step": 45
    },
    {
      "epoch": 0.00046,
      "grad_norm": 0.9061319871747918,
      "learning_rate": 0.000138,
      "loss": 10.2171,
      "step": 46
    },
    {
      "epoch": 0.00047,
      "grad_norm": 0.9134144094551727,
      "learning_rate": 0.000141,
      "loss": 10.1906,
      "step": 47
    },
    {
      "epoch": 0.00048,
      "grad_norm": 0.9111015850262806,
      "learning_rate": 0.000144,
      "loss": 10.1669,
      "step": 48
    },
    {
      "epoch": 0.00049,
      "grad_norm": 0.9106086112424904,
      "learning_rate": 0.000147,
      "loss": 10.1439,
      "step": 49
    },
    {
      "epoch": 0.0005,
      "grad_norm": 0.9135108001899231,
      "learning_rate": 0.00015000000000000001,
      "loss": 10.1178,
      "step": 50
    },
    {
      "epoch": 0.00051,
      "grad_norm": 0.9152031108656089,
      "learning_rate": 0.000153,
      "loss": 10.0918,
      "step": 51
    },
    {
      "epoch": 0.00052,
      "grad_norm": 0.9133043896843657,
      "learning_rate": 0.000156,
      "loss": 10.0658,
      "step": 52
    },
    {
      "epoch": 0.00053,
      "grad_norm": 0.9039676544194273,
      "learning_rate": 0.000159,
      "loss": 10.0419,
      "step": 53
    },
    {
      "epoch": 0.00054,
      "grad_norm": 0.9245050218484777,
      "learning_rate": 0.000162,
      "loss": 10.0099,
      "step": 54
    },
    {
      "epoch": 0.00055,
      "grad_norm": 0.915653013423474,
      "learning_rate": 0.000165,
      "loss": 9.9858,
      "step": 55
    },
    {
      "epoch": 0.00056,
      "grad_norm": 0.90743999026624,
      "learning_rate": 0.00016800000000000002,
      "loss": 9.9567,
      "step": 56
    },
    {
      "epoch": 0.00057,
      "grad_norm": 0.9125740935300273,
      "learning_rate": 0.000171,
      "loss": 9.9292,
      "step": 57
    },
    {
      "epoch": 0.00058,
      "grad_norm": 0.9103134473221595,
      "learning_rate": 0.000174,
      "loss": 9.9046,
      "step": 58
    },
    {
      "epoch": 0.00059,
      "grad_norm": 0.91502039796166,
      "learning_rate": 0.000177,
      "loss": 9.8727,
      "step": 59
    },
    {
      "epoch": 0.0006,
      "grad_norm": 0.902549468432534,
      "learning_rate": 0.00017999999999999998,
      "loss": 9.8467,
      "step": 60
    },
    {
      "epoch": 0.00061,
      "grad_norm": 0.9011753674575653,
      "learning_rate": 0.000183,
      "loss": 9.8184,
      "step": 61
    },
    {
      "epoch": 0.00062,
      "grad_norm": 0.9073829944096542,
      "learning_rate": 0.000186,
      "loss": 9.7865,
      "step": 62
    },
    {
      "epoch": 0.00063,
      "grad_norm": 0.900277782228101,
      "learning_rate": 0.000189,
      "loss": 9.7594,
      "step": 63
    },
    {
      "epoch": 0.00064,
      "grad_norm": 0.8964836174343672,
      "learning_rate": 0.000192,
      "loss": 9.7292,
      "step": 64
    },
    {
      "epoch": 0.00065,
      "grad_norm": 0.9040253232587138,
      "learning_rate": 0.00019500000000000002,
      "loss": 9.6969,
      "step": 65
    },
    {
      "epoch": 0.00066,
      "grad_norm": 0.8969700410935529,
      "learning_rate": 0.00019800000000000002,
      "loss": 9.6728,
      "step": 66
    },
    {
      "epoch": 0.00067,
      "grad_norm": 0.8981910064021,
      "learning_rate": 0.000201,
      "loss": 9.643,
      "step": 67
    },
    {
      "epoch": 0.00068,
      "grad_norm": 0.9049113812894196,
      "learning_rate": 0.00020400000000000003,
      "loss": 9.61,
      "step": 68
    },
    {
      "epoch": 0.00069,
      "grad_norm": 0.8903730719674341,
      "learning_rate": 0.00020700000000000002,
      "loss": 9.5824,
      "step": 69
    },
    {
      "epoch": 0.0007,
      "grad_norm": 0.9034192157313848,
      "learning_rate": 0.00021000000000000004,
      "loss": 9.5456,
      "step": 70
    },
    {
      "epoch": 0.00071,
      "grad_norm": 0.8933343163190056,
      "learning_rate": 0.00021299999999999997,
      "loss": 9.5189,
      "step": 71
    },
    {
      "epoch": 0.00072,
      "grad_norm": 0.8968593008835964,
      "learning_rate": 0.000216,
      "loss": 9.4914,
      "step": 72
    },
    {
      "epoch": 0.00073,
      "grad_norm": 0.8960167287531013,
      "learning_rate": 0.00021899999999999998,
      "loss": 9.4574,
      "step": 73
    },
    {
      "epoch": 0.00074,
      "grad_norm": 0.8965354004907367,
      "learning_rate": 0.000222,
      "loss": 9.4304,
      "step": 74
    },
    {
      "epoch": 0.00075,
      "grad_norm": 0.8923958382533757,
      "learning_rate": 0.000225,
      "loss": 9.3946,
      "step": 75
    },
    {
      "epoch": 0.00076,
      "grad_norm": 0.9011054851249423,
      "learning_rate": 0.000228,
      "loss": 9.3667,
      "step": 76
    },
    {
      "epoch": 0.00077,
      "grad_norm": 0.8959973023769764,
      "learning_rate": 0.000231,
      "loss": 9.3374,
      "step": 77
    },
    {
      "epoch": 0.00078,
      "grad_norm": 0.8901362352309407,
      "learning_rate": 0.000234,
      "loss": 9.3055,
      "step": 78
    },
    {
      "epoch": 0.00079,
      "grad_norm": 0.8931237127830396,
      "learning_rate": 0.00023700000000000001,
      "loss": 9.2694,
      "step": 79
    },
    {
      "epoch": 0.0008,
      "grad_norm": 0.8937972757596185,
      "learning_rate": 0.00024,
      "loss": 9.2355,
      "step": 80
    },
    {
      "epoch": 0.00081,
      "grad_norm": 0.8984186266430717,
      "learning_rate": 0.00024300000000000002,
      "loss": 9.2033,
      "step": 81
    },
    {
      "epoch": 0.00082,
      "grad_norm": 0.9060711515558022,
      "learning_rate": 0.000246,
      "loss": 9.1789,
      "step": 82
    },
    {
      "epoch": 0.00083,
      "grad_norm": 0.8964035250431631,
      "learning_rate": 0.00024900000000000004,
      "loss": 9.1434,
      "step": 83
    },
    {
      "epoch": 0.00084,
      "grad_norm": 0.8986377804082708,
      "learning_rate": 0.000252,
      "loss": 9.1119,
      "step": 84
    },
    {
      "epoch": 0.00085,
      "grad_norm": 0.8933390971667627,
      "learning_rate": 0.000255,
      "loss": 9.0885,
      "step": 85
    },
    {
      "epoch": 0.00086,
      "grad_norm": 0.8953786211031305,
      "learning_rate": 0.000258,
      "loss": 9.0521,
      "step": 86
    },
    {
      "epoch": 0.00087,
      "grad_norm": 0.8876773666743288,
      "learning_rate": 0.000261,
      "loss": 9.0302,
      "step": 87
    },
    {
      "epoch": 0.00088,
      "grad_norm": 0.8929152197347487,
      "learning_rate": 0.00026399999999999997,
      "loss": 8.9927,
      "step": 88
    },
    {
      "epoch": 0.00089,
      "grad_norm": 0.8872457913370299,
      "learning_rate": 0.000267,
      "loss": 8.9669,
      "step": 89
    },
    {
      "epoch": 0.0009,
      "grad_norm": 0.8862509419810688,
      "learning_rate": 0.00027,
      "loss": 8.9393,
      "step": 90
    },
    {
      "epoch": 0.00091,
      "grad_norm": 0.8807880266520192,
      "learning_rate": 0.000273,
      "loss": 8.9089,
      "step": 91
    },
    {
      "epoch": 0.00092,
      "grad_norm": 0.8831207589668301,
      "learning_rate": 0.000276,
      "loss": 8.876,
      "step": 92
    },
    {
      "epoch": 0.00093,
      "grad_norm": 0.883866988348631,
      "learning_rate": 0.000279,
      "loss": 8.8462,
      "step": 93
    },
    {
      "epoch": 0.00094,
      "grad_norm": 0.8824853161719922,
      "learning_rate": 0.000282,
      "loss": 8.8199,
      "step": 94
    },
    {
      "epoch": 0.00095,
      "grad_norm": 0.8794572899807177,
      "learning_rate": 0.000285,
      "loss": 8.789,
      "step": 95
    },
    {
      "epoch": 0.00096,
      "grad_norm": 0.8879520784944948,
      "learning_rate": 0.000288,
      "loss": 8.7571,
      "step": 96
    },
    {
      "epoch": 0.00097,
      "grad_norm": 0.8798042444972031,
      "learning_rate": 0.000291,
      "loss": 8.7321,
      "step": 97
    },
    {
      "epoch": 0.00098,
      "grad_norm": 0.8802597894834375,
      "learning_rate": 0.000294,
      "loss": 8.6954,
      "step": 98
    },
    {
      "epoch": 0.00099,
      "grad_norm": 0.8806466323910314,
      "learning_rate": 0.000297,
      "loss": 8.6749,
      "step": 99
    },
    {
      "epoch": 0.001,
      "grad_norm": 0.8777097574069823,
      "learning_rate": 0.00030000000000000003,
      "loss": 8.6485,
      "step": 100
    },
    {
      "epoch": 0.00101,
      "grad_norm": 0.8786755147609817,
      "learning_rate": 0.00030300000000000005,
      "loss": 8.614,
      "step": 101
    },
    {
      "epoch": 0.00102,
      "grad_norm": 0.8680143868447665,
      "learning_rate": 0.000306,
      "loss": 8.5949,
      "step": 102
    },
    {
      "epoch": 0.00103,
      "grad_norm": 0.8748449452730288,
      "learning_rate": 0.000309,
      "loss": 8.5706,
      "step": 103
    },
    {
      "epoch": 0.00104,
      "grad_norm": 0.8670215859388973,
      "learning_rate": 0.000312,
      "loss": 8.5498,
      "step": 104
    },
    {
      "epoch": 0.00105,
      "grad_norm": 0.8687292008731472,
      "learning_rate": 0.000315,
      "loss": 8.5231,
      "step": 105
    },
    {
      "epoch": 0.00106,
      "grad_norm": 0.8611486197845404,
      "learning_rate": 0.000318,
      "loss": 8.4945,
      "step": 106
    },
    {
      "epoch": 0.00107,
      "grad_norm": 0.8521924348129856,
      "learning_rate": 0.000321,
      "loss": 8.4693,
      "step": 107
    },
    {
      "epoch": 0.00108,
      "grad_norm": 0.8581933475380797,
      "learning_rate": 0.000324,
      "loss": 8.4407,
      "step": 108
    },
    {
      "epoch": 0.00109,
      "grad_norm": 0.8524146875206363,
      "learning_rate": 0.000327,
      "loss": 8.421,
      "step": 109
    },
    {
      "epoch": 0.0011,
      "grad_norm": 0.8682563584613229,
      "learning_rate": 0.00033,
      "loss": 8.3983,
      "step": 110
    },
    {
      "epoch": 0.00111,
      "grad_norm": 0.8871469803064874,
      "learning_rate": 0.000333,
      "loss": 8.3685,
      "step": 111
    },
    {
      "epoch": 0.00112,
      "grad_norm": 0.9236879668329372,
      "learning_rate": 0.00033600000000000004,
      "loss": 8.3463,
      "step": 112
    },
    {
      "epoch": 0.00113,
      "grad_norm": 0.9129864456504505,
      "learning_rate": 0.000339,
      "loss": 8.3063,
      "step": 113
    },
    {
      "epoch": 0.00114,
      "grad_norm": 0.8352263501003522,
      "learning_rate": 0.000342,
      "loss": 8.2966,
      "step": 114
    },
    {
      "epoch": 0.00115,
      "grad_norm": 0.8592375580090957,
      "learning_rate": 0.00034500000000000004,
      "loss": 8.2718,
      "step": 115
    },
    {
      "epoch": 0.00116,
      "grad_norm": 0.8674294753896091,
      "learning_rate": 0.000348,
      "loss": 8.2506,
      "step": 116
    },
    {
      "epoch": 0.00117,
      "grad_norm": 0.8276917493567776,
      "learning_rate": 0.000351,
      "loss": 8.2188,
      "step": 117
    },
    {
      "epoch": 0.00118,
      "grad_norm": 0.8476744963131545,
      "learning_rate": 0.000354,
      "loss": 8.2045,
      "step": 118
    },
    {
      "epoch": 0.00119,
      "grad_norm": 0.844358071644388,
      "learning_rate": 0.000357,
      "loss": 8.1926,
      "step": 119
    },
    {
      "epoch": 0.0012,
      "grad_norm": 0.8166594269287538,
      "learning_rate": 0.00035999999999999997,
      "loss": 8.1658,
      "step": 120
    },
    {
      "epoch": 0.00121,
      "grad_norm": 0.8239930081024902,
      "learning_rate": 0.000363,
      "loss": 8.1389,
      "step": 121
    },
    {
      "epoch": 0.00122,
      "grad_norm": 0.8099951959348987,
      "learning_rate": 0.000366,
      "loss": 8.1225,
      "step": 122
    },
    {
      "epoch": 0.00123,
      "grad_norm": 0.830800320388625,
      "learning_rate": 0.000369,
      "loss": 8.1005,
      "step": 123
    },
    {
      "epoch": 0.00124,
      "grad_norm": 0.8139169053139192,
      "learning_rate": 0.000372,
      "loss": 8.0791,
      "step": 124
    },
    {
      "epoch": 0.00125,
      "grad_norm": 0.8112246790149765,
      "learning_rate": 0.000375,
      "loss": 8.0547,
      "step": 125
    },
    {
      "epoch": 0.00126,
      "grad_norm": 0.7922278873371895,
      "learning_rate": 0.000378,
      "loss": 8.0424,
      "step": 126
    },
    {
      "epoch": 0.00127,
      "grad_norm": 0.7755075943975184,
      "learning_rate": 0.000381,
      "loss": 8.0182,
      "step": 127
    },
    {
      "epoch": 0.00128,
      "grad_norm": 0.8028212720713388,
      "learning_rate": 0.000384,
      "loss": 8.0,
      "step": 128
    },
    {
      "epoch": 0.00129,
      "grad_norm": 0.8765129391436198,
      "learning_rate": 0.00038700000000000003,
      "loss": 7.9904,
      "step": 129
    },
    {
      "epoch": 0.0013,
      "grad_norm": 1.1486399983200042,
      "learning_rate": 0.00039000000000000005,
      "loss": 7.9724,
      "step": 130
    },
    {
      "epoch": 0.00131,
      "grad_norm": 1.0507777578095836,
      "learning_rate": 0.000393,
      "loss": 7.9382,
      "step": 131
    },
    {
      "epoch": 0.00132,
      "grad_norm": 0.7511305165281239,
      "learning_rate": 0.00039600000000000003,
      "loss": 7.9248,
      "step": 132
    },
    {
      "epoch": 0.00133,
      "grad_norm": 0.8108077692812641,
      "learning_rate": 0.00039900000000000005,
      "loss": 7.906,
      "step": 133
    },
    {
      "epoch": 0.00134,
      "grad_norm": 0.8301313874662418,
      "learning_rate": 0.000402,
      "loss": 7.8952,
      "step": 134
    },
    {
      "epoch": 0.00135,
      "grad_norm": 0.7123568807247732,
      "learning_rate": 0.00040500000000000003,
      "loss": 7.8651,
      "step": 135
    },
    {
      "epoch": 0.00136,
      "grad_norm": 0.7697556904537746,
      "learning_rate": 0.00040800000000000005,
      "loss": 7.8515,
      "step": 136
    },
    {
      "epoch": 0.00137,
      "grad_norm": 0.7190977621725152,
      "learning_rate": 0.000411,
      "loss": 7.8299,
      "step": 137
    },
    {
      "epoch": 0.00138,
      "grad_norm": 0.7147305335216294,
      "learning_rate": 0.00041400000000000003,
      "loss": 7.8066,
      "step": 138
    },
    {
      "epoch": 0.00139,
      "grad_norm": 0.747529428120578,
      "learning_rate": 0.00041700000000000005,
      "loss": 7.7954,
      "step": 139
    },
    {
      "epoch": 0.0014,
      "grad_norm": 0.6748198408281931,
      "learning_rate": 0.00042000000000000007,
      "loss": 7.7774,
      "step": 140
    },
    {
      "epoch": 0.00141,
      "grad_norm": 0.662142968172009,
      "learning_rate": 0.000423,
      "loss": 7.7644,
      "step": 141
    },
    {
      "epoch": 0.00142,
      "grad_norm": 0.6859796391897652,
      "learning_rate": 0.00042599999999999995,
      "loss": 7.7534,
      "step": 142
    },
    {
      "epoch": 0.00143,
      "grad_norm": 0.6828514310354903,
      "learning_rate": 0.00042899999999999997,
      "loss": 7.7255,
      "step": 143
    },
    {
      "epoch": 0.00144,
      "grad_norm": 0.6490687938821236,
      "learning_rate": 0.000432,
      "loss": 7.7078,
      "step": 144
    },
    {
      "epoch": 0.00145,
      "grad_norm": 0.6453156151137228,
      "learning_rate": 0.000435,
      "loss": 7.7035,
      "step": 145
    },
    {
      "epoch": 0.00146,
      "grad_norm": 0.7166050341593803,
      "learning_rate": 0.00043799999999999997,
      "loss": 7.6672,
      "step": 146
    },
    {
      "epoch": 0.00147,
      "grad_norm": 0.7443563124698165,
      "learning_rate": 0.000441,
      "loss": 7.6627,
      "step": 147
    },
    {
      "epoch": 0.00148,
      "grad_norm": 0.7124453009383569,
      "learning_rate": 0.000444,
      "loss": 7.648,
      "step": 148
    },
    {
      "epoch": 0.00149,
      "grad_norm": 0.6775306354557482,
      "learning_rate": 0.00044699999999999997,
      "loss": 7.6419,
      "step": 149
    },
    {
      "epoch": 0.0015,
      "grad_norm": 0.8906380813028638,
      "learning_rate": 0.00045,
      "loss": 7.6135,
      "step": 150
    },
    {
      "epoch": 0.00151,
      "grad_norm": 1.2892607147030477,
      "learning_rate": 0.000453,
      "loss": 7.605,
      "step": 151
    },
    {
      "epoch": 0.00152,
      "grad_norm": 0.6492280537852009,
      "learning_rate": 0.000456,
      "loss": 7.5875,
      "step": 152
    },
    {
      "epoch": 0.00153,
      "grad_norm": 0.8226748559002907,
      "learning_rate": 0.000459,
      "loss": 7.5783,
      "step": 153
    },
    {
      "epoch": 0.00154,
      "grad_norm": 1.073498340899344,
      "learning_rate": 0.000462,
      "loss": 7.5662,
      "step": 154
    },
    {
      "epoch": 0.00155,
      "grad_norm": 0.772416311968319,
      "learning_rate": 0.000465,
      "loss": 7.5481,
      "step": 155
    },
    {
      "epoch": 0.00156,
      "grad_norm": 0.7999364848474875,
      "learning_rate": 0.000468,
      "loss": 7.5209,
      "step": 156
    },
    {
      "epoch": 0.00157,
      "grad_norm": 0.9435570004081977,
      "learning_rate": 0.000471,
      "loss": 7.5196,
      "step": 157
    },
    {
      "epoch": 0.00158,
      "grad_norm": 0.8364296006508578,
      "learning_rate": 0.00047400000000000003,
      "loss": 7.4997,
      "step": 158
    },
    {
      "epoch": 0.00159,
      "grad_norm": 0.5596750978008624,
      "learning_rate": 0.000477,
      "loss": 7.4825,
      "step": 159
    },
    {
      "epoch": 0.0016,
      "grad_norm": 0.5748856794890025,
      "learning_rate": 0.00048,
      "loss": 7.4616,
      "step": 160
    },
    {
      "epoch": 0.00161,
      "grad_norm": 0.5986411025981033,
      "learning_rate": 0.00048300000000000003,
      "loss": 7.4419,
      "step": 161
    },
    {
      "epoch": 0.00162,
      "grad_norm": 0.5117747222458712,
      "learning_rate": 0.00048600000000000005,
      "loss": 7.4429,
      "step": 162
    },
    {
      "epoch": 0.00163,
      "grad_norm": 0.5509815158764758,
      "learning_rate": 0.0004890000000000001,
      "loss": 7.4259,
      "step": 163
    },
    {
      "epoch": 0.00164,
      "grad_norm": 0.46393808675226217,
      "learning_rate": 0.000492,
      "loss": 7.4236,
      "step": 164
    },
    {
      "epoch": 0.00165,
      "grad_norm": 0.5390679177469344,
      "learning_rate": 0.000495,
      "loss": 7.4006,
      "step": 165
    },
    {
      "epoch": 0.00166,
      "grad_norm": 0.645354949444588,
      "learning_rate": 0.0004980000000000001,
      "loss": 7.3773,
      "step": 166
    },
    {
      "epoch": 0.00167,
      "grad_norm": 1.0877276724000633,
      "learning_rate": 0.000501,
      "loss": 7.3741,
      "step": 167
    },
    {
      "epoch": 0.00168,
      "grad_norm": 1.0909247587015876,
      "learning_rate": 0.000504,
      "loss": 7.3697,
      "step": 168
    },
    {
      "epoch": 0.00169,
      "grad_norm": 0.5084368948335112,
      "learning_rate": 0.0005070000000000001,
      "loss": 7.3418,
      "step": 169
    },
    {
      "epoch": 0.0017,
      "grad_norm": 1.5053615947271437,
      "learning_rate": 0.00051,
      "loss": 7.3492,
      "step": 170
    },
    {
      "epoch": 0.00171,
      "grad_norm": 0.6120595065498764,
      "learning_rate": 0.000513,
      "loss": 7.3094,
      "step": 171
    },
    {
      "epoch": 0.00172,
      "grad_norm": 0.9401489103136018,
      "learning_rate": 0.000516,
      "loss": 7.3189,
      "step": 172
    },
    {
      "epoch": 0.00173,
      "grad_norm": 0.5558816443971099,
      "learning_rate": 0.0005189999999999999,
      "loss": 7.2906,
      "step": 173
    },
    {
      "epoch": 0.00174,
      "grad_norm": 0.7810658223347703,
      "learning_rate": 0.000522,
      "loss": 7.314,
      "step": 174
    },
    {
      "epoch": 0.00175,
      "grad_norm": 0.5409187817835382,
      "learning_rate": 0.000525,
      "loss": 7.2727,
      "step": 175
    },
    {
      "epoch": 0.00176,
      "grad_norm": 0.630636606756854,
      "learning_rate": 0.0005279999999999999,
      "loss": 7.2484,
      "step": 176
    },
    {
      "epoch": 0.00177,
      "grad_norm": 0.5206138592215499,
      "learning_rate": 0.000531,
      "loss": 7.2507,
      "step": 177
    },
    {
      "epoch": 0.00178,
      "grad_norm": 0.5577166707239251,
      "learning_rate": 0.000534,
      "loss": 7.2428,
      "step": 178
    },
    {
      "epoch": 0.00179,
      "grad_norm": 0.4980094016420717,
      "learning_rate": 0.000537,
      "loss": 7.2333,
      "step": 179
    },
    {
      "epoch": 0.0018,
      "grad_norm": 0.5436132024149352,
      "learning_rate": 0.00054,
      "loss": 7.2053,
      "step": 180
    },
    {
      "epoch": 0.00181,
      "grad_norm": 0.4850436660416719,
      "learning_rate": 0.000543,
      "loss": 7.1915,
      "step": 181
    },
    {
      "epoch": 0.00182,
      "grad_norm": 0.686701132410735,
      "learning_rate": 0.000546,
      "loss": 7.1909,
      "step": 182
    },
    {
      "epoch": 0.00183,
      "grad_norm": 0.68400997444651,
      "learning_rate": 0.000549,
      "loss": 7.1802,
      "step": 183
    },
    {
      "epoch": 0.00184,
      "grad_norm": 0.8396027257351396,
      "learning_rate": 0.000552,
      "loss": 7.171,
      "step": 184
    },
    {
      "epoch": 0.00185,
      "grad_norm": 0.7665365037410753,
      "learning_rate": 0.000555,
      "loss": 7.1476,
      "step": 185
    },
    {
      "epoch": 0.00186,
      "grad_norm": 0.5359175667027454,
      "learning_rate": 0.000558,
      "loss": 7.1506,
      "step": 186
    },
    {
      "epoch": 0.00187,
      "grad_norm": 0.5513800735908173,
      "learning_rate": 0.000561,
      "loss": 7.1261,
      "step": 187
    },
    {
      "epoch": 0.00188,
      "grad_norm": 0.5352093004255375,
      "learning_rate": 0.000564,
      "loss": 7.1044,
      "step": 188
    },
    {
      "epoch": 0.00189,
      "grad_norm": 0.5938457818726526,
      "learning_rate": 0.000567,
      "loss": 7.1004,
      "step": 189
    },
    {
      "epoch": 0.0019,
      "grad_norm": 0.5273842405001533,
      "learning_rate": 0.00057,
      "loss": 7.0834,
      "step": 190
    },
    {
      "epoch": 0.00191,
      "grad_norm": 0.47487115515279366,
      "learning_rate": 0.000573,
      "loss": 7.0721,
      "step": 191
    },
    {
      "epoch": 0.00192,
      "grad_norm": 0.6036837698051599,
      "learning_rate": 0.000576,
      "loss": 7.0655,
      "step": 192
    },
    {
      "epoch": 0.00193,
      "grad_norm": 0.3951184212196986,
      "learning_rate": 0.000579,
      "loss": 7.061,
      "step": 193
    },
    {
      "epoch": 0.00194,
      "grad_norm": 0.442083567688087,
      "learning_rate": 0.000582,
      "loss": 7.0548,
      "step": 194
    },
    {
      "epoch": 0.00195,
      "grad_norm": 0.439546882682468,
      "learning_rate": 0.000585,
      "loss": 7.0348,
      "step": 195
    },
    {
      "epoch": 0.00196,
      "grad_norm": 0.46247531692771043,
      "learning_rate": 0.000588,
      "loss": 7.0228,
      "step": 196
    },
    {
      "epoch": 0.00197,
      "grad_norm": 0.4140335072217301,
      "learning_rate": 0.000591,
      "loss": 7.0171,
      "step": 197
    },
    {
      "epoch": 0.00198,
      "grad_norm": 0.3685986320410548,
      "learning_rate": 0.000594,
      "loss": 7.0081,
      "step": 198
    },
    {
      "epoch": 0.00199,
      "grad_norm": 0.4020373564129086,
      "learning_rate": 0.0005970000000000001,
      "loss": 6.9898,
      "step": 199
    },
    {
      "epoch": 0.002,
      "grad_norm": 0.37126410475941546,
      "learning_rate": 0.0006000000000000001,
      "loss": 6.9867,
      "step": 200
    },
    {
      "epoch": 0.00201,
      "grad_norm": 0.3773154493828028,
      "learning_rate": 0.000603,
      "loss": 6.9617,
      "step": 201
    },
    {
      "epoch": 0.00202,
      "grad_norm": 0.3540017416986532,
      "learning_rate": 0.0006060000000000001,
      "loss": 6.9491,
      "step": 202
    },
    {
      "epoch": 0.00203,
      "grad_norm": 0.403279648640721,
      "learning_rate": 0.0006090000000000001,
      "loss": 6.9534,
      "step": 203
    },
    {
      "epoch": 0.00204,
      "grad_norm": 0.5112949618253247,
      "learning_rate": 0.000612,
      "loss": 6.9385,
      "step": 204
    },
    {
      "epoch": 0.00205,
      "grad_norm": 0.7200998739972175,
      "learning_rate": 0.000615,
      "loss": 6.931,
      "step": 205
    },
    {
      "epoch": 0.00206,
      "grad_norm": 1.209379808685074,
      "learning_rate": 0.000618,
      "loss": 6.9351,
      "step": 206
    },
    {
      "epoch": 0.00207,
      "grad_norm": 1.0040442357645134,
      "learning_rate": 0.000621,
      "loss": 6.9279,
      "step": 207
    },
    {
      "epoch": 0.00208,
      "grad_norm": 0.7878464521205251,
      "learning_rate": 0.000624,
      "loss": 6.9129,
      "step": 208
    },
    {
      "epoch": 0.00209,
      "grad_norm": 1.3096135062236434,
      "learning_rate": 0.000627,
      "loss": 6.9067,
      "step": 209
    },
    {
      "epoch": 0.0021,
      "grad_norm": 0.910504669978176,
      "learning_rate": 0.00063,
      "loss": 6.8878,
      "step": 210
    },
    {
      "epoch": 0.00211,
      "grad_norm": 1.3015817966038044,
      "learning_rate": 0.000633,
      "loss": 6.8987,
      "step": 211
    },
    {
      "epoch": 0.00212,
      "grad_norm": 0.5587575104994011,
      "learning_rate": 0.000636,
      "loss": 6.8696,
      "step": 212
    },
    {
      "epoch": 0.00213,
      "grad_norm": 0.8081412049208773,
      "learning_rate": 0.000639,
      "loss": 6.8608,
      "step": 213
    },
    {
      "epoch": 0.00214,
      "grad_norm": 0.6397075273457759,
      "learning_rate": 0.000642,
      "loss": 6.8511,
      "step": 214
    },
    {
      "epoch": 0.00215,
      "grad_norm": 0.5875043250740225,
      "learning_rate": 0.000645,
      "loss": 6.8482,
      "step": 215
    },
    {
      "epoch": 0.00216,
      "grad_norm": 0.6060774535893669,
      "learning_rate": 0.000648,
      "loss": 6.8413,
      "step": 216
    },
    {
      "epoch": 0.00217,
      "grad_norm": 0.5183751970166313,
      "learning_rate": 0.000651,
      "loss": 6.8119,
      "step": 217
    },
    {
      "epoch": 0.00218,
      "grad_norm": 0.5539011900924167,
      "learning_rate": 0.000654,
      "loss": 6.8171,
      "step": 218
    },
    {
      "epoch": 0.00219,
      "grad_norm": 0.5004312163301685,
      "learning_rate": 0.000657,
      "loss": 6.7882,
      "step": 219
    },
    {
      "epoch": 0.0022,
      "grad_norm": 0.4762494747244133,
      "learning_rate": 0.00066,
      "loss": 6.8062,
      "step": 220
    },
    {
      "epoch": 0.00221,
      "grad_norm": 0.353452768224107,
      "learning_rate": 0.0006630000000000001,
      "loss": 6.7814,
      "step": 221
    },
    {
      "epoch": 0.00222,
      "grad_norm": 0.3879096289808107,
      "learning_rate": 0.000666,
      "loss": 6.7696,
      "step": 222
    },
    {
      "epoch": 0.00223,
      "grad_norm": 0.4253518811476648,
      "learning_rate": 0.000669,
      "loss": 6.7658,
      "step": 223
    },
    {
      "epoch": 0.00224,
      "grad_norm": 0.4076338977034843,
      "learning_rate": 0.0006720000000000001,
      "loss": 6.7609,
      "step": 224
    },
    {
      "epoch": 0.00225,
      "grad_norm": 0.4491881195721512,
      "learning_rate": 0.000675,
      "loss": 6.7489,
      "step": 225
    },
    {
      "epoch": 0.00226,
      "grad_norm": 0.4195112098951784,
      "learning_rate": 0.000678,
      "loss": 6.7444,
      "step": 226
    },
    {
      "epoch": 0.00227,
      "grad_norm": 0.371663908330708,
      "learning_rate": 0.0006810000000000001,
      "loss": 6.7174,
      "step": 227
    },
    {
      "epoch": 0.00228,
      "grad_norm": 0.3462952066328263,
      "learning_rate": 0.000684,
      "loss": 6.7197,
      "step": 228
    },
    {
      "epoch": 0.00229,
      "grad_norm": 0.5288966810878937,
      "learning_rate": 0.000687,
      "loss": 6.7178,
      "step": 229
    },
    {
      "epoch": 0.0023,
      "grad_norm": 0.7623592789505088,
      "learning_rate": 0.0006900000000000001,
      "loss": 6.6993,
      "step": 230
    },
    {
      "epoch": 0.00231,
      "grad_norm": 0.9880780315432149,
      "learning_rate": 0.000693,
      "loss": 6.6923,
      "step": 231
    },
    {
      "epoch": 0.00232,
      "grad_norm": 1.028691756937799,
      "learning_rate": 0.000696,
      "loss": 6.7142,
      "step": 232
    },
    {
      "epoch": 0.00233,
      "grad_norm": 0.9501761132787978,
      "learning_rate": 0.0006990000000000001,
      "loss": 6.6946,
      "step": 233
    },
    {
      "epoch": 0.00234,
      "grad_norm": 0.9999298841530961,
      "learning_rate": 0.000702,
      "loss": 6.6948,
      "step": 234
    },
    {
      "epoch": 0.00235,
      "grad_norm": 0.6945930996245869,
      "learning_rate": 0.000705,
      "loss": 6.6675,
      "step": 235
    },
    {
      "epoch": 0.00236,
      "grad_norm": 0.6422608041910052,
      "learning_rate": 0.000708,
      "loss": 6.6513,
      "step": 236
    },
    {
      "epoch": 0.00237,
      "grad_norm": 0.47985797682797676,
      "learning_rate": 0.0007109999999999999,
      "loss": 6.657,
      "step": 237
    },
    {
      "epoch": 0.00238,
      "grad_norm": 0.6191094472441181,
      "learning_rate": 0.000714,
      "loss": 6.6482,
      "step": 238
    },
    {
      "epoch": 0.00239,
      "grad_norm": 0.5181523072026278,
      "learning_rate": 0.000717,
      "loss": 6.628,
      "step": 239
    },
    {
      "epoch": 0.0024,
      "grad_norm": 0.5894783578801835,
      "learning_rate": 0.0007199999999999999,
      "loss": 6.645,
      "step": 240
    },
    {
      "epoch": 0.00241,
      "grad_norm": 0.5639698176641863,
      "learning_rate": 0.000723,
      "loss": 6.6279,
      "step": 241
    },
    {
      "epoch": 0.00242,
      "grad_norm": 0.9095170652542525,
      "learning_rate": 0.000726,
      "loss": 6.6087,
      "step": 242
    },
    {
      "epoch": 0.00243,
      "grad_norm": 1.3373514416355459,
      "learning_rate": 0.000729,
      "loss": 6.6191,
      "step": 243
    },
    {
      "epoch": 0.00244,
      "grad_norm": 0.5837452319331187,
      "learning_rate": 0.000732,
      "loss": 6.5991,
      "step": 244
    },
    {
      "epoch": 0.00245,
      "grad_norm": 1.0261084178764917,
      "learning_rate": 0.000735,
      "loss": 6.6035,
      "step": 245
    },
    {
      "epoch": 0.00246,
      "grad_norm": 0.5663979442820745,
      "learning_rate": 0.000738,
      "loss": 6.574,
      "step": 246
    },
    {
      "epoch": 0.00247,
      "grad_norm": 0.5896636858778472,
      "learning_rate": 0.000741,
      "loss": 6.5719,
      "step": 247
    },
    {
      "epoch": 0.00248,
      "grad_norm": 0.47326080430149503,
      "learning_rate": 0.000744,
      "loss": 6.5757,
      "step": 248
    },
    {
      "epoch": 0.00249,
      "grad_norm": 0.6248164732961499,
      "learning_rate": 0.000747,
      "loss": 6.5582,
      "step": 249
    },
    {
      "epoch": 0.0025,
      "grad_norm": 0.48362025373458484,
      "learning_rate": 0.00075,
      "loss": 6.5567,
      "step": 250
    },
    {
      "epoch": 0.00251,
      "grad_norm": 0.4858269118610639,
      "learning_rate": 0.000753,
      "loss": 6.5227,
      "step": 251
    },
    {
      "epoch": 0.00252,
      "grad_norm": 0.41862369731289734,
      "learning_rate": 0.000756,
      "loss": 6.5302,
      "step": 252
    },
    {
      "epoch": 0.00253,
      "grad_norm": 0.47200180330590646,
      "learning_rate": 0.000759,
      "loss": 6.522,
      "step": 253
    },
    {
      "epoch": 0.00254,
      "grad_norm": 0.3651546358356223,
      "learning_rate": 0.000762,
      "loss": 6.5136,
      "step": 254
    },
    {
      "epoch": 0.00255,
      "grad_norm": 0.4285192093499382,
      "learning_rate": 0.0007650000000000001,
      "loss": 6.5149,
      "step": 255
    },
    {
      "epoch": 0.00256,
      "grad_norm": 0.42767490819993825,
      "learning_rate": 0.000768,
      "loss": 6.4917,
      "step": 256
    },
    {
      "epoch": 0.00257,
      "grad_norm": 0.48143237273687123,
      "learning_rate": 0.000771,
      "loss": 6.4974,
      "step": 257
    },
    {
      "epoch": 0.00258,
      "grad_norm": 0.5467090762573645,
      "learning_rate": 0.0007740000000000001,
      "loss": 6.4744,
      "step": 258
    },
    {
      "epoch": 0.00259,
      "grad_norm": 0.6750367098607939,
      "learning_rate": 0.000777,
      "loss": 6.4781,
      "step": 259
    },
    {
      "epoch": 0.0026,
      "grad_norm": 0.8320211245129605,
      "learning_rate": 0.0007800000000000001,
      "loss": 6.4681,
      "step": 260
    },
    {
      "epoch": 0.00261,
      "grad_norm": 0.905841412497731,
      "learning_rate": 0.0007830000000000001,
      "loss": 6.4809,
      "step": 261
    },
    {
      "epoch": 0.00262,
      "grad_norm": 1.0687360869781928,
      "learning_rate": 0.000786,
      "loss": 6.4647,
      "step": 262
    },
    {
      "epoch": 0.00263,
      "grad_norm": 1.2051408681263374,
      "learning_rate": 0.0007890000000000001,
      "loss": 6.4874,
      "step": 263
    },
    {
      "epoch": 0.00264,
      "grad_norm": 0.8690142485653533,
      "learning_rate": 0.0007920000000000001,
      "loss": 6.455,
      "step": 264
    },
    {
      "epoch": 0.00265,
      "grad_norm": 1.2774066489819682,
      "learning_rate": 0.000795,
      "loss": 6.4641,
      "step": 265
    },
    {
      "epoch": 0.00266,
      "grad_norm": 0.7288978979341997,
      "learning_rate": 0.0007980000000000001,
      "loss": 6.4454,
      "step": 266
    },
    {
      "epoch": 0.00267,
      "grad_norm": 0.6506287971604123,
      "learning_rate": 0.0008010000000000001,
      "loss": 6.4371,
      "step": 267
    },
    {
      "epoch": 0.00268,
      "grad_norm": 0.6866545943797145,
      "learning_rate": 0.000804,
      "loss": 6.4338,
      "step": 268
    },
    {
      "epoch": 0.00269,
      "grad_norm": 1.0440114151727509,
      "learning_rate": 0.0008070000000000001,
      "loss": 6.4227,
      "step": 269
    },
    {
      "epoch": 0.0027,
      "grad_norm": 1.170351969791303,
      "learning_rate": 0.0008100000000000001,
      "loss": 6.4362,
      "step": 270
    },
    {
      "epoch": 0.00271,
      "grad_norm": 0.6832222674646221,
      "learning_rate": 0.000813,
      "loss": 6.4253,
      "step": 271
    },
    {
      "epoch": 0.00272,
      "grad_norm": 0.617936452008115,
      "learning_rate": 0.0008160000000000001,
      "loss": 6.4063,
      "step": 272
    },
    {
      "epoch": 0.00273,
      "grad_norm": 0.6750066600068259,
      "learning_rate": 0.0008190000000000001,
      "loss": 6.4008,
      "step": 273
    },
    {
      "epoch": 0.00274,
      "grad_norm": 0.6315560378177079,
      "learning_rate": 0.000822,
      "loss": 6.379,
      "step": 274
    },
    {
      "epoch": 0.00275,
      "grad_norm": 0.6774266517760174,
      "learning_rate": 0.0008250000000000001,
      "loss": 6.3852,
      "step": 275
    },
    {
      "epoch": 0.00276,
      "grad_norm": 0.7272895822836024,
      "learning_rate": 0.0008280000000000001,
      "loss": 6.3794,
      "step": 276
    },
    {
      "epoch": 0.00277,
      "grad_norm": 0.6752471037637485,
      "learning_rate": 0.0008310000000000001,
      "loss": 6.3735,
      "step": 277
    },
    {
      "epoch": 0.00278,
      "grad_norm": 0.5678457826039285,
      "learning_rate": 0.0008340000000000001,
      "loss": 6.354,
      "step": 278
    },
    {
      "epoch": 0.00279,
      "grad_norm": 0.4611700189072147,
      "learning_rate": 0.0008370000000000001,
      "loss": 6.3529,
      "step": 279
    },
    {
      "epoch": 0.0028,
      "grad_norm": 0.343285643042232,
      "learning_rate": 0.0008400000000000001,
      "loss": 6.3329,
      "step": 280
    },
    {
      "epoch": 0.00281,
      "grad_norm": 0.4519631747446028,
      "learning_rate": 0.0008430000000000001,
      "loss": 6.3253,
      "step": 281
    },
    {
      "epoch": 0.00282,
      "grad_norm": 0.3255189118052276,
      "learning_rate": 0.000846,
      "loss": 6.3232,
      "step": 282
    },
    {
      "epoch": 0.00283,
      "grad_norm": 0.4297016476682907,
      "learning_rate": 0.0008489999999999999,
      "loss": 6.3149,
      "step": 283
    },
    {
      "epoch": 0.00284,
      "grad_norm": 0.37515936714697207,
      "learning_rate": 0.0008519999999999999,
      "loss": 6.3058,
      "step": 284
    },
    {
      "epoch": 0.00285,
      "grad_norm": 0.3458870104505622,
      "learning_rate": 0.000855,
      "loss": 6.3075,
      "step": 285
    },
    {
      "epoch": 0.00286,
      "grad_norm": 0.39943359732325345,
      "learning_rate": 0.0008579999999999999,
      "loss": 6.2857,
      "step": 286
    },
    {
      "epoch": 0.00287,
      "grad_norm": 0.3631751365570726,
      "learning_rate": 0.000861,
      "loss": 6.2875,
      "step": 287
    },
    {
      "epoch": 0.00288,
      "grad_norm": 0.552280032213235,
      "learning_rate": 0.000864,
      "loss": 6.2707,
      "step": 288
    },
    {
      "epoch": 0.00289,
      "grad_norm": 0.9396362484724781,
      "learning_rate": 0.0008669999999999999,
      "loss": 6.2923,
      "step": 289
    },
    {
      "epoch": 0.0029,
      "grad_norm": 1.3959756273937387,
      "learning_rate": 0.00087,
      "loss": 6.2945,
      "step": 290
    },
    {
      "epoch": 0.00291,
      "grad_norm": 0.6791923155853262,
      "learning_rate": 0.000873,
      "loss": 6.2663,
      "step": 291
    },
    {
      "epoch": 0.00292,
      "grad_norm": 0.9879732835250642,
      "learning_rate": 0.0008759999999999999,
      "loss": 6.2794,
      "step": 292
    },
    {
      "epoch": 0.00293,
      "grad_norm": 1.0140384817226566,
      "learning_rate": 0.000879,
      "loss": 6.2652,
      "step": 293
    },
    {
      "epoch": 0.00294,
      "grad_norm": 0.9889225551341856,
      "learning_rate": 0.000882,
      "loss": 6.2822,
      "step": 294
    },
    {
      "epoch": 0.00295,
      "grad_norm": 1.0429477871581094,
      "learning_rate": 0.0008849999999999999,
      "loss": 6.2468,
      "step": 295
    },
    {
      "epoch": 0.00296,
      "grad_norm": 1.1774929945794055,
      "learning_rate": 0.000888,
      "loss": 6.2705,
      "step": 296
    },
    {
      "epoch": 0.00297,
      "grad_norm": 0.7364156162079134,
      "learning_rate": 0.000891,
      "loss": 6.2278,
      "step": 297
    },
    {
      "epoch": 0.00298,
      "grad_norm": 0.8424638476384282,
      "learning_rate": 0.0008939999999999999,
      "loss": 6.2455,
      "step": 298
    },
    {
      "epoch": 0.00299,
      "grad_norm": 0.8668489286879963,
      "learning_rate": 0.000897,
      "loss": 6.2225,
      "step": 299
    },
    {
      "epoch": 0.003,
      "grad_norm": 0.9039057952602142,
      "learning_rate": 0.0009,
      "loss": 6.2236,
      "step": 300
    },
    {
      "epoch": 0.00301,
      "grad_norm": 1.0344858724084711,
      "learning_rate": 0.0009029999999999999,
      "loss": 6.2222,
      "step": 301
    },
    {
      "epoch": 0.00302,
      "grad_norm": 0.801399966246171,
      "learning_rate": 0.000906,
      "loss": 6.2007,
      "step": 302
    },
    {
      "epoch": 0.00303,
      "grad_norm": 0.7276255151675343,
      "learning_rate": 0.000909,
      "loss": 6.2106,
      "step": 303
    },
    {
      "epoch": 0.00304,
      "grad_norm": 0.8306162070729353,
      "learning_rate": 0.000912,
      "loss": 6.205,
      "step": 304
    },
    {
      "epoch": 0.00305,
      "grad_norm": 0.7650178489304597,
      "learning_rate": 0.000915,
      "loss": 6.2045,
      "step": 305
    },
    {
      "epoch": 0.00306,
      "grad_norm": 0.7024963687074245,
      "learning_rate": 0.000918,
      "loss": 6.1878,
      "step": 306
    },
    {
      "epoch": 0.00307,
      "grad_norm": 0.5687961336654864,
      "learning_rate": 0.000921,
      "loss": 6.1778,
      "step": 307
    },
    {
      "epoch": 0.00308,
      "grad_norm": 0.4515866134049927,
      "learning_rate": 0.000924,
      "loss": 6.1586,
      "step": 308
    },
    {
      "epoch": 0.00309,
      "grad_norm": 0.5454081565882548,
      "learning_rate": 0.000927,
      "loss": 6.1763,
      "step": 309
    },
    {
      "epoch": 0.0031,
      "grad_norm": 0.4033959865123679,
      "learning_rate": 0.00093,
      "loss": 6.1549,
      "step": 310
    },
    {
      "epoch": 0.00311,
      "grad_norm": 0.5004494164305024,
      "learning_rate": 0.000933,
      "loss": 6.1478,
      "step": 311
    },
    {
      "epoch": 0.00312,
      "grad_norm": 0.470361361901935,
      "learning_rate": 0.000936,
      "loss": 6.1275,
      "step": 312
    },
    {
      "epoch": 0.00313,
      "grad_norm": 0.4973667290148138,
      "learning_rate": 0.0009390000000000001,
      "loss": 6.1399,
      "step": 313
    },
    {
      "epoch": 0.00314,
      "grad_norm": 0.47398584855820086,
      "learning_rate": 0.000942,
      "loss": 6.1453,
      "step": 314
    },
    {
      "epoch": 0.00315,
      "grad_norm": 0.39081338977861474,
      "learning_rate": 0.000945,
      "loss": 6.1206,
      "step": 315
    },
    {
      "epoch": 0.00316,
      "grad_norm": 0.308452368547838,
      "learning_rate": 0.0009480000000000001,
      "loss": 6.0925,
      "step": 316
    },
    {
      "epoch": 0.00317,
      "grad_norm": 0.33735261718955184,
      "learning_rate": 0.000951,
      "loss": 6.1112,
      "step": 317
    },
    {
      "epoch": 0.00318,
      "grad_norm": 0.3843792856632324,
      "learning_rate": 0.000954,
      "loss": 6.1055,
      "step": 318
    },
    {
      "epoch": 0.00319,
      "grad_norm": 0.45015697169720664,
      "learning_rate": 0.0009570000000000001,
      "loss": 6.0951,
      "step": 319
    },
    {
      "epoch": 0.0032,
      "grad_norm": 0.6347836654615971,
      "learning_rate": 0.00096,
      "loss": 6.097,
      "step": 320
    },
    {
      "epoch": 0.00321,
      "grad_norm": 0.9264517947757075,
      "learning_rate": 0.000963,
      "loss": 6.0955,
      "step": 321
    },
    {
      "epoch": 0.00322,
      "grad_norm": 1.3311662164937155,
      "learning_rate": 0.0009660000000000001,
      "loss": 6.0984,
      "step": 322
    },
    {
      "epoch": 0.00323,
      "grad_norm": 0.7192768486088142,
      "learning_rate": 0.000969,
      "loss": 6.081,
      "step": 323
    },
    {
      "epoch": 0.00324,
      "grad_norm": 0.8815686637890671,
      "learning_rate": 0.0009720000000000001,
      "loss": 6.0921,
      "step": 324
    },
    {
      "epoch": 0.00325,
      "grad_norm": 0.9399377971403509,
      "learning_rate": 0.0009750000000000001,
      "loss": 6.0559,
      "step": 325
    },
    {
      "epoch": 0.00326,
      "grad_norm": 1.1161084705724094,
      "learning_rate": 0.0009780000000000001,
      "loss": 6.0866,
      "step": 326
    },
    {
      "epoch": 0.00327,
      "grad_norm": 0.9798577225908143,
      "learning_rate": 0.000981,
      "loss": 6.08,
      "step": 327
    },
    {
      "epoch": 0.00328,
      "grad_norm": 1.4594200076446016,
      "learning_rate": 0.000984,
      "loss": 6.0949,
      "step": 328
    },
    {
      "epoch": 0.00329,
      "grad_norm": 0.7530414187502621,
      "learning_rate": 0.000987,
      "loss": 6.0592,
      "step": 329
    },
    {
      "epoch": 0.0033,
      "grad_norm": 0.9983029089238832,
      "learning_rate": 0.00099,
      "loss": 6.0707,
      "step": 330
    },
    {
      "epoch": 0.00331,
      "grad_norm": 0.835334238631822,
      "learning_rate": 0.0009930000000000002,
      "loss": 6.0609,
      "step": 331
    },
    {
      "epoch": 0.00332,
      "grad_norm": 1.2633352311024129,
      "learning_rate": 0.0009960000000000001,
      "loss": 6.0632,
      "step": 332
    },
    {
      "epoch": 0.00333,
      "grad_norm": 0.9415989578137678,
      "learning_rate": 0.000999,
      "loss": 6.0388,
      "step": 333
    },
    {
      "epoch": 0.00334,
      "grad_norm": 0.752057040577036,
      "learning_rate": 0.001002,
      "loss": 6.0434,
      "step": 334
    },
    {
      "epoch": 0.00335,
      "grad_norm": 0.6368712711427076,
      "learning_rate": 0.001005,
      "loss": 6.039,
      "step": 335
    },
    {
      "epoch": 0.00336,
      "grad_norm": 0.5277321252477392,
      "learning_rate": 0.001008,
      "loss": 6.0255,
      "step": 336
    },
    {
      "epoch": 0.00337,
      "grad_norm": 0.545640031716998,
      "learning_rate": 0.0010110000000000002,
      "loss": 6.0051,
      "step": 337
    },
    {
      "epoch": 0.00338,
      "grad_norm": 0.5605817392374447,
      "learning_rate": 0.0010140000000000001,
      "loss": 6.0183,
      "step": 338
    },
    {
      "epoch": 0.00339,
      "grad_norm": 0.4963166785854256,
      "learning_rate": 0.0010170000000000001,
      "loss": 6.0214,
      "step": 339
    },
    {
      "epoch": 0.0034,
      "grad_norm": 0.496225629593103,
      "learning_rate": 0.00102,
      "loss": 5.9915,
      "step": 340
    },
    {
      "epoch": 0.00341,
      "grad_norm": 0.45013704339915594,
      "learning_rate": 0.001023,
      "loss": 5.9719,
      "step": 341
    },
    {
      "epoch": 0.00342,
      "grad_norm": 0.37899658405778774,
      "learning_rate": 0.001026,
      "loss": 5.9765,
      "step": 342
    },
    {
      "epoch": 0.00343,
      "grad_norm": 0.4216072327604619,
      "learning_rate": 0.0010290000000000002,
      "loss": 5.9773,
      "step": 343
    },
    {
      "epoch": 0.00344,
      "grad_norm": 0.4964437378733662,
      "learning_rate": 0.001032,
      "loss": 5.9678,
      "step": 344
    },
    {
      "epoch": 0.00345,
      "grad_norm": 0.6031142308234815,
      "learning_rate": 0.001035,
      "loss": 5.9605,
      "step": 345
    },
    {
      "epoch": 0.00346,
      "grad_norm": 0.6349778009023861,
      "learning_rate": 0.0010379999999999999,
      "loss": 5.9675,
      "step": 346
    },
    {
      "epoch": 0.00347,
      "grad_norm": 0.6213366174245898,
      "learning_rate": 0.001041,
      "loss": 5.9393,
      "step": 347
    },
    {
      "epoch": 0.00348,
      "grad_norm": 0.684857817555668,
      "learning_rate": 0.001044,
      "loss": 5.9499,
      "step": 348
    },
    {
      "epoch": 0.00349,
      "grad_norm": 0.8421661368559449,
      "learning_rate": 0.001047,
      "loss": 5.9398,
      "step": 349
    },
    {
      "epoch": 0.0035,
      "grad_norm": 0.8062802732253019,
      "learning_rate": 0.00105,
      "loss": 5.9568,
      "step": 350
    },
    {
      "epoch": 0.00351,
      "grad_norm": 0.7669273491138234,
      "learning_rate": 0.001053,
      "loss": 5.9549,
      "step": 351
    },
    {
      "epoch": 0.00352,
      "grad_norm": 1.0673384228730578,
      "learning_rate": 0.0010559999999999999,
      "loss": 5.9491,
      "step": 352
    },
    {
      "epoch": 0.00353,
      "grad_norm": 1.313057737158452,
      "learning_rate": 0.001059,
      "loss": 5.9596,
      "step": 353
    },
    {
      "epoch": 0.00354,
      "grad_norm": 0.8748293544965601,
      "learning_rate": 0.001062,
      "loss": 5.9228,
      "step": 354
    },
    {
      "epoch": 0.00355,
      "grad_norm": 0.7617258072087358,
      "learning_rate": 0.001065,
      "loss": 5.9348,
      "step": 355
    },
    {
      "epoch": 0.00356,
      "grad_norm": 0.9561923051718775,
      "learning_rate": 0.001068,
      "loss": 5.9306,
      "step": 356
    },
    {
      "epoch": 0.00357,
      "grad_norm": 0.8631216371911231,
      "learning_rate": 0.001071,
      "loss": 5.9199,
      "step": 357
    },
    {
      "epoch": 0.00358,
      "grad_norm": 0.9525953351992685,
      "learning_rate": 0.001074,
      "loss": 5.9184,
      "step": 358
    },
    {
      "epoch": 0.00359,
      "grad_norm": 0.6858307664974747,
      "learning_rate": 0.001077,
      "loss": 5.8973,
      "step": 359
    },
    {
      "epoch": 0.0036,
      "grad_norm": 0.7167037627907079,
      "learning_rate": 0.00108,
      "loss": 5.9065,
      "step": 360
    },
    {
      "epoch": 0.00361,
      "grad_norm": 0.7969047134484166,
      "learning_rate": 0.001083,
      "loss": 5.8986,
      "step": 361
    },
    {
      "epoch": 0.00362,
      "grad_norm": 0.8186204145394074,
      "learning_rate": 0.001086,
      "loss": 5.888,
      "step": 362
    },
    {
      "epoch": 0.00363,
      "grad_norm": 0.7436182013517663,
      "learning_rate": 0.001089,
      "loss": 5.8766,
      "step": 363
    },
    {
      "epoch": 0.00364,
      "grad_norm": 0.6658023261534547,
      "learning_rate": 0.001092,
      "loss": 5.8872,
      "step": 364
    },
    {
      "epoch": 0.00365,
      "grad_norm": 0.7206668214896482,
      "learning_rate": 0.001095,
      "loss": 5.8872,
      "step": 365
    },
    {
      "epoch": 0.00366,
      "grad_norm": 0.6345413922647961,
      "learning_rate": 0.001098,
      "loss": 5.8617,
      "step": 366
    },
    {
      "epoch": 0.00367,
      "grad_norm": 0.6094924231218852,
      "learning_rate": 0.001101,
      "loss": 5.8618,
      "step": 367
    },
    {
      "epoch": 0.00368,
      "grad_norm": 0.7732260654201254,
      "learning_rate": 0.001104,
      "loss": 5.8653,
      "step": 368
    },
    {
      "epoch": 0.00369,
      "grad_norm": 0.9451132874875877,
      "learning_rate": 0.001107,
      "loss": 5.8779,
      "step": 369
    },
    {
      "epoch": 0.0037,
      "grad_norm": 1.047387946938651,
      "learning_rate": 0.00111,
      "loss": 5.8603,
      "step": 370
    },
    {
      "epoch": 0.00371,
      "grad_norm": 0.8709497379832931,
      "learning_rate": 0.001113,
      "loss": 5.8262,
      "step": 371
    },
    {
      "epoch": 0.00372,
      "grad_norm": 0.6993621521276565,
      "learning_rate": 0.001116,
      "loss": 5.8434,
      "step": 372
    },
    {
      "epoch": 0.00373,
      "grad_norm": 0.6889019297226876,
      "learning_rate": 0.001119,
      "loss": 5.8344,
      "step": 373
    },
    {
      "epoch": 0.00374,
      "grad_norm": 0.8187808414499578,
      "learning_rate": 0.001122,
      "loss": 5.8424,
      "step": 374
    },
    {
      "epoch": 0.00375,
      "grad_norm": 0.7395556739972736,
      "learning_rate": 0.0011250000000000001,
      "loss": 5.8404,
      "step": 375
    },
    {
      "epoch": 0.00376,
      "grad_norm": 0.5027446756427529,
      "learning_rate": 0.001128,
      "loss": 5.8296,
      "step": 376
    },
    {
      "epoch": 0.00377,
      "grad_norm": 0.45297103786338255,
      "learning_rate": 0.001131,
      "loss": 5.8239,
      "step": 377
    },
    {
      "epoch": 0.00378,
      "grad_norm": 0.5150298222384522,
      "learning_rate": 0.001134,
      "loss": 5.82,
      "step": 378
    },
    {
      "epoch": 0.00379,
      "grad_norm": 0.4216428009711753,
      "learning_rate": 0.001137,
      "loss": 5.8036,
      "step": 379
    },
    {
      "epoch": 0.0038,
      "grad_norm": 0.43574801532624385,
      "learning_rate": 0.00114,
      "loss": 5.8234,
      "step": 380
    },
    {
      "epoch": 0.00381,
      "grad_norm": 0.4737821231317218,
      "learning_rate": 0.0011430000000000001,
      "loss": 5.7908,
      "step": 381
    },
    {
      "epoch": 0.00382,
      "grad_norm": 0.5236457098681065,
      "learning_rate": 0.001146,
      "loss": 5.7778,
      "step": 382
    },
    {
      "epoch": 0.00383,
      "grad_norm": 0.5006340354259897,
      "learning_rate": 0.001149,
      "loss": 5.787,
      "step": 383
    },
    {
      "epoch": 0.00384,
      "grad_norm": 0.5762330042414852,
      "learning_rate": 0.001152,
      "loss": 5.7915,
      "step": 384
    },
    {
      "epoch": 0.00385,
      "grad_norm": 0.6623935085819848,
      "learning_rate": 0.001155,
      "loss": 5.7808,
      "step": 385
    },
    {
      "epoch": 0.00386,
      "grad_norm": 0.6780931079980513,
      "learning_rate": 0.001158,
      "loss": 5.7864,
      "step": 386
    },
    {
      "epoch": 0.00387,
      "grad_norm": 0.6283557380392781,
      "learning_rate": 0.0011610000000000001,
      "loss": 5.76,
      "step": 387
    },
    {
      "epoch": 0.00388,
      "grad_norm": 0.8661947460887196,
      "learning_rate": 0.001164,
      "loss": 5.7642,
      "step": 388
    },
    {
      "epoch": 0.00389,
      "grad_norm": 1.1877567285678448,
      "learning_rate": 0.001167,
      "loss": 5.7898,
      "step": 389
    },
    {
      "epoch": 0.0039,
      "grad_norm": 0.5285139722189788,
      "learning_rate": 0.00117,
      "loss": 5.7491,
      "step": 390
    },
    {
      "epoch": 0.00391,
      "grad_norm": 0.6430118692881639,
      "learning_rate": 0.001173,
      "loss": 5.7622,
      "step": 391
    },
    {
      "epoch": 0.00392,
      "grad_norm": 0.6748100403552353,
      "learning_rate": 0.001176,
      "loss": 5.7713,
      "step": 392
    },
    {
      "epoch": 0.00393,
      "grad_norm": 0.7344199345621275,
      "learning_rate": 0.0011790000000000001,
      "loss": 5.7448,
      "step": 393
    },
    {
      "epoch": 0.00394,
      "grad_norm": 0.8611653440542995,
      "learning_rate": 0.001182,
      "loss": 5.7377,
      "step": 394
    },
    {
      "epoch": 0.00395,
      "grad_norm": 1.0200403565527223,
      "learning_rate": 0.001185,
      "loss": 5.7297,
      "step": 395
    },
    {
      "epoch": 0.00396,
      "grad_norm": 1.3219082304761296,
      "learning_rate": 0.001188,
      "loss": 5.7571,
      "step": 396
    },
    {
      "epoch": 0.00397,
      "grad_norm": 1.0648841185793536,
      "learning_rate": 0.001191,
      "loss": 5.7452,
      "step": 397
    },
    {
      "epoch": 0.00398,
      "grad_norm": 0.8818852137031271,
      "learning_rate": 0.0011940000000000002,
      "loss": 5.7476,
      "step": 398
    },
    {
      "epoch": 0.00399,
      "grad_norm": 0.7229082271352473,
      "learning_rate": 0.0011970000000000001,
      "loss": 5.7455,
      "step": 399
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.7464785543636239,
      "learning_rate": 0.0012000000000000001,
      "loss": 5.7335,
      "step": 400
    },
    {
      "epoch": 0.00401,
      "grad_norm": 0.788008045277313,
      "learning_rate": 0.001203,
      "loss": 5.7147,
      "step": 401
    },
    {
      "epoch": 0.00402,
      "grad_norm": 0.8844811042429299,
      "learning_rate": 0.001206,
      "loss": 5.7255,
      "step": 402
    },
    {
      "epoch": 0.00403,
      "grad_norm": 0.8786909754931423,
      "learning_rate": 0.001209,
      "loss": 5.7201,
      "step": 403
    },
    {
      "epoch": 0.00404,
      "grad_norm": 0.9763167280044874,
      "learning_rate": 0.0012120000000000002,
      "loss": 5.7188,
      "step": 404
    },
    {
      "epoch": 0.00405,
      "grad_norm": 0.9708490732732808,
      "learning_rate": 0.0012150000000000002,
      "loss": 5.7216,
      "step": 405
    },
    {
      "epoch": 0.00406,
      "grad_norm": 0.7953769652219763,
      "learning_rate": 0.0012180000000000001,
      "loss": 5.6987,
      "step": 406
    },
    {
      "epoch": 0.00407,
      "grad_norm": 0.7231086032909518,
      "learning_rate": 0.0012209999999999999,
      "loss": 5.6945,
      "step": 407
    },
    {
      "epoch": 0.00408,
      "grad_norm": 0.6539281657127057,
      "learning_rate": 0.001224,
      "loss": 5.7108,
      "step": 408
    },
    {
      "epoch": 0.00409,
      "grad_norm": 0.48725126694443294,
      "learning_rate": 0.001227,
      "loss": 5.6827,
      "step": 409
    },
    {
      "epoch": 0.0041,
      "grad_norm": 0.5935989194477762,
      "learning_rate": 0.00123,
      "loss": 5.6771,
      "step": 410
    },
    {
      "epoch": 0.00411,
      "grad_norm": 0.6524672694061662,
      "learning_rate": 0.001233,
      "loss": 5.6945,
      "step": 411
    },
    {
      "epoch": 0.00412,
      "grad_norm": 0.5686252583506046,
      "learning_rate": 0.001236,
      "loss": 5.6796,
      "step": 412
    },
    {
      "epoch": 0.00413,
      "grad_norm": 0.5206794027062916,
      "learning_rate": 0.0012389999999999999,
      "loss": 5.676,
      "step": 413
    },
    {
      "epoch": 0.00414,
      "grad_norm": 0.5100681259570212,
      "learning_rate": 0.001242,
      "loss": 5.6561,
      "step": 414
    },
    {
      "epoch": 0.00415,
      "grad_norm": 0.5323494958966052,
      "learning_rate": 0.001245,
      "loss": 5.6572,
      "step": 415
    },
    {
      "epoch": 0.00416,
      "grad_norm": 0.593833682795931,
      "learning_rate": 0.001248,
      "loss": 5.6419,
      "step": 416
    },
    {
      "epoch": 0.00417,
      "grad_norm": 0.5881502982050868,
      "learning_rate": 0.001251,
      "loss": 5.6711,
      "step": 417
    },
    {
      "epoch": 0.00418,
      "grad_norm": 0.6571826832460801,
      "learning_rate": 0.001254,
      "loss": 5.6451,
      "step": 418
    },
    {
      "epoch": 0.00419,
      "grad_norm": 0.8242389183582979,
      "learning_rate": 0.0012569999999999999,
      "loss": 5.6393,
      "step": 419
    },
    {
      "epoch": 0.0042,
      "grad_norm": 0.7322278119135938,
      "learning_rate": 0.00126,
      "loss": 5.6531,
      "step": 420
    },
    {
      "epoch": 0.00421,
      "grad_norm": 0.6080486142741918,
      "learning_rate": 0.001263,
      "loss": 5.6415,
      "step": 421
    },
    {
      "epoch": 0.00422,
      "grad_norm": 0.6271805214461816,
      "learning_rate": 0.001266,
      "loss": 5.6327,
      "step": 422
    },
    {
      "epoch": 0.00423,
      "grad_norm": 0.5934268893913589,
      "learning_rate": 0.001269,
      "loss": 5.6286,
      "step": 423
    },
    {
      "epoch": 0.00424,
      "grad_norm": 0.6795296945522938,
      "learning_rate": 0.001272,
      "loss": 5.6299,
      "step": 424
    },
    {
      "epoch": 0.00425,
      "grad_norm": 0.7845276448161947,
      "learning_rate": 0.001275,
      "loss": 5.6344,
      "step": 425
    },
    {
      "epoch": 0.00426,
      "grad_norm": 0.9176562117983922,
      "learning_rate": 0.001278,
      "loss": 5.6402,
      "step": 426
    },
    {
      "epoch": 0.00427,
      "grad_norm": 1.0306205830362438,
      "learning_rate": 0.001281,
      "loss": 5.6293,
      "step": 427
    },
    {
      "epoch": 0.00428,
      "grad_norm": 1.0252381178312269,
      "learning_rate": 0.001284,
      "loss": 5.6086,
      "step": 428
    },
    {
      "epoch": 0.00429,
      "grad_norm": 1.3332162612637855,
      "learning_rate": 0.001287,
      "loss": 5.6393,
      "step": 429
    },
    {
      "epoch": 0.0043,
      "grad_norm": 0.8174291079939794,
      "learning_rate": 0.00129,
      "loss": 5.6202,
      "step": 430
    },
    {
      "epoch": 0.00431,
      "grad_norm": 0.821028463418781,
      "learning_rate": 0.001293,
      "loss": 5.6102,
      "step": 431
    },
    {
      "epoch": 0.00432,
      "grad_norm": 0.7475137161763143,
      "learning_rate": 0.001296,
      "loss": 5.6163,
      "step": 432
    },
    {
      "epoch": 0.00433,
      "grad_norm": 0.7571870052992741,
      "learning_rate": 0.001299,
      "loss": 5.604,
      "step": 433
    },
    {
      "epoch": 0.00434,
      "grad_norm": 1.0543560255015263,
      "learning_rate": 0.001302,
      "loss": 5.6184,
      "step": 434
    },
    {
      "epoch": 0.00435,
      "grad_norm": 0.8758758304266553,
      "learning_rate": 0.001305,
      "loss": 5.6066,
      "step": 435
    },
    {
      "epoch": 0.00436,
      "grad_norm": 0.9137062421440546,
      "learning_rate": 0.001308,
      "loss": 5.5859,
      "step": 436
    },
    {
      "epoch": 0.00437,
      "grad_norm": 1.301736875083812,
      "learning_rate": 0.001311,
      "loss": 5.6173,
      "step": 437
    },
    {
      "epoch": 0.00438,
      "grad_norm": 0.9356953917037294,
      "learning_rate": 0.001314,
      "loss": 5.5919,
      "step": 438
    },
    {
      "epoch": 0.00439,
      "grad_norm": 0.8522821992819578,
      "learning_rate": 0.001317,
      "loss": 5.6158,
      "step": 439
    },
    {
      "epoch": 0.0044,
      "grad_norm": 0.7182761753028103,
      "learning_rate": 0.00132,
      "loss": 5.5821,
      "step": 440
    },
    {
      "epoch": 0.00441,
      "grad_norm": 0.6293266892726601,
      "learning_rate": 0.001323,
      "loss": 5.577,
      "step": 441
    },
    {
      "epoch": 0.00442,
      "grad_norm": 0.8189921873128356,
      "learning_rate": 0.0013260000000000001,
      "loss": 5.5762,
      "step": 442
    },
    {
      "epoch": 0.00443,
      "grad_norm": 1.0212422821054057,
      "learning_rate": 0.001329,
      "loss": 5.5904,
      "step": 443
    },
    {
      "epoch": 0.00444,
      "grad_norm": 0.8951785018901781,
      "learning_rate": 0.001332,
      "loss": 5.5851,
      "step": 444
    },
    {
      "epoch": 0.00445,
      "grad_norm": 0.7868769140150608,
      "learning_rate": 0.001335,
      "loss": 5.5661,
      "step": 445
    },
    {
      "epoch": 0.00446,
      "grad_norm": 0.790102612629763,
      "learning_rate": 0.001338,
      "loss": 5.5718,
      "step": 446
    },
    {
      "epoch": 0.00447,
      "grad_norm": 0.8396194874372788,
      "learning_rate": 0.001341,
      "loss": 5.5716,
      "step": 447
    },
    {
      "epoch": 0.00448,
      "grad_norm": 0.9120841535821665,
      "learning_rate": 0.0013440000000000001,
      "loss": 5.5589,
      "step": 448
    },
    {
      "epoch": 0.00449,
      "grad_norm": 0.8573073152890212,
      "learning_rate": 0.001347,
      "loss": 5.5603,
      "step": 449
    },
    {
      "epoch": 0.0045,
      "grad_norm": 0.7420732830276576,
      "learning_rate": 0.00135,
      "loss": 5.5551,
      "step": 450
    },
    {
      "epoch": 0.00451,
      "grad_norm": 0.7054051762730813,
      "learning_rate": 0.001353,
      "loss": 5.5451,
      "step": 451
    },
    {
      "epoch": 0.00452,
      "grad_norm": 0.5383194985068459,
      "learning_rate": 0.001356,
      "loss": 5.5229,
      "step": 452
    },
    {
      "epoch": 0.00453,
      "grad_norm": 0.5845224529530345,
      "learning_rate": 0.001359,
      "loss": 5.5144,
      "step": 453
    },
    {
      "epoch": 0.00454,
      "grad_norm": 0.5197277965757966,
      "learning_rate": 0.0013620000000000001,
      "loss": 5.5269,
      "step": 454
    },
    {
      "epoch": 0.00455,
      "grad_norm": 0.44901215490386587,
      "learning_rate": 0.0013650000000000001,
      "loss": 5.5227,
      "step": 455
    },
    {
      "epoch": 0.00456,
      "grad_norm": 0.40823042735319937,
      "learning_rate": 0.001368,
      "loss": 5.5155,
      "step": 456
    },
    {
      "epoch": 0.00457,
      "grad_norm": 0.4077054091161063,
      "learning_rate": 0.001371,
      "loss": 5.5045,
      "step": 457
    },
    {
      "epoch": 0.00458,
      "grad_norm": 0.4051721280659754,
      "learning_rate": 0.001374,
      "loss": 5.5115,
      "step": 458
    },
    {
      "epoch": 0.00459,
      "grad_norm": 0.3819222938644956,
      "learning_rate": 0.0013770000000000002,
      "loss": 5.4986,
      "step": 459
    },
    {
      "epoch": 0.0046,
      "grad_norm": 0.4096791477469412,
      "learning_rate": 0.0013800000000000002,
      "loss": 5.4851,
      "step": 460
    },
    {
      "epoch": 0.00461,
      "grad_norm": 0.44167018915391465,
      "learning_rate": 0.0013830000000000001,
      "loss": 5.4805,
      "step": 461
    },
    {
      "epoch": 0.00462,
      "grad_norm": 0.49346826910496583,
      "learning_rate": 0.001386,
      "loss": 5.4928,
      "step": 462
    },
    {
      "epoch": 0.00463,
      "grad_norm": 0.463489633996114,
      "learning_rate": 0.001389,
      "loss": 5.4652,
      "step": 463
    },
    {
      "epoch": 0.00464,
      "grad_norm": 0.507205937400668,
      "learning_rate": 0.001392,
      "loss": 5.4859,
      "step": 464
    },
    {
      "epoch": 0.00465,
      "grad_norm": 0.5318639728571777,
      "learning_rate": 0.0013950000000000002,
      "loss": 5.4572,
      "step": 465
    },
    {
      "epoch": 0.00466,
      "grad_norm": 0.6489252257104292,
      "learning_rate": 0.0013980000000000002,
      "loss": 5.4792,
      "step": 466
    },
    {
      "epoch": 0.00467,
      "grad_norm": 0.7957789059873086,
      "learning_rate": 0.0014010000000000001,
      "loss": 5.4727,
      "step": 467
    },
    {
      "epoch": 0.00468,
      "grad_norm": 0.7621042481808248,
      "learning_rate": 0.001404,
      "loss": 5.462,
      "step": 468
    },
    {
      "epoch": 0.00469,
      "grad_norm": 0.573822393217689,
      "learning_rate": 0.001407,
      "loss": 5.457,
      "step": 469
    },
    {
      "epoch": 0.0047,
      "grad_norm": 0.7543746323219513,
      "learning_rate": 0.00141,
      "loss": 5.4709,
      "step": 470
    },
    {
      "epoch": 0.00471,
      "grad_norm": 0.8641651762100534,
      "learning_rate": 0.001413,
      "loss": 5.4567,
      "step": 471
    },
    {
      "epoch": 0.00472,
      "grad_norm": 0.9026674726231502,
      "learning_rate": 0.001416,
      "loss": 5.4699,
      "step": 472
    },
    {
      "epoch": 0.00473,
      "grad_norm": 0.8212885101559565,
      "learning_rate": 0.001419,
      "loss": 5.4778,
      "step": 473
    },
    {
      "epoch": 0.00474,
      "grad_norm": 0.8914030740906659,
      "learning_rate": 0.0014219999999999999,
      "loss": 5.461,
      "step": 474
    },
    {
      "epoch": 0.00475,
      "grad_norm": 1.0570094425693455,
      "learning_rate": 0.001425,
      "loss": 5.4652,
      "step": 475
    },
    {
      "epoch": 0.00476,
      "grad_norm": 0.9736444976311589,
      "learning_rate": 0.001428,
      "loss": 5.4875,
      "step": 476
    },
    {
      "epoch": 0.00477,
      "grad_norm": 1.1550380737092787,
      "learning_rate": 0.001431,
      "loss": 5.4568,
      "step": 477
    },
    {
      "epoch": 0.00478,
      "grad_norm": 0.848321570803796,
      "learning_rate": 0.001434,
      "loss": 5.4695,
      "step": 478
    },
    {
      "epoch": 0.00479,
      "grad_norm": 0.9517827225501269,
      "learning_rate": 0.001437,
      "loss": 5.4501,
      "step": 479
    },
    {
      "epoch": 0.0048,
      "grad_norm": 1.0883787540754652,
      "learning_rate": 0.0014399999999999999,
      "loss": 5.4562,
      "step": 480
    },
    {
      "epoch": 0.00481,
      "grad_norm": 0.9422991164230814,
      "learning_rate": 0.001443,
      "loss": 5.4516,
      "step": 481
    },
    {
      "epoch": 0.00482,
      "grad_norm": 0.9159499791385636,
      "learning_rate": 0.001446,
      "loss": 5.4273,
      "step": 482
    },
    {
      "epoch": 0.00483,
      "grad_norm": 0.9688645055474768,
      "learning_rate": 0.001449,
      "loss": 5.44,
      "step": 483
    },
    {
      "epoch": 0.00484,
      "grad_norm": 1.1114303023214132,
      "learning_rate": 0.001452,
      "loss": 5.4474,
      "step": 484
    },
    {
      "epoch": 0.00485,
      "grad_norm": 0.9051569573634253,
      "learning_rate": 0.001455,
      "loss": 5.4468,
      "step": 485
    },
    {
      "epoch": 0.00486,
      "grad_norm": 0.9247454458721566,
      "learning_rate": 0.001458,
      "loss": 5.4217,
      "step": 486
    },
    {
      "epoch": 0.00487,
      "grad_norm": 0.852928162562673,
      "learning_rate": 0.001461,
      "loss": 5.4339,
      "step": 487
    },
    {
      "epoch": 0.00488,
      "grad_norm": 0.8142513932978471,
      "learning_rate": 0.001464,
      "loss": 5.41,
      "step": 488
    },
    {
      "epoch": 0.00489,
      "grad_norm": 0.9329231917883894,
      "learning_rate": 0.001467,
      "loss": 5.403,
      "step": 489
    },
    {
      "epoch": 0.0049,
      "grad_norm": 0.9855383918059676,
      "learning_rate": 0.00147,
      "loss": 5.4217,
      "step": 490
    },
    {
      "epoch": 0.00491,
      "grad_norm": 0.7649189111405409,
      "learning_rate": 0.001473,
      "loss": 5.4131,
      "step": 491
    },
    {
      "epoch": 0.00492,
      "grad_norm": 0.7909135941762935,
      "learning_rate": 0.001476,
      "loss": 5.4088,
      "step": 492
    },
    {
      "epoch": 0.00493,
      "grad_norm": 0.9895574066894314,
      "learning_rate": 0.001479,
      "loss": 5.412,
      "step": 493
    },
    {
      "epoch": 0.00494,
      "grad_norm": 1.132991334341666,
      "learning_rate": 0.001482,
      "loss": 5.429,
      "step": 494
    },
    {
      "epoch": 0.00495,
      "grad_norm": 0.761507737933228,
      "learning_rate": 0.001485,
      "loss": 5.4134,
      "step": 495
    },
    {
      "epoch": 0.00496,
      "grad_norm": 0.9726159326361932,
      "learning_rate": 0.001488,
      "loss": 5.4067,
      "step": 496
    },
    {
      "epoch": 0.00497,
      "grad_norm": 1.1482864163713484,
      "learning_rate": 0.001491,
      "loss": 5.3997,
      "step": 497
    },
    {
      "epoch": 0.00498,
      "grad_norm": 0.8327332046897746,
      "learning_rate": 0.001494,
      "loss": 5.4147,
      "step": 498
    },
    {
      "epoch": 0.00499,
      "grad_norm": 0.725916611519047,
      "learning_rate": 0.001497,
      "loss": 5.3792,
      "step": 499
    },
    {
      "epoch": 0.005,
      "grad_norm": 0.6719782846245154,
      "learning_rate": 0.0015,
      "loss": 5.3842,
      "step": 500
    },
    {
      "epoch": 0.00501,
      "grad_norm": 0.7463076928465905,
      "learning_rate": 0.001503,
      "loss": 5.3796,
      "step": 501
    },
    {
      "epoch": 0.00502,
      "grad_norm": 0.8408104186601356,
      "learning_rate": 0.001506,
      "loss": 5.389,
      "step": 502
    },
    {
      "epoch": 0.00503,
      "grad_norm": 0.860909611402231,
      "learning_rate": 0.0015090000000000001,
      "loss": 5.3926,
      "step": 503
    },
    {
      "epoch": 0.00504,
      "grad_norm": 0.7026490274993983,
      "learning_rate": 0.001512,
      "loss": 5.3646,
      "step": 504
    },
    {
      "epoch": 0.00505,
      "grad_norm": 0.7184807991697565,
      "learning_rate": 0.001515,
      "loss": 5.3547,
      "step": 505
    },
    {
      "epoch": 0.00506,
      "grad_norm": 0.7839475253802514,
      "learning_rate": 0.001518,
      "loss": 5.3611,
      "step": 506
    },
    {
      "epoch": 0.00507,
      "grad_norm": 0.7039499110993044,
      "learning_rate": 0.001521,
      "loss": 5.3522,
      "step": 507
    },
    {
      "epoch": 0.00508,
      "grad_norm": 0.5587706859739108,
      "learning_rate": 0.001524,
      "loss": 5.3512,
      "step": 508
    },
    {
      "epoch": 0.00509,
      "grad_norm": 0.4952941325228141,
      "learning_rate": 0.0015270000000000001,
      "loss": 5.326,
      "step": 509
    },
    {
      "epoch": 0.0051,
      "grad_norm": 0.5131490664852795,
      "learning_rate": 0.0015300000000000001,
      "loss": 5.3428,
      "step": 510
    },
    {
      "epoch": 0.00511,
      "grad_norm": 0.5539213487194597,
      "learning_rate": 0.001533,
      "loss": 5.3196,
      "step": 511
    },
    {
      "epoch": 0.00512,
      "grad_norm": 0.5937876471208409,
      "learning_rate": 0.001536,
      "loss": 5.3158,
      "step": 512
    },
    {
      "epoch": 0.00513,
      "grad_norm": 0.5441672327389838,
      "learning_rate": 0.001539,
      "loss": 5.327,
      "step": 513
    },
    {
      "epoch": 0.00514,
      "grad_norm": 0.5371789671410057,
      "learning_rate": 0.001542,
      "loss": 5.3038,
      "step": 514
    },
    {
      "epoch": 0.00515,
      "grad_norm": 0.5194765862771661,
      "learning_rate": 0.0015450000000000001,
      "loss": 5.3109,
      "step": 515
    },
    {
      "epoch": 0.00516,
      "grad_norm": 0.5575198815834714,
      "learning_rate": 0.0015480000000000001,
      "loss": 5.3083,
      "step": 516
    },
    {
      "epoch": 0.00517,
      "grad_norm": 0.5237583962443445,
      "learning_rate": 0.001551,
      "loss": 5.2962,
      "step": 517
    },
    {
      "epoch": 0.00518,
      "grad_norm": 0.5237181167506534,
      "learning_rate": 0.001554,
      "loss": 5.2786,
      "step": 518
    },
    {
      "epoch": 0.00519,
      "grad_norm": 0.6652146969315359,
      "learning_rate": 0.001557,
      "loss": 5.2847,
      "step": 519
    },
    {
      "epoch": 0.0052,
      "grad_norm": 0.8817258231902962,
      "learning_rate": 0.0015600000000000002,
      "loss": 5.2966,
      "step": 520
    },
    {
      "epoch": 0.00521,
      "grad_norm": 0.9047592596397305,
      "learning_rate": 0.0015630000000000002,
      "loss": 5.2732,
      "step": 521
    },
    {
      "epoch": 0.00522,
      "grad_norm": 0.6139926424217688,
      "learning_rate": 0.0015660000000000001,
      "loss": 5.2701,
      "step": 522
    },
    {
      "epoch": 0.00523,
      "grad_norm": 0.6292464658556638,
      "learning_rate": 0.001569,
      "loss": 5.2826,
      "step": 523
    },
    {
      "epoch": 0.00524,
      "grad_norm": 0.752030715547053,
      "learning_rate": 0.001572,
      "loss": 5.2856,
      "step": 524
    },
    {
      "epoch": 0.00525,
      "grad_norm": 0.929770536798091,
      "learning_rate": 0.001575,
      "loss": 5.287,
      "step": 525
    },
    {
      "epoch": 0.00526,
      "grad_norm": 0.887370520628206,
      "learning_rate": 0.0015780000000000002,
      "loss": 5.2635,
      "step": 526
    },
    {
      "epoch": 0.00527,
      "grad_norm": 0.7819104471156305,
      "learning_rate": 0.0015810000000000002,
      "loss": 5.2728,
      "step": 527
    },
    {
      "epoch": 0.00528,
      "grad_norm": 0.9038037239389326,
      "learning_rate": 0.0015840000000000001,
      "loss": 5.2534,
      "step": 528
    },
    {
      "epoch": 0.00529,
      "grad_norm": 0.8898068580069259,
      "learning_rate": 0.001587,
      "loss": 5.265,
      "step": 529
    },
    {
      "epoch": 0.0053,
      "grad_norm": 1.0010848916209774,
      "learning_rate": 0.00159,
      "loss": 5.2764,
      "step": 530
    },
    {
      "epoch": 0.00531,
      "grad_norm": 1.0010984765594055,
      "learning_rate": 0.001593,
      "loss": 5.2677,
      "step": 531
    },
    {
      "epoch": 0.00532,
      "grad_norm": 1.026224264336229,
      "learning_rate": 0.0015960000000000002,
      "loss": 5.2779,
      "step": 532
    },
    {
      "epoch": 0.00533,
      "grad_norm": 0.9759289805508353,
      "learning_rate": 0.0015990000000000002,
      "loss": 5.2678,
      "step": 533
    },
    {
      "epoch": 0.00534,
      "grad_norm": 1.1376558518204782,
      "learning_rate": 0.0016020000000000001,
      "loss": 5.2612,
      "step": 534
    },
    {
      "epoch": 0.00535,
      "grad_norm": 1.0517455067238486,
      "learning_rate": 0.001605,
      "loss": 5.2672,
      "step": 535
    },
    {
      "epoch": 0.00536,
      "grad_norm": 0.9398243649272562,
      "learning_rate": 0.001608,
      "loss": 5.2627,
      "step": 536
    },
    {
      "epoch": 0.00537,
      "grad_norm": 0.9512995727424398,
      "learning_rate": 0.0016110000000000002,
      "loss": 5.2558,
      "step": 537
    },
    {
      "epoch": 0.00538,
      "grad_norm": 1.2746657535312511,
      "learning_rate": 0.0016140000000000002,
      "loss": 5.2451,
      "step": 538
    },
    {
      "epoch": 0.00539,
      "grad_norm": 1.067518036012326,
      "learning_rate": 0.0016170000000000002,
      "loss": 5.2415,
      "step": 539
    },
    {
      "epoch": 0.0054,
      "grad_norm": 1.1859628541053806,
      "learning_rate": 0.0016200000000000001,
      "loss": 5.2618,
      "step": 540
    },
    {
      "epoch": 0.00541,
      "grad_norm": 0.8882936824028492,
      "learning_rate": 0.001623,
      "loss": 5.2308,
      "step": 541
    },
    {
      "epoch": 0.00542,
      "grad_norm": 0.8517075205384302,
      "learning_rate": 0.001626,
      "loss": 5.2545,
      "step": 542
    },
    {
      "epoch": 0.00543,
      "grad_norm": 0.8283552605034004,
      "learning_rate": 0.0016290000000000002,
      "loss": 5.2098,
      "step": 543
    },
    {
      "epoch": 0.00544,
      "grad_norm": 0.9087829134911761,
      "learning_rate": 0.0016320000000000002,
      "loss": 5.2265,
      "step": 544
    },
    {
      "epoch": 0.00545,
      "grad_norm": 0.8034144620978907,
      "learning_rate": 0.0016350000000000002,
      "loss": 5.2348,
      "step": 545
    },
    {
      "epoch": 0.00546,
      "grad_norm": 0.7091235133563132,
      "learning_rate": 0.0016380000000000001,
      "loss": 5.2004,
      "step": 546
    },
    {
      "epoch": 0.00547,
      "grad_norm": 0.6683331586466694,
      "learning_rate": 0.001641,
      "loss": 5.1915,
      "step": 547
    },
    {
      "epoch": 0.00548,
      "grad_norm": 0.5441552662279447,
      "learning_rate": 0.001644,
      "loss": 5.1843,
      "step": 548
    },
    {
      "epoch": 0.00549,
      "grad_norm": 0.5462993858197037,
      "learning_rate": 0.0016470000000000002,
      "loss": 5.1806,
      "step": 549
    },
    {
      "epoch": 0.0055,
      "grad_norm": 0.5691406737163984,
      "learning_rate": 0.0016500000000000002,
      "loss": 5.1914,
      "step": 550
    },
    {
      "epoch": 0.00551,
      "grad_norm": 0.57998143312047,
      "learning_rate": 0.0016530000000000002,
      "loss": 5.1789,
      "step": 551
    },
    {
      "epoch": 0.00552,
      "grad_norm": 0.5286482487653069,
      "learning_rate": 0.0016560000000000001,
      "loss": 5.158,
      "step": 552
    },
    {
      "epoch": 0.00553,
      "grad_norm": 0.48759014943874474,
      "learning_rate": 0.001659,
      "loss": 5.1465,
      "step": 553
    },
    {
      "epoch": 0.00554,
      "grad_norm": 0.4792524526847805,
      "learning_rate": 0.0016620000000000003,
      "loss": 5.1537,
      "step": 554
    },
    {
      "epoch": 0.00555,
      "grad_norm": 0.5054837577144806,
      "learning_rate": 0.0016650000000000002,
      "loss": 5.1496,
      "step": 555
    },
    {
      "epoch": 0.00556,
      "grad_norm": 0.5148667639200912,
      "learning_rate": 0.0016680000000000002,
      "loss": 5.1288,
      "step": 556
    },
    {
      "epoch": 0.00557,
      "grad_norm": 0.5122706189056161,
      "learning_rate": 0.0016710000000000002,
      "loss": 5.1153,
      "step": 557
    },
    {
      "epoch": 0.00558,
      "grad_norm": 0.6046224327964763,
      "learning_rate": 0.0016740000000000001,
      "loss": 5.1358,
      "step": 558
    },
    {
      "epoch": 0.00559,
      "grad_norm": 0.6851636176151574,
      "learning_rate": 0.001677,
      "loss": 5.1268,
      "step": 559
    },
    {
      "epoch": 0.0056,
      "grad_norm": 0.6922547745331437,
      "learning_rate": 0.0016800000000000003,
      "loss": 5.096,
      "step": 560
    },
    {
      "epoch": 0.00561,
      "grad_norm": 0.7411075733344746,
      "learning_rate": 0.0016830000000000003,
      "loss": 5.1176,
      "step": 561
    },
    {
      "epoch": 0.00562,
      "grad_norm": 0.8981627852593407,
      "learning_rate": 0.0016860000000000002,
      "loss": 5.1206,
      "step": 562
    },
    {
      "epoch": 0.00563,
      "grad_norm": 1.1260200882948381,
      "learning_rate": 0.001689,
      "loss": 5.1239,
      "step": 563
    },
    {
      "epoch": 0.00564,
      "grad_norm": 1.1027210513289374,
      "learning_rate": 0.001692,
      "loss": 5.113,
      "step": 564
    },
    {
      "epoch": 0.00565,
      "grad_norm": 0.890213024695838,
      "learning_rate": 0.001695,
      "loss": 5.1071,
      "step": 565
    },
    {
      "epoch": 0.00566,
      "grad_norm": 0.8627979775394023,
      "learning_rate": 0.0016979999999999999,
      "loss": 5.0976,
      "step": 566
    },
    {
      "epoch": 0.00567,
      "grad_norm": 0.9823945407746334,
      "learning_rate": 0.0017009999999999998,
      "loss": 5.101,
      "step": 567
    },
    {
      "epoch": 0.00568,
      "grad_norm": 0.9715849000742567,
      "learning_rate": 0.0017039999999999998,
      "loss": 5.1139,
      "step": 568
    },
    {
      "epoch": 0.00569,
      "grad_norm": 0.9989913016974431,
      "learning_rate": 0.001707,
      "loss": 5.0861,
      "step": 569
    },
    {
      "epoch": 0.0057,
      "grad_norm": 1.0927877338999235,
      "learning_rate": 0.00171,
      "loss": 5.1262,
      "step": 570
    },
    {
      "epoch": 0.00571,
      "grad_norm": 1.0328737042176641,
      "learning_rate": 0.001713,
      "loss": 5.1192,
      "step": 571
    },
    {
      "epoch": 0.00572,
      "grad_norm": 1.3722388350456287,
      "learning_rate": 0.0017159999999999999,
      "loss": 5.1049,
      "step": 572
    },
    {
      "epoch": 0.00573,
      "grad_norm": 0.9525979203379623,
      "learning_rate": 0.0017189999999999998,
      "loss": 5.1081,
      "step": 573
    },
    {
      "epoch": 0.00574,
      "grad_norm": 1.0626128534442882,
      "learning_rate": 0.001722,
      "loss": 5.1048,
      "step": 574
    },
    {
      "epoch": 0.00575,
      "grad_norm": 0.9331527599734185,
      "learning_rate": 0.001725,
      "loss": 5.074,
      "step": 575
    },
    {
      "epoch": 0.00576,
      "grad_norm": 0.9277735357052385,
      "learning_rate": 0.001728,
      "loss": 5.0843,
      "step": 576
    },
    {
      "epoch": 0.00577,
      "grad_norm": 0.9070967517565243,
      "learning_rate": 0.001731,
      "loss": 5.0908,
      "step": 577
    },
    {
      "epoch": 0.00578,
      "grad_norm": 0.8451551366430134,
      "learning_rate": 0.0017339999999999999,
      "loss": 5.0704,
      "step": 578
    },
    {
      "epoch": 0.00579,
      "grad_norm": 0.7590562073625285,
      "learning_rate": 0.0017369999999999998,
      "loss": 5.058,
      "step": 579
    },
    {
      "epoch": 0.0058,
      "grad_norm": 0.6385326977156373,
      "learning_rate": 0.00174,
      "loss": 5.0662,
      "step": 580
    },
    {
      "epoch": 0.00581,
      "grad_norm": 0.5982129206576257,
      "learning_rate": 0.001743,
      "loss": 5.0624,
      "step": 581
    },
    {
      "epoch": 0.00582,
      "grad_norm": 0.7358321954275717,
      "learning_rate": 0.001746,
      "loss": 5.0371,
      "step": 582
    },
    {
      "epoch": 0.00583,
      "grad_norm": 0.790704997209863,
      "learning_rate": 0.001749,
      "loss": 5.0641,
      "step": 583
    },
    {
      "epoch": 0.00584,
      "grad_norm": 0.7351931257233056,
      "learning_rate": 0.0017519999999999999,
      "loss": 5.0549,
      "step": 584
    },
    {
      "epoch": 0.00585,
      "grad_norm": 0.6167540927458872,
      "learning_rate": 0.0017549999999999998,
      "loss": 5.0362,
      "step": 585
    },
    {
      "epoch": 0.00586,
      "grad_norm": 0.6143004439139178,
      "learning_rate": 0.001758,
      "loss": 5.0316,
      "step": 586
    },
    {
      "epoch": 0.00587,
      "grad_norm": 0.619830997492515,
      "learning_rate": 0.001761,
      "loss": 5.018,
      "step": 587
    },
    {
      "epoch": 0.00588,
      "grad_norm": 0.676754014852622,
      "learning_rate": 0.001764,
      "loss": 5.0113,
      "step": 588
    },
    {
      "epoch": 0.00589,
      "grad_norm": 0.8152367183756798,
      "learning_rate": 0.001767,
      "loss": 5.0063,
      "step": 589
    },
    {
      "epoch": 0.0059,
      "grad_norm": 0.8067730505459064,
      "learning_rate": 0.0017699999999999999,
      "loss": 5.0009,
      "step": 590
    },
    {
      "epoch": 0.00591,
      "grad_norm": 0.7857115694134444,
      "learning_rate": 0.001773,
      "loss": 5.0014,
      "step": 591
    },
    {
      "epoch": 0.00592,
      "grad_norm": 1.0305248086573016,
      "learning_rate": 0.001776,
      "loss": 5.0384,
      "step": 592
    },
    {
      "epoch": 0.00593,
      "grad_norm": 1.1034322728795254,
      "learning_rate": 0.001779,
      "loss": 5.0387,
      "step": 593
    },
    {
      "epoch": 0.00594,
      "grad_norm": 0.796782800390064,
      "learning_rate": 0.001782,
      "loss": 5.0138,
      "step": 594
    },
    {
      "epoch": 0.00595,
      "grad_norm": 0.947951394101208,
      "learning_rate": 0.001785,
      "loss": 5.0288,
      "step": 595
    },
    {
      "epoch": 0.00596,
      "grad_norm": 0.948001541672118,
      "learning_rate": 0.0017879999999999999,
      "loss": 5.0236,
      "step": 596
    },
    {
      "epoch": 0.00597,
      "grad_norm": 0.8614311642183788,
      "learning_rate": 0.001791,
      "loss": 4.983,
      "step": 597
    },
    {
      "epoch": 0.00598,
      "grad_norm": 0.7822502963641336,
      "learning_rate": 0.001794,
      "loss": 5.0032,
      "step": 598
    },
    {
      "epoch": 0.00599,
      "grad_norm": 0.662502314672726,
      "learning_rate": 0.001797,
      "loss": 4.9824,
      "step": 599
    },
    {
      "epoch": 0.006,
      "grad_norm": 0.612691797581944,
      "learning_rate": 0.0018,
      "loss": 4.9865,
      "step": 600
    },
    {
      "epoch": 0.00601,
      "grad_norm": 0.6872791409552744,
      "learning_rate": 0.001803,
      "loss": 4.9768,
      "step": 601
    },
    {
      "epoch": 0.00602,
      "grad_norm": 0.6805787880992017,
      "learning_rate": 0.0018059999999999999,
      "loss": 4.9611,
      "step": 602
    },
    {
      "epoch": 0.00603,
      "grad_norm": 0.6726518476757681,
      "learning_rate": 0.001809,
      "loss": 4.9863,
      "step": 603
    },
    {
      "epoch": 0.00604,
      "grad_norm": 0.6569683723636671,
      "learning_rate": 0.001812,
      "loss": 4.9594,
      "step": 604
    },
    {
      "epoch": 0.00605,
      "grad_norm": 0.7635898698708745,
      "learning_rate": 0.001815,
      "loss": 4.9733,
      "step": 605
    },
    {
      "epoch": 0.00606,
      "grad_norm": 0.7455554600622951,
      "learning_rate": 0.001818,
      "loss": 4.9743,
      "step": 606
    },
    {
      "epoch": 0.00607,
      "grad_norm": 0.7493825131119356,
      "learning_rate": 0.001821,
      "loss": 4.968,
      "step": 607
    },
    {
      "epoch": 0.00608,
      "grad_norm": 0.7957828614510537,
      "learning_rate": 0.001824,
      "loss": 4.9536,
      "step": 608
    },
    {
      "epoch": 0.00609,
      "grad_norm": 0.8304899179729943,
      "learning_rate": 0.001827,
      "loss": 4.9564,
      "step": 609
    },
    {
      "epoch": 0.0061,
      "grad_norm": 0.8109281284885322,
      "learning_rate": 0.00183,
      "loss": 4.9141,
      "step": 610
    },
    {
      "epoch": 0.00611,
      "grad_norm": 0.7476158062529945,
      "learning_rate": 0.001833,
      "loss": 4.936,
      "step": 611
    },
    {
      "epoch": 0.00612,
      "grad_norm": 0.7704555466663776,
      "learning_rate": 0.001836,
      "loss": 4.9262,
      "step": 612
    },
    {
      "epoch": 0.00613,
      "grad_norm": 0.7179188189545482,
      "learning_rate": 0.001839,
      "loss": 4.9285,
      "step": 613
    },
    {
      "epoch": 0.00614,
      "grad_norm": 0.6487628040476678,
      "learning_rate": 0.001842,
      "loss": 4.9483,
      "step": 614
    },
    {
      "epoch": 0.00615,
      "grad_norm": 0.6346341426047015,
      "learning_rate": 0.001845,
      "loss": 4.9336,
      "step": 615
    },
    {
      "epoch": 0.00616,
      "grad_norm": 0.6919482296512011,
      "learning_rate": 0.001848,
      "loss": 4.92,
      "step": 616
    },
    {
      "epoch": 0.00617,
      "grad_norm": 0.8470324326487232,
      "learning_rate": 0.001851,
      "loss": 4.937,
      "step": 617
    },
    {
      "epoch": 0.00618,
      "grad_norm": 0.9780451420741652,
      "learning_rate": 0.001854,
      "loss": 4.918,
      "step": 618
    },
    {
      "epoch": 0.00619,
      "grad_norm": 1.0393940937639112,
      "learning_rate": 0.001857,
      "loss": 4.9393,
      "step": 619
    },
    {
      "epoch": 0.0062,
      "grad_norm": 1.0864888594503201,
      "learning_rate": 0.00186,
      "loss": 4.9315,
      "step": 620
    },
    {
      "epoch": 0.00621,
      "grad_norm": 1.1243573748169302,
      "learning_rate": 0.001863,
      "loss": 4.931,
      "step": 621
    },
    {
      "epoch": 0.00622,
      "grad_norm": 1.2701399731469665,
      "learning_rate": 0.001866,
      "loss": 4.9502,
      "step": 622
    },
    {
      "epoch": 0.00623,
      "grad_norm": 0.8273605340089976,
      "learning_rate": 0.001869,
      "loss": 4.9139,
      "step": 623
    },
    {
      "epoch": 0.00624,
      "grad_norm": 0.9643311427939416,
      "learning_rate": 0.001872,
      "loss": 4.9253,
      "step": 624
    },
    {
      "epoch": 0.00625,
      "grad_norm": 1.2310098040523034,
      "learning_rate": 0.001875,
      "loss": 4.9539,
      "step": 625
    },
    {
      "epoch": 0.00626,
      "grad_norm": 0.8268848993203812,
      "learning_rate": 0.0018780000000000001,
      "loss": 4.9199,
      "step": 626
    },
    {
      "epoch": 0.00627,
      "grad_norm": 0.9575639973355906,
      "learning_rate": 0.001881,
      "loss": 4.909,
      "step": 627
    },
    {
      "epoch": 0.00628,
      "grad_norm": 0.9391450767019313,
      "learning_rate": 0.001884,
      "loss": 4.9487,
      "step": 628
    },
    {
      "epoch": 0.00629,
      "grad_norm": 0.8387975784827085,
      "learning_rate": 0.001887,
      "loss": 4.8976,
      "step": 629
    },
    {
      "epoch": 0.0063,
      "grad_norm": 0.7745062985579545,
      "learning_rate": 0.00189,
      "loss": 4.9062,
      "step": 630
    },
    {
      "epoch": 0.00631,
      "grad_norm": 0.8213366097020988,
      "learning_rate": 0.0018930000000000002,
      "loss": 4.9006,
      "step": 631
    },
    {
      "epoch": 0.00632,
      "grad_norm": 0.8670608316656828,
      "learning_rate": 0.0018960000000000001,
      "loss": 4.9226,
      "step": 632
    },
    {
      "epoch": 0.00633,
      "grad_norm": 0.9183940724690223,
      "learning_rate": 0.001899,
      "loss": 4.9055,
      "step": 633
    },
    {
      "epoch": 0.00634,
      "grad_norm": 0.8953336214105924,
      "learning_rate": 0.001902,
      "loss": 4.8931,
      "step": 634
    },
    {
      "epoch": 0.00635,
      "grad_norm": 0.8869843611928877,
      "learning_rate": 0.001905,
      "loss": 4.8851,
      "step": 635
    },
    {
      "epoch": 0.00636,
      "grad_norm": 1.014617544478078,
      "learning_rate": 0.001908,
      "loss": 4.911,
      "step": 636
    },
    {
      "epoch": 0.00637,
      "grad_norm": 0.8681767489690893,
      "learning_rate": 0.0019110000000000002,
      "loss": 4.909,
      "step": 637
    },
    {
      "epoch": 0.00638,
      "grad_norm": 0.9792692499883113,
      "learning_rate": 0.0019140000000000001,
      "loss": 4.9073,
      "step": 638
    },
    {
      "epoch": 0.00639,
      "grad_norm": 0.8702384519183142,
      "learning_rate": 0.001917,
      "loss": 4.8724,
      "step": 639
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.7656979903996737,
      "learning_rate": 0.00192,
      "loss": 4.8809,
      "step": 640
    },
    {
      "epoch": 0.00641,
      "grad_norm": 0.7895438458212233,
      "learning_rate": 0.001923,
      "loss": 4.8762,
      "step": 641
    },
    {
      "epoch": 0.00642,
      "grad_norm": 0.8281699495124627,
      "learning_rate": 0.001926,
      "loss": 4.8749,
      "step": 642
    },
    {
      "epoch": 0.00643,
      "grad_norm": 0.7845360432833325,
      "learning_rate": 0.0019290000000000002,
      "loss": 4.8856,
      "step": 643
    },
    {
      "epoch": 0.00644,
      "grad_norm": 0.6857666802048429,
      "learning_rate": 0.0019320000000000001,
      "loss": 4.8583,
      "step": 644
    },
    {
      "epoch": 0.00645,
      "grad_norm": 0.5056941566313082,
      "learning_rate": 0.001935,
      "loss": 4.8522,
      "step": 645
    },
    {
      "epoch": 0.00646,
      "grad_norm": 0.47964630717910517,
      "learning_rate": 0.001938,
      "loss": 4.8628,
      "step": 646
    },
    {
      "epoch": 0.00647,
      "grad_norm": 0.519046406898008,
      "learning_rate": 0.001941,
      "loss": 4.8552,
      "step": 647
    },
    {
      "epoch": 0.00648,
      "grad_norm": 0.4935185080219269,
      "learning_rate": 0.0019440000000000002,
      "loss": 4.8488,
      "step": 648
    },
    {
      "epoch": 0.00649,
      "grad_norm": 0.5205599594018984,
      "learning_rate": 0.0019470000000000002,
      "loss": 4.8333,
      "step": 649
    },
    {
      "epoch": 0.0065,
      "grad_norm": 0.45295352715192905,
      "learning_rate": 0.0019500000000000001,
      "loss": 4.8617,
      "step": 650
    },
    {
      "epoch": 0.00651,
      "grad_norm": 0.4378817965282719,
      "learning_rate": 0.001953,
      "loss": 4.84,
      "step": 651
    },
    {
      "epoch": 0.00652,
      "grad_norm": 0.48395025642257355,
      "learning_rate": 0.0019560000000000003,
      "loss": 4.8266,
      "step": 652
    },
    {
      "epoch": 0.00653,
      "grad_norm": 0.5194845024092328,
      "learning_rate": 0.0019590000000000002,
      "loss": 4.8294,
      "step": 653
    },
    {
      "epoch": 0.00654,
      "grad_norm": 0.5707068593609262,
      "learning_rate": 0.001962,
      "loss": 4.8026,
      "step": 654
    },
    {
      "epoch": 0.00655,
      "grad_norm": 0.6725172209521785,
      "learning_rate": 0.001965,
      "loss": 4.834,
      "step": 655
    },
    {
      "epoch": 0.00656,
      "grad_norm": 0.7729568432311581,
      "learning_rate": 0.001968,
      "loss": 4.8398,
      "step": 656
    },
    {
      "epoch": 0.00657,
      "grad_norm": 0.9182317536069756,
      "learning_rate": 0.001971,
      "loss": 4.8175,
      "step": 657
    },
    {
      "epoch": 0.00658,
      "grad_norm": 0.9241834469814327,
      "learning_rate": 0.001974,
      "loss": 4.8482,
      "step": 658
    },
    {
      "epoch": 0.00659,
      "grad_norm": 0.7554289117841486,
      "learning_rate": 0.001977,
      "loss": 4.8199,
      "step": 659
    },
    {
      "epoch": 0.0066,
      "grad_norm": 0.786222644002383,
      "learning_rate": 0.00198,
      "loss": 4.8336,
      "step": 660
    },
    {
      "epoch": 0.00661,
      "grad_norm": 0.7486956414167568,
      "learning_rate": 0.001983,
      "loss": 4.8035,
      "step": 661
    },
    {
      "epoch": 0.00662,
      "grad_norm": 0.8224936329580866,
      "learning_rate": 0.0019860000000000004,
      "loss": 4.8038,
      "step": 662
    },
    {
      "epoch": 0.00663,
      "grad_norm": 1.0333891900276588,
      "learning_rate": 0.0019890000000000003,
      "loss": 4.8216,
      "step": 663
    },
    {
      "epoch": 0.00664,
      "grad_norm": 1.0454184895583787,
      "learning_rate": 0.0019920000000000003,
      "loss": 4.826,
      "step": 664
    },
    {
      "epoch": 0.00665,
      "grad_norm": 0.9684168839706867,
      "learning_rate": 0.0019950000000000002,
      "loss": 4.7952,
      "step": 665
    },
    {
      "epoch": 0.00666,
      "grad_norm": 1.0353944208678731,
      "learning_rate": 0.001998,
      "loss": 4.8213,
      "step": 666
    },
    {
      "epoch": 0.00667,
      "grad_norm": 0.9360230286599108,
      "learning_rate": 0.002001,
      "loss": 4.8166,
      "step": 667
    },
    {
      "epoch": 0.00668,
      "grad_norm": 0.7256915823313501,
      "learning_rate": 0.002004,
      "loss": 4.83,
      "step": 668
    },
    {
      "epoch": 0.00669,
      "grad_norm": 0.8394069732991961,
      "learning_rate": 0.002007,
      "loss": 4.8494,
      "step": 669
    },
    {
      "epoch": 0.0067,
      "grad_norm": 1.077577977883947,
      "learning_rate": 0.00201,
      "loss": 4.8532,
      "step": 670
    },
    {
      "epoch": 0.00671,
      "grad_norm": 0.944781077919628,
      "learning_rate": 0.002013,
      "loss": 4.8236,
      "step": 671
    },
    {
      "epoch": 0.00672,
      "grad_norm": 0.7024086289087288,
      "learning_rate": 0.002016,
      "loss": 4.7966,
      "step": 672
    },
    {
      "epoch": 0.00673,
      "grad_norm": 0.6931174948166909,
      "learning_rate": 0.002019,
      "loss": 4.809,
      "step": 673
    },
    {
      "epoch": 0.00674,
      "grad_norm": 0.6108100930625565,
      "learning_rate": 0.0020220000000000004,
      "loss": 4.8005,
      "step": 674
    },
    {
      "epoch": 0.00675,
      "grad_norm": 0.5498762136107717,
      "learning_rate": 0.0020250000000000003,
      "loss": 4.8171,
      "step": 675
    },
    {
      "epoch": 0.00676,
      "grad_norm": 0.6244881777878202,
      "learning_rate": 0.0020280000000000003,
      "loss": 4.7951,
      "step": 676
    },
    {
      "epoch": 0.00677,
      "grad_norm": 0.5815667573949113,
      "learning_rate": 0.0020310000000000003,
      "loss": 4.7913,
      "step": 677
    },
    {
      "epoch": 0.00678,
      "grad_norm": 0.521491074875552,
      "learning_rate": 0.0020340000000000002,
      "loss": 4.7959,
      "step": 678
    },
    {
      "epoch": 0.00679,
      "grad_norm": 0.5801192243287052,
      "learning_rate": 0.002037,
      "loss": 4.7864,
      "step": 679
    },
    {
      "epoch": 0.0068,
      "grad_norm": 0.6398788453868361,
      "learning_rate": 0.00204,
      "loss": 4.7875,
      "step": 680
    },
    {
      "epoch": 0.00681,
      "grad_norm": 0.6456567912241706,
      "learning_rate": 0.002043,
      "loss": 4.7807,
      "step": 681
    },
    {
      "epoch": 0.00682,
      "grad_norm": 0.6889535888719373,
      "learning_rate": 0.002046,
      "loss": 4.7776,
      "step": 682
    },
    {
      "epoch": 0.00683,
      "grad_norm": 0.68132819276425,
      "learning_rate": 0.002049,
      "loss": 4.7612,
      "step": 683
    },
    {
      "epoch": 0.00684,
      "grad_norm": 0.6724100911918068,
      "learning_rate": 0.002052,
      "loss": 4.7814,
      "step": 684
    },
    {
      "epoch": 0.00685,
      "grad_norm": 0.815928121483196,
      "learning_rate": 0.0020550000000000004,
      "loss": 4.7913,
      "step": 685
    },
    {
      "epoch": 0.00686,
      "grad_norm": 0.7537150884203632,
      "learning_rate": 0.0020580000000000004,
      "loss": 4.7584,
      "step": 686
    },
    {
      "epoch": 0.00687,
      "grad_norm": 0.7326770278926477,
      "learning_rate": 0.0020610000000000003,
      "loss": 4.7655,
      "step": 687
    },
    {
      "epoch": 0.00688,
      "grad_norm": 0.7145131916015852,
      "learning_rate": 0.002064,
      "loss": 4.7631,
      "step": 688
    },
    {
      "epoch": 0.00689,
      "grad_norm": 0.694828893634318,
      "learning_rate": 0.002067,
      "loss": 4.7685,
      "step": 689
    },
    {
      "epoch": 0.0069,
      "grad_norm": 0.7402628986076095,
      "learning_rate": 0.00207,
      "loss": 4.7629,
      "step": 690
    },
    {
      "epoch": 0.00691,
      "grad_norm": 0.7809038473743037,
      "learning_rate": 0.0020729999999999998,
      "loss": 4.734,
      "step": 691
    },
    {
      "epoch": 0.00692,
      "grad_norm": 0.9321737974219415,
      "learning_rate": 0.0020759999999999997,
      "loss": 4.7623,
      "step": 692
    },
    {
      "epoch": 0.00693,
      "grad_norm": 1.0940875671633443,
      "learning_rate": 0.0020789999999999997,
      "loss": 4.788,
      "step": 693
    },
    {
      "epoch": 0.00694,
      "grad_norm": 0.9030072391970948,
      "learning_rate": 0.002082,
      "loss": 4.7582,
      "step": 694
    },
    {
      "epoch": 0.00695,
      "grad_norm": 0.8857004379777869,
      "learning_rate": 0.002085,
      "loss": 4.7649,
      "step": 695
    },
    {
      "epoch": 0.00696,
      "grad_norm": 0.8991843577664433,
      "learning_rate": 0.002088,
      "loss": 4.7712,
      "step": 696
    },
    {
      "epoch": 0.00697,
      "grad_norm": 1.0094726318481275,
      "learning_rate": 0.002091,
      "loss": 4.7869,
      "step": 697
    },
    {
      "epoch": 0.00698,
      "grad_norm": 1.09376996979269,
      "learning_rate": 0.002094,
      "loss": 4.7773,
      "step": 698
    },
    {
      "epoch": 0.00699,
      "grad_norm": 0.8766228601427865,
      "learning_rate": 0.002097,
      "loss": 4.7548,
      "step": 699
    },
    {
      "epoch": 0.007,
      "grad_norm": 0.8580408760742435,
      "learning_rate": 0.0021,
      "loss": 4.7811,
      "step": 700
    },
    {
      "epoch": 0.00701,
      "grad_norm": 0.7398805390269545,
      "learning_rate": 0.002103,
      "loss": 4.7496,
      "step": 701
    },
    {
      "epoch": 0.00702,
      "grad_norm": 0.8329034135969645,
      "learning_rate": 0.002106,
      "loss": 4.762,
      "step": 702
    },
    {
      "epoch": 0.00703,
      "grad_norm": 0.8505784581176613,
      "learning_rate": 0.0021089999999999998,
      "loss": 4.7427,
      "step": 703
    },
    {
      "epoch": 0.00704,
      "grad_norm": 0.8003313240374758,
      "learning_rate": 0.0021119999999999997,
      "loss": 4.7516,
      "step": 704
    },
    {
      "epoch": 0.00705,
      "grad_norm": 0.7109629500256387,
      "learning_rate": 0.002115,
      "loss": 4.7635,
      "step": 705
    },
    {
      "epoch": 0.00706,
      "grad_norm": 0.6289559048526784,
      "learning_rate": 0.002118,
      "loss": 4.7473,
      "step": 706
    },
    {
      "epoch": 0.00707,
      "grad_norm": 0.5482382237711074,
      "learning_rate": 0.002121,
      "loss": 4.7193,
      "step": 707
    },
    {
      "epoch": 0.00708,
      "grad_norm": 0.5598390325768088,
      "learning_rate": 0.002124,
      "loss": 4.7015,
      "step": 708
    },
    {
      "epoch": 0.00709,
      "grad_norm": 0.5529551760214286,
      "learning_rate": 0.002127,
      "loss": 4.7331,
      "step": 709
    },
    {
      "epoch": 0.0071,
      "grad_norm": 0.6671084870548735,
      "learning_rate": 0.00213,
      "loss": 4.7271,
      "step": 710
    },
    {
      "epoch": 0.00711,
      "grad_norm": 1.1336515667563218,
      "learning_rate": 0.002133,
      "loss": 4.7225,
      "step": 711
    },
    {
      "epoch": 0.00712,
      "grad_norm": 0.8878745395415352,
      "learning_rate": 0.002136,
      "loss": 4.761,
      "step": 712
    },
    {
      "epoch": 0.00713,
      "grad_norm": 0.6767692274074788,
      "learning_rate": 0.002139,
      "loss": 4.7547,
      "step": 713
    },
    {
      "epoch": 0.00714,
      "grad_norm": 0.7698487468076475,
      "learning_rate": 0.002142,
      "loss": 4.7544,
      "step": 714
    },
    {
      "epoch": 0.00715,
      "grad_norm": 0.6189384208707256,
      "learning_rate": 0.0021449999999999998,
      "loss": 4.7077,
      "step": 715
    },
    {
      "epoch": 0.00716,
      "grad_norm": 0.5968344999809513,
      "learning_rate": 0.002148,
      "loss": 4.7064,
      "step": 716
    },
    {
      "epoch": 0.00717,
      "grad_norm": 0.5117629580972962,
      "learning_rate": 0.002151,
      "loss": 4.7111,
      "step": 717
    },
    {
      "epoch": 0.00718,
      "grad_norm": 0.5586203984201494,
      "learning_rate": 0.002154,
      "loss": 4.7079,
      "step": 718
    },
    {
      "epoch": 0.00719,
      "grad_norm": 0.5075479633130432,
      "learning_rate": 0.002157,
      "loss": 4.7121,
      "step": 719
    },
    {
      "epoch": 0.0072,
      "grad_norm": 0.4840284295584723,
      "learning_rate": 0.00216,
      "loss": 4.7045,
      "step": 720
    },
    {
      "epoch": 0.00721,
      "grad_norm": 0.4971167150526394,
      "learning_rate": 0.002163,
      "loss": 4.7056,
      "step": 721
    },
    {
      "epoch": 0.00722,
      "grad_norm": 0.5347204422021664,
      "learning_rate": 0.002166,
      "loss": 4.6807,
      "step": 722
    },
    {
      "epoch": 0.00723,
      "grad_norm": 0.5424206330484346,
      "learning_rate": 0.002169,
      "loss": 4.6942,
      "step": 723
    },
    {
      "epoch": 0.00724,
      "grad_norm": 0.5219918511718452,
      "learning_rate": 0.002172,
      "loss": 4.6912,
      "step": 724
    },
    {
      "epoch": 0.00725,
      "grad_norm": 0.4975444213381179,
      "learning_rate": 0.002175,
      "loss": 4.6834,
      "step": 725
    },
    {
      "epoch": 0.00726,
      "grad_norm": 0.5840922524665152,
      "learning_rate": 0.002178,
      "loss": 4.7008,
      "step": 726
    },
    {
      "epoch": 0.00727,
      "grad_norm": 0.7633698368434719,
      "learning_rate": 0.0021809999999999998,
      "loss": 4.6777,
      "step": 727
    },
    {
      "epoch": 0.00728,
      "grad_norm": 0.9052599955462425,
      "learning_rate": 0.002184,
      "loss": 4.7084,
      "step": 728
    },
    {
      "epoch": 0.00729,
      "grad_norm": 0.9052862109699649,
      "learning_rate": 0.002187,
      "loss": 4.7065,
      "step": 729
    },
    {
      "epoch": 0.0073,
      "grad_norm": 0.9788098741089467,
      "learning_rate": 0.00219,
      "loss": 4.7163,
      "step": 730
    },
    {
      "epoch": 0.00731,
      "grad_norm": 0.9218858714566311,
      "learning_rate": 0.002193,
      "loss": 4.7193,
      "step": 731
    },
    {
      "epoch": 0.00732,
      "grad_norm": 1.1412145043722297,
      "learning_rate": 0.002196,
      "loss": 4.7513,
      "step": 732
    },
    {
      "epoch": 0.00733,
      "grad_norm": 1.0851884555194036,
      "learning_rate": 0.002199,
      "loss": 4.7135,
      "step": 733
    },
    {
      "epoch": 0.00734,
      "grad_norm": 0.9295060612046938,
      "learning_rate": 0.002202,
      "loss": 4.745,
      "step": 734
    },
    {
      "epoch": 0.00735,
      "grad_norm": 1.1744848346988581,
      "learning_rate": 0.002205,
      "loss": 4.766,
      "step": 735
    },
    {
      "epoch": 0.00736,
      "grad_norm": 0.8278931321658332,
      "learning_rate": 0.002208,
      "loss": 4.7413,
      "step": 736
    },
    {
      "epoch": 0.00737,
      "grad_norm": 0.8170748474782394,
      "learning_rate": 0.002211,
      "loss": 4.701,
      "step": 737
    },
    {
      "epoch": 0.00738,
      "grad_norm": 0.772640277783885,
      "learning_rate": 0.002214,
      "loss": 4.7083,
      "step": 738
    },
    {
      "epoch": 0.00739,
      "grad_norm": 0.6328057525928463,
      "learning_rate": 0.0022170000000000002,
      "loss": 4.6886,
      "step": 739
    },
    {
      "epoch": 0.0074,
      "grad_norm": 0.6456777453177256,
      "learning_rate": 0.00222,
      "loss": 4.7001,
      "step": 740
    },
    {
      "epoch": 0.00741,
      "grad_norm": 0.5912580097221507,
      "learning_rate": 0.002223,
      "loss": 4.6652,
      "step": 741
    },
    {
      "epoch": 0.00742,
      "grad_norm": 0.4622238730689768,
      "learning_rate": 0.002226,
      "loss": 4.6972,
      "step": 742
    },
    {
      "epoch": 0.00743,
      "grad_norm": 0.4904966133460914,
      "learning_rate": 0.002229,
      "loss": 4.6585,
      "step": 743
    },
    {
      "epoch": 0.00744,
      "grad_norm": 0.5611948954376664,
      "learning_rate": 0.002232,
      "loss": 4.6643,
      "step": 744
    },
    {
      "epoch": 0.00745,
      "grad_norm": 0.6222946629538655,
      "learning_rate": 0.002235,
      "loss": 4.6777,
      "step": 745
    },
    {
      "epoch": 0.00746,
      "grad_norm": 0.6678221674623648,
      "learning_rate": 0.002238,
      "loss": 4.6735,
      "step": 746
    },
    {
      "epoch": 0.00747,
      "grad_norm": 0.7945455347777014,
      "learning_rate": 0.002241,
      "loss": 4.6637,
      "step": 747
    },
    {
      "epoch": 0.00748,
      "grad_norm": 0.9027593299924651,
      "learning_rate": 0.002244,
      "loss": 4.683,
      "step": 748
    },
    {
      "epoch": 0.00749,
      "grad_norm": 0.8224517051186598,
      "learning_rate": 0.002247,
      "loss": 4.6612,
      "step": 749
    },
    {
      "epoch": 0.0075,
      "grad_norm": 0.6737194715747743,
      "learning_rate": 0.0022500000000000003,
      "loss": 4.6955,
      "step": 750
    },
    {
      "epoch": 0.00751,
      "grad_norm": 0.7232822830562855,
      "learning_rate": 0.0022530000000000002,
      "loss": 4.6893,
      "step": 751
    },
    {
      "epoch": 0.00752,
      "grad_norm": 0.6449225627821611,
      "learning_rate": 0.002256,
      "loss": 4.637,
      "step": 752
    },
    {
      "epoch": 0.00753,
      "grad_norm": 0.6558342053271724,
      "learning_rate": 0.002259,
      "loss": 4.6782,
      "step": 753
    },
    {
      "epoch": 0.00754,
      "grad_norm": 0.7056102529212358,
      "learning_rate": 0.002262,
      "loss": 4.6906,
      "step": 754
    },
    {
      "epoch": 0.00755,
      "grad_norm": 0.9043185481447295,
      "learning_rate": 0.002265,
      "loss": 4.6606,
      "step": 755
    },
    {
      "epoch": 0.00756,
      "grad_norm": 0.8770375990036015,
      "learning_rate": 0.002268,
      "loss": 4.6551,
      "step": 756
    },
    {
      "epoch": 0.00757,
      "grad_norm": 0.784655485815756,
      "learning_rate": 0.002271,
      "loss": 4.6655,
      "step": 757
    },
    {
      "epoch": 0.00758,
      "grad_norm": 0.6912079950422929,
      "learning_rate": 0.002274,
      "loss": 4.655,
      "step": 758
    },
    {
      "epoch": 0.00759,
      "grad_norm": 0.6281371231708264,
      "learning_rate": 0.002277,
      "loss": 4.6629,
      "step": 759
    },
    {
      "epoch": 0.0076,
      "grad_norm": 0.5992557531443571,
      "learning_rate": 0.00228,
      "loss": 4.6463,
      "step": 760
    },
    {
      "epoch": 0.00761,
      "grad_norm": 0.6034804931159479,
      "learning_rate": 0.002283,
      "loss": 4.6146,
      "step": 761
    },
    {
      "epoch": 0.00762,
      "grad_norm": 0.6060668832918386,
      "learning_rate": 0.0022860000000000003,
      "loss": 4.5953,
      "step": 762
    },
    {
      "epoch": 0.00763,
      "grad_norm": 0.5816892837581542,
      "learning_rate": 0.0022890000000000002,
      "loss": 4.6459,
      "step": 763
    },
    {
      "epoch": 0.00764,
      "grad_norm": 0.599786845424844,
      "learning_rate": 0.002292,
      "loss": 4.6554,
      "step": 764
    },
    {
      "epoch": 0.00765,
      "grad_norm": 0.6640243443171511,
      "learning_rate": 0.002295,
      "loss": 4.6451,
      "step": 765
    },
    {
      "epoch": 0.00766,
      "grad_norm": 0.6663176934335011,
      "learning_rate": 0.002298,
      "loss": 4.6438,
      "step": 766
    },
    {
      "epoch": 0.00767,
      "grad_norm": 0.6384162747773746,
      "learning_rate": 0.002301,
      "loss": 4.6392,
      "step": 767
    },
    {
      "epoch": 0.00768,
      "grad_norm": 0.8537658256313299,
      "learning_rate": 0.002304,
      "loss": 4.6457,
      "step": 768
    },
    {
      "epoch": 0.00769,
      "grad_norm": 1.0098120726516953,
      "learning_rate": 0.002307,
      "loss": 4.6694,
      "step": 769
    },
    {
      "epoch": 0.0077,
      "grad_norm": 0.7423003820485509,
      "learning_rate": 0.00231,
      "loss": 4.6223,
      "step": 770
    },
    {
      "epoch": 0.00771,
      "grad_norm": 0.6691351211965346,
      "learning_rate": 0.002313,
      "loss": 4.6569,
      "step": 771
    },
    {
      "epoch": 0.00772,
      "grad_norm": 0.8185609860643415,
      "learning_rate": 0.002316,
      "loss": 4.6531,
      "step": 772
    },
    {
      "epoch": 0.00773,
      "grad_norm": 1.0039613061235502,
      "learning_rate": 0.0023190000000000003,
      "loss": 4.6664,
      "step": 773
    },
    {
      "epoch": 0.00774,
      "grad_norm": 0.8500607234716588,
      "learning_rate": 0.0023220000000000003,
      "loss": 4.6845,
      "step": 774
    },
    {
      "epoch": 0.00775,
      "grad_norm": 0.7057192127032131,
      "learning_rate": 0.0023250000000000002,
      "loss": 4.6688,
      "step": 775
    },
    {
      "epoch": 0.00776,
      "grad_norm": 0.8527617729361273,
      "learning_rate": 0.002328,
      "loss": 4.6732,
      "step": 776
    },
    {
      "epoch": 0.00777,
      "grad_norm": 0.7987783133918631,
      "learning_rate": 0.002331,
      "loss": 4.6469,
      "step": 777
    },
    {
      "epoch": 0.00778,
      "grad_norm": 0.8221364860163118,
      "learning_rate": 0.002334,
      "loss": 4.6645,
      "step": 778
    },
    {
      "epoch": 0.00779,
      "grad_norm": 0.8901185821524535,
      "learning_rate": 0.002337,
      "loss": 4.6243,
      "step": 779
    },
    {
      "epoch": 0.0078,
      "grad_norm": 0.889956887452623,
      "learning_rate": 0.00234,
      "loss": 4.646,
      "step": 780
    },
    {
      "epoch": 0.00781,
      "grad_norm": 0.763195026878423,
      "learning_rate": 0.002343,
      "loss": 4.6465,
      "step": 781
    },
    {
      "epoch": 0.00782,
      "grad_norm": 0.6941360203492353,
      "learning_rate": 0.002346,
      "loss": 4.6378,
      "step": 782
    },
    {
      "epoch": 0.00783,
      "grad_norm": 0.6504570243922513,
      "learning_rate": 0.002349,
      "loss": 4.6195,
      "step": 783
    },
    {
      "epoch": 0.00784,
      "grad_norm": 0.5612286515783087,
      "learning_rate": 0.002352,
      "loss": 4.6293,
      "step": 784
    },
    {
      "epoch": 0.00785,
      "grad_norm": 0.6491841258536649,
      "learning_rate": 0.0023550000000000003,
      "loss": 4.6258,
      "step": 785
    },
    {
      "epoch": 0.00786,
      "grad_norm": 0.7618604140557245,
      "learning_rate": 0.0023580000000000003,
      "loss": 4.6353,
      "step": 786
    },
    {
      "epoch": 0.00787,
      "grad_norm": 0.9289251225559676,
      "learning_rate": 0.0023610000000000003,
      "loss": 4.6478,
      "step": 787
    },
    {
      "epoch": 0.00788,
      "grad_norm": 1.0160410825626025,
      "learning_rate": 0.002364,
      "loss": 4.6569,
      "step": 788
    },
    {
      "epoch": 0.00789,
      "grad_norm": 0.9904632547103471,
      "learning_rate": 0.002367,
      "loss": 4.6841,
      "step": 789
    },
    {
      "epoch": 0.0079,
      "grad_norm": 0.978549262142993,
      "learning_rate": 0.00237,
      "loss": 4.6525,
      "step": 790
    },
    {
      "epoch": 0.00791,
      "grad_norm": 0.7626000247604355,
      "learning_rate": 0.002373,
      "loss": 4.6008,
      "step": 791
    },
    {
      "epoch": 0.00792,
      "grad_norm": 0.7032433100729657,
      "learning_rate": 0.002376,
      "loss": 4.6361,
      "step": 792
    },
    {
      "epoch": 0.00793,
      "grad_norm": 0.6164393933780811,
      "learning_rate": 0.002379,
      "loss": 4.6263,
      "step": 793
    },
    {
      "epoch": 0.00794,
      "grad_norm": 0.6849838631141278,
      "learning_rate": 0.002382,
      "loss": 4.6402,
      "step": 794
    },
    {
      "epoch": 0.00795,
      "grad_norm": 0.7541453624973187,
      "learning_rate": 0.002385,
      "loss": 4.6181,
      "step": 795
    },
    {
      "epoch": 0.00796,
      "grad_norm": 0.7900314117496404,
      "learning_rate": 0.0023880000000000004,
      "loss": 4.6154,
      "step": 796
    },
    {
      "epoch": 0.00797,
      "grad_norm": 0.8032302751261208,
      "learning_rate": 0.0023910000000000003,
      "loss": 4.6598,
      "step": 797
    },
    {
      "epoch": 0.00798,
      "grad_norm": 0.7037506066431396,
      "learning_rate": 0.0023940000000000003,
      "loss": 4.6206,
      "step": 798
    },
    {
      "epoch": 0.00799,
      "grad_norm": 0.6873143294249761,
      "learning_rate": 0.0023970000000000003,
      "loss": 4.6565,
      "step": 799
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.5611968184575689,
      "learning_rate": 0.0024000000000000002,
      "loss": 4.6002,
      "step": 800
    },
    {
      "epoch": 0.00801,
      "grad_norm": 0.6431524925156966,
      "learning_rate": 0.002403,
      "loss": 4.6326,
      "step": 801
    },
    {
      "epoch": 0.00802,
      "grad_norm": 0.664361851854149,
      "learning_rate": 0.002406,
      "loss": 4.624,
      "step": 802
    },
    {
      "epoch": 0.00803,
      "grad_norm": 0.7385532765716779,
      "learning_rate": 0.002409,
      "loss": 4.6034,
      "step": 803
    },
    {
      "epoch": 0.00804,
      "grad_norm": 0.6763586866376322,
      "learning_rate": 0.002412,
      "loss": 4.609,
      "step": 804
    },
    {
      "epoch": 0.00805,
      "grad_norm": 0.4984516329675046,
      "learning_rate": 0.002415,
      "loss": 4.5958,
      "step": 805
    },
    {
      "epoch": 0.00806,
      "grad_norm": 0.6422151194456196,
      "learning_rate": 0.002418,
      "loss": 4.6202,
      "step": 806
    },
    {
      "epoch": 0.00807,
      "grad_norm": 0.6520347813599764,
      "learning_rate": 0.0024210000000000004,
      "loss": 4.6111,
      "step": 807
    },
    {
      "epoch": 0.00808,
      "grad_norm": 0.6466117231395049,
      "learning_rate": 0.0024240000000000004,
      "loss": 4.5938,
      "step": 808
    },
    {
      "epoch": 0.00809,
      "grad_norm": 0.7173763198404414,
      "learning_rate": 0.0024270000000000003,
      "loss": 4.5875,
      "step": 809
    },
    {
      "epoch": 0.0081,
      "grad_norm": 0.7630708316428362,
      "learning_rate": 0.0024300000000000003,
      "loss": 4.6257,
      "step": 810
    },
    {
      "epoch": 0.00811,
      "grad_norm": 0.6597258634486322,
      "learning_rate": 0.0024330000000000003,
      "loss": 4.5916,
      "step": 811
    },
    {
      "epoch": 0.00812,
      "grad_norm": 0.5596135689444884,
      "learning_rate": 0.0024360000000000002,
      "loss": 4.5639,
      "step": 812
    },
    {
      "epoch": 0.00813,
      "grad_norm": 0.5455291997717118,
      "learning_rate": 0.0024389999999999998,
      "loss": 4.6039,
      "step": 813
    },
    {
      "epoch": 0.00814,
      "grad_norm": 0.643394229982758,
      "learning_rate": 0.0024419999999999997,
      "loss": 4.6125,
      "step": 814
    },
    {
      "epoch": 0.00815,
      "grad_norm": 0.7154153020141445,
      "learning_rate": 0.0024449999999999997,
      "loss": 4.6204,
      "step": 815
    },
    {
      "epoch": 0.00816,
      "grad_norm": 0.7061775094598466,
      "learning_rate": 0.002448,
      "loss": 4.6005,
      "step": 816
    },
    {
      "epoch": 0.00817,
      "grad_norm": 0.7497949470997279,
      "learning_rate": 0.002451,
      "loss": 4.6054,
      "step": 817
    },
    {
      "epoch": 0.00818,
      "grad_norm": 0.8303821582765404,
      "learning_rate": 0.002454,
      "loss": 4.6047,
      "step": 818
    },
    {
      "epoch": 0.00819,
      "grad_norm": 0.7870772660210968,
      "learning_rate": 0.002457,
      "loss": 4.6129,
      "step": 819
    },
    {
      "epoch": 0.0082,
      "grad_norm": 0.8045799279983024,
      "learning_rate": 0.00246,
      "loss": 4.5959,
      "step": 820
    },
    {
      "epoch": 0.00821,
      "grad_norm": 0.6847814476141592,
      "learning_rate": 0.002463,
      "loss": 4.5753,
      "step": 821
    },
    {
      "epoch": 0.00822,
      "grad_norm": 0.6767342952639601,
      "learning_rate": 0.002466,
      "loss": 4.5743,
      "step": 822
    },
    {
      "epoch": 0.00823,
      "grad_norm": 0.7814972414925468,
      "learning_rate": 0.002469,
      "loss": 4.5892,
      "step": 823
    },
    {
      "epoch": 0.00824,
      "grad_norm": 0.7970591329920511,
      "learning_rate": 0.002472,
      "loss": 4.6124,
      "step": 824
    },
    {
      "epoch": 0.00825,
      "grad_norm": 0.7237059608391628,
      "learning_rate": 0.0024749999999999998,
      "loss": 4.6158,
      "step": 825
    },
    {
      "epoch": 0.00826,
      "grad_norm": 0.8825063151832095,
      "learning_rate": 0.0024779999999999997,
      "loss": 4.5838,
      "step": 826
    },
    {
      "epoch": 0.00827,
      "grad_norm": 0.895197450200386,
      "learning_rate": 0.002481,
      "loss": 4.6224,
      "step": 827
    },
    {
      "epoch": 0.00828,
      "grad_norm": 0.7760924275661016,
      "learning_rate": 0.002484,
      "loss": 4.6087,
      "step": 828
    },
    {
      "epoch": 0.00829,
      "grad_norm": 0.8835354563964355,
      "learning_rate": 0.002487,
      "loss": 4.6157,
      "step": 829
    },
    {
      "epoch": 0.0083,
      "grad_norm": 0.8984013262238549,
      "learning_rate": 0.00249,
      "loss": 4.6232,
      "step": 830
    },
    {
      "epoch": 0.00831,
      "grad_norm": 0.8653834946214679,
      "learning_rate": 0.002493,
      "loss": 4.6038,
      "step": 831
    },
    {
      "epoch": 0.00832,
      "grad_norm": 0.8641563564291244,
      "learning_rate": 0.002496,
      "loss": 4.6089,
      "step": 832
    },
    {
      "epoch": 0.00833,
      "grad_norm": 0.8849940351412556,
      "learning_rate": 0.002499,
      "loss": 4.6069,
      "step": 833
    },
    {
      "epoch": 0.00834,
      "grad_norm": 0.9333438912141526,
      "learning_rate": 0.002502,
      "loss": 4.6165,
      "step": 834
    },
    {
      "epoch": 0.00835,
      "grad_norm": 0.8853225829488065,
      "learning_rate": 0.002505,
      "loss": 4.6108,
      "step": 835
    },
    {
      "epoch": 0.00836,
      "grad_norm": 0.9042165265638229,
      "learning_rate": 0.002508,
      "loss": 4.6433,
      "step": 836
    },
    {
      "epoch": 0.00837,
      "grad_norm": 0.8034248166459536,
      "learning_rate": 0.0025109999999999998,
      "loss": 4.5745,
      "step": 837
    },
    {
      "epoch": 0.00838,
      "grad_norm": 0.8086998635281932,
      "learning_rate": 0.0025139999999999997,
      "loss": 4.6114,
      "step": 838
    },
    {
      "epoch": 0.00839,
      "grad_norm": 0.587375456181757,
      "learning_rate": 0.002517,
      "loss": 4.5954,
      "step": 839
    },
    {
      "epoch": 0.0084,
      "grad_norm": 0.6561997865962053,
      "learning_rate": 0.00252,
      "loss": 4.5818,
      "step": 840
    },
    {
      "epoch": 0.00841,
      "grad_norm": 0.6860624256233325,
      "learning_rate": 0.002523,
      "loss": 4.5813,
      "step": 841
    },
    {
      "epoch": 0.00842,
      "grad_norm": 0.6384378914075728,
      "learning_rate": 0.002526,
      "loss": 4.5783,
      "step": 842
    },
    {
      "epoch": 0.00843,
      "grad_norm": 0.5397302637540236,
      "learning_rate": 0.002529,
      "loss": 4.5728,
      "step": 843
    },
    {
      "epoch": 0.00844,
      "grad_norm": 0.5753282450552862,
      "learning_rate": 0.002532,
      "loss": 4.5912,
      "step": 844
    },
    {
      "epoch": 0.00845,
      "grad_norm": 0.605188801053789,
      "learning_rate": 0.002535,
      "loss": 4.5588,
      "step": 845
    },
    {
      "epoch": 0.00846,
      "grad_norm": 0.523231421872222,
      "learning_rate": 0.002538,
      "loss": 4.5771,
      "step": 846
    },
    {
      "epoch": 0.00847,
      "grad_norm": 0.4442194357732046,
      "learning_rate": 0.002541,
      "loss": 4.5539,
      "step": 847
    },
    {
      "epoch": 0.00848,
      "grad_norm": 0.4340595323225716,
      "learning_rate": 0.002544,
      "loss": 4.5562,
      "step": 848
    },
    {
      "epoch": 0.00849,
      "grad_norm": 0.40987686572717924,
      "learning_rate": 0.002547,
      "loss": 4.5471,
      "step": 849
    },
    {
      "epoch": 0.0085,
      "grad_norm": 0.38823479672015115,
      "learning_rate": 0.00255,
      "loss": 4.5547,
      "step": 850
    },
    {
      "epoch": 0.00851,
      "grad_norm": 0.42997075207568086,
      "learning_rate": 0.002553,
      "loss": 4.5733,
      "step": 851
    },
    {
      "epoch": 0.00852,
      "grad_norm": 0.5192269912554525,
      "learning_rate": 0.002556,
      "loss": 4.5624,
      "step": 852
    },
    {
      "epoch": 0.00853,
      "grad_norm": 0.6626364019198889,
      "learning_rate": 0.002559,
      "loss": 4.5465,
      "step": 853
    },
    {
      "epoch": 0.00854,
      "grad_norm": 0.9475750496728129,
      "learning_rate": 0.002562,
      "loss": 4.5544,
      "step": 854
    },
    {
      "epoch": 0.00855,
      "grad_norm": 0.9196861875673602,
      "learning_rate": 0.002565,
      "loss": 4.5554,
      "step": 855
    },
    {
      "epoch": 0.00856,
      "grad_norm": 0.6972970172901616,
      "learning_rate": 0.002568,
      "loss": 4.5747,
      "step": 856
    },
    {
      "epoch": 0.00857,
      "grad_norm": 0.7425487517589463,
      "learning_rate": 0.002571,
      "loss": 4.5717,
      "step": 857
    },
    {
      "epoch": 0.00858,
      "grad_norm": 0.678911102843075,
      "learning_rate": 0.002574,
      "loss": 4.5576,
      "step": 858
    },
    {
      "epoch": 0.00859,
      "grad_norm": 0.6680695283315139,
      "learning_rate": 0.002577,
      "loss": 4.5494,
      "step": 859
    },
    {
      "epoch": 0.0086,
      "grad_norm": 0.7323389776740598,
      "learning_rate": 0.00258,
      "loss": 4.5797,
      "step": 860
    },
    {
      "epoch": 0.00861,
      "grad_norm": 0.6564756555111457,
      "learning_rate": 0.0025830000000000002,
      "loss": 4.5928,
      "step": 861
    },
    {
      "epoch": 0.00862,
      "grad_norm": 0.6458920384469554,
      "learning_rate": 0.002586,
      "loss": 4.5402,
      "step": 862
    },
    {
      "epoch": 0.00863,
      "grad_norm": 0.56307833723133,
      "learning_rate": 0.002589,
      "loss": 4.5659,
      "step": 863
    },
    {
      "epoch": 0.00864,
      "grad_norm": 0.6301527121328465,
      "learning_rate": 0.002592,
      "loss": 4.5169,
      "step": 864
    },
    {
      "epoch": 0.00865,
      "grad_norm": 0.5741592962502369,
      "learning_rate": 0.002595,
      "loss": 4.5342,
      "step": 865
    },
    {
      "epoch": 0.00866,
      "grad_norm": 0.6185631950272453,
      "learning_rate": 0.002598,
      "loss": 4.5624,
      "step": 866
    },
    {
      "epoch": 0.00867,
      "grad_norm": 0.7226342011827854,
      "learning_rate": 0.002601,
      "loss": 4.5538,
      "step": 867
    },
    {
      "epoch": 0.00868,
      "grad_norm": 0.6871063796143311,
      "learning_rate": 0.002604,
      "loss": 4.5354,
      "step": 868
    },
    {
      "epoch": 0.00869,
      "grad_norm": 0.6146544871717051,
      "learning_rate": 0.002607,
      "loss": 4.5411,
      "step": 869
    },
    {
      "epoch": 0.0087,
      "grad_norm": 0.5850681754527672,
      "learning_rate": 0.00261,
      "loss": 4.5279,
      "step": 870
    },
    {
      "epoch": 0.00871,
      "grad_norm": 0.6619050810997609,
      "learning_rate": 0.002613,
      "loss": 4.5496,
      "step": 871
    },
    {
      "epoch": 0.00872,
      "grad_norm": 0.6309510333250544,
      "learning_rate": 0.002616,
      "loss": 4.5312,
      "step": 872
    },
    {
      "epoch": 0.00873,
      "grad_norm": 0.620458951387438,
      "learning_rate": 0.0026190000000000002,
      "loss": 4.5409,
      "step": 873
    },
    {
      "epoch": 0.00874,
      "grad_norm": 0.7675582961358233,
      "learning_rate": 0.002622,
      "loss": 4.5679,
      "step": 874
    },
    {
      "epoch": 0.00875,
      "grad_norm": 0.8413199287183839,
      "learning_rate": 0.002625,
      "loss": 4.5217,
      "step": 875
    },
    {
      "epoch": 0.00876,
      "grad_norm": 0.6439305106538762,
      "learning_rate": 0.002628,
      "loss": 4.5435,
      "step": 876
    },
    {
      "epoch": 0.00877,
      "grad_norm": 0.6155973146282673,
      "learning_rate": 0.002631,
      "loss": 4.5607,
      "step": 877
    },
    {
      "epoch": 0.00878,
      "grad_norm": 0.6572664983016872,
      "learning_rate": 0.002634,
      "loss": 4.4959,
      "step": 878
    },
    {
      "epoch": 0.00879,
      "grad_norm": 0.678308084092591,
      "learning_rate": 0.002637,
      "loss": 4.526,
      "step": 879
    },
    {
      "epoch": 0.0088,
      "grad_norm": 0.682644138366955,
      "learning_rate": 0.00264,
      "loss": 4.5397,
      "step": 880
    },
    {
      "epoch": 0.00881,
      "grad_norm": 0.6916960221367301,
      "learning_rate": 0.002643,
      "loss": 4.5424,
      "step": 881
    },
    {
      "epoch": 0.00882,
      "grad_norm": 0.8625989079873627,
      "learning_rate": 0.002646,
      "loss": 4.5523,
      "step": 882
    },
    {
      "epoch": 0.00883,
      "grad_norm": 0.9115056000231379,
      "learning_rate": 0.002649,
      "loss": 4.5538,
      "step": 883
    },
    {
      "epoch": 0.00884,
      "grad_norm": 1.1386535317654105,
      "learning_rate": 0.0026520000000000003,
      "loss": 4.5528,
      "step": 884
    },
    {
      "epoch": 0.00885,
      "grad_norm": 0.9401738564991132,
      "learning_rate": 0.0026550000000000002,
      "loss": 4.5478,
      "step": 885
    },
    {
      "epoch": 0.00886,
      "grad_norm": 1.0261918984073035,
      "learning_rate": 0.002658,
      "loss": 4.5785,
      "step": 886
    },
    {
      "epoch": 0.00887,
      "grad_norm": 1.2427377192417122,
      "learning_rate": 0.002661,
      "loss": 4.6,
      "step": 887
    },
    {
      "epoch": 0.00888,
      "grad_norm": 1.061546131863703,
      "learning_rate": 0.002664,
      "loss": 4.5885,
      "step": 888
    },
    {
      "epoch": 0.00889,
      "grad_norm": 0.907593068920597,
      "learning_rate": 0.002667,
      "loss": 4.535,
      "step": 889
    },
    {
      "epoch": 0.0089,
      "grad_norm": 0.8129313603021157,
      "learning_rate": 0.00267,
      "loss": 4.5855,
      "step": 890
    },
    {
      "epoch": 0.00891,
      "grad_norm": 0.8985699151758613,
      "learning_rate": 0.002673,
      "loss": 4.5762,
      "step": 891
    },
    {
      "epoch": 0.00892,
      "grad_norm": 0.9328845387585828,
      "learning_rate": 0.002676,
      "loss": 4.5962,
      "step": 892
    },
    {
      "epoch": 0.00893,
      "grad_norm": 0.9658499469408992,
      "learning_rate": 0.002679,
      "loss": 4.5743,
      "step": 893
    },
    {
      "epoch": 0.00894,
      "grad_norm": 0.9262060176852961,
      "learning_rate": 0.002682,
      "loss": 4.5624,
      "step": 894
    },
    {
      "epoch": 0.00895,
      "grad_norm": 1.0340643070320608,
      "learning_rate": 0.0026850000000000003,
      "loss": 4.5798,
      "step": 895
    },
    {
      "epoch": 0.00896,
      "grad_norm": 0.961568519948311,
      "learning_rate": 0.0026880000000000003,
      "loss": 4.5724,
      "step": 896
    },
    {
      "epoch": 0.00897,
      "grad_norm": 0.9522979094666822,
      "learning_rate": 0.0026910000000000002,
      "loss": 4.5901,
      "step": 897
    },
    {
      "epoch": 0.00898,
      "grad_norm": 0.90558868323236,
      "learning_rate": 0.002694,
      "loss": 4.5802,
      "step": 898
    },
    {
      "epoch": 0.00899,
      "grad_norm": 0.7752291080346148,
      "learning_rate": 0.002697,
      "loss": 4.5848,
      "step": 899
    },
    {
      "epoch": 0.009,
      "grad_norm": 0.7198757797557334,
      "learning_rate": 0.0027,
      "loss": 4.5998,
      "step": 900
    },
    {
      "epoch": 0.00901,
      "grad_norm": 0.7328950523365488,
      "learning_rate": 0.002703,
      "loss": 4.5552,
      "step": 901
    },
    {
      "epoch": 0.00902,
      "grad_norm": 0.5974121750232518,
      "learning_rate": 0.002706,
      "loss": 4.558,
      "step": 902
    },
    {
      "epoch": 0.00903,
      "grad_norm": 0.5753314236304967,
      "learning_rate": 0.002709,
      "loss": 4.5534,
      "step": 903
    },
    {
      "epoch": 0.00904,
      "grad_norm": 0.6128643778394687,
      "learning_rate": 0.002712,
      "loss": 4.5329,
      "step": 904
    },
    {
      "epoch": 0.00905,
      "grad_norm": 0.5691916638541509,
      "learning_rate": 0.002715,
      "loss": 4.5446,
      "step": 905
    },
    {
      "epoch": 0.00906,
      "grad_norm": 0.5122473095411203,
      "learning_rate": 0.002718,
      "loss": 4.5538,
      "step": 906
    },
    {
      "epoch": 0.00907,
      "grad_norm": 0.4479956748150508,
      "learning_rate": 0.0027210000000000003,
      "loss": 4.5603,
      "step": 907
    },
    {
      "epoch": 0.00908,
      "grad_norm": 0.4387217017050332,
      "learning_rate": 0.0027240000000000003,
      "loss": 4.5304,
      "step": 908
    },
    {
      "epoch": 0.00909,
      "grad_norm": 0.4358600325368994,
      "learning_rate": 0.0027270000000000003,
      "loss": 4.5454,
      "step": 909
    },
    {
      "epoch": 0.0091,
      "grad_norm": 0.3849373702796595,
      "learning_rate": 0.0027300000000000002,
      "loss": 4.5477,
      "step": 910
    },
    {
      "epoch": 0.00911,
      "grad_norm": 0.4574591880095763,
      "learning_rate": 0.002733,
      "loss": 4.5038,
      "step": 911
    },
    {
      "epoch": 0.00912,
      "grad_norm": 0.6121326743360266,
      "learning_rate": 0.002736,
      "loss": 4.5131,
      "step": 912
    },
    {
      "epoch": 0.00913,
      "grad_norm": 0.6845161679319078,
      "learning_rate": 0.002739,
      "loss": 4.513,
      "step": 913
    },
    {
      "epoch": 0.00914,
      "grad_norm": 0.7333722528690678,
      "learning_rate": 0.002742,
      "loss": 4.5432,
      "step": 914
    },
    {
      "epoch": 0.00915,
      "grad_norm": 0.6714442153045657,
      "learning_rate": 0.002745,
      "loss": 4.5048,
      "step": 915
    },
    {
      "epoch": 0.00916,
      "grad_norm": 0.5861682792098877,
      "learning_rate": 0.002748,
      "loss": 4.4768,
      "step": 916
    },
    {
      "epoch": 0.00917,
      "grad_norm": 0.7568015257717708,
      "learning_rate": 0.002751,
      "loss": 4.4985,
      "step": 917
    },
    {
      "epoch": 0.00918,
      "grad_norm": 0.67845085346177,
      "learning_rate": 0.0027540000000000004,
      "loss": 4.5185,
      "step": 918
    },
    {
      "epoch": 0.00919,
      "grad_norm": 0.6376249584521229,
      "learning_rate": 0.0027570000000000003,
      "loss": 4.4976,
      "step": 919
    },
    {
      "epoch": 0.0092,
      "grad_norm": 0.5953295490629925,
      "learning_rate": 0.0027600000000000003,
      "loss": 4.5104,
      "step": 920
    },
    {
      "epoch": 0.00921,
      "grad_norm": 0.5470113372090761,
      "learning_rate": 0.0027630000000000003,
      "loss": 4.4892,
      "step": 921
    },
    {
      "epoch": 0.00922,
      "grad_norm": 0.5586775976924615,
      "learning_rate": 0.0027660000000000002,
      "loss": 4.5007,
      "step": 922
    },
    {
      "epoch": 0.00923,
      "grad_norm": 0.5602976125001545,
      "learning_rate": 0.002769,
      "loss": 4.4767,
      "step": 923
    },
    {
      "epoch": 0.00924,
      "grad_norm": 0.5512257756150963,
      "learning_rate": 0.002772,
      "loss": 4.4992,
      "step": 924
    },
    {
      "epoch": 0.00925,
      "grad_norm": 0.6104014154605217,
      "learning_rate": 0.002775,
      "loss": 4.4804,
      "step": 925
    },
    {
      "epoch": 0.00926,
      "grad_norm": 0.627882987139063,
      "learning_rate": 0.002778,
      "loss": 4.4994,
      "step": 926
    },
    {
      "epoch": 0.00927,
      "grad_norm": 0.5637389395797978,
      "learning_rate": 0.002781,
      "loss": 4.4582,
      "step": 927
    },
    {
      "epoch": 0.00928,
      "grad_norm": 0.5793012970445044,
      "learning_rate": 0.002784,
      "loss": 4.505,
      "step": 928
    },
    {
      "epoch": 0.00929,
      "grad_norm": 0.5538600481661593,
      "learning_rate": 0.0027870000000000004,
      "loss": 4.4867,
      "step": 929
    },
    {
      "epoch": 0.0093,
      "grad_norm": 0.4915384406563099,
      "learning_rate": 0.0027900000000000004,
      "loss": 4.4652,
      "step": 930
    },
    {
      "epoch": 0.00931,
      "grad_norm": 0.5291025545236148,
      "learning_rate": 0.0027930000000000003,
      "loss": 4.4796,
      "step": 931
    },
    {
      "epoch": 0.00932,
      "grad_norm": 0.5189072536981689,
      "learning_rate": 0.0027960000000000003,
      "loss": 4.4881,
      "step": 932
    },
    {
      "epoch": 0.00933,
      "grad_norm": 0.4391692113712953,
      "learning_rate": 0.0027990000000000003,
      "loss": 4.4495,
      "step": 933
    },
    {
      "epoch": 0.00934,
      "grad_norm": 0.5067956527040228,
      "learning_rate": 0.0028020000000000002,
      "loss": 4.4811,
      "step": 934
    },
    {
      "epoch": 0.00935,
      "grad_norm": 0.5752082887405049,
      "learning_rate": 0.002805,
      "loss": 4.4893,
      "step": 935
    },
    {
      "epoch": 0.00936,
      "grad_norm": 0.8597610157431208,
      "learning_rate": 0.002808,
      "loss": 4.5012,
      "step": 936
    },
    {
      "epoch": 0.00937,
      "grad_norm": 1.0923006968336397,
      "learning_rate": 0.002811,
      "loss": 4.5462,
      "step": 937
    },
    {
      "epoch": 0.00938,
      "grad_norm": 0.8211742236965874,
      "learning_rate": 0.002814,
      "loss": 4.5052,
      "step": 938
    },
    {
      "epoch": 0.00939,
      "grad_norm": 0.699189062953733,
      "learning_rate": 0.002817,
      "loss": 4.4989,
      "step": 939
    },
    {
      "epoch": 0.0094,
      "grad_norm": 0.6251265662758575,
      "learning_rate": 0.00282,
      "loss": 4.495,
      "step": 940
    },
    {
      "epoch": 0.00941,
      "grad_norm": 0.8614570213641495,
      "learning_rate": 0.002823,
      "loss": 4.4823,
      "step": 941
    },
    {
      "epoch": 0.00942,
      "grad_norm": 0.9710835908546839,
      "learning_rate": 0.002826,
      "loss": 4.5282,
      "step": 942
    },
    {
      "epoch": 0.00943,
      "grad_norm": 0.8255156288963836,
      "learning_rate": 0.002829,
      "loss": 4.5175,
      "step": 943
    },
    {
      "epoch": 0.00944,
      "grad_norm": 0.8529303551550831,
      "learning_rate": 0.002832,
      "loss": 4.5233,
      "step": 944
    },
    {
      "epoch": 0.00945,
      "grad_norm": 0.8805636588256177,
      "learning_rate": 0.002835,
      "loss": 4.5307,
      "step": 945
    },
    {
      "epoch": 0.00946,
      "grad_norm": 0.8815394682869104,
      "learning_rate": 0.002838,
      "loss": 4.545,
      "step": 946
    },
    {
      "epoch": 0.00947,
      "grad_norm": 0.9792537736888025,
      "learning_rate": 0.0028409999999999998,
      "loss": 4.5259,
      "step": 947
    },
    {
      "epoch": 0.00948,
      "grad_norm": 1.16480627354709,
      "learning_rate": 0.0028439999999999997,
      "loss": 4.5258,
      "step": 948
    },
    {
      "epoch": 0.00949,
      "grad_norm": 0.9311558570406092,
      "learning_rate": 0.002847,
      "loss": 4.5678,
      "step": 949
    },
    {
      "epoch": 0.0095,
      "grad_norm": 0.9151500308490546,
      "learning_rate": 0.00285,
      "loss": 4.5566,
      "step": 950
    },
    {
      "epoch": 0.00951,
      "grad_norm": 0.8788038193040383,
      "learning_rate": 0.002853,
      "loss": 4.539,
      "step": 951
    },
    {
      "epoch": 0.00952,
      "grad_norm": 0.6883343465241535,
      "learning_rate": 0.002856,
      "loss": 4.5492,
      "step": 952
    },
    {
      "epoch": 0.00953,
      "grad_norm": 0.7428316161957196,
      "learning_rate": 0.002859,
      "loss": 4.4912,
      "step": 953
    },
    {
      "epoch": 0.00954,
      "grad_norm": 0.6589075049765231,
      "learning_rate": 0.002862,
      "loss": 4.5478,
      "step": 954
    },
    {
      "epoch": 0.00955,
      "grad_norm": 0.6374906159512967,
      "learning_rate": 0.002865,
      "loss": 4.5305,
      "step": 955
    },
    {
      "epoch": 0.00956,
      "grad_norm": 0.7037593758453927,
      "learning_rate": 0.002868,
      "loss": 4.5123,
      "step": 956
    },
    {
      "epoch": 0.00957,
      "grad_norm": 0.7575837270625414,
      "learning_rate": 0.002871,
      "loss": 4.5364,
      "step": 957
    },
    {
      "epoch": 0.00958,
      "grad_norm": 0.8136537306721928,
      "learning_rate": 0.002874,
      "loss": 4.5221,
      "step": 958
    },
    {
      "epoch": 0.00959,
      "grad_norm": 0.7529303499273822,
      "learning_rate": 0.002877,
      "loss": 4.5134,
      "step": 959
    },
    {
      "epoch": 0.0096,
      "grad_norm": 0.7723176518474248,
      "learning_rate": 0.0028799999999999997,
      "loss": 4.5128,
      "step": 960
    },
    {
      "epoch": 0.00961,
      "grad_norm": 0.6826701565045171,
      "learning_rate": 0.002883,
      "loss": 4.4802,
      "step": 961
    },
    {
      "epoch": 0.00962,
      "grad_norm": 0.6022941634910601,
      "learning_rate": 0.002886,
      "loss": 4.5086,
      "step": 962
    },
    {
      "epoch": 0.00963,
      "grad_norm": 0.5423609760386227,
      "learning_rate": 0.002889,
      "loss": 4.5025,
      "step": 963
    },
    {
      "epoch": 0.00964,
      "grad_norm": 0.5430502092899187,
      "learning_rate": 0.002892,
      "loss": 4.493,
      "step": 964
    },
    {
      "epoch": 0.00965,
      "grad_norm": 0.5046734063237629,
      "learning_rate": 0.002895,
      "loss": 4.4643,
      "step": 965
    },
    {
      "epoch": 0.00966,
      "grad_norm": 0.4625107724617308,
      "learning_rate": 0.002898,
      "loss": 4.4758,
      "step": 966
    },
    {
      "epoch": 0.00967,
      "grad_norm": 0.4403845150464526,
      "learning_rate": 0.002901,
      "loss": 4.5117,
      "step": 967
    },
    {
      "epoch": 0.00968,
      "grad_norm": 0.40739936008519134,
      "learning_rate": 0.002904,
      "loss": 4.4717,
      "step": 968
    },
    {
      "epoch": 0.00969,
      "grad_norm": 0.41620271754109195,
      "learning_rate": 0.002907,
      "loss": 4.5059,
      "step": 969
    },
    {
      "epoch": 0.0097,
      "grad_norm": 0.3884080596123629,
      "learning_rate": 0.00291,
      "loss": 4.4561,
      "step": 970
    },
    {
      "epoch": 0.00971,
      "grad_norm": 0.40963591049792675,
      "learning_rate": 0.002913,
      "loss": 4.4624,
      "step": 971
    },
    {
      "epoch": 0.00972,
      "grad_norm": 0.427860611994679,
      "learning_rate": 0.002916,
      "loss": 4.4693,
      "step": 972
    },
    {
      "epoch": 0.00973,
      "grad_norm": 0.6266830722962007,
      "learning_rate": 0.002919,
      "loss": 4.4689,
      "step": 973
    },
    {
      "epoch": 0.00974,
      "grad_norm": 0.7795241673951117,
      "learning_rate": 0.002922,
      "loss": 4.4734,
      "step": 974
    },
    {
      "epoch": 0.00975,
      "grad_norm": 0.7392088002205494,
      "learning_rate": 0.002925,
      "loss": 4.4915,
      "step": 975
    },
    {
      "epoch": 0.00976,
      "grad_norm": 0.6090919745791438,
      "learning_rate": 0.002928,
      "loss": 4.4688,
      "step": 976
    },
    {
      "epoch": 0.00977,
      "grad_norm": 0.6721667952819499,
      "learning_rate": 0.002931,
      "loss": 4.5114,
      "step": 977
    },
    {
      "epoch": 0.00978,
      "grad_norm": 0.7157395676490952,
      "learning_rate": 0.002934,
      "loss": 4.4828,
      "step": 978
    },
    {
      "epoch": 0.00979,
      "grad_norm": 0.6673604586868628,
      "learning_rate": 0.002937,
      "loss": 4.4724,
      "step": 979
    },
    {
      "epoch": 0.0098,
      "grad_norm": 0.6241827666303977,
      "learning_rate": 0.00294,
      "loss": 4.4641,
      "step": 980
    },
    {
      "epoch": 0.00981,
      "grad_norm": 0.6477735028588306,
      "learning_rate": 0.002943,
      "loss": 4.479,
      "step": 981
    },
    {
      "epoch": 0.00982,
      "grad_norm": 0.5903229373466107,
      "learning_rate": 0.002946,
      "loss": 4.4717,
      "step": 982
    },
    {
      "epoch": 0.00983,
      "grad_norm": 0.5367081021491611,
      "learning_rate": 0.0029490000000000002,
      "loss": 4.4684,
      "step": 983
    },
    {
      "epoch": 0.00984,
      "grad_norm": 0.55855952891459,
      "learning_rate": 0.002952,
      "loss": 4.455,
      "step": 984
    },
    {
      "epoch": 0.00985,
      "grad_norm": 0.6566800838357816,
      "learning_rate": 0.002955,
      "loss": 4.4789,
      "step": 985
    },
    {
      "epoch": 0.00986,
      "grad_norm": 0.7602203626787503,
      "learning_rate": 0.002958,
      "loss": 4.4461,
      "step": 986
    },
    {
      "epoch": 0.00987,
      "grad_norm": 0.7515002206199092,
      "learning_rate": 0.002961,
      "loss": 4.4656,
      "step": 987
    },
    {
      "epoch": 0.00988,
      "grad_norm": 0.7510028045825011,
      "learning_rate": 0.002964,
      "loss": 4.497,
      "step": 988
    },
    {
      "epoch": 0.00989,
      "grad_norm": 0.6686524266681442,
      "learning_rate": 0.002967,
      "loss": 4.438,
      "step": 989
    },
    {
      "epoch": 0.0099,
      "grad_norm": 0.7110181433203292,
      "learning_rate": 0.00297,
      "loss": 4.4759,
      "step": 990
    },
    {
      "epoch": 0.00991,
      "grad_norm": 0.9487700940206629,
      "learning_rate": 0.002973,
      "loss": 4.4909,
      "step": 991
    },
    {
      "epoch": 0.00992,
      "grad_norm": 1.0128664486589338,
      "learning_rate": 0.002976,
      "loss": 4.4936,
      "step": 992
    },
    {
      "epoch": 0.00993,
      "grad_norm": 0.776315556008045,
      "learning_rate": 0.002979,
      "loss": 4.4927,
      "step": 993
    },
    {
      "epoch": 0.00994,
      "grad_norm": 0.6954881784010936,
      "learning_rate": 0.002982,
      "loss": 4.4894,
      "step": 994
    },
    {
      "epoch": 0.00995,
      "grad_norm": 0.8343170542469196,
      "learning_rate": 0.0029850000000000002,
      "loss": 4.4966,
      "step": 995
    },
    {
      "epoch": 0.00996,
      "grad_norm": 0.8064256733311401,
      "learning_rate": 0.002988,
      "loss": 4.4974,
      "step": 996
    },
    {
      "epoch": 0.00997,
      "grad_norm": 0.7600837252115415,
      "learning_rate": 0.002991,
      "loss": 4.4859,
      "step": 997
    },
    {
      "epoch": 0.00998,
      "grad_norm": 0.7413851219310601,
      "learning_rate": 0.002994,
      "loss": 4.4878,
      "step": 998
    },
    {
      "epoch": 0.00999,
      "grad_norm": 0.7789142980889687,
      "learning_rate": 0.002997,
      "loss": 4.4774,
      "step": 999
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5973093879700552,
      "learning_rate": 0.003,
      "loss": 4.4903,
      "step": 1000
    },
    {
      "epoch": 0.01001,
      "grad_norm": 0.5861183186896064,
      "learning_rate": 0.003,
      "loss": 4.4874,
      "step": 1001
    },
    {
      "epoch": 0.01002,
      "grad_norm": 0.6212400302116947,
      "learning_rate": 0.003,
      "loss": 4.4557,
      "step": 1002
    },
    {
      "epoch": 0.01003,
      "grad_norm": 0.6083357500633214,
      "learning_rate": 0.003,
      "loss": 4.4779,
      "step": 1003
    },
    {
      "epoch": 0.01004,
      "grad_norm": 0.5830404519486425,
      "learning_rate": 0.003,
      "loss": 4.4653,
      "step": 1004
    },
    {
      "epoch": 0.01005,
      "grad_norm": 0.7477664572876674,
      "learning_rate": 0.003,
      "loss": 4.4465,
      "step": 1005
    },
    {
      "epoch": 0.01006,
      "grad_norm": 0.7841384646860249,
      "learning_rate": 0.003,
      "loss": 4.449,
      "step": 1006
    },
    {
      "epoch": 0.01007,
      "grad_norm": 0.7235135066769743,
      "learning_rate": 0.003,
      "loss": 4.4927,
      "step": 1007
    },
    {
      "epoch": 0.01008,
      "grad_norm": 0.6474532468386683,
      "learning_rate": 0.003,
      "loss": 4.4993,
      "step": 1008
    },
    {
      "epoch": 0.01009,
      "grad_norm": 0.5980408215414467,
      "learning_rate": 0.003,
      "loss": 4.4651,
      "step": 1009
    },
    {
      "epoch": 0.0101,
      "grad_norm": 0.5537282779479755,
      "learning_rate": 0.003,
      "loss": 4.4693,
      "step": 1010
    },
    {
      "epoch": 0.01011,
      "grad_norm": 0.5708819324924145,
      "learning_rate": 0.003,
      "loss": 4.4785,
      "step": 1011
    },
    {
      "epoch": 0.01012,
      "grad_norm": 0.4974883261450977,
      "learning_rate": 0.003,
      "loss": 4.4576,
      "step": 1012
    },
    {
      "epoch": 0.01013,
      "grad_norm": 0.4665247551568978,
      "learning_rate": 0.003,
      "loss": 4.4798,
      "step": 1013
    },
    {
      "epoch": 0.01014,
      "grad_norm": 0.42165803422689113,
      "learning_rate": 0.003,
      "loss": 4.4182,
      "step": 1014
    },
    {
      "epoch": 0.01015,
      "grad_norm": 0.4067134475964037,
      "learning_rate": 0.003,
      "loss": 4.4608,
      "step": 1015
    },
    {
      "epoch": 0.01016,
      "grad_norm": 0.5123147411688123,
      "learning_rate": 0.003,
      "loss": 4.4714,
      "step": 1016
    },
    {
      "epoch": 0.01017,
      "grad_norm": 0.6275035755668664,
      "learning_rate": 0.003,
      "loss": 4.4159,
      "step": 1017
    },
    {
      "epoch": 0.01018,
      "grad_norm": 0.8581692725167155,
      "learning_rate": 0.003,
      "loss": 4.4817,
      "step": 1018
    },
    {
      "epoch": 0.01019,
      "grad_norm": 0.8393947155453974,
      "learning_rate": 0.003,
      "loss": 4.4755,
      "step": 1019
    },
    {
      "epoch": 0.0102,
      "grad_norm": 0.6933428051970444,
      "learning_rate": 0.003,
      "loss": 4.4592,
      "step": 1020
    },
    {
      "epoch": 0.01021,
      "grad_norm": 0.7990025577528214,
      "learning_rate": 0.003,
      "loss": 4.4512,
      "step": 1021
    },
    {
      "epoch": 0.01022,
      "grad_norm": 0.8110623623420127,
      "learning_rate": 0.003,
      "loss": 4.4643,
      "step": 1022
    },
    {
      "epoch": 0.01023,
      "grad_norm": 0.7434736091722897,
      "learning_rate": 0.003,
      "loss": 4.4609,
      "step": 1023
    },
    {
      "epoch": 0.01024,
      "grad_norm": 0.7213203816141717,
      "learning_rate": 0.003,
      "loss": 4.417,
      "step": 1024
    },
    {
      "epoch": 0.01025,
      "grad_norm": 0.7010589428906292,
      "learning_rate": 0.003,
      "loss": 4.4518,
      "step": 1025
    },
    {
      "epoch": 0.01026,
      "grad_norm": 0.6280712957780686,
      "learning_rate": 0.003,
      "loss": 4.44,
      "step": 1026
    },
    {
      "epoch": 0.01027,
      "grad_norm": 0.5410145302803399,
      "learning_rate": 0.003,
      "loss": 4.4579,
      "step": 1027
    },
    {
      "epoch": 0.01028,
      "grad_norm": 0.4828232064846382,
      "learning_rate": 0.003,
      "loss": 4.4353,
      "step": 1028
    },
    {
      "epoch": 0.01029,
      "grad_norm": 0.4739043118914256,
      "learning_rate": 0.003,
      "loss": 4.412,
      "step": 1029
    },
    {
      "epoch": 0.0103,
      "grad_norm": 0.45225551629540334,
      "learning_rate": 0.003,
      "loss": 4.4408,
      "step": 1030
    },
    {
      "epoch": 0.01031,
      "grad_norm": 0.40901639056794215,
      "learning_rate": 0.003,
      "loss": 4.4521,
      "step": 1031
    },
    {
      "epoch": 0.01032,
      "grad_norm": 0.4266157306452721,
      "learning_rate": 0.003,
      "loss": 4.4241,
      "step": 1032
    },
    {
      "epoch": 0.01033,
      "grad_norm": 0.476318572377966,
      "learning_rate": 0.003,
      "loss": 4.4623,
      "step": 1033
    },
    {
      "epoch": 0.01034,
      "grad_norm": 0.4976367752405493,
      "learning_rate": 0.003,
      "loss": 4.4286,
      "step": 1034
    },
    {
      "epoch": 0.01035,
      "grad_norm": 0.48706616162522565,
      "learning_rate": 0.003,
      "loss": 4.427,
      "step": 1035
    },
    {
      "epoch": 0.01036,
      "grad_norm": 0.5540613528135478,
      "learning_rate": 0.003,
      "loss": 4.4185,
      "step": 1036
    },
    {
      "epoch": 0.01037,
      "grad_norm": 0.5585932570414466,
      "learning_rate": 0.003,
      "loss": 4.4458,
      "step": 1037
    },
    {
      "epoch": 0.01038,
      "grad_norm": 0.5014561937587317,
      "learning_rate": 0.003,
      "loss": 4.44,
      "step": 1038
    },
    {
      "epoch": 0.01039,
      "grad_norm": 0.5070723079813054,
      "learning_rate": 0.003,
      "loss": 4.4296,
      "step": 1039
    },
    {
      "epoch": 0.0104,
      "grad_norm": 0.5609823517547058,
      "learning_rate": 0.003,
      "loss": 4.4462,
      "step": 1040
    },
    {
      "epoch": 0.01041,
      "grad_norm": 0.664341991841828,
      "learning_rate": 0.003,
      "loss": 4.4332,
      "step": 1041
    },
    {
      "epoch": 0.01042,
      "grad_norm": 0.631896992045698,
      "learning_rate": 0.003,
      "loss": 4.458,
      "step": 1042
    },
    {
      "epoch": 0.01043,
      "grad_norm": 0.8655430264592057,
      "learning_rate": 0.003,
      "loss": 4.4103,
      "step": 1043
    },
    {
      "epoch": 0.01044,
      "grad_norm": 1.0843670364695495,
      "learning_rate": 0.003,
      "loss": 4.4702,
      "step": 1044
    },
    {
      "epoch": 0.01045,
      "grad_norm": 0.9544830822145469,
      "learning_rate": 0.003,
      "loss": 4.4986,
      "step": 1045
    },
    {
      "epoch": 0.01046,
      "grad_norm": 0.9579295900139632,
      "learning_rate": 0.003,
      "loss": 4.4718,
      "step": 1046
    },
    {
      "epoch": 0.01047,
      "grad_norm": 0.9275123969574217,
      "learning_rate": 0.003,
      "loss": 4.4903,
      "step": 1047
    },
    {
      "epoch": 0.01048,
      "grad_norm": 0.8417384302831535,
      "learning_rate": 0.003,
      "loss": 4.4694,
      "step": 1048
    },
    {
      "epoch": 0.01049,
      "grad_norm": 0.7711730353324007,
      "learning_rate": 0.003,
      "loss": 4.449,
      "step": 1049
    },
    {
      "epoch": 0.0105,
      "grad_norm": 0.747892326408584,
      "learning_rate": 0.003,
      "loss": 4.4932,
      "step": 1050
    },
    {
      "epoch": 0.01051,
      "grad_norm": 0.8100969209243005,
      "learning_rate": 0.003,
      "loss": 4.4724,
      "step": 1051
    },
    {
      "epoch": 0.01052,
      "grad_norm": 0.7211696952483453,
      "learning_rate": 0.003,
      "loss": 4.4859,
      "step": 1052
    },
    {
      "epoch": 0.01053,
      "grad_norm": 0.761937042029771,
      "learning_rate": 0.003,
      "loss": 4.4907,
      "step": 1053
    },
    {
      "epoch": 0.01054,
      "grad_norm": 0.791018085034848,
      "learning_rate": 0.003,
      "loss": 4.4849,
      "step": 1054
    },
    {
      "epoch": 0.01055,
      "grad_norm": 0.76513637315221,
      "learning_rate": 0.003,
      "loss": 4.464,
      "step": 1055
    },
    {
      "epoch": 0.01056,
      "grad_norm": 0.844999763652046,
      "learning_rate": 0.003,
      "loss": 4.4686,
      "step": 1056
    },
    {
      "epoch": 0.01057,
      "grad_norm": 0.7706046759321704,
      "learning_rate": 0.003,
      "loss": 4.476,
      "step": 1057
    },
    {
      "epoch": 0.01058,
      "grad_norm": 0.6536852085679028,
      "learning_rate": 0.003,
      "loss": 4.4607,
      "step": 1058
    },
    {
      "epoch": 0.01059,
      "grad_norm": 0.7239862733740466,
      "learning_rate": 0.003,
      "loss": 4.4656,
      "step": 1059
    },
    {
      "epoch": 0.0106,
      "grad_norm": 0.8167528488457484,
      "learning_rate": 0.003,
      "loss": 4.4597,
      "step": 1060
    },
    {
      "epoch": 0.01061,
      "grad_norm": 0.6791440431932436,
      "learning_rate": 0.003,
      "loss": 4.4551,
      "step": 1061
    },
    {
      "epoch": 0.01062,
      "grad_norm": 0.7024673414145854,
      "learning_rate": 0.003,
      "loss": 4.4488,
      "step": 1062
    },
    {
      "epoch": 0.01063,
      "grad_norm": 0.5950427080817663,
      "learning_rate": 0.003,
      "loss": 4.4376,
      "step": 1063
    },
    {
      "epoch": 0.01064,
      "grad_norm": 0.4616517916448735,
      "learning_rate": 0.003,
      "loss": 4.4487,
      "step": 1064
    },
    {
      "epoch": 0.01065,
      "grad_norm": 0.43744796242075984,
      "learning_rate": 0.003,
      "loss": 4.4449,
      "step": 1065
    },
    {
      "epoch": 0.01066,
      "grad_norm": 0.3954250450266796,
      "learning_rate": 0.003,
      "loss": 4.4441,
      "step": 1066
    },
    {
      "epoch": 0.01067,
      "grad_norm": 0.44911980407136426,
      "learning_rate": 0.003,
      "loss": 4.4535,
      "step": 1067
    },
    {
      "epoch": 0.01068,
      "grad_norm": 0.6121110106038351,
      "learning_rate": 0.003,
      "loss": 4.4381,
      "step": 1068
    },
    {
      "epoch": 0.01069,
      "grad_norm": 0.8069569048435549,
      "learning_rate": 0.003,
      "loss": 4.4677,
      "step": 1069
    },
    {
      "epoch": 0.0107,
      "grad_norm": 0.9567755776549173,
      "learning_rate": 0.003,
      "loss": 4.5257,
      "step": 1070
    },
    {
      "epoch": 0.01071,
      "grad_norm": 0.7710306613088521,
      "learning_rate": 0.003,
      "loss": 4.4413,
      "step": 1071
    },
    {
      "epoch": 0.01072,
      "grad_norm": 0.6442072811444085,
      "learning_rate": 0.003,
      "loss": 4.4742,
      "step": 1072
    },
    {
      "epoch": 0.01073,
      "grad_norm": 0.6426873609153096,
      "learning_rate": 0.003,
      "loss": 4.464,
      "step": 1073
    },
    {
      "epoch": 0.01074,
      "grad_norm": 0.5296718591337678,
      "learning_rate": 0.003,
      "loss": 4.4244,
      "step": 1074
    },
    {
      "epoch": 0.01075,
      "grad_norm": 0.4598820284284391,
      "learning_rate": 0.003,
      "loss": 4.4589,
      "step": 1075
    },
    {
      "epoch": 0.01076,
      "grad_norm": 0.4671476178414801,
      "learning_rate": 0.003,
      "loss": 4.4514,
      "step": 1076
    },
    {
      "epoch": 0.01077,
      "grad_norm": 0.5378166305898745,
      "learning_rate": 0.003,
      "loss": 4.4483,
      "step": 1077
    },
    {
      "epoch": 0.01078,
      "grad_norm": 0.4897385837475064,
      "learning_rate": 0.003,
      "loss": 4.3993,
      "step": 1078
    },
    {
      "epoch": 0.01079,
      "grad_norm": 0.40845103276318817,
      "learning_rate": 0.003,
      "loss": 4.4297,
      "step": 1079
    },
    {
      "epoch": 0.0108,
      "grad_norm": 0.36771124599267696,
      "learning_rate": 0.003,
      "loss": 4.4028,
      "step": 1080
    },
    {
      "epoch": 0.01081,
      "grad_norm": 0.4088680521129332,
      "learning_rate": 0.003,
      "loss": 4.41,
      "step": 1081
    },
    {
      "epoch": 0.01082,
      "grad_norm": 0.41501548740150246,
      "learning_rate": 0.003,
      "loss": 4.4279,
      "step": 1082
    },
    {
      "epoch": 0.01083,
      "grad_norm": 0.3846754433134812,
      "learning_rate": 0.003,
      "loss": 4.4131,
      "step": 1083
    },
    {
      "epoch": 0.01084,
      "grad_norm": 0.3759434314813712,
      "learning_rate": 0.003,
      "loss": 4.4174,
      "step": 1084
    },
    {
      "epoch": 0.01085,
      "grad_norm": 0.38199974093520966,
      "learning_rate": 0.003,
      "loss": 4.4048,
      "step": 1085
    },
    {
      "epoch": 0.01086,
      "grad_norm": 0.417068769159195,
      "learning_rate": 0.003,
      "loss": 4.3806,
      "step": 1086
    },
    {
      "epoch": 0.01087,
      "grad_norm": 0.4600653555422747,
      "learning_rate": 0.003,
      "loss": 4.3971,
      "step": 1087
    },
    {
      "epoch": 0.01088,
      "grad_norm": 0.46093681942570613,
      "learning_rate": 0.003,
      "loss": 4.4326,
      "step": 1088
    },
    {
      "epoch": 0.01089,
      "grad_norm": 0.3917178547896987,
      "learning_rate": 0.003,
      "loss": 4.3982,
      "step": 1089
    },
    {
      "epoch": 0.0109,
      "grad_norm": 0.40807888361788897,
      "learning_rate": 0.003,
      "loss": 4.4241,
      "step": 1090
    },
    {
      "epoch": 0.01091,
      "grad_norm": 0.48077039467346944,
      "learning_rate": 0.003,
      "loss": 4.3944,
      "step": 1091
    },
    {
      "epoch": 0.01092,
      "grad_norm": 0.5412841894896883,
      "learning_rate": 0.003,
      "loss": 4.3919,
      "step": 1092
    },
    {
      "epoch": 0.01093,
      "grad_norm": 0.6050990959157666,
      "learning_rate": 0.003,
      "loss": 4.4164,
      "step": 1093
    },
    {
      "epoch": 0.01094,
      "grad_norm": 0.7349015923043832,
      "learning_rate": 0.003,
      "loss": 4.4107,
      "step": 1094
    },
    {
      "epoch": 0.01095,
      "grad_norm": 0.8311905656610761,
      "learning_rate": 0.003,
      "loss": 4.4503,
      "step": 1095
    },
    {
      "epoch": 0.01096,
      "grad_norm": 0.7514639212089553,
      "learning_rate": 0.003,
      "loss": 4.4205,
      "step": 1096
    },
    {
      "epoch": 0.01097,
      "grad_norm": 0.7811392154893482,
      "learning_rate": 0.003,
      "loss": 4.4143,
      "step": 1097
    },
    {
      "epoch": 0.01098,
      "grad_norm": 0.784974458960817,
      "learning_rate": 0.003,
      "loss": 4.4358,
      "step": 1098
    },
    {
      "epoch": 0.01099,
      "grad_norm": 0.8366574206737059,
      "learning_rate": 0.003,
      "loss": 4.4105,
      "step": 1099
    },
    {
      "epoch": 0.011,
      "grad_norm": 0.8308538935944765,
      "learning_rate": 0.003,
      "loss": 4.4197,
      "step": 1100
    },
    {
      "epoch": 0.01101,
      "grad_norm": 0.7744395626168694,
      "learning_rate": 0.003,
      "loss": 4.4297,
      "step": 1101
    },
    {
      "epoch": 0.01102,
      "grad_norm": 0.790296389339303,
      "learning_rate": 0.003,
      "loss": 4.4394,
      "step": 1102
    },
    {
      "epoch": 0.01103,
      "grad_norm": 0.888052985760201,
      "learning_rate": 0.003,
      "loss": 4.4459,
      "step": 1103
    },
    {
      "epoch": 0.01104,
      "grad_norm": 0.7149547126246749,
      "learning_rate": 0.003,
      "loss": 4.429,
      "step": 1104
    },
    {
      "epoch": 0.01105,
      "grad_norm": 0.6372550706190871,
      "learning_rate": 0.003,
      "loss": 4.4469,
      "step": 1105
    },
    {
      "epoch": 0.01106,
      "grad_norm": 0.6914402582964255,
      "learning_rate": 0.003,
      "loss": 4.4208,
      "step": 1106
    },
    {
      "epoch": 0.01107,
      "grad_norm": 0.6242116385034774,
      "learning_rate": 0.003,
      "loss": 4.4302,
      "step": 1107
    },
    {
      "epoch": 0.01108,
      "grad_norm": 0.5046709947552265,
      "learning_rate": 0.003,
      "loss": 4.3916,
      "step": 1108
    },
    {
      "epoch": 0.01109,
      "grad_norm": 0.44296219090623196,
      "learning_rate": 0.003,
      "loss": 4.4076,
      "step": 1109
    },
    {
      "epoch": 0.0111,
      "grad_norm": 0.42911746175989035,
      "learning_rate": 0.003,
      "loss": 4.4253,
      "step": 1110
    },
    {
      "epoch": 0.01111,
      "grad_norm": 0.4399471864730289,
      "learning_rate": 0.003,
      "loss": 4.401,
      "step": 1111
    },
    {
      "epoch": 0.01112,
      "grad_norm": 0.4608476260695391,
      "learning_rate": 0.003,
      "loss": 4.4197,
      "step": 1112
    },
    {
      "epoch": 0.01113,
      "grad_norm": 0.48813138086989816,
      "learning_rate": 0.003,
      "loss": 4.388,
      "step": 1113
    },
    {
      "epoch": 0.01114,
      "grad_norm": 0.5565547407761932,
      "learning_rate": 0.003,
      "loss": 4.4024,
      "step": 1114
    },
    {
      "epoch": 0.01115,
      "grad_norm": 0.6052109602996041,
      "learning_rate": 0.003,
      "loss": 4.4297,
      "step": 1115
    },
    {
      "epoch": 0.01116,
      "grad_norm": 0.6641979249916196,
      "learning_rate": 0.003,
      "loss": 4.4207,
      "step": 1116
    },
    {
      "epoch": 0.01117,
      "grad_norm": 0.6083566793907378,
      "learning_rate": 0.003,
      "loss": 4.4056,
      "step": 1117
    },
    {
      "epoch": 0.01118,
      "grad_norm": 0.5238007850801012,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1118
    },
    {
      "epoch": 0.01119,
      "grad_norm": 0.5793839332710249,
      "learning_rate": 0.003,
      "loss": 4.4112,
      "step": 1119
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.5969906941980586,
      "learning_rate": 0.003,
      "loss": 4.3958,
      "step": 1120
    },
    {
      "epoch": 0.01121,
      "grad_norm": 0.6030781006503133,
      "learning_rate": 0.003,
      "loss": 4.4075,
      "step": 1121
    },
    {
      "epoch": 0.01122,
      "grad_norm": 0.5551009392133712,
      "learning_rate": 0.003,
      "loss": 4.382,
      "step": 1122
    },
    {
      "epoch": 0.01123,
      "grad_norm": 0.5109047416253846,
      "learning_rate": 0.003,
      "loss": 4.3689,
      "step": 1123
    },
    {
      "epoch": 0.01124,
      "grad_norm": 0.5407283441136803,
      "learning_rate": 0.003,
      "loss": 4.4164,
      "step": 1124
    },
    {
      "epoch": 0.01125,
      "grad_norm": 0.538491909735447,
      "learning_rate": 0.003,
      "loss": 4.3877,
      "step": 1125
    },
    {
      "epoch": 0.01126,
      "grad_norm": 0.4826610539173529,
      "learning_rate": 0.003,
      "loss": 4.392,
      "step": 1126
    },
    {
      "epoch": 0.01127,
      "grad_norm": 0.41526400192632873,
      "learning_rate": 0.003,
      "loss": 4.421,
      "step": 1127
    },
    {
      "epoch": 0.01128,
      "grad_norm": 0.4898630750070376,
      "learning_rate": 0.003,
      "loss": 4.4001,
      "step": 1128
    },
    {
      "epoch": 0.01129,
      "grad_norm": 0.6900647874251847,
      "learning_rate": 0.003,
      "loss": 4.3988,
      "step": 1129
    },
    {
      "epoch": 0.0113,
      "grad_norm": 0.6715183668037129,
      "learning_rate": 0.003,
      "loss": 4.3803,
      "step": 1130
    },
    {
      "epoch": 0.01131,
      "grad_norm": 0.5369210919725328,
      "learning_rate": 0.003,
      "loss": 4.3895,
      "step": 1131
    },
    {
      "epoch": 0.01132,
      "grad_norm": 0.7090208352393726,
      "learning_rate": 0.003,
      "loss": 4.4216,
      "step": 1132
    },
    {
      "epoch": 0.01133,
      "grad_norm": 0.8026720884346313,
      "learning_rate": 0.003,
      "loss": 4.3951,
      "step": 1133
    },
    {
      "epoch": 0.01134,
      "grad_norm": 0.7235729334025307,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1134
    },
    {
      "epoch": 0.01135,
      "grad_norm": 0.5789287370870784,
      "learning_rate": 0.003,
      "loss": 4.4255,
      "step": 1135
    },
    {
      "epoch": 0.01136,
      "grad_norm": 0.5872111107371502,
      "learning_rate": 0.003,
      "loss": 4.4037,
      "step": 1136
    },
    {
      "epoch": 0.01137,
      "grad_norm": 0.5973670624507018,
      "learning_rate": 0.003,
      "loss": 4.4236,
      "step": 1137
    },
    {
      "epoch": 0.01138,
      "grad_norm": 0.5814805523631724,
      "learning_rate": 0.003,
      "loss": 4.3808,
      "step": 1138
    },
    {
      "epoch": 0.01139,
      "grad_norm": 0.5749927487336652,
      "learning_rate": 0.003,
      "loss": 4.389,
      "step": 1139
    },
    {
      "epoch": 0.0114,
      "grad_norm": 0.5177155580469588,
      "learning_rate": 0.003,
      "loss": 4.3725,
      "step": 1140
    },
    {
      "epoch": 0.01141,
      "grad_norm": 0.5098128826528755,
      "learning_rate": 0.003,
      "loss": 4.3956,
      "step": 1141
    },
    {
      "epoch": 0.01142,
      "grad_norm": 0.5805509550413942,
      "learning_rate": 0.003,
      "loss": 4.3764,
      "step": 1142
    },
    {
      "epoch": 0.01143,
      "grad_norm": 0.5868263165469564,
      "learning_rate": 0.003,
      "loss": 4.3885,
      "step": 1143
    },
    {
      "epoch": 0.01144,
      "grad_norm": 0.6024984248815738,
      "learning_rate": 0.003,
      "loss": 4.3595,
      "step": 1144
    },
    {
      "epoch": 0.01145,
      "grad_norm": 0.6011422862217553,
      "learning_rate": 0.003,
      "loss": 4.396,
      "step": 1145
    },
    {
      "epoch": 0.01146,
      "grad_norm": 0.6291389752386858,
      "learning_rate": 0.003,
      "loss": 4.3845,
      "step": 1146
    },
    {
      "epoch": 0.01147,
      "grad_norm": 0.5858383738783732,
      "learning_rate": 0.003,
      "loss": 4.3734,
      "step": 1147
    },
    {
      "epoch": 0.01148,
      "grad_norm": 0.5690631266524724,
      "learning_rate": 0.003,
      "loss": 4.3732,
      "step": 1148
    },
    {
      "epoch": 0.01149,
      "grad_norm": 0.47603449917616164,
      "learning_rate": 0.003,
      "loss": 4.3879,
      "step": 1149
    },
    {
      "epoch": 0.0115,
      "grad_norm": 0.5321591521167512,
      "learning_rate": 0.003,
      "loss": 4.3815,
      "step": 1150
    },
    {
      "epoch": 0.01151,
      "grad_norm": 0.5902114027403508,
      "learning_rate": 0.003,
      "loss": 4.3804,
      "step": 1151
    },
    {
      "epoch": 0.01152,
      "grad_norm": 0.720836087411827,
      "learning_rate": 0.003,
      "loss": 4.3994,
      "step": 1152
    },
    {
      "epoch": 0.01153,
      "grad_norm": 0.858014811127527,
      "learning_rate": 0.003,
      "loss": 4.4233,
      "step": 1153
    },
    {
      "epoch": 0.01154,
      "grad_norm": 0.8276250061884273,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1154
    },
    {
      "epoch": 0.01155,
      "grad_norm": 0.9316719148571102,
      "learning_rate": 0.003,
      "loss": 4.4003,
      "step": 1155
    },
    {
      "epoch": 0.01156,
      "grad_norm": 0.9492720138035956,
      "learning_rate": 0.003,
      "loss": 4.4137,
      "step": 1156
    },
    {
      "epoch": 0.01157,
      "grad_norm": 1.1562300430530366,
      "learning_rate": 0.003,
      "loss": 4.471,
      "step": 1157
    },
    {
      "epoch": 0.01158,
      "grad_norm": 0.8082778562361257,
      "learning_rate": 0.003,
      "loss": 4.4212,
      "step": 1158
    },
    {
      "epoch": 0.01159,
      "grad_norm": 0.8296228637182822,
      "learning_rate": 0.003,
      "loss": 4.4254,
      "step": 1159
    },
    {
      "epoch": 0.0116,
      "grad_norm": 0.9729894065836546,
      "learning_rate": 0.003,
      "loss": 4.4435,
      "step": 1160
    },
    {
      "epoch": 0.01161,
      "grad_norm": 0.986208959900233,
      "learning_rate": 0.003,
      "loss": 4.4396,
      "step": 1161
    },
    {
      "epoch": 0.01162,
      "grad_norm": 0.9917874948238292,
      "learning_rate": 0.003,
      "loss": 4.4412,
      "step": 1162
    },
    {
      "epoch": 0.01163,
      "grad_norm": 1.022093528985829,
      "learning_rate": 0.003,
      "loss": 4.4772,
      "step": 1163
    },
    {
      "epoch": 0.01164,
      "grad_norm": 0.8274710934079806,
      "learning_rate": 0.003,
      "loss": 4.4588,
      "step": 1164
    },
    {
      "epoch": 0.01165,
      "grad_norm": 0.8544376883862705,
      "learning_rate": 0.003,
      "loss": 4.4518,
      "step": 1165
    },
    {
      "epoch": 0.01166,
      "grad_norm": 0.9826443233479919,
      "learning_rate": 0.003,
      "loss": 4.4596,
      "step": 1166
    },
    {
      "epoch": 0.01167,
      "grad_norm": 0.8638656323292462,
      "learning_rate": 0.003,
      "loss": 4.427,
      "step": 1167
    },
    {
      "epoch": 0.01168,
      "grad_norm": 0.7965515530800152,
      "learning_rate": 0.003,
      "loss": 4.4285,
      "step": 1168
    },
    {
      "epoch": 0.01169,
      "grad_norm": 0.621289788302785,
      "learning_rate": 0.003,
      "loss": 4.4631,
      "step": 1169
    },
    {
      "epoch": 0.0117,
      "grad_norm": 0.5251543844281131,
      "learning_rate": 0.003,
      "loss": 4.441,
      "step": 1170
    },
    {
      "epoch": 0.01171,
      "grad_norm": 0.6074391771886244,
      "learning_rate": 0.003,
      "loss": 4.423,
      "step": 1171
    },
    {
      "epoch": 0.01172,
      "grad_norm": 0.5839489329030457,
      "learning_rate": 0.003,
      "loss": 4.4084,
      "step": 1172
    },
    {
      "epoch": 0.01173,
      "grad_norm": 0.561004783856832,
      "learning_rate": 0.003,
      "loss": 4.4404,
      "step": 1173
    },
    {
      "epoch": 0.01174,
      "grad_norm": 0.5821159441803843,
      "learning_rate": 0.003,
      "loss": 4.4125,
      "step": 1174
    },
    {
      "epoch": 0.01175,
      "grad_norm": 0.5941266163389948,
      "learning_rate": 0.003,
      "loss": 4.3949,
      "step": 1175
    },
    {
      "epoch": 0.01176,
      "grad_norm": 0.6502157975630524,
      "learning_rate": 0.003,
      "loss": 4.432,
      "step": 1176
    },
    {
      "epoch": 0.01177,
      "grad_norm": 0.8335241720178251,
      "learning_rate": 0.003,
      "loss": 4.4041,
      "step": 1177
    },
    {
      "epoch": 0.01178,
      "grad_norm": 1.0000155108324966,
      "learning_rate": 0.003,
      "loss": 4.4462,
      "step": 1178
    },
    {
      "epoch": 0.01179,
      "grad_norm": 0.9223184266441969,
      "learning_rate": 0.003,
      "loss": 4.4309,
      "step": 1179
    },
    {
      "epoch": 0.0118,
      "grad_norm": 0.6248971968290589,
      "learning_rate": 0.003,
      "loss": 4.4722,
      "step": 1180
    },
    {
      "epoch": 0.01181,
      "grad_norm": 0.6327502868796897,
      "learning_rate": 0.003,
      "loss": 4.4606,
      "step": 1181
    },
    {
      "epoch": 0.01182,
      "grad_norm": 0.6032057741631047,
      "learning_rate": 0.003,
      "loss": 4.4179,
      "step": 1182
    },
    {
      "epoch": 0.01183,
      "grad_norm": 0.6177419756766389,
      "learning_rate": 0.003,
      "loss": 4.4068,
      "step": 1183
    },
    {
      "epoch": 0.01184,
      "grad_norm": 0.4603977151305031,
      "learning_rate": 0.003,
      "loss": 4.4388,
      "step": 1184
    },
    {
      "epoch": 0.01185,
      "grad_norm": 0.39775931274104387,
      "learning_rate": 0.003,
      "loss": 4.4039,
      "step": 1185
    },
    {
      "epoch": 0.01186,
      "grad_norm": 0.43457323577233337,
      "learning_rate": 0.003,
      "loss": 4.4263,
      "step": 1186
    },
    {
      "epoch": 0.01187,
      "grad_norm": 0.3445355147627072,
      "learning_rate": 0.003,
      "loss": 4.3841,
      "step": 1187
    },
    {
      "epoch": 0.01188,
      "grad_norm": 0.31570813535234143,
      "learning_rate": 0.003,
      "loss": 4.3952,
      "step": 1188
    },
    {
      "epoch": 0.01189,
      "grad_norm": 0.3234807855930626,
      "learning_rate": 0.003,
      "loss": 4.3874,
      "step": 1189
    },
    {
      "epoch": 0.0119,
      "grad_norm": 0.3007248052516205,
      "learning_rate": 0.003,
      "loss": 4.3694,
      "step": 1190
    },
    {
      "epoch": 0.01191,
      "grad_norm": 0.289692411083719,
      "learning_rate": 0.003,
      "loss": 4.3476,
      "step": 1191
    },
    {
      "epoch": 0.01192,
      "grad_norm": 0.2772752758842908,
      "learning_rate": 0.003,
      "loss": 4.3779,
      "step": 1192
    },
    {
      "epoch": 0.01193,
      "grad_norm": 0.2637988935747097,
      "learning_rate": 0.003,
      "loss": 4.3787,
      "step": 1193
    },
    {
      "epoch": 0.01194,
      "grad_norm": 0.2480894935347364,
      "learning_rate": 0.003,
      "loss": 4.3791,
      "step": 1194
    },
    {
      "epoch": 0.01195,
      "grad_norm": 0.2744828686479978,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1195
    },
    {
      "epoch": 0.01196,
      "grad_norm": 0.2694264980010376,
      "learning_rate": 0.003,
      "loss": 4.3608,
      "step": 1196
    },
    {
      "epoch": 0.01197,
      "grad_norm": 0.2911995649935099,
      "learning_rate": 0.003,
      "loss": 4.3243,
      "step": 1197
    },
    {
      "epoch": 0.01198,
      "grad_norm": 0.3049686361769555,
      "learning_rate": 0.003,
      "loss": 4.3978,
      "step": 1198
    },
    {
      "epoch": 0.01199,
      "grad_norm": 0.4091366608598886,
      "learning_rate": 0.003,
      "loss": 4.3744,
      "step": 1199
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.5260409537460498,
      "learning_rate": 0.003,
      "loss": 4.3673,
      "step": 1200
    },
    {
      "epoch": 0.01201,
      "grad_norm": 0.6358041499118771,
      "learning_rate": 0.003,
      "loss": 4.4018,
      "step": 1201
    },
    {
      "epoch": 0.01202,
      "grad_norm": 0.5923034972204151,
      "learning_rate": 0.003,
      "loss": 4.3715,
      "step": 1202
    },
    {
      "epoch": 0.01203,
      "grad_norm": 0.49351792103508907,
      "learning_rate": 0.003,
      "loss": 4.3737,
      "step": 1203
    },
    {
      "epoch": 0.01204,
      "grad_norm": 0.6405227463764683,
      "learning_rate": 0.003,
      "loss": 4.3772,
      "step": 1204
    },
    {
      "epoch": 0.01205,
      "grad_norm": 0.7035300072883678,
      "learning_rate": 0.003,
      "loss": 4.3759,
      "step": 1205
    },
    {
      "epoch": 0.01206,
      "grad_norm": 0.6623021094722211,
      "learning_rate": 0.003,
      "loss": 4.4176,
      "step": 1206
    },
    {
      "epoch": 0.01207,
      "grad_norm": 0.6906429556810797,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1207
    },
    {
      "epoch": 0.01208,
      "grad_norm": 0.7416830674542866,
      "learning_rate": 0.003,
      "loss": 4.4101,
      "step": 1208
    },
    {
      "epoch": 0.01209,
      "grad_norm": 0.7431220077552587,
      "learning_rate": 0.003,
      "loss": 4.3922,
      "step": 1209
    },
    {
      "epoch": 0.0121,
      "grad_norm": 0.7626206113161277,
      "learning_rate": 0.003,
      "loss": 4.397,
      "step": 1210
    },
    {
      "epoch": 0.01211,
      "grad_norm": 0.7273957054715338,
      "learning_rate": 0.003,
      "loss": 4.381,
      "step": 1211
    },
    {
      "epoch": 0.01212,
      "grad_norm": 0.5840029568104971,
      "learning_rate": 0.003,
      "loss": 4.4002,
      "step": 1212
    },
    {
      "epoch": 0.01213,
      "grad_norm": 0.5534819591366041,
      "learning_rate": 0.003,
      "loss": 4.3747,
      "step": 1213
    },
    {
      "epoch": 0.01214,
      "grad_norm": 0.5479838122682379,
      "learning_rate": 0.003,
      "loss": 4.3596,
      "step": 1214
    },
    {
      "epoch": 0.01215,
      "grad_norm": 0.6541197102272803,
      "learning_rate": 0.003,
      "loss": 4.3736,
      "step": 1215
    },
    {
      "epoch": 0.01216,
      "grad_norm": 0.6998905075239406,
      "learning_rate": 0.003,
      "loss": 4.369,
      "step": 1216
    },
    {
      "epoch": 0.01217,
      "grad_norm": 0.6272982074936522,
      "learning_rate": 0.003,
      "loss": 4.3725,
      "step": 1217
    },
    {
      "epoch": 0.01218,
      "grad_norm": 0.6037550089398132,
      "learning_rate": 0.003,
      "loss": 4.3687,
      "step": 1218
    },
    {
      "epoch": 0.01219,
      "grad_norm": 0.5872809556780119,
      "learning_rate": 0.003,
      "loss": 4.3682,
      "step": 1219
    },
    {
      "epoch": 0.0122,
      "grad_norm": 0.45035452365127926,
      "learning_rate": 0.003,
      "loss": 4.3537,
      "step": 1220
    },
    {
      "epoch": 0.01221,
      "grad_norm": 0.41160354363285073,
      "learning_rate": 0.003,
      "loss": 4.3666,
      "step": 1221
    },
    {
      "epoch": 0.01222,
      "grad_norm": 0.5075830142490442,
      "learning_rate": 0.003,
      "loss": 4.3511,
      "step": 1222
    },
    {
      "epoch": 0.01223,
      "grad_norm": 0.5662251248767171,
      "learning_rate": 0.003,
      "loss": 4.3541,
      "step": 1223
    },
    {
      "epoch": 0.01224,
      "grad_norm": 0.690794501870828,
      "learning_rate": 0.003,
      "loss": 4.3444,
      "step": 1224
    },
    {
      "epoch": 0.01225,
      "grad_norm": 0.675909010637242,
      "learning_rate": 0.003,
      "loss": 4.3698,
      "step": 1225
    },
    {
      "epoch": 0.01226,
      "grad_norm": 0.6653173755559965,
      "learning_rate": 0.003,
      "loss": 4.36,
      "step": 1226
    },
    {
      "epoch": 0.01227,
      "grad_norm": 0.8124014201072856,
      "learning_rate": 0.003,
      "loss": 4.4022,
      "step": 1227
    },
    {
      "epoch": 0.01228,
      "grad_norm": 0.8601590167217709,
      "learning_rate": 0.003,
      "loss": 4.3869,
      "step": 1228
    },
    {
      "epoch": 0.01229,
      "grad_norm": 0.7816229118417856,
      "learning_rate": 0.003,
      "loss": 4.3791,
      "step": 1229
    },
    {
      "epoch": 0.0123,
      "grad_norm": 0.7167415795127342,
      "learning_rate": 0.003,
      "loss": 4.4104,
      "step": 1230
    },
    {
      "epoch": 0.01231,
      "grad_norm": 0.7227237760755019,
      "learning_rate": 0.003,
      "loss": 4.3789,
      "step": 1231
    },
    {
      "epoch": 0.01232,
      "grad_norm": 0.6338303976227229,
      "learning_rate": 0.003,
      "loss": 4.3516,
      "step": 1232
    },
    {
      "epoch": 0.01233,
      "grad_norm": 0.5121760575897172,
      "learning_rate": 0.003,
      "loss": 4.3548,
      "step": 1233
    },
    {
      "epoch": 0.01234,
      "grad_norm": 0.562103643632394,
      "learning_rate": 0.003,
      "loss": 4.3735,
      "step": 1234
    },
    {
      "epoch": 0.01235,
      "grad_norm": 0.5094370853466236,
      "learning_rate": 0.003,
      "loss": 4.3603,
      "step": 1235
    },
    {
      "epoch": 0.01236,
      "grad_norm": 0.42756368516441545,
      "learning_rate": 0.003,
      "loss": 4.3622,
      "step": 1236
    },
    {
      "epoch": 0.01237,
      "grad_norm": 0.41715621646232526,
      "learning_rate": 0.003,
      "loss": 4.3707,
      "step": 1237
    },
    {
      "epoch": 0.01238,
      "grad_norm": 0.41900290536826335,
      "learning_rate": 0.003,
      "loss": 4.3664,
      "step": 1238
    },
    {
      "epoch": 0.01239,
      "grad_norm": 0.4057196989115332,
      "learning_rate": 0.003,
      "loss": 4.3485,
      "step": 1239
    },
    {
      "epoch": 0.0124,
      "grad_norm": 0.41828480066605506,
      "learning_rate": 0.003,
      "loss": 4.331,
      "step": 1240
    },
    {
      "epoch": 0.01241,
      "grad_norm": 0.4114484855240935,
      "learning_rate": 0.003,
      "loss": 4.3417,
      "step": 1241
    },
    {
      "epoch": 0.01242,
      "grad_norm": 0.4335820493661237,
      "learning_rate": 0.003,
      "loss": 4.348,
      "step": 1242
    },
    {
      "epoch": 0.01243,
      "grad_norm": 0.5022671727550078,
      "learning_rate": 0.003,
      "loss": 4.3316,
      "step": 1243
    },
    {
      "epoch": 0.01244,
      "grad_norm": 0.5275944783156026,
      "learning_rate": 0.003,
      "loss": 4.3445,
      "step": 1244
    },
    {
      "epoch": 0.01245,
      "grad_norm": 0.5485058077885037,
      "learning_rate": 0.003,
      "loss": 4.3378,
      "step": 1245
    },
    {
      "epoch": 0.01246,
      "grad_norm": 0.5308550603908434,
      "learning_rate": 0.003,
      "loss": 4.312,
      "step": 1246
    },
    {
      "epoch": 0.01247,
      "grad_norm": 0.5603455205230277,
      "learning_rate": 0.003,
      "loss": 4.373,
      "step": 1247
    },
    {
      "epoch": 0.01248,
      "grad_norm": 0.6616614283266387,
      "learning_rate": 0.003,
      "loss": 4.3622,
      "step": 1248
    },
    {
      "epoch": 0.01249,
      "grad_norm": 0.6950989703790047,
      "learning_rate": 0.003,
      "loss": 4.3579,
      "step": 1249
    },
    {
      "epoch": 0.0125,
      "grad_norm": 0.6807657788060711,
      "learning_rate": 0.003,
      "loss": 4.3266,
      "step": 1250
    },
    {
      "epoch": 0.01251,
      "grad_norm": 0.7044512298181651,
      "learning_rate": 0.003,
      "loss": 4.3715,
      "step": 1251
    },
    {
      "epoch": 0.01252,
      "grad_norm": 0.7187283948109732,
      "learning_rate": 0.003,
      "loss": 4.3491,
      "step": 1252
    },
    {
      "epoch": 0.01253,
      "grad_norm": 0.6168851566630172,
      "learning_rate": 0.003,
      "loss": 4.3238,
      "step": 1253
    },
    {
      "epoch": 0.01254,
      "grad_norm": 0.5482483685536774,
      "learning_rate": 0.003,
      "loss": 4.3671,
      "step": 1254
    },
    {
      "epoch": 0.01255,
      "grad_norm": 0.5158716427473324,
      "learning_rate": 0.003,
      "loss": 4.3154,
      "step": 1255
    },
    {
      "epoch": 0.01256,
      "grad_norm": 0.46371154610018284,
      "learning_rate": 0.003,
      "loss": 4.3421,
      "step": 1256
    },
    {
      "epoch": 0.01257,
      "grad_norm": 0.4261160552146568,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1257
    },
    {
      "epoch": 0.01258,
      "grad_norm": 0.507565461332445,
      "learning_rate": 0.003,
      "loss": 4.3375,
      "step": 1258
    },
    {
      "epoch": 0.01259,
      "grad_norm": 0.7023530299208391,
      "learning_rate": 0.003,
      "loss": 4.3461,
      "step": 1259
    },
    {
      "epoch": 0.0126,
      "grad_norm": 0.8862638702569765,
      "learning_rate": 0.003,
      "loss": 4.3945,
      "step": 1260
    },
    {
      "epoch": 0.01261,
      "grad_norm": 0.92528404873397,
      "learning_rate": 0.003,
      "loss": 4.3876,
      "step": 1261
    },
    {
      "epoch": 0.01262,
      "grad_norm": 0.9810348621626278,
      "learning_rate": 0.003,
      "loss": 4.3924,
      "step": 1262
    },
    {
      "epoch": 0.01263,
      "grad_norm": 0.7811680454044853,
      "learning_rate": 0.003,
      "loss": 4.3835,
      "step": 1263
    },
    {
      "epoch": 0.01264,
      "grad_norm": 0.6458262464633842,
      "learning_rate": 0.003,
      "loss": 4.365,
      "step": 1264
    },
    {
      "epoch": 0.01265,
      "grad_norm": 0.6726321620447105,
      "learning_rate": 0.003,
      "loss": 4.3777,
      "step": 1265
    },
    {
      "epoch": 0.01266,
      "grad_norm": 0.7462208985516781,
      "learning_rate": 0.003,
      "loss": 4.3808,
      "step": 1266
    },
    {
      "epoch": 0.01267,
      "grad_norm": 0.8334608532491081,
      "learning_rate": 0.003,
      "loss": 4.4053,
      "step": 1267
    },
    {
      "epoch": 0.01268,
      "grad_norm": 0.722253192550455,
      "learning_rate": 0.003,
      "loss": 4.3914,
      "step": 1268
    },
    {
      "epoch": 0.01269,
      "grad_norm": 0.5434607483157148,
      "learning_rate": 0.003,
      "loss": 4.3468,
      "step": 1269
    },
    {
      "epoch": 0.0127,
      "grad_norm": 0.4389640950286333,
      "learning_rate": 0.003,
      "loss": 4.3757,
      "step": 1270
    },
    {
      "epoch": 0.01271,
      "grad_norm": 0.4493097096002282,
      "learning_rate": 0.003,
      "loss": 4.357,
      "step": 1271
    },
    {
      "epoch": 0.01272,
      "grad_norm": 0.4602878891635523,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1272
    },
    {
      "epoch": 0.01273,
      "grad_norm": 0.4816441128313513,
      "learning_rate": 0.003,
      "loss": 4.3755,
      "step": 1273
    },
    {
      "epoch": 0.01274,
      "grad_norm": 0.5275730605295648,
      "learning_rate": 0.003,
      "loss": 4.3705,
      "step": 1274
    },
    {
      "epoch": 0.01275,
      "grad_norm": 0.5285253035410874,
      "learning_rate": 0.003,
      "loss": 4.3384,
      "step": 1275
    },
    {
      "epoch": 0.01276,
      "grad_norm": 0.42608260843601453,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1276
    },
    {
      "epoch": 0.01277,
      "grad_norm": 0.394113210363875,
      "learning_rate": 0.003,
      "loss": 4.3444,
      "step": 1277
    },
    {
      "epoch": 0.01278,
      "grad_norm": 0.485427057016395,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1278
    },
    {
      "epoch": 0.01279,
      "grad_norm": 0.4912898619415999,
      "learning_rate": 0.003,
      "loss": 4.3362,
      "step": 1279
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.4970947767829747,
      "learning_rate": 0.003,
      "loss": 4.3331,
      "step": 1280
    },
    {
      "epoch": 0.01281,
      "grad_norm": 0.45661885236926897,
      "learning_rate": 0.003,
      "loss": 4.3167,
      "step": 1281
    },
    {
      "epoch": 0.01282,
      "grad_norm": 0.3777414491844073,
      "learning_rate": 0.003,
      "loss": 4.3562,
      "step": 1282
    },
    {
      "epoch": 0.01283,
      "grad_norm": 0.4194385613191725,
      "learning_rate": 0.003,
      "loss": 4.3264,
      "step": 1283
    },
    {
      "epoch": 0.01284,
      "grad_norm": 0.455784660804564,
      "learning_rate": 0.003,
      "loss": 4.328,
      "step": 1284
    },
    {
      "epoch": 0.01285,
      "grad_norm": 0.5164075234519105,
      "learning_rate": 0.003,
      "loss": 4.3319,
      "step": 1285
    },
    {
      "epoch": 0.01286,
      "grad_norm": 0.6148100521816782,
      "learning_rate": 0.003,
      "loss": 4.3568,
      "step": 1286
    },
    {
      "epoch": 0.01287,
      "grad_norm": 0.8435250802638579,
      "learning_rate": 0.003,
      "loss": 4.3378,
      "step": 1287
    },
    {
      "epoch": 0.01288,
      "grad_norm": 0.8948772567277533,
      "learning_rate": 0.003,
      "loss": 4.3616,
      "step": 1288
    },
    {
      "epoch": 0.01289,
      "grad_norm": 0.5768385697936498,
      "learning_rate": 0.003,
      "loss": 4.3412,
      "step": 1289
    },
    {
      "epoch": 0.0129,
      "grad_norm": 0.48744560346189664,
      "learning_rate": 0.003,
      "loss": 4.3621,
      "step": 1290
    },
    {
      "epoch": 0.01291,
      "grad_norm": 0.5888239794628021,
      "learning_rate": 0.003,
      "loss": 4.3458,
      "step": 1291
    },
    {
      "epoch": 0.01292,
      "grad_norm": 0.5667412086864034,
      "learning_rate": 0.003,
      "loss": 4.3773,
      "step": 1292
    },
    {
      "epoch": 0.01293,
      "grad_norm": 0.6066487822204776,
      "learning_rate": 0.003,
      "loss": 4.3648,
      "step": 1293
    },
    {
      "epoch": 0.01294,
      "grad_norm": 0.7065627866796411,
      "learning_rate": 0.003,
      "loss": 4.3371,
      "step": 1294
    },
    {
      "epoch": 0.01295,
      "grad_norm": 0.7389325253179186,
      "learning_rate": 0.003,
      "loss": 4.3679,
      "step": 1295
    },
    {
      "epoch": 0.01296,
      "grad_norm": 0.7660662423328507,
      "learning_rate": 0.003,
      "loss": 4.3399,
      "step": 1296
    },
    {
      "epoch": 0.01297,
      "grad_norm": 0.7847737463247922,
      "learning_rate": 0.003,
      "loss": 4.3573,
      "step": 1297
    },
    {
      "epoch": 0.01298,
      "grad_norm": 0.8150967050528192,
      "learning_rate": 0.003,
      "loss": 4.3384,
      "step": 1298
    },
    {
      "epoch": 0.01299,
      "grad_norm": 0.7237792681987708,
      "learning_rate": 0.003,
      "loss": 4.3533,
      "step": 1299
    },
    {
      "epoch": 0.013,
      "grad_norm": 0.6452098996944176,
      "learning_rate": 0.003,
      "loss": 4.3355,
      "step": 1300
    },
    {
      "epoch": 0.01301,
      "grad_norm": 0.6428524108659182,
      "learning_rate": 0.003,
      "loss": 4.3495,
      "step": 1301
    },
    {
      "epoch": 0.01302,
      "grad_norm": 0.7475054826389519,
      "learning_rate": 0.003,
      "loss": 4.4013,
      "step": 1302
    },
    {
      "epoch": 0.01303,
      "grad_norm": 0.9085755519721904,
      "learning_rate": 0.003,
      "loss": 4.3576,
      "step": 1303
    },
    {
      "epoch": 0.01304,
      "grad_norm": 0.8565196075884696,
      "learning_rate": 0.003,
      "loss": 4.3934,
      "step": 1304
    },
    {
      "epoch": 0.01305,
      "grad_norm": 0.8916073679840805,
      "learning_rate": 0.003,
      "loss": 4.3789,
      "step": 1305
    },
    {
      "epoch": 0.01306,
      "grad_norm": 0.7910970693035262,
      "learning_rate": 0.003,
      "loss": 4.3681,
      "step": 1306
    },
    {
      "epoch": 0.01307,
      "grad_norm": 0.7398667792205602,
      "learning_rate": 0.003,
      "loss": 4.3875,
      "step": 1307
    },
    {
      "epoch": 0.01308,
      "grad_norm": 0.6800690310649735,
      "learning_rate": 0.003,
      "loss": 4.3597,
      "step": 1308
    },
    {
      "epoch": 0.01309,
      "grad_norm": 0.610129543183007,
      "learning_rate": 0.003,
      "loss": 4.3978,
      "step": 1309
    },
    {
      "epoch": 0.0131,
      "grad_norm": 0.6342445693889617,
      "learning_rate": 0.003,
      "loss": 4.3435,
      "step": 1310
    },
    {
      "epoch": 0.01311,
      "grad_norm": 0.5612187898652411,
      "learning_rate": 0.003,
      "loss": 4.3787,
      "step": 1311
    },
    {
      "epoch": 0.01312,
      "grad_norm": 0.6173335576080043,
      "learning_rate": 0.003,
      "loss": 4.3472,
      "step": 1312
    },
    {
      "epoch": 0.01313,
      "grad_norm": 0.5958431337675327,
      "learning_rate": 0.003,
      "loss": 4.3387,
      "step": 1313
    },
    {
      "epoch": 0.01314,
      "grad_norm": 0.6624135563680444,
      "learning_rate": 0.003,
      "loss": 4.3367,
      "step": 1314
    },
    {
      "epoch": 0.01315,
      "grad_norm": 0.6490907209314275,
      "learning_rate": 0.003,
      "loss": 4.368,
      "step": 1315
    },
    {
      "epoch": 0.01316,
      "grad_norm": 0.5500794253188088,
      "learning_rate": 0.003,
      "loss": 4.3457,
      "step": 1316
    },
    {
      "epoch": 0.01317,
      "grad_norm": 0.6275958010748339,
      "learning_rate": 0.003,
      "loss": 4.3746,
      "step": 1317
    },
    {
      "epoch": 0.01318,
      "grad_norm": 0.5648473326010665,
      "learning_rate": 0.003,
      "loss": 4.3513,
      "step": 1318
    },
    {
      "epoch": 0.01319,
      "grad_norm": 0.45510587840319355,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1319
    },
    {
      "epoch": 0.0132,
      "grad_norm": 0.44264387716031117,
      "learning_rate": 0.003,
      "loss": 4.3472,
      "step": 1320
    },
    {
      "epoch": 0.01321,
      "grad_norm": 0.40376640077570203,
      "learning_rate": 0.003,
      "loss": 4.322,
      "step": 1321
    },
    {
      "epoch": 0.01322,
      "grad_norm": 0.3877836192326668,
      "learning_rate": 0.003,
      "loss": 4.3502,
      "step": 1322
    },
    {
      "epoch": 0.01323,
      "grad_norm": 0.3937183906523392,
      "learning_rate": 0.003,
      "loss": 4.341,
      "step": 1323
    },
    {
      "epoch": 0.01324,
      "grad_norm": 0.4100022537274817,
      "learning_rate": 0.003,
      "loss": 4.3255,
      "step": 1324
    },
    {
      "epoch": 0.01325,
      "grad_norm": 0.46279809510243225,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1325
    },
    {
      "epoch": 0.01326,
      "grad_norm": 0.4767280607611473,
      "learning_rate": 0.003,
      "loss": 4.3173,
      "step": 1326
    },
    {
      "epoch": 0.01327,
      "grad_norm": 0.5123257836661763,
      "learning_rate": 0.003,
      "loss": 4.3366,
      "step": 1327
    },
    {
      "epoch": 0.01328,
      "grad_norm": 0.5434476080816805,
      "learning_rate": 0.003,
      "loss": 4.3286,
      "step": 1328
    },
    {
      "epoch": 0.01329,
      "grad_norm": 0.48979784037018337,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1329
    },
    {
      "epoch": 0.0133,
      "grad_norm": 0.42591921766495494,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1330
    },
    {
      "epoch": 0.01331,
      "grad_norm": 0.48229010910172454,
      "learning_rate": 0.003,
      "loss": 4.318,
      "step": 1331
    },
    {
      "epoch": 0.01332,
      "grad_norm": 0.514388044498456,
      "learning_rate": 0.003,
      "loss": 4.3386,
      "step": 1332
    },
    {
      "epoch": 0.01333,
      "grad_norm": 0.5443479600217087,
      "learning_rate": 0.003,
      "loss": 4.3145,
      "step": 1333
    },
    {
      "epoch": 0.01334,
      "grad_norm": 0.5537269490906184,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1334
    },
    {
      "epoch": 0.01335,
      "grad_norm": 0.5158317522709531,
      "learning_rate": 0.003,
      "loss": 4.3253,
      "step": 1335
    },
    {
      "epoch": 0.01336,
      "grad_norm": 0.5108167456714484,
      "learning_rate": 0.003,
      "loss": 4.3223,
      "step": 1336
    },
    {
      "epoch": 0.01337,
      "grad_norm": 0.6228714832065689,
      "learning_rate": 0.003,
      "loss": 4.3418,
      "step": 1337
    },
    {
      "epoch": 0.01338,
      "grad_norm": 0.7769527340926119,
      "learning_rate": 0.003,
      "loss": 4.3169,
      "step": 1338
    },
    {
      "epoch": 0.01339,
      "grad_norm": 1.0163715747866278,
      "learning_rate": 0.003,
      "loss": 4.3527,
      "step": 1339
    },
    {
      "epoch": 0.0134,
      "grad_norm": 0.8875484689453775,
      "learning_rate": 0.003,
      "loss": 4.3448,
      "step": 1340
    },
    {
      "epoch": 0.01341,
      "grad_norm": 0.7593448848803411,
      "learning_rate": 0.003,
      "loss": 4.3557,
      "step": 1341
    },
    {
      "epoch": 0.01342,
      "grad_norm": 0.7038826124517171,
      "learning_rate": 0.003,
      "loss": 4.3272,
      "step": 1342
    },
    {
      "epoch": 0.01343,
      "grad_norm": 0.6570836329978591,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1343
    },
    {
      "epoch": 0.01344,
      "grad_norm": 0.7840667137315036,
      "learning_rate": 0.003,
      "loss": 4.3338,
      "step": 1344
    },
    {
      "epoch": 0.01345,
      "grad_norm": 0.8057220270427362,
      "learning_rate": 0.003,
      "loss": 4.3808,
      "step": 1345
    },
    {
      "epoch": 0.01346,
      "grad_norm": 0.7768442613998496,
      "learning_rate": 0.003,
      "loss": 4.3555,
      "step": 1346
    },
    {
      "epoch": 0.01347,
      "grad_norm": 0.7491448291061306,
      "learning_rate": 0.003,
      "loss": 4.3463,
      "step": 1347
    },
    {
      "epoch": 0.01348,
      "grad_norm": 0.7233252261262297,
      "learning_rate": 0.003,
      "loss": 4.3476,
      "step": 1348
    },
    {
      "epoch": 0.01349,
      "grad_norm": 0.7147253457500066,
      "learning_rate": 0.003,
      "loss": 4.3662,
      "step": 1349
    },
    {
      "epoch": 0.0135,
      "grad_norm": 0.8413435804466313,
      "learning_rate": 0.003,
      "loss": 4.3673,
      "step": 1350
    },
    {
      "epoch": 0.01351,
      "grad_norm": 0.9169233695908002,
      "learning_rate": 0.003,
      "loss": 4.406,
      "step": 1351
    },
    {
      "epoch": 0.01352,
      "grad_norm": 0.7136998484165762,
      "learning_rate": 0.003,
      "loss": 4.371,
      "step": 1352
    },
    {
      "epoch": 0.01353,
      "grad_norm": 0.7015996798304995,
      "learning_rate": 0.003,
      "loss": 4.3575,
      "step": 1353
    },
    {
      "epoch": 0.01354,
      "grad_norm": 0.7349461226651943,
      "learning_rate": 0.003,
      "loss": 4.3427,
      "step": 1354
    },
    {
      "epoch": 0.01355,
      "grad_norm": 0.7908101359678147,
      "learning_rate": 0.003,
      "loss": 4.3615,
      "step": 1355
    },
    {
      "epoch": 0.01356,
      "grad_norm": 0.6826667144170498,
      "learning_rate": 0.003,
      "loss": 4.3532,
      "step": 1356
    },
    {
      "epoch": 0.01357,
      "grad_norm": 0.5539967821197357,
      "learning_rate": 0.003,
      "loss": 4.3308,
      "step": 1357
    },
    {
      "epoch": 0.01358,
      "grad_norm": 0.6087032954279707,
      "learning_rate": 0.003,
      "loss": 4.3295,
      "step": 1358
    },
    {
      "epoch": 0.01359,
      "grad_norm": 0.7945356614677406,
      "learning_rate": 0.003,
      "loss": 4.3753,
      "step": 1359
    },
    {
      "epoch": 0.0136,
      "grad_norm": 0.7371837851861581,
      "learning_rate": 0.003,
      "loss": 4.3479,
      "step": 1360
    },
    {
      "epoch": 0.01361,
      "grad_norm": 0.5619004170613455,
      "learning_rate": 0.003,
      "loss": 4.3527,
      "step": 1361
    },
    {
      "epoch": 0.01362,
      "grad_norm": 0.5285167943277829,
      "learning_rate": 0.003,
      "loss": 4.3621,
      "step": 1362
    },
    {
      "epoch": 0.01363,
      "grad_norm": 0.5076817025344985,
      "learning_rate": 0.003,
      "loss": 4.3625,
      "step": 1363
    },
    {
      "epoch": 0.01364,
      "grad_norm": 0.45959981115038123,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1364
    },
    {
      "epoch": 0.01365,
      "grad_norm": 0.4790054557157507,
      "learning_rate": 0.003,
      "loss": 4.3663,
      "step": 1365
    },
    {
      "epoch": 0.01366,
      "grad_norm": 0.47982525149920063,
      "learning_rate": 0.003,
      "loss": 4.3186,
      "step": 1366
    },
    {
      "epoch": 0.01367,
      "grad_norm": 0.4158841510813793,
      "learning_rate": 0.003,
      "loss": 4.3634,
      "step": 1367
    },
    {
      "epoch": 0.01368,
      "grad_norm": 0.40516006301845847,
      "learning_rate": 0.003,
      "loss": 4.3337,
      "step": 1368
    },
    {
      "epoch": 0.01369,
      "grad_norm": 0.4130828938355455,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1369
    },
    {
      "epoch": 0.0137,
      "grad_norm": 0.4124457908402306,
      "learning_rate": 0.003,
      "loss": 4.3486,
      "step": 1370
    },
    {
      "epoch": 0.01371,
      "grad_norm": 0.3816075969714427,
      "learning_rate": 0.003,
      "loss": 4.3376,
      "step": 1371
    },
    {
      "epoch": 0.01372,
      "grad_norm": 0.360968745189086,
      "learning_rate": 0.003,
      "loss": 4.2843,
      "step": 1372
    },
    {
      "epoch": 0.01373,
      "grad_norm": 0.3546042076579899,
      "learning_rate": 0.003,
      "loss": 4.3232,
      "step": 1373
    },
    {
      "epoch": 0.01374,
      "grad_norm": 0.35618421664945743,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1374
    },
    {
      "epoch": 0.01375,
      "grad_norm": 0.3353008720630168,
      "learning_rate": 0.003,
      "loss": 4.2881,
      "step": 1375
    },
    {
      "epoch": 0.01376,
      "grad_norm": 0.3199771170841034,
      "learning_rate": 0.003,
      "loss": 4.3127,
      "step": 1376
    },
    {
      "epoch": 0.01377,
      "grad_norm": 0.3779659083446569,
      "learning_rate": 0.003,
      "loss": 4.3119,
      "step": 1377
    },
    {
      "epoch": 0.01378,
      "grad_norm": 0.5745139158210405,
      "learning_rate": 0.003,
      "loss": 4.3459,
      "step": 1378
    },
    {
      "epoch": 0.01379,
      "grad_norm": 0.8747330770544977,
      "learning_rate": 0.003,
      "loss": 4.3327,
      "step": 1379
    },
    {
      "epoch": 0.0138,
      "grad_norm": 1.1013712635151454,
      "learning_rate": 0.003,
      "loss": 4.3554,
      "step": 1380
    },
    {
      "epoch": 0.01381,
      "grad_norm": 0.6702950145107445,
      "learning_rate": 0.003,
      "loss": 4.3457,
      "step": 1381
    },
    {
      "epoch": 0.01382,
      "grad_norm": 0.674972596014113,
      "learning_rate": 0.003,
      "loss": 4.3437,
      "step": 1382
    },
    {
      "epoch": 0.01383,
      "grad_norm": 0.6997537624642522,
      "learning_rate": 0.003,
      "loss": 4.3369,
      "step": 1383
    },
    {
      "epoch": 0.01384,
      "grad_norm": 0.6469176723359975,
      "learning_rate": 0.003,
      "loss": 4.3638,
      "step": 1384
    },
    {
      "epoch": 0.01385,
      "grad_norm": 0.5530980119483301,
      "learning_rate": 0.003,
      "loss": 4.3456,
      "step": 1385
    },
    {
      "epoch": 0.01386,
      "grad_norm": 0.6135404549370512,
      "learning_rate": 0.003,
      "loss": 4.3261,
      "step": 1386
    },
    {
      "epoch": 0.01387,
      "grad_norm": 0.552332815368935,
      "learning_rate": 0.003,
      "loss": 4.3116,
      "step": 1387
    },
    {
      "epoch": 0.01388,
      "grad_norm": 0.49081872927843356,
      "learning_rate": 0.003,
      "loss": 4.3469,
      "step": 1388
    },
    {
      "epoch": 0.01389,
      "grad_norm": 0.495957727461612,
      "learning_rate": 0.003,
      "loss": 4.3269,
      "step": 1389
    },
    {
      "epoch": 0.0139,
      "grad_norm": 0.4896458918623629,
      "learning_rate": 0.003,
      "loss": 4.3267,
      "step": 1390
    },
    {
      "epoch": 0.01391,
      "grad_norm": 0.49070092673030247,
      "learning_rate": 0.003,
      "loss": 4.3285,
      "step": 1391
    },
    {
      "epoch": 0.01392,
      "grad_norm": 0.475140973661053,
      "learning_rate": 0.003,
      "loss": 4.3381,
      "step": 1392
    },
    {
      "epoch": 0.01393,
      "grad_norm": 0.45720050629487835,
      "learning_rate": 0.003,
      "loss": 4.3042,
      "step": 1393
    },
    {
      "epoch": 0.01394,
      "grad_norm": 0.5119120302085031,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1394
    },
    {
      "epoch": 0.01395,
      "grad_norm": 0.6190916799083034,
      "learning_rate": 0.003,
      "loss": 4.3279,
      "step": 1395
    },
    {
      "epoch": 0.01396,
      "grad_norm": 0.6508817551577758,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1396
    },
    {
      "epoch": 0.01397,
      "grad_norm": 0.6075070397922983,
      "learning_rate": 0.003,
      "loss": 4.3359,
      "step": 1397
    },
    {
      "epoch": 0.01398,
      "grad_norm": 0.6196238264215417,
      "learning_rate": 0.003,
      "loss": 4.3165,
      "step": 1398
    },
    {
      "epoch": 0.01399,
      "grad_norm": 0.6249218455396569,
      "learning_rate": 0.003,
      "loss": 4.3396,
      "step": 1399
    },
    {
      "epoch": 0.014,
      "grad_norm": 0.7148437815408427,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1400
    },
    {
      "epoch": 0.01401,
      "grad_norm": 0.6723818092453638,
      "learning_rate": 0.003,
      "loss": 4.3228,
      "step": 1401
    },
    {
      "epoch": 0.01402,
      "grad_norm": 0.7455518751558462,
      "learning_rate": 0.003,
      "loss": 4.3392,
      "step": 1402
    },
    {
      "epoch": 0.01403,
      "grad_norm": 0.6611806665670097,
      "learning_rate": 0.003,
      "loss": 4.3114,
      "step": 1403
    },
    {
      "epoch": 0.01404,
      "grad_norm": 0.6150645931793568,
      "learning_rate": 0.003,
      "loss": 4.3339,
      "step": 1404
    },
    {
      "epoch": 0.01405,
      "grad_norm": 0.591029520159785,
      "learning_rate": 0.003,
      "loss": 4.3493,
      "step": 1405
    },
    {
      "epoch": 0.01406,
      "grad_norm": 0.6587405723247819,
      "learning_rate": 0.003,
      "loss": 4.3328,
      "step": 1406
    },
    {
      "epoch": 0.01407,
      "grad_norm": 0.7112635318635802,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1407
    },
    {
      "epoch": 0.01408,
      "grad_norm": 0.7760050777299251,
      "learning_rate": 0.003,
      "loss": 4.3163,
      "step": 1408
    },
    {
      "epoch": 0.01409,
      "grad_norm": 0.7313606609426694,
      "learning_rate": 0.003,
      "loss": 4.3489,
      "step": 1409
    },
    {
      "epoch": 0.0141,
      "grad_norm": 0.7213404812022338,
      "learning_rate": 0.003,
      "loss": 4.326,
      "step": 1410
    },
    {
      "epoch": 0.01411,
      "grad_norm": 0.6616036396606474,
      "learning_rate": 0.003,
      "loss": 4.3016,
      "step": 1411
    },
    {
      "epoch": 0.01412,
      "grad_norm": 0.6008069011437791,
      "learning_rate": 0.003,
      "loss": 4.3378,
      "step": 1412
    },
    {
      "epoch": 0.01413,
      "grad_norm": 0.7020614194036953,
      "learning_rate": 0.003,
      "loss": 4.3356,
      "step": 1413
    },
    {
      "epoch": 0.01414,
      "grad_norm": 0.6966655821964342,
      "learning_rate": 0.003,
      "loss": 4.322,
      "step": 1414
    },
    {
      "epoch": 0.01415,
      "grad_norm": 0.7878825992848615,
      "learning_rate": 0.003,
      "loss": 4.3182,
      "step": 1415
    },
    {
      "epoch": 0.01416,
      "grad_norm": 0.9133447782042373,
      "learning_rate": 0.003,
      "loss": 4.3606,
      "step": 1416
    },
    {
      "epoch": 0.01417,
      "grad_norm": 1.0833856715197905,
      "learning_rate": 0.003,
      "loss": 4.3655,
      "step": 1417
    },
    {
      "epoch": 0.01418,
      "grad_norm": 0.8776183907401212,
      "learning_rate": 0.003,
      "loss": 4.3779,
      "step": 1418
    },
    {
      "epoch": 0.01419,
      "grad_norm": 0.8716355229415133,
      "learning_rate": 0.003,
      "loss": 4.3745,
      "step": 1419
    },
    {
      "epoch": 0.0142,
      "grad_norm": 0.6900477550364122,
      "learning_rate": 0.003,
      "loss": 4.3678,
      "step": 1420
    },
    {
      "epoch": 0.01421,
      "grad_norm": 0.6518139426774641,
      "learning_rate": 0.003,
      "loss": 4.3347,
      "step": 1421
    },
    {
      "epoch": 0.01422,
      "grad_norm": 0.6549157122857109,
      "learning_rate": 0.003,
      "loss": 4.3382,
      "step": 1422
    },
    {
      "epoch": 0.01423,
      "grad_norm": 0.6092791265519953,
      "learning_rate": 0.003,
      "loss": 4.3402,
      "step": 1423
    },
    {
      "epoch": 0.01424,
      "grad_norm": 0.5943123649374407,
      "learning_rate": 0.003,
      "loss": 4.3537,
      "step": 1424
    },
    {
      "epoch": 0.01425,
      "grad_norm": 0.5086164834561862,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1425
    },
    {
      "epoch": 0.01426,
      "grad_norm": 0.45109246822140664,
      "learning_rate": 0.003,
      "loss": 4.3278,
      "step": 1426
    },
    {
      "epoch": 0.01427,
      "grad_norm": 0.43801724179324136,
      "learning_rate": 0.003,
      "loss": 4.3209,
      "step": 1427
    },
    {
      "epoch": 0.01428,
      "grad_norm": 0.5018488931475374,
      "learning_rate": 0.003,
      "loss": 4.3159,
      "step": 1428
    },
    {
      "epoch": 0.01429,
      "grad_norm": 0.47066094753828186,
      "learning_rate": 0.003,
      "loss": 4.3234,
      "step": 1429
    },
    {
      "epoch": 0.0143,
      "grad_norm": 0.4403238802934058,
      "learning_rate": 0.003,
      "loss": 4.3319,
      "step": 1430
    },
    {
      "epoch": 0.01431,
      "grad_norm": 0.44039117208650785,
      "learning_rate": 0.003,
      "loss": 4.3183,
      "step": 1431
    },
    {
      "epoch": 0.01432,
      "grad_norm": 0.477019489570791,
      "learning_rate": 0.003,
      "loss": 4.3074,
      "step": 1432
    },
    {
      "epoch": 0.01433,
      "grad_norm": 0.4264410276196829,
      "learning_rate": 0.003,
      "loss": 4.3017,
      "step": 1433
    },
    {
      "epoch": 0.01434,
      "grad_norm": 0.3944694636285646,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1434
    },
    {
      "epoch": 0.01435,
      "grad_norm": 0.3924802801708403,
      "learning_rate": 0.003,
      "loss": 4.2942,
      "step": 1435
    },
    {
      "epoch": 0.01436,
      "grad_norm": 0.4172733802501552,
      "learning_rate": 0.003,
      "loss": 4.3128,
      "step": 1436
    },
    {
      "epoch": 0.01437,
      "grad_norm": 0.3851734831243834,
      "learning_rate": 0.003,
      "loss": 4.3237,
      "step": 1437
    },
    {
      "epoch": 0.01438,
      "grad_norm": 0.35404107026104975,
      "learning_rate": 0.003,
      "loss": 4.3163,
      "step": 1438
    },
    {
      "epoch": 0.01439,
      "grad_norm": 0.3935862503777362,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 1439
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.48537387134962556,
      "learning_rate": 0.003,
      "loss": 4.2941,
      "step": 1440
    },
    {
      "epoch": 0.01441,
      "grad_norm": 0.6205869970599557,
      "learning_rate": 0.003,
      "loss": 4.3081,
      "step": 1441
    },
    {
      "epoch": 0.01442,
      "grad_norm": 0.733625820231546,
      "learning_rate": 0.003,
      "loss": 4.2954,
      "step": 1442
    },
    {
      "epoch": 0.01443,
      "grad_norm": 0.6539626450528481,
      "learning_rate": 0.003,
      "loss": 4.3597,
      "step": 1443
    },
    {
      "epoch": 0.01444,
      "grad_norm": 0.4715932337839517,
      "learning_rate": 0.003,
      "loss": 4.3144,
      "step": 1444
    },
    {
      "epoch": 0.01445,
      "grad_norm": 0.5734177290330968,
      "learning_rate": 0.003,
      "loss": 4.2799,
      "step": 1445
    },
    {
      "epoch": 0.01446,
      "grad_norm": 0.6345599077429956,
      "learning_rate": 0.003,
      "loss": 4.3,
      "step": 1446
    },
    {
      "epoch": 0.01447,
      "grad_norm": 0.7159211854064349,
      "learning_rate": 0.003,
      "loss": 4.3124,
      "step": 1447
    },
    {
      "epoch": 0.01448,
      "grad_norm": 0.847805509409058,
      "learning_rate": 0.003,
      "loss": 4.3346,
      "step": 1448
    },
    {
      "epoch": 0.01449,
      "grad_norm": 0.8334367502615841,
      "learning_rate": 0.003,
      "loss": 4.3316,
      "step": 1449
    },
    {
      "epoch": 0.0145,
      "grad_norm": 0.8387485961034247,
      "learning_rate": 0.003,
      "loss": 4.3521,
      "step": 1450
    },
    {
      "epoch": 0.01451,
      "grad_norm": 0.7069735242725316,
      "learning_rate": 0.003,
      "loss": 4.3254,
      "step": 1451
    },
    {
      "epoch": 0.01452,
      "grad_norm": 0.7128306367496979,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1452
    },
    {
      "epoch": 0.01453,
      "grad_norm": 0.7607386324594154,
      "learning_rate": 0.003,
      "loss": 4.3465,
      "step": 1453
    },
    {
      "epoch": 0.01454,
      "grad_norm": 0.8531443855411419,
      "learning_rate": 0.003,
      "loss": 4.3152,
      "step": 1454
    },
    {
      "epoch": 0.01455,
      "grad_norm": 0.9072653937920705,
      "learning_rate": 0.003,
      "loss": 4.3602,
      "step": 1455
    },
    {
      "epoch": 0.01456,
      "grad_norm": 0.9678032867174141,
      "learning_rate": 0.003,
      "loss": 4.3448,
      "step": 1456
    },
    {
      "epoch": 0.01457,
      "grad_norm": 0.8263837469314969,
      "learning_rate": 0.003,
      "loss": 4.3358,
      "step": 1457
    },
    {
      "epoch": 0.01458,
      "grad_norm": 0.7586740850669937,
      "learning_rate": 0.003,
      "loss": 4.3274,
      "step": 1458
    },
    {
      "epoch": 0.01459,
      "grad_norm": 0.7104303096772606,
      "learning_rate": 0.003,
      "loss": 4.3506,
      "step": 1459
    },
    {
      "epoch": 0.0146,
      "grad_norm": 0.7105635489785329,
      "learning_rate": 0.003,
      "loss": 4.358,
      "step": 1460
    },
    {
      "epoch": 0.01461,
      "grad_norm": 0.6051123991598398,
      "learning_rate": 0.003,
      "loss": 4.3255,
      "step": 1461
    },
    {
      "epoch": 0.01462,
      "grad_norm": 0.49818465497872577,
      "learning_rate": 0.003,
      "loss": 4.3277,
      "step": 1462
    },
    {
      "epoch": 0.01463,
      "grad_norm": 0.4705387398623177,
      "learning_rate": 0.003,
      "loss": 4.3234,
      "step": 1463
    },
    {
      "epoch": 0.01464,
      "grad_norm": 0.43420725887330847,
      "learning_rate": 0.003,
      "loss": 4.2947,
      "step": 1464
    },
    {
      "epoch": 0.01465,
      "grad_norm": 0.416394344592158,
      "learning_rate": 0.003,
      "loss": 4.3376,
      "step": 1465
    },
    {
      "epoch": 0.01466,
      "grad_norm": 0.46274333085308,
      "learning_rate": 0.003,
      "loss": 4.3061,
      "step": 1466
    },
    {
      "epoch": 0.01467,
      "grad_norm": 0.412473133729069,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1467
    },
    {
      "epoch": 0.01468,
      "grad_norm": 0.4311218055740771,
      "learning_rate": 0.003,
      "loss": 4.3146,
      "step": 1468
    },
    {
      "epoch": 0.01469,
      "grad_norm": 0.4074958191436831,
      "learning_rate": 0.003,
      "loss": 4.312,
      "step": 1469
    },
    {
      "epoch": 0.0147,
      "grad_norm": 0.43399248954304653,
      "learning_rate": 0.003,
      "loss": 4.3033,
      "step": 1470
    },
    {
      "epoch": 0.01471,
      "grad_norm": 0.41361534167776814,
      "learning_rate": 0.003,
      "loss": 4.3141,
      "step": 1471
    },
    {
      "epoch": 0.01472,
      "grad_norm": 0.4291933422709367,
      "learning_rate": 0.003,
      "loss": 4.2989,
      "step": 1472
    },
    {
      "epoch": 0.01473,
      "grad_norm": 0.4534239206252403,
      "learning_rate": 0.003,
      "loss": 4.3122,
      "step": 1473
    },
    {
      "epoch": 0.01474,
      "grad_norm": 0.5901119500480289,
      "learning_rate": 0.003,
      "loss": 4.3004,
      "step": 1474
    },
    {
      "epoch": 0.01475,
      "grad_norm": 0.893255157155039,
      "learning_rate": 0.003,
      "loss": 4.3326,
      "step": 1475
    },
    {
      "epoch": 0.01476,
      "grad_norm": 1.4183061783370785,
      "learning_rate": 0.003,
      "loss": 4.3574,
      "step": 1476
    },
    {
      "epoch": 0.01477,
      "grad_norm": 0.6731343966041683,
      "learning_rate": 0.003,
      "loss": 4.3345,
      "step": 1477
    },
    {
      "epoch": 0.01478,
      "grad_norm": 0.6759295392700941,
      "learning_rate": 0.003,
      "loss": 4.3246,
      "step": 1478
    },
    {
      "epoch": 0.01479,
      "grad_norm": 0.5808036667947172,
      "learning_rate": 0.003,
      "loss": 4.3486,
      "step": 1479
    },
    {
      "epoch": 0.0148,
      "grad_norm": 0.5015217636867718,
      "learning_rate": 0.003,
      "loss": 4.3142,
      "step": 1480
    },
    {
      "epoch": 0.01481,
      "grad_norm": 0.4567088802488701,
      "learning_rate": 0.003,
      "loss": 4.3125,
      "step": 1481
    },
    {
      "epoch": 0.01482,
      "grad_norm": 0.4183399036149268,
      "learning_rate": 0.003,
      "loss": 4.2956,
      "step": 1482
    },
    {
      "epoch": 0.01483,
      "grad_norm": 0.4450691330955631,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1483
    },
    {
      "epoch": 0.01484,
      "grad_norm": 0.3765600896203587,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1484
    },
    {
      "epoch": 0.01485,
      "grad_norm": 0.3992681409647294,
      "learning_rate": 0.003,
      "loss": 4.3029,
      "step": 1485
    },
    {
      "epoch": 0.01486,
      "grad_norm": 0.3969999829277036,
      "learning_rate": 0.003,
      "loss": 4.2984,
      "step": 1486
    },
    {
      "epoch": 0.01487,
      "grad_norm": 0.3308465500623018,
      "learning_rate": 0.003,
      "loss": 4.2984,
      "step": 1487
    },
    {
      "epoch": 0.01488,
      "grad_norm": 0.2814090123328206,
      "learning_rate": 0.003,
      "loss": 4.3052,
      "step": 1488
    },
    {
      "epoch": 0.01489,
      "grad_norm": 0.30005415710473554,
      "learning_rate": 0.003,
      "loss": 4.2986,
      "step": 1489
    },
    {
      "epoch": 0.0149,
      "grad_norm": 0.32421623348009687,
      "learning_rate": 0.003,
      "loss": 4.2912,
      "step": 1490
    },
    {
      "epoch": 0.01491,
      "grad_norm": 0.35115426496092417,
      "learning_rate": 0.003,
      "loss": 4.3043,
      "step": 1491
    },
    {
      "epoch": 0.01492,
      "grad_norm": 0.37018314484045245,
      "learning_rate": 0.003,
      "loss": 4.2686,
      "step": 1492
    },
    {
      "epoch": 0.01493,
      "grad_norm": 0.4075858968364619,
      "learning_rate": 0.003,
      "loss": 4.3015,
      "step": 1493
    },
    {
      "epoch": 0.01494,
      "grad_norm": 0.4912830051602376,
      "learning_rate": 0.003,
      "loss": 4.2818,
      "step": 1494
    },
    {
      "epoch": 0.01495,
      "grad_norm": 0.5565886808339906,
      "learning_rate": 0.003,
      "loss": 4.2955,
      "step": 1495
    },
    {
      "epoch": 0.01496,
      "grad_norm": 0.6104515927182645,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1496
    },
    {
      "epoch": 0.01497,
      "grad_norm": 0.6049970308529968,
      "learning_rate": 0.003,
      "loss": 4.2492,
      "step": 1497
    },
    {
      "epoch": 0.01498,
      "grad_norm": 0.5392848754302815,
      "learning_rate": 0.003,
      "loss": 4.304,
      "step": 1498
    },
    {
      "epoch": 0.01499,
      "grad_norm": 0.544325685587062,
      "learning_rate": 0.003,
      "loss": 4.3252,
      "step": 1499
    },
    {
      "epoch": 0.015,
      "grad_norm": 0.5744687962127912,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1500
    },
    {
      "epoch": 0.01501,
      "grad_norm": 0.6650575718497268,
      "learning_rate": 0.003,
      "loss": 4.2836,
      "step": 1501
    },
    {
      "epoch": 0.01502,
      "grad_norm": 0.762494815469482,
      "learning_rate": 0.003,
      "loss": 4.3344,
      "step": 1502
    },
    {
      "epoch": 0.01503,
      "grad_norm": 0.8042887596944258,
      "learning_rate": 0.003,
      "loss": 4.3203,
      "step": 1503
    },
    {
      "epoch": 0.01504,
      "grad_norm": 0.8888233604685658,
      "learning_rate": 0.003,
      "loss": 4.3247,
      "step": 1504
    },
    {
      "epoch": 0.01505,
      "grad_norm": 0.9445171252403785,
      "learning_rate": 0.003,
      "loss": 4.3231,
      "step": 1505
    },
    {
      "epoch": 0.01506,
      "grad_norm": 1.0444284185336012,
      "learning_rate": 0.003,
      "loss": 4.3641,
      "step": 1506
    },
    {
      "epoch": 0.01507,
      "grad_norm": 1.1261023785455877,
      "learning_rate": 0.003,
      "loss": 4.3931,
      "step": 1507
    },
    {
      "epoch": 0.01508,
      "grad_norm": 1.234913329930944,
      "learning_rate": 0.003,
      "loss": 4.3763,
      "step": 1508
    },
    {
      "epoch": 0.01509,
      "grad_norm": 0.8497166266198067,
      "learning_rate": 0.003,
      "loss": 4.3805,
      "step": 1509
    },
    {
      "epoch": 0.0151,
      "grad_norm": 0.8376711335758371,
      "learning_rate": 0.003,
      "loss": 4.3542,
      "step": 1510
    },
    {
      "epoch": 0.01511,
      "grad_norm": 0.724530376603094,
      "learning_rate": 0.003,
      "loss": 4.3387,
      "step": 1511
    },
    {
      "epoch": 0.01512,
      "grad_norm": 0.6831424127546084,
      "learning_rate": 0.003,
      "loss": 4.3429,
      "step": 1512
    },
    {
      "epoch": 0.01513,
      "grad_norm": 0.5603527644059662,
      "learning_rate": 0.003,
      "loss": 4.3405,
      "step": 1513
    },
    {
      "epoch": 0.01514,
      "grad_norm": 0.5930589812891277,
      "learning_rate": 0.003,
      "loss": 4.3465,
      "step": 1514
    },
    {
      "epoch": 0.01515,
      "grad_norm": 0.7232891849880219,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1515
    },
    {
      "epoch": 0.01516,
      "grad_norm": 0.7759860995114962,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1516
    },
    {
      "epoch": 0.01517,
      "grad_norm": 0.6404044980036384,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 1517
    },
    {
      "epoch": 0.01518,
      "grad_norm": 0.640364805854496,
      "learning_rate": 0.003,
      "loss": 4.3263,
      "step": 1518
    },
    {
      "epoch": 0.01519,
      "grad_norm": 0.7392318945229942,
      "learning_rate": 0.003,
      "loss": 4.3661,
      "step": 1519
    },
    {
      "epoch": 0.0152,
      "grad_norm": 0.9450167317883038,
      "learning_rate": 0.003,
      "loss": 4.3284,
      "step": 1520
    },
    {
      "epoch": 0.01521,
      "grad_norm": 0.9107576371118652,
      "learning_rate": 0.003,
      "loss": 4.3495,
      "step": 1521
    },
    {
      "epoch": 0.01522,
      "grad_norm": 0.7612821914439016,
      "learning_rate": 0.003,
      "loss": 4.3394,
      "step": 1522
    },
    {
      "epoch": 0.01523,
      "grad_norm": 0.6613113244902528,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1523
    },
    {
      "epoch": 0.01524,
      "grad_norm": 0.5975485724244859,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1524
    },
    {
      "epoch": 0.01525,
      "grad_norm": 0.49953522198986544,
      "learning_rate": 0.003,
      "loss": 4.349,
      "step": 1525
    },
    {
      "epoch": 0.01526,
      "grad_norm": 0.5020786027972391,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1526
    },
    {
      "epoch": 0.01527,
      "grad_norm": 0.4272044641011338,
      "learning_rate": 0.003,
      "loss": 4.301,
      "step": 1527
    },
    {
      "epoch": 0.01528,
      "grad_norm": 0.38333031076180046,
      "learning_rate": 0.003,
      "loss": 4.3172,
      "step": 1528
    },
    {
      "epoch": 0.01529,
      "grad_norm": 0.37710300633649013,
      "learning_rate": 0.003,
      "loss": 4.3078,
      "step": 1529
    },
    {
      "epoch": 0.0153,
      "grad_norm": 0.3539206796857484,
      "learning_rate": 0.003,
      "loss": 4.3136,
      "step": 1530
    },
    {
      "epoch": 0.01531,
      "grad_norm": 0.31579499408524375,
      "learning_rate": 0.003,
      "loss": 4.3044,
      "step": 1531
    },
    {
      "epoch": 0.01532,
      "grad_norm": 0.28740282558130736,
      "learning_rate": 0.003,
      "loss": 4.2793,
      "step": 1532
    },
    {
      "epoch": 0.01533,
      "grad_norm": 0.32298125914434667,
      "learning_rate": 0.003,
      "loss": 4.3053,
      "step": 1533
    },
    {
      "epoch": 0.01534,
      "grad_norm": 0.34152544816126584,
      "learning_rate": 0.003,
      "loss": 4.302,
      "step": 1534
    },
    {
      "epoch": 0.01535,
      "grad_norm": 0.40345568055827374,
      "learning_rate": 0.003,
      "loss": 4.3133,
      "step": 1535
    },
    {
      "epoch": 0.01536,
      "grad_norm": 0.47716276358652043,
      "learning_rate": 0.003,
      "loss": 4.2948,
      "step": 1536
    },
    {
      "epoch": 0.01537,
      "grad_norm": 0.5673450395929736,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 1537
    },
    {
      "epoch": 0.01538,
      "grad_norm": 0.5662986079723249,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1538
    },
    {
      "epoch": 0.01539,
      "grad_norm": 0.5250656872774493,
      "learning_rate": 0.003,
      "loss": 4.2993,
      "step": 1539
    },
    {
      "epoch": 0.0154,
      "grad_norm": 0.4759070450942551,
      "learning_rate": 0.003,
      "loss": 4.3011,
      "step": 1540
    },
    {
      "epoch": 0.01541,
      "grad_norm": 0.525208601732405,
      "learning_rate": 0.003,
      "loss": 4.2841,
      "step": 1541
    },
    {
      "epoch": 0.01542,
      "grad_norm": 0.6195424842603703,
      "learning_rate": 0.003,
      "loss": 4.3158,
      "step": 1542
    },
    {
      "epoch": 0.01543,
      "grad_norm": 0.7066044834019704,
      "learning_rate": 0.003,
      "loss": 4.3015,
      "step": 1543
    },
    {
      "epoch": 0.01544,
      "grad_norm": 0.6094492242185264,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1544
    },
    {
      "epoch": 0.01545,
      "grad_norm": 0.6542596992859665,
      "learning_rate": 0.003,
      "loss": 4.3554,
      "step": 1545
    },
    {
      "epoch": 0.01546,
      "grad_norm": 0.7470834345800471,
      "learning_rate": 0.003,
      "loss": 4.3329,
      "step": 1546
    },
    {
      "epoch": 0.01547,
      "grad_norm": 0.7200211733418024,
      "learning_rate": 0.003,
      "loss": 4.3307,
      "step": 1547
    },
    {
      "epoch": 0.01548,
      "grad_norm": 0.6295943931814879,
      "learning_rate": 0.003,
      "loss": 4.2934,
      "step": 1548
    },
    {
      "epoch": 0.01549,
      "grad_norm": 0.54945766010764,
      "learning_rate": 0.003,
      "loss": 4.2766,
      "step": 1549
    },
    {
      "epoch": 0.0155,
      "grad_norm": 0.5950909013773433,
      "learning_rate": 0.003,
      "loss": 4.297,
      "step": 1550
    },
    {
      "epoch": 0.01551,
      "grad_norm": 0.6224936273193388,
      "learning_rate": 0.003,
      "loss": 4.2776,
      "step": 1551
    },
    {
      "epoch": 0.01552,
      "grad_norm": 0.6624794177801069,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 1552
    },
    {
      "epoch": 0.01553,
      "grad_norm": 0.7863145855853837,
      "learning_rate": 0.003,
      "loss": 4.335,
      "step": 1553
    },
    {
      "epoch": 0.01554,
      "grad_norm": 0.7964168301162362,
      "learning_rate": 0.003,
      "loss": 4.3213,
      "step": 1554
    },
    {
      "epoch": 0.01555,
      "grad_norm": 0.6686376199306979,
      "learning_rate": 0.003,
      "loss": 4.292,
      "step": 1555
    },
    {
      "epoch": 0.01556,
      "grad_norm": 0.6587445991969533,
      "learning_rate": 0.003,
      "loss": 4.327,
      "step": 1556
    },
    {
      "epoch": 0.01557,
      "grad_norm": 0.8168668532513341,
      "learning_rate": 0.003,
      "loss": 4.3101,
      "step": 1557
    },
    {
      "epoch": 0.01558,
      "grad_norm": 0.6341929215834538,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 1558
    },
    {
      "epoch": 0.01559,
      "grad_norm": 0.5639732491475488,
      "learning_rate": 0.003,
      "loss": 4.3265,
      "step": 1559
    },
    {
      "epoch": 0.0156,
      "grad_norm": 0.6344606998016725,
      "learning_rate": 0.003,
      "loss": 4.3189,
      "step": 1560
    },
    {
      "epoch": 0.01561,
      "grad_norm": 0.6020928854417142,
      "learning_rate": 0.003,
      "loss": 4.301,
      "step": 1561
    },
    {
      "epoch": 0.01562,
      "grad_norm": 0.6447473041378132,
      "learning_rate": 0.003,
      "loss": 4.2957,
      "step": 1562
    },
    {
      "epoch": 0.01563,
      "grad_norm": 0.637527615370107,
      "learning_rate": 0.003,
      "loss": 4.3012,
      "step": 1563
    },
    {
      "epoch": 0.01564,
      "grad_norm": 0.6065116381126623,
      "learning_rate": 0.003,
      "loss": 4.3275,
      "step": 1564
    },
    {
      "epoch": 0.01565,
      "grad_norm": 0.5530540273129289,
      "learning_rate": 0.003,
      "loss": 4.3008,
      "step": 1565
    },
    {
      "epoch": 0.01566,
      "grad_norm": 0.523767221886022,
      "learning_rate": 0.003,
      "loss": 4.3185,
      "step": 1566
    },
    {
      "epoch": 0.01567,
      "grad_norm": 0.44666292802850066,
      "learning_rate": 0.003,
      "loss": 4.3073,
      "step": 1567
    },
    {
      "epoch": 0.01568,
      "grad_norm": 0.4825649782380437,
      "learning_rate": 0.003,
      "loss": 4.2912,
      "step": 1568
    },
    {
      "epoch": 0.01569,
      "grad_norm": 0.4768074953920202,
      "learning_rate": 0.003,
      "loss": 4.2946,
      "step": 1569
    },
    {
      "epoch": 0.0157,
      "grad_norm": 0.4756489341732161,
      "learning_rate": 0.003,
      "loss": 4.3112,
      "step": 1570
    },
    {
      "epoch": 0.01571,
      "grad_norm": 0.5079994555205132,
      "learning_rate": 0.003,
      "loss": 4.2965,
      "step": 1571
    },
    {
      "epoch": 0.01572,
      "grad_norm": 0.5289839099051613,
      "learning_rate": 0.003,
      "loss": 4.3054,
      "step": 1572
    },
    {
      "epoch": 0.01573,
      "grad_norm": 0.5222081857718813,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1573
    },
    {
      "epoch": 0.01574,
      "grad_norm": 0.6451289760332761,
      "learning_rate": 0.003,
      "loss": 4.3076,
      "step": 1574
    },
    {
      "epoch": 0.01575,
      "grad_norm": 0.7877974193239016,
      "learning_rate": 0.003,
      "loss": 4.2813,
      "step": 1575
    },
    {
      "epoch": 0.01576,
      "grad_norm": 0.8607804040917739,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1576
    },
    {
      "epoch": 0.01577,
      "grad_norm": 0.8142185085462708,
      "learning_rate": 0.003,
      "loss": 4.2736,
      "step": 1577
    },
    {
      "epoch": 0.01578,
      "grad_norm": 0.7720878068888671,
      "learning_rate": 0.003,
      "loss": 4.3026,
      "step": 1578
    },
    {
      "epoch": 0.01579,
      "grad_norm": 0.7192241887142894,
      "learning_rate": 0.003,
      "loss": 4.2996,
      "step": 1579
    },
    {
      "epoch": 0.0158,
      "grad_norm": 0.606159564012465,
      "learning_rate": 0.003,
      "loss": 4.3013,
      "step": 1580
    },
    {
      "epoch": 0.01581,
      "grad_norm": 0.4880751313224698,
      "learning_rate": 0.003,
      "loss": 4.3071,
      "step": 1581
    },
    {
      "epoch": 0.01582,
      "grad_norm": 0.5304647614551631,
      "learning_rate": 0.003,
      "loss": 4.2891,
      "step": 1582
    },
    {
      "epoch": 0.01583,
      "grad_norm": 0.46382514166554145,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 1583
    },
    {
      "epoch": 0.01584,
      "grad_norm": 0.47578841605412814,
      "learning_rate": 0.003,
      "loss": 4.3048,
      "step": 1584
    },
    {
      "epoch": 0.01585,
      "grad_norm": 0.4985689869447637,
      "learning_rate": 0.003,
      "loss": 4.2972,
      "step": 1585
    },
    {
      "epoch": 0.01586,
      "grad_norm": 0.4793261849715971,
      "learning_rate": 0.003,
      "loss": 4.2718,
      "step": 1586
    },
    {
      "epoch": 0.01587,
      "grad_norm": 0.4867805801070396,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1587
    },
    {
      "epoch": 0.01588,
      "grad_norm": 0.49512882436423494,
      "learning_rate": 0.003,
      "loss": 4.3139,
      "step": 1588
    },
    {
      "epoch": 0.01589,
      "grad_norm": 0.5289005946834403,
      "learning_rate": 0.003,
      "loss": 4.306,
      "step": 1589
    },
    {
      "epoch": 0.0159,
      "grad_norm": 0.6096773535511837,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1590
    },
    {
      "epoch": 0.01591,
      "grad_norm": 0.6229667469755624,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1591
    },
    {
      "epoch": 0.01592,
      "grad_norm": 0.4936084502041978,
      "learning_rate": 0.003,
      "loss": 4.2967,
      "step": 1592
    },
    {
      "epoch": 0.01593,
      "grad_norm": 0.42446523481816184,
      "learning_rate": 0.003,
      "loss": 4.2946,
      "step": 1593
    },
    {
      "epoch": 0.01594,
      "grad_norm": 0.40613386955274233,
      "learning_rate": 0.003,
      "loss": 4.2814,
      "step": 1594
    },
    {
      "epoch": 0.01595,
      "grad_norm": 0.4303780684833319,
      "learning_rate": 0.003,
      "loss": 4.308,
      "step": 1595
    },
    {
      "epoch": 0.01596,
      "grad_norm": 0.4038673300031157,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1596
    },
    {
      "epoch": 0.01597,
      "grad_norm": 0.42824275182777244,
      "learning_rate": 0.003,
      "loss": 4.2611,
      "step": 1597
    },
    {
      "epoch": 0.01598,
      "grad_norm": 0.4949313860176736,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1598
    },
    {
      "epoch": 0.01599,
      "grad_norm": 0.5005123390361799,
      "learning_rate": 0.003,
      "loss": 4.2803,
      "step": 1599
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.5929838548199676,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 1600
    },
    {
      "epoch": 0.01601,
      "grad_norm": 0.8301504230585989,
      "learning_rate": 0.003,
      "loss": 4.315,
      "step": 1601
    },
    {
      "epoch": 0.01602,
      "grad_norm": 0.9853750506753596,
      "learning_rate": 0.003,
      "loss": 4.3061,
      "step": 1602
    },
    {
      "epoch": 0.01603,
      "grad_norm": 1.003124318602637,
      "learning_rate": 0.003,
      "loss": 4.3258,
      "step": 1603
    },
    {
      "epoch": 0.01604,
      "grad_norm": 0.912578535311941,
      "learning_rate": 0.003,
      "loss": 4.327,
      "step": 1604
    },
    {
      "epoch": 0.01605,
      "grad_norm": 0.8301067621469277,
      "learning_rate": 0.003,
      "loss": 4.3002,
      "step": 1605
    },
    {
      "epoch": 0.01606,
      "grad_norm": 0.7646428413230717,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1606
    },
    {
      "epoch": 0.01607,
      "grad_norm": 0.8210614746463125,
      "learning_rate": 0.003,
      "loss": 4.3396,
      "step": 1607
    },
    {
      "epoch": 0.01608,
      "grad_norm": 0.8162978830688478,
      "learning_rate": 0.003,
      "loss": 4.3084,
      "step": 1608
    },
    {
      "epoch": 0.01609,
      "grad_norm": 0.7940470937374723,
      "learning_rate": 0.003,
      "loss": 4.3225,
      "step": 1609
    },
    {
      "epoch": 0.0161,
      "grad_norm": 0.7532673019192272,
      "learning_rate": 0.003,
      "loss": 4.2882,
      "step": 1610
    },
    {
      "epoch": 0.01611,
      "grad_norm": 0.77086180161133,
      "learning_rate": 0.003,
      "loss": 4.3117,
      "step": 1611
    },
    {
      "epoch": 0.01612,
      "grad_norm": 0.6754537663457288,
      "learning_rate": 0.003,
      "loss": 4.2961,
      "step": 1612
    },
    {
      "epoch": 0.01613,
      "grad_norm": 0.599175162495873,
      "learning_rate": 0.003,
      "loss": 4.3386,
      "step": 1613
    },
    {
      "epoch": 0.01614,
      "grad_norm": 0.5411062109149707,
      "learning_rate": 0.003,
      "loss": 4.3566,
      "step": 1614
    },
    {
      "epoch": 0.01615,
      "grad_norm": 0.512685672879493,
      "learning_rate": 0.003,
      "loss": 4.3203,
      "step": 1615
    },
    {
      "epoch": 0.01616,
      "grad_norm": 0.49714926242353863,
      "learning_rate": 0.003,
      "loss": 4.3009,
      "step": 1616
    },
    {
      "epoch": 0.01617,
      "grad_norm": 0.4964011411644138,
      "learning_rate": 0.003,
      "loss": 4.2889,
      "step": 1617
    },
    {
      "epoch": 0.01618,
      "grad_norm": 0.5061608293930215,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1618
    },
    {
      "epoch": 0.01619,
      "grad_norm": 0.5494277838056353,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1619
    },
    {
      "epoch": 0.0162,
      "grad_norm": 0.6308404917903175,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 1620
    },
    {
      "epoch": 0.01621,
      "grad_norm": 0.731307523044738,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1621
    },
    {
      "epoch": 0.01622,
      "grad_norm": 0.9046402251963843,
      "learning_rate": 0.003,
      "loss": 4.3346,
      "step": 1622
    },
    {
      "epoch": 0.01623,
      "grad_norm": 0.7735409772436379,
      "learning_rate": 0.003,
      "loss": 4.3147,
      "step": 1623
    },
    {
      "epoch": 0.01624,
      "grad_norm": 0.672645275853148,
      "learning_rate": 0.003,
      "loss": 4.3229,
      "step": 1624
    },
    {
      "epoch": 0.01625,
      "grad_norm": 0.6719787877984026,
      "learning_rate": 0.003,
      "loss": 4.2983,
      "step": 1625
    },
    {
      "epoch": 0.01626,
      "grad_norm": 0.6708647627350693,
      "learning_rate": 0.003,
      "loss": 4.3168,
      "step": 1626
    },
    {
      "epoch": 0.01627,
      "grad_norm": 0.649063452794678,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1627
    },
    {
      "epoch": 0.01628,
      "grad_norm": 0.6161891488716968,
      "learning_rate": 0.003,
      "loss": 4.2901,
      "step": 1628
    },
    {
      "epoch": 0.01629,
      "grad_norm": 0.5565589154413988,
      "learning_rate": 0.003,
      "loss": 4.2903,
      "step": 1629
    },
    {
      "epoch": 0.0163,
      "grad_norm": 0.5030071975984001,
      "learning_rate": 0.003,
      "loss": 4.2755,
      "step": 1630
    },
    {
      "epoch": 0.01631,
      "grad_norm": 0.48590911922639535,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1631
    },
    {
      "epoch": 0.01632,
      "grad_norm": 0.5725399795731132,
      "learning_rate": 0.003,
      "loss": 4.3061,
      "step": 1632
    },
    {
      "epoch": 0.01633,
      "grad_norm": 0.7370492556479813,
      "learning_rate": 0.003,
      "loss": 4.2846,
      "step": 1633
    },
    {
      "epoch": 0.01634,
      "grad_norm": 0.8641758267856038,
      "learning_rate": 0.003,
      "loss": 4.3131,
      "step": 1634
    },
    {
      "epoch": 0.01635,
      "grad_norm": 0.7344051689167792,
      "learning_rate": 0.003,
      "loss": 4.2907,
      "step": 1635
    },
    {
      "epoch": 0.01636,
      "grad_norm": 0.5215072843138687,
      "learning_rate": 0.003,
      "loss": 4.2982,
      "step": 1636
    },
    {
      "epoch": 0.01637,
      "grad_norm": 0.4749653803767528,
      "learning_rate": 0.003,
      "loss": 4.28,
      "step": 1637
    },
    {
      "epoch": 0.01638,
      "grad_norm": 0.4605003028840984,
      "learning_rate": 0.003,
      "loss": 4.2856,
      "step": 1638
    },
    {
      "epoch": 0.01639,
      "grad_norm": 0.42552013547580053,
      "learning_rate": 0.003,
      "loss": 4.3073,
      "step": 1639
    },
    {
      "epoch": 0.0164,
      "grad_norm": 0.4220228883023831,
      "learning_rate": 0.003,
      "loss": 4.2853,
      "step": 1640
    },
    {
      "epoch": 0.01641,
      "grad_norm": 0.48264096364388565,
      "learning_rate": 0.003,
      "loss": 4.2601,
      "step": 1641
    },
    {
      "epoch": 0.01642,
      "grad_norm": 0.5522814057439278,
      "learning_rate": 0.003,
      "loss": 4.3056,
      "step": 1642
    },
    {
      "epoch": 0.01643,
      "grad_norm": 0.5448290909435038,
      "learning_rate": 0.003,
      "loss": 4.2899,
      "step": 1643
    },
    {
      "epoch": 0.01644,
      "grad_norm": 0.4693252715517831,
      "learning_rate": 0.003,
      "loss": 4.3048,
      "step": 1644
    },
    {
      "epoch": 0.01645,
      "grad_norm": 0.38397009629291196,
      "learning_rate": 0.003,
      "loss": 4.2478,
      "step": 1645
    },
    {
      "epoch": 0.01646,
      "grad_norm": 0.47479208422272845,
      "learning_rate": 0.003,
      "loss": 4.2841,
      "step": 1646
    },
    {
      "epoch": 0.01647,
      "grad_norm": 0.5095257415638307,
      "learning_rate": 0.003,
      "loss": 4.2814,
      "step": 1647
    },
    {
      "epoch": 0.01648,
      "grad_norm": 0.5223808020316233,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1648
    },
    {
      "epoch": 0.01649,
      "grad_norm": 0.4909186651871344,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1649
    },
    {
      "epoch": 0.0165,
      "grad_norm": 0.4445354228997141,
      "learning_rate": 0.003,
      "loss": 4.2687,
      "step": 1650
    },
    {
      "epoch": 0.01651,
      "grad_norm": 0.47462201592081604,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1651
    },
    {
      "epoch": 0.01652,
      "grad_norm": 0.5712842642113961,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 1652
    },
    {
      "epoch": 0.01653,
      "grad_norm": 0.6633586551393839,
      "learning_rate": 0.003,
      "loss": 4.2872,
      "step": 1653
    },
    {
      "epoch": 0.01654,
      "grad_norm": 0.6882337845469141,
      "learning_rate": 0.003,
      "loss": 4.2797,
      "step": 1654
    },
    {
      "epoch": 0.01655,
      "grad_norm": 0.5883903260155589,
      "learning_rate": 0.003,
      "loss": 4.2855,
      "step": 1655
    },
    {
      "epoch": 0.01656,
      "grad_norm": 0.692175129507891,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 1656
    },
    {
      "epoch": 0.01657,
      "grad_norm": 0.7208024346780962,
      "learning_rate": 0.003,
      "loss": 4.285,
      "step": 1657
    },
    {
      "epoch": 0.01658,
      "grad_norm": 0.7405524865262961,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 1658
    },
    {
      "epoch": 0.01659,
      "grad_norm": 0.6893828101785304,
      "learning_rate": 0.003,
      "loss": 4.2781,
      "step": 1659
    },
    {
      "epoch": 0.0166,
      "grad_norm": 0.6598372867680238,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1660
    },
    {
      "epoch": 0.01661,
      "grad_norm": 0.47529948561931934,
      "learning_rate": 0.003,
      "loss": 4.2595,
      "step": 1661
    },
    {
      "epoch": 0.01662,
      "grad_norm": 0.46023265898975174,
      "learning_rate": 0.003,
      "loss": 4.2646,
      "step": 1662
    },
    {
      "epoch": 0.01663,
      "grad_norm": 0.4507377258313933,
      "learning_rate": 0.003,
      "loss": 4.2842,
      "step": 1663
    },
    {
      "epoch": 0.01664,
      "grad_norm": 0.4200561869900329,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 1664
    },
    {
      "epoch": 0.01665,
      "grad_norm": 0.4244798072607477,
      "learning_rate": 0.003,
      "loss": 4.2616,
      "step": 1665
    },
    {
      "epoch": 0.01666,
      "grad_norm": 0.5427314965460833,
      "learning_rate": 0.003,
      "loss": 4.2661,
      "step": 1666
    },
    {
      "epoch": 0.01667,
      "grad_norm": 0.5950166710073447,
      "learning_rate": 0.003,
      "loss": 4.2869,
      "step": 1667
    },
    {
      "epoch": 0.01668,
      "grad_norm": 0.7033617268618768,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1668
    },
    {
      "epoch": 0.01669,
      "grad_norm": 0.7337795007439258,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1669
    },
    {
      "epoch": 0.0167,
      "grad_norm": 0.7228924020186139,
      "learning_rate": 0.003,
      "loss": 4.2759,
      "step": 1670
    },
    {
      "epoch": 0.01671,
      "grad_norm": 0.9293602473006644,
      "learning_rate": 0.003,
      "loss": 4.2785,
      "step": 1671
    },
    {
      "epoch": 0.01672,
      "grad_norm": 1.029746366393395,
      "learning_rate": 0.003,
      "loss": 4.3064,
      "step": 1672
    },
    {
      "epoch": 0.01673,
      "grad_norm": 0.9727787844526897,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1673
    },
    {
      "epoch": 0.01674,
      "grad_norm": 0.9554074542733949,
      "learning_rate": 0.003,
      "loss": 4.3312,
      "step": 1674
    },
    {
      "epoch": 0.01675,
      "grad_norm": 0.9706031573401521,
      "learning_rate": 0.003,
      "loss": 4.3194,
      "step": 1675
    },
    {
      "epoch": 0.01676,
      "grad_norm": 0.9822377528346978,
      "learning_rate": 0.003,
      "loss": 4.3319,
      "step": 1676
    },
    {
      "epoch": 0.01677,
      "grad_norm": 0.9324715324675055,
      "learning_rate": 0.003,
      "loss": 4.335,
      "step": 1677
    },
    {
      "epoch": 0.01678,
      "grad_norm": 0.8555988190467146,
      "learning_rate": 0.003,
      "loss": 4.3399,
      "step": 1678
    },
    {
      "epoch": 0.01679,
      "grad_norm": 0.9906267827738694,
      "learning_rate": 0.003,
      "loss": 4.3344,
      "step": 1679
    },
    {
      "epoch": 0.0168,
      "grad_norm": 0.9961431022000035,
      "learning_rate": 0.003,
      "loss": 4.3626,
      "step": 1680
    },
    {
      "epoch": 0.01681,
      "grad_norm": 0.8880584596730197,
      "learning_rate": 0.003,
      "loss": 4.362,
      "step": 1681
    },
    {
      "epoch": 0.01682,
      "grad_norm": 0.864035612954157,
      "learning_rate": 0.003,
      "loss": 4.3519,
      "step": 1682
    },
    {
      "epoch": 0.01683,
      "grad_norm": 0.830453292981266,
      "learning_rate": 0.003,
      "loss": 4.3671,
      "step": 1683
    },
    {
      "epoch": 0.01684,
      "grad_norm": 0.8911963046678297,
      "learning_rate": 0.003,
      "loss": 4.3454,
      "step": 1684
    },
    {
      "epoch": 0.01685,
      "grad_norm": 0.8195919434333439,
      "learning_rate": 0.003,
      "loss": 4.3594,
      "step": 1685
    },
    {
      "epoch": 0.01686,
      "grad_norm": 0.8629708344208772,
      "learning_rate": 0.003,
      "loss": 4.3459,
      "step": 1686
    },
    {
      "epoch": 0.01687,
      "grad_norm": 0.7564331005412712,
      "learning_rate": 0.003,
      "loss": 4.3417,
      "step": 1687
    },
    {
      "epoch": 0.01688,
      "grad_norm": 0.5556779425751791,
      "learning_rate": 0.003,
      "loss": 4.3178,
      "step": 1688
    },
    {
      "epoch": 0.01689,
      "grad_norm": 0.48730261859047036,
      "learning_rate": 0.003,
      "loss": 4.3272,
      "step": 1689
    },
    {
      "epoch": 0.0169,
      "grad_norm": 0.44730185947220713,
      "learning_rate": 0.003,
      "loss": 4.3108,
      "step": 1690
    },
    {
      "epoch": 0.01691,
      "grad_norm": 0.4517455858191217,
      "learning_rate": 0.003,
      "loss": 4.3185,
      "step": 1691
    },
    {
      "epoch": 0.01692,
      "grad_norm": 0.3859623506260394,
      "learning_rate": 0.003,
      "loss": 4.3134,
      "step": 1692
    },
    {
      "epoch": 0.01693,
      "grad_norm": 0.4280684753336336,
      "learning_rate": 0.003,
      "loss": 4.3374,
      "step": 1693
    },
    {
      "epoch": 0.01694,
      "grad_norm": 0.430827914877835,
      "learning_rate": 0.003,
      "loss": 4.2839,
      "step": 1694
    },
    {
      "epoch": 0.01695,
      "grad_norm": 0.43375392594165474,
      "learning_rate": 0.003,
      "loss": 4.3125,
      "step": 1695
    },
    {
      "epoch": 0.01696,
      "grad_norm": 0.38718080857219184,
      "learning_rate": 0.003,
      "loss": 4.2992,
      "step": 1696
    },
    {
      "epoch": 0.01697,
      "grad_norm": 0.30363972361127406,
      "learning_rate": 0.003,
      "loss": 4.272,
      "step": 1697
    },
    {
      "epoch": 0.01698,
      "grad_norm": 0.31558190433180666,
      "learning_rate": 0.003,
      "loss": 4.2652,
      "step": 1698
    },
    {
      "epoch": 0.01699,
      "grad_norm": 0.3101672448572644,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 1699
    },
    {
      "epoch": 0.017,
      "grad_norm": 0.28571559628804677,
      "learning_rate": 0.003,
      "loss": 4.2885,
      "step": 1700
    },
    {
      "epoch": 0.01701,
      "grad_norm": 0.31584250090208377,
      "learning_rate": 0.003,
      "loss": 4.2694,
      "step": 1701
    },
    {
      "epoch": 0.01702,
      "grad_norm": 0.36960398708639597,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1702
    },
    {
      "epoch": 0.01703,
      "grad_norm": 0.4412523037970288,
      "learning_rate": 0.003,
      "loss": 4.277,
      "step": 1703
    },
    {
      "epoch": 0.01704,
      "grad_norm": 0.6716654424542439,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 1704
    },
    {
      "epoch": 0.01705,
      "grad_norm": 1.1072937714524183,
      "learning_rate": 0.003,
      "loss": 4.2932,
      "step": 1705
    },
    {
      "epoch": 0.01706,
      "grad_norm": 0.8287243662947943,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1706
    },
    {
      "epoch": 0.01707,
      "grad_norm": 0.6387856390419926,
      "learning_rate": 0.003,
      "loss": 4.3079,
      "step": 1707
    },
    {
      "epoch": 0.01708,
      "grad_norm": 0.7665689860690651,
      "learning_rate": 0.003,
      "loss": 4.2925,
      "step": 1708
    },
    {
      "epoch": 0.01709,
      "grad_norm": 0.6842592484083714,
      "learning_rate": 0.003,
      "loss": 4.3099,
      "step": 1709
    },
    {
      "epoch": 0.0171,
      "grad_norm": 0.5640128407150151,
      "learning_rate": 0.003,
      "loss": 4.2554,
      "step": 1710
    },
    {
      "epoch": 0.01711,
      "grad_norm": 0.5706999470034969,
      "learning_rate": 0.003,
      "loss": 4.2824,
      "step": 1711
    },
    {
      "epoch": 0.01712,
      "grad_norm": 0.4795902281641074,
      "learning_rate": 0.003,
      "loss": 4.2884,
      "step": 1712
    },
    {
      "epoch": 0.01713,
      "grad_norm": 0.425849272235493,
      "learning_rate": 0.003,
      "loss": 4.2609,
      "step": 1713
    },
    {
      "epoch": 0.01714,
      "grad_norm": 0.38297971865347935,
      "learning_rate": 0.003,
      "loss": 4.2951,
      "step": 1714
    },
    {
      "epoch": 0.01715,
      "grad_norm": 0.3866137138438359,
      "learning_rate": 0.003,
      "loss": 4.3021,
      "step": 1715
    },
    {
      "epoch": 0.01716,
      "grad_norm": 0.35314982927049987,
      "learning_rate": 0.003,
      "loss": 4.2748,
      "step": 1716
    },
    {
      "epoch": 0.01717,
      "grad_norm": 0.31552154478040295,
      "learning_rate": 0.003,
      "loss": 4.3044,
      "step": 1717
    },
    {
      "epoch": 0.01718,
      "grad_norm": 0.30638930868708425,
      "learning_rate": 0.003,
      "loss": 4.2709,
      "step": 1718
    },
    {
      "epoch": 0.01719,
      "grad_norm": 0.3127248594754382,
      "learning_rate": 0.003,
      "loss": 4.2905,
      "step": 1719
    },
    {
      "epoch": 0.0172,
      "grad_norm": 0.33132313581376,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 1720
    },
    {
      "epoch": 0.01721,
      "grad_norm": 0.3400058323027828,
      "learning_rate": 0.003,
      "loss": 4.2635,
      "step": 1721
    },
    {
      "epoch": 0.01722,
      "grad_norm": 0.3468025060877176,
      "learning_rate": 0.003,
      "loss": 4.2559,
      "step": 1722
    },
    {
      "epoch": 0.01723,
      "grad_norm": 0.35695313771503173,
      "learning_rate": 0.003,
      "loss": 4.2636,
      "step": 1723
    },
    {
      "epoch": 0.01724,
      "grad_norm": 0.34758213513255676,
      "learning_rate": 0.003,
      "loss": 4.2648,
      "step": 1724
    },
    {
      "epoch": 0.01725,
      "grad_norm": 0.40361756493758844,
      "learning_rate": 0.003,
      "loss": 4.2508,
      "step": 1725
    },
    {
      "epoch": 0.01726,
      "grad_norm": 0.5053798954640991,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 1726
    },
    {
      "epoch": 0.01727,
      "grad_norm": 0.6257283402319211,
      "learning_rate": 0.003,
      "loss": 4.2558,
      "step": 1727
    },
    {
      "epoch": 0.01728,
      "grad_norm": 0.5898179215522085,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1728
    },
    {
      "epoch": 0.01729,
      "grad_norm": 0.5313524252553607,
      "learning_rate": 0.003,
      "loss": 4.2643,
      "step": 1729
    },
    {
      "epoch": 0.0173,
      "grad_norm": 0.6103904325056944,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 1730
    },
    {
      "epoch": 0.01731,
      "grad_norm": 0.498773172096663,
      "learning_rate": 0.003,
      "loss": 4.2592,
      "step": 1731
    },
    {
      "epoch": 0.01732,
      "grad_norm": 0.6373537928005123,
      "learning_rate": 0.003,
      "loss": 4.2725,
      "step": 1732
    },
    {
      "epoch": 0.01733,
      "grad_norm": 0.7241473038492131,
      "learning_rate": 0.003,
      "loss": 4.257,
      "step": 1733
    },
    {
      "epoch": 0.01734,
      "grad_norm": 0.8759884069698398,
      "learning_rate": 0.003,
      "loss": 4.2776,
      "step": 1734
    },
    {
      "epoch": 0.01735,
      "grad_norm": 0.8320108473030087,
      "learning_rate": 0.003,
      "loss": 4.2551,
      "step": 1735
    },
    {
      "epoch": 0.01736,
      "grad_norm": 0.6879588721711866,
      "learning_rate": 0.003,
      "loss": 4.2513,
      "step": 1736
    },
    {
      "epoch": 0.01737,
      "grad_norm": 0.6180790162276626,
      "learning_rate": 0.003,
      "loss": 4.2782,
      "step": 1737
    },
    {
      "epoch": 0.01738,
      "grad_norm": 0.6570007295297866,
      "learning_rate": 0.003,
      "loss": 4.2689,
      "step": 1738
    },
    {
      "epoch": 0.01739,
      "grad_norm": 0.7234192244460489,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1739
    },
    {
      "epoch": 0.0174,
      "grad_norm": 0.646718379715682,
      "learning_rate": 0.003,
      "loss": 4.2832,
      "step": 1740
    },
    {
      "epoch": 0.01741,
      "grad_norm": 0.7667295546998241,
      "learning_rate": 0.003,
      "loss": 4.2989,
      "step": 1741
    },
    {
      "epoch": 0.01742,
      "grad_norm": 0.6671673777635381,
      "learning_rate": 0.003,
      "loss": 4.2839,
      "step": 1742
    },
    {
      "epoch": 0.01743,
      "grad_norm": 0.6506146600686363,
      "learning_rate": 0.003,
      "loss": 4.2938,
      "step": 1743
    },
    {
      "epoch": 0.01744,
      "grad_norm": 0.6573560282994302,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1744
    },
    {
      "epoch": 0.01745,
      "grad_norm": 0.640684434992135,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 1745
    },
    {
      "epoch": 0.01746,
      "grad_norm": 0.657793650803233,
      "learning_rate": 0.003,
      "loss": 4.2671,
      "step": 1746
    },
    {
      "epoch": 0.01747,
      "grad_norm": 0.7709605353943576,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1747
    },
    {
      "epoch": 0.01748,
      "grad_norm": 0.8278007400279545,
      "learning_rate": 0.003,
      "loss": 4.2678,
      "step": 1748
    },
    {
      "epoch": 0.01749,
      "grad_norm": 0.7466841943172032,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 1749
    },
    {
      "epoch": 0.0175,
      "grad_norm": 0.763703983777172,
      "learning_rate": 0.003,
      "loss": 4.2821,
      "step": 1750
    },
    {
      "epoch": 0.01751,
      "grad_norm": 0.7380887125320881,
      "learning_rate": 0.003,
      "loss": 4.2741,
      "step": 1751
    },
    {
      "epoch": 0.01752,
      "grad_norm": 0.6415685853882744,
      "learning_rate": 0.003,
      "loss": 4.314,
      "step": 1752
    },
    {
      "epoch": 0.01753,
      "grad_norm": 0.6180937585023022,
      "learning_rate": 0.003,
      "loss": 4.2575,
      "step": 1753
    },
    {
      "epoch": 0.01754,
      "grad_norm": 0.6258159102980642,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 1754
    },
    {
      "epoch": 0.01755,
      "grad_norm": 0.5258294344568889,
      "learning_rate": 0.003,
      "loss": 4.2275,
      "step": 1755
    },
    {
      "epoch": 0.01756,
      "grad_norm": 0.4224710169236117,
      "learning_rate": 0.003,
      "loss": 4.3014,
      "step": 1756
    },
    {
      "epoch": 0.01757,
      "grad_norm": 0.4609603622580685,
      "learning_rate": 0.003,
      "loss": 4.2799,
      "step": 1757
    },
    {
      "epoch": 0.01758,
      "grad_norm": 0.5088646105086777,
      "learning_rate": 0.003,
      "loss": 4.2854,
      "step": 1758
    },
    {
      "epoch": 0.01759,
      "grad_norm": 0.6116749982391423,
      "learning_rate": 0.003,
      "loss": 4.2559,
      "step": 1759
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.6421141472355831,
      "learning_rate": 0.003,
      "loss": 4.2809,
      "step": 1760
    },
    {
      "epoch": 0.01761,
      "grad_norm": 0.683100042759788,
      "learning_rate": 0.003,
      "loss": 4.2664,
      "step": 1761
    },
    {
      "epoch": 0.01762,
      "grad_norm": 0.6507618810640869,
      "learning_rate": 0.003,
      "loss": 4.2868,
      "step": 1762
    },
    {
      "epoch": 0.01763,
      "grad_norm": 0.6268432713329561,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1763
    },
    {
      "epoch": 0.01764,
      "grad_norm": 0.5913417648906798,
      "learning_rate": 0.003,
      "loss": 4.2433,
      "step": 1764
    },
    {
      "epoch": 0.01765,
      "grad_norm": 0.6076223877479996,
      "learning_rate": 0.003,
      "loss": 4.2712,
      "step": 1765
    },
    {
      "epoch": 0.01766,
      "grad_norm": 0.7266026161134381,
      "learning_rate": 0.003,
      "loss": 4.2888,
      "step": 1766
    },
    {
      "epoch": 0.01767,
      "grad_norm": 0.7022487568636244,
      "learning_rate": 0.003,
      "loss": 4.2548,
      "step": 1767
    },
    {
      "epoch": 0.01768,
      "grad_norm": 0.5934615837163861,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1768
    },
    {
      "epoch": 0.01769,
      "grad_norm": 0.4361100704507323,
      "learning_rate": 0.003,
      "loss": 4.2511,
      "step": 1769
    },
    {
      "epoch": 0.0177,
      "grad_norm": 0.48611120007740843,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 1770
    },
    {
      "epoch": 0.01771,
      "grad_norm": 0.45731585839886957,
      "learning_rate": 0.003,
      "loss": 4.266,
      "step": 1771
    },
    {
      "epoch": 0.01772,
      "grad_norm": 0.4581498365355611,
      "learning_rate": 0.003,
      "loss": 4.2403,
      "step": 1772
    },
    {
      "epoch": 0.01773,
      "grad_norm": 0.3835918309901609,
      "learning_rate": 0.003,
      "loss": 4.2284,
      "step": 1773
    },
    {
      "epoch": 0.01774,
      "grad_norm": 0.4161002686293629,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 1774
    },
    {
      "epoch": 0.01775,
      "grad_norm": 0.48314560217147945,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 1775
    },
    {
      "epoch": 0.01776,
      "grad_norm": 0.550180805967614,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 1776
    },
    {
      "epoch": 0.01777,
      "grad_norm": 0.570528755486745,
      "learning_rate": 0.003,
      "loss": 4.2588,
      "step": 1777
    },
    {
      "epoch": 0.01778,
      "grad_norm": 0.6830277922868914,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1778
    },
    {
      "epoch": 0.01779,
      "grad_norm": 0.8262794189445563,
      "learning_rate": 0.003,
      "loss": 4.281,
      "step": 1779
    },
    {
      "epoch": 0.0178,
      "grad_norm": 0.85849663326723,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1780
    },
    {
      "epoch": 0.01781,
      "grad_norm": 0.7027313657081399,
      "learning_rate": 0.003,
      "loss": 4.2685,
      "step": 1781
    },
    {
      "epoch": 0.01782,
      "grad_norm": 0.5620584115129139,
      "learning_rate": 0.003,
      "loss": 4.2878,
      "step": 1782
    },
    {
      "epoch": 0.01783,
      "grad_norm": 0.5096347371759039,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 1783
    },
    {
      "epoch": 0.01784,
      "grad_norm": 0.5297569056185621,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 1784
    },
    {
      "epoch": 0.01785,
      "grad_norm": 0.6034694109958456,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 1785
    },
    {
      "epoch": 0.01786,
      "grad_norm": 0.6309039198553906,
      "learning_rate": 0.003,
      "loss": 4.2565,
      "step": 1786
    },
    {
      "epoch": 0.01787,
      "grad_norm": 0.6879917544738386,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 1787
    },
    {
      "epoch": 0.01788,
      "grad_norm": 0.6818187529366616,
      "learning_rate": 0.003,
      "loss": 4.2789,
      "step": 1788
    },
    {
      "epoch": 0.01789,
      "grad_norm": 0.5930877660332807,
      "learning_rate": 0.003,
      "loss": 4.2445,
      "step": 1789
    },
    {
      "epoch": 0.0179,
      "grad_norm": 0.5104382089041954,
      "learning_rate": 0.003,
      "loss": 4.2605,
      "step": 1790
    },
    {
      "epoch": 0.01791,
      "grad_norm": 0.48343390667073793,
      "learning_rate": 0.003,
      "loss": 4.2706,
      "step": 1791
    },
    {
      "epoch": 0.01792,
      "grad_norm": 0.5502398127361193,
      "learning_rate": 0.003,
      "loss": 4.2742,
      "step": 1792
    },
    {
      "epoch": 0.01793,
      "grad_norm": 0.514623976410524,
      "learning_rate": 0.003,
      "loss": 4.235,
      "step": 1793
    },
    {
      "epoch": 0.01794,
      "grad_norm": 0.5328860529823664,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 1794
    },
    {
      "epoch": 0.01795,
      "grad_norm": 0.5324866930035196,
      "learning_rate": 0.003,
      "loss": 4.2511,
      "step": 1795
    },
    {
      "epoch": 0.01796,
      "grad_norm": 0.4431327306726926,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1796
    },
    {
      "epoch": 0.01797,
      "grad_norm": 0.4163239838007479,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 1797
    },
    {
      "epoch": 0.01798,
      "grad_norm": 0.4680057225647005,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 1798
    },
    {
      "epoch": 0.01799,
      "grad_norm": 0.5257929933308816,
      "learning_rate": 0.003,
      "loss": 4.2569,
      "step": 1799
    },
    {
      "epoch": 0.018,
      "grad_norm": 0.6818533479906342,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1800
    },
    {
      "epoch": 0.01801,
      "grad_norm": 0.8347036722912681,
      "learning_rate": 0.003,
      "loss": 4.2531,
      "step": 1801
    },
    {
      "epoch": 0.01802,
      "grad_norm": 0.8911873404291202,
      "learning_rate": 0.003,
      "loss": 4.2851,
      "step": 1802
    },
    {
      "epoch": 0.01803,
      "grad_norm": 0.8674521299479792,
      "learning_rate": 0.003,
      "loss": 4.2905,
      "step": 1803
    },
    {
      "epoch": 0.01804,
      "grad_norm": 0.8602615225653094,
      "learning_rate": 0.003,
      "loss": 4.2682,
      "step": 1804
    },
    {
      "epoch": 0.01805,
      "grad_norm": 0.8301928044098125,
      "learning_rate": 0.003,
      "loss": 4.2953,
      "step": 1805
    },
    {
      "epoch": 0.01806,
      "grad_norm": 0.7981063012004259,
      "learning_rate": 0.003,
      "loss": 4.2503,
      "step": 1806
    },
    {
      "epoch": 0.01807,
      "grad_norm": 0.8614656914400735,
      "learning_rate": 0.003,
      "loss": 4.2917,
      "step": 1807
    },
    {
      "epoch": 0.01808,
      "grad_norm": 0.7448619096800406,
      "learning_rate": 0.003,
      "loss": 4.2867,
      "step": 1808
    },
    {
      "epoch": 0.01809,
      "grad_norm": 0.6456900848105672,
      "learning_rate": 0.003,
      "loss": 4.2963,
      "step": 1809
    },
    {
      "epoch": 0.0181,
      "grad_norm": 0.5841374638529157,
      "learning_rate": 0.003,
      "loss": 4.2907,
      "step": 1810
    },
    {
      "epoch": 0.01811,
      "grad_norm": 0.5584017956424331,
      "learning_rate": 0.003,
      "loss": 4.2718,
      "step": 1811
    },
    {
      "epoch": 0.01812,
      "grad_norm": 0.5780683839817958,
      "learning_rate": 0.003,
      "loss": 4.2546,
      "step": 1812
    },
    {
      "epoch": 0.01813,
      "grad_norm": 0.5210766747257082,
      "learning_rate": 0.003,
      "loss": 4.298,
      "step": 1813
    },
    {
      "epoch": 0.01814,
      "grad_norm": 0.5061687855504291,
      "learning_rate": 0.003,
      "loss": 4.2617,
      "step": 1814
    },
    {
      "epoch": 0.01815,
      "grad_norm": 0.5646752857396716,
      "learning_rate": 0.003,
      "loss": 4.2497,
      "step": 1815
    },
    {
      "epoch": 0.01816,
      "grad_norm": 0.6329676204068563,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1816
    },
    {
      "epoch": 0.01817,
      "grad_norm": 0.576517926530056,
      "learning_rate": 0.003,
      "loss": 4.2664,
      "step": 1817
    },
    {
      "epoch": 0.01818,
      "grad_norm": 0.4950065288657199,
      "learning_rate": 0.003,
      "loss": 4.2615,
      "step": 1818
    },
    {
      "epoch": 0.01819,
      "grad_norm": 0.5161367831332595,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 1819
    },
    {
      "epoch": 0.0182,
      "grad_norm": 0.5814618735674658,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 1820
    },
    {
      "epoch": 0.01821,
      "grad_norm": 0.556938739734867,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 1821
    },
    {
      "epoch": 0.01822,
      "grad_norm": 0.5708870221750882,
      "learning_rate": 0.003,
      "loss": 4.2636,
      "step": 1822
    },
    {
      "epoch": 0.01823,
      "grad_norm": 0.5398260260215829,
      "learning_rate": 0.003,
      "loss": 4.2764,
      "step": 1823
    },
    {
      "epoch": 0.01824,
      "grad_norm": 0.5454416785979428,
      "learning_rate": 0.003,
      "loss": 4.2571,
      "step": 1824
    },
    {
      "epoch": 0.01825,
      "grad_norm": 0.6656252297688491,
      "learning_rate": 0.003,
      "loss": 4.2668,
      "step": 1825
    },
    {
      "epoch": 0.01826,
      "grad_norm": 0.9269512928572629,
      "learning_rate": 0.003,
      "loss": 4.289,
      "step": 1826
    },
    {
      "epoch": 0.01827,
      "grad_norm": 1.076217745090645,
      "learning_rate": 0.003,
      "loss": 4.2819,
      "step": 1827
    },
    {
      "epoch": 0.01828,
      "grad_norm": 0.650029561792675,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1828
    },
    {
      "epoch": 0.01829,
      "grad_norm": 0.5962218186152778,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1829
    },
    {
      "epoch": 0.0183,
      "grad_norm": 0.6672932732108263,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1830
    },
    {
      "epoch": 0.01831,
      "grad_norm": 0.6051971108876394,
      "learning_rate": 0.003,
      "loss": 4.2699,
      "step": 1831
    },
    {
      "epoch": 0.01832,
      "grad_norm": 0.5439007581448924,
      "learning_rate": 0.003,
      "loss": 4.2598,
      "step": 1832
    },
    {
      "epoch": 0.01833,
      "grad_norm": 0.55902144104146,
      "learning_rate": 0.003,
      "loss": 4.2938,
      "step": 1833
    },
    {
      "epoch": 0.01834,
      "grad_norm": 0.5728509198658349,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1834
    },
    {
      "epoch": 0.01835,
      "grad_norm": 0.5340971547409412,
      "learning_rate": 0.003,
      "loss": 4.2588,
      "step": 1835
    },
    {
      "epoch": 0.01836,
      "grad_norm": 0.47778636042203493,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 1836
    },
    {
      "epoch": 0.01837,
      "grad_norm": 0.4158033301420617,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 1837
    },
    {
      "epoch": 0.01838,
      "grad_norm": 0.5117944575544118,
      "learning_rate": 0.003,
      "loss": 4.2549,
      "step": 1838
    },
    {
      "epoch": 0.01839,
      "grad_norm": 0.6067550068103876,
      "learning_rate": 0.003,
      "loss": 4.2791,
      "step": 1839
    },
    {
      "epoch": 0.0184,
      "grad_norm": 0.6159872555117356,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 1840
    },
    {
      "epoch": 0.01841,
      "grad_norm": 0.5609724495757069,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 1841
    },
    {
      "epoch": 0.01842,
      "grad_norm": 0.5051094597061737,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 1842
    },
    {
      "epoch": 0.01843,
      "grad_norm": 0.5100819064264481,
      "learning_rate": 0.003,
      "loss": 4.228,
      "step": 1843
    },
    {
      "epoch": 0.01844,
      "grad_norm": 0.5582900536993392,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 1844
    },
    {
      "epoch": 0.01845,
      "grad_norm": 0.6111866156562342,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 1845
    },
    {
      "epoch": 0.01846,
      "grad_norm": 0.6619002433974149,
      "learning_rate": 0.003,
      "loss": 4.2608,
      "step": 1846
    },
    {
      "epoch": 0.01847,
      "grad_norm": 1.0200537693834115,
      "learning_rate": 0.003,
      "loss": 4.2703,
      "step": 1847
    },
    {
      "epoch": 0.01848,
      "grad_norm": 1.0513050095404686,
      "learning_rate": 0.003,
      "loss": 4.3125,
      "step": 1848
    },
    {
      "epoch": 0.01849,
      "grad_norm": 0.8927900194204563,
      "learning_rate": 0.003,
      "loss": 4.3063,
      "step": 1849
    },
    {
      "epoch": 0.0185,
      "grad_norm": 0.9502466432757984,
      "learning_rate": 0.003,
      "loss": 4.316,
      "step": 1850
    },
    {
      "epoch": 0.01851,
      "grad_norm": 0.9099370930692962,
      "learning_rate": 0.003,
      "loss": 4.2922,
      "step": 1851
    },
    {
      "epoch": 0.01852,
      "grad_norm": 0.7734254610813497,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1852
    },
    {
      "epoch": 0.01853,
      "grad_norm": 0.7300466041466253,
      "learning_rate": 0.003,
      "loss": 4.2883,
      "step": 1853
    },
    {
      "epoch": 0.01854,
      "grad_norm": 0.7064556599969221,
      "learning_rate": 0.003,
      "loss": 4.319,
      "step": 1854
    },
    {
      "epoch": 0.01855,
      "grad_norm": 0.5941104488816943,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1855
    },
    {
      "epoch": 0.01856,
      "grad_norm": 0.5639252296572004,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 1856
    },
    {
      "epoch": 0.01857,
      "grad_norm": 0.5497157641923732,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 1857
    },
    {
      "epoch": 0.01858,
      "grad_norm": 0.5999940363832145,
      "learning_rate": 0.003,
      "loss": 4.2675,
      "step": 1858
    },
    {
      "epoch": 0.01859,
      "grad_norm": 0.6524431060775852,
      "learning_rate": 0.003,
      "loss": 4.2673,
      "step": 1859
    },
    {
      "epoch": 0.0186,
      "grad_norm": 0.5786076880789475,
      "learning_rate": 0.003,
      "loss": 4.2435,
      "step": 1860
    },
    {
      "epoch": 0.01861,
      "grad_norm": 0.5737732368480609,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1861
    },
    {
      "epoch": 0.01862,
      "grad_norm": 0.6266977494361998,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1862
    },
    {
      "epoch": 0.01863,
      "grad_norm": 0.5490443064279302,
      "learning_rate": 0.003,
      "loss": 4.2575,
      "step": 1863
    },
    {
      "epoch": 0.01864,
      "grad_norm": 0.4951546054862347,
      "learning_rate": 0.003,
      "loss": 4.2885,
      "step": 1864
    },
    {
      "epoch": 0.01865,
      "grad_norm": 0.4870782718763887,
      "learning_rate": 0.003,
      "loss": 4.2626,
      "step": 1865
    },
    {
      "epoch": 0.01866,
      "grad_norm": 0.46735035399196706,
      "learning_rate": 0.003,
      "loss": 4.2603,
      "step": 1866
    },
    {
      "epoch": 0.01867,
      "grad_norm": 0.4716752041768468,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 1867
    },
    {
      "epoch": 0.01868,
      "grad_norm": 0.44705830429737475,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 1868
    },
    {
      "epoch": 0.01869,
      "grad_norm": 0.4991544420243406,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1869
    },
    {
      "epoch": 0.0187,
      "grad_norm": 0.4942905986952634,
      "learning_rate": 0.003,
      "loss": 4.2547,
      "step": 1870
    },
    {
      "epoch": 0.01871,
      "grad_norm": 0.5121508182199742,
      "learning_rate": 0.003,
      "loss": 4.2287,
      "step": 1871
    },
    {
      "epoch": 0.01872,
      "grad_norm": 0.5132249094210571,
      "learning_rate": 0.003,
      "loss": 4.2747,
      "step": 1872
    },
    {
      "epoch": 0.01873,
      "grad_norm": 0.45833884531988217,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 1873
    },
    {
      "epoch": 0.01874,
      "grad_norm": 0.4075899028314144,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1874
    },
    {
      "epoch": 0.01875,
      "grad_norm": 0.4250746194260601,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 1875
    },
    {
      "epoch": 0.01876,
      "grad_norm": 0.4231054620615194,
      "learning_rate": 0.003,
      "loss": 4.2577,
      "step": 1876
    },
    {
      "epoch": 0.01877,
      "grad_norm": 0.39967365013750716,
      "learning_rate": 0.003,
      "loss": 4.2584,
      "step": 1877
    },
    {
      "epoch": 0.01878,
      "grad_norm": 0.4152490558819513,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 1878
    },
    {
      "epoch": 0.01879,
      "grad_norm": 0.42538683613055095,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 1879
    },
    {
      "epoch": 0.0188,
      "grad_norm": 0.44463665765346116,
      "learning_rate": 0.003,
      "loss": 4.2318,
      "step": 1880
    },
    {
      "epoch": 0.01881,
      "grad_norm": 0.4351959985696958,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 1881
    },
    {
      "epoch": 0.01882,
      "grad_norm": 0.5024559892307489,
      "learning_rate": 0.003,
      "loss": 4.2542,
      "step": 1882
    },
    {
      "epoch": 0.01883,
      "grad_norm": 0.6063994875754019,
      "learning_rate": 0.003,
      "loss": 4.2323,
      "step": 1883
    },
    {
      "epoch": 0.01884,
      "grad_norm": 0.7335676135390363,
      "learning_rate": 0.003,
      "loss": 4.2381,
      "step": 1884
    },
    {
      "epoch": 0.01885,
      "grad_norm": 0.9057741844151627,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 1885
    },
    {
      "epoch": 0.01886,
      "grad_norm": 1.0977669981210454,
      "learning_rate": 0.003,
      "loss": 4.2863,
      "step": 1886
    },
    {
      "epoch": 0.01887,
      "grad_norm": 0.6506964260225092,
      "learning_rate": 0.003,
      "loss": 4.2622,
      "step": 1887
    },
    {
      "epoch": 0.01888,
      "grad_norm": 0.7018734791941831,
      "learning_rate": 0.003,
      "loss": 4.2552,
      "step": 1888
    },
    {
      "epoch": 0.01889,
      "grad_norm": 0.6726610800126339,
      "learning_rate": 0.003,
      "loss": 4.2655,
      "step": 1889
    },
    {
      "epoch": 0.0189,
      "grad_norm": 0.7428766906603239,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 1890
    },
    {
      "epoch": 0.01891,
      "grad_norm": 0.684421659513949,
      "learning_rate": 0.003,
      "loss": 4.2969,
      "step": 1891
    },
    {
      "epoch": 0.01892,
      "grad_norm": 0.6909566019627704,
      "learning_rate": 0.003,
      "loss": 4.2744,
      "step": 1892
    },
    {
      "epoch": 0.01893,
      "grad_norm": 0.6520197616765968,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 1893
    },
    {
      "epoch": 0.01894,
      "grad_norm": 0.6739359420697347,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1894
    },
    {
      "epoch": 0.01895,
      "grad_norm": 0.697458296569538,
      "learning_rate": 0.003,
      "loss": 4.2498,
      "step": 1895
    },
    {
      "epoch": 0.01896,
      "grad_norm": 0.5936568619706635,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1896
    },
    {
      "epoch": 0.01897,
      "grad_norm": 0.5494055764115975,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1897
    },
    {
      "epoch": 0.01898,
      "grad_norm": 0.5676106882870927,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 1898
    },
    {
      "epoch": 0.01899,
      "grad_norm": 0.6101424369862174,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 1899
    },
    {
      "epoch": 0.019,
      "grad_norm": 0.6545357483725045,
      "learning_rate": 0.003,
      "loss": 4.2443,
      "step": 1900
    },
    {
      "epoch": 0.01901,
      "grad_norm": 0.6221145886969454,
      "learning_rate": 0.003,
      "loss": 4.29,
      "step": 1901
    },
    {
      "epoch": 0.01902,
      "grad_norm": 0.47771631002661874,
      "learning_rate": 0.003,
      "loss": 4.2605,
      "step": 1902
    },
    {
      "epoch": 0.01903,
      "grad_norm": 0.503444994520635,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 1903
    },
    {
      "epoch": 0.01904,
      "grad_norm": 0.5071441864936951,
      "learning_rate": 0.003,
      "loss": 4.2711,
      "step": 1904
    },
    {
      "epoch": 0.01905,
      "grad_norm": 0.5754503645956631,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 1905
    },
    {
      "epoch": 0.01906,
      "grad_norm": 0.6654408282834325,
      "learning_rate": 0.003,
      "loss": 4.2694,
      "step": 1906
    },
    {
      "epoch": 0.01907,
      "grad_norm": 0.7906613760805888,
      "learning_rate": 0.003,
      "loss": 4.2662,
      "step": 1907
    },
    {
      "epoch": 0.01908,
      "grad_norm": 0.8850433480932498,
      "learning_rate": 0.003,
      "loss": 4.2906,
      "step": 1908
    },
    {
      "epoch": 0.01909,
      "grad_norm": 0.806456676126049,
      "learning_rate": 0.003,
      "loss": 4.2573,
      "step": 1909
    },
    {
      "epoch": 0.0191,
      "grad_norm": 0.5544806973924623,
      "learning_rate": 0.003,
      "loss": 4.256,
      "step": 1910
    },
    {
      "epoch": 0.01911,
      "grad_norm": 0.6152192065162949,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 1911
    },
    {
      "epoch": 0.01912,
      "grad_norm": 0.5774523882886098,
      "learning_rate": 0.003,
      "loss": 4.2416,
      "step": 1912
    },
    {
      "epoch": 0.01913,
      "grad_norm": 0.5526940825090116,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 1913
    },
    {
      "epoch": 0.01914,
      "grad_norm": 0.5202691625530954,
      "learning_rate": 0.003,
      "loss": 4.2739,
      "step": 1914
    },
    {
      "epoch": 0.01915,
      "grad_norm": 0.4828726686790076,
      "learning_rate": 0.003,
      "loss": 4.257,
      "step": 1915
    },
    {
      "epoch": 0.01916,
      "grad_norm": 0.4684476049653884,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 1916
    },
    {
      "epoch": 0.01917,
      "grad_norm": 0.47660708071333646,
      "learning_rate": 0.003,
      "loss": 4.2552,
      "step": 1917
    },
    {
      "epoch": 0.01918,
      "grad_norm": 0.4979902159298128,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 1918
    },
    {
      "epoch": 0.01919,
      "grad_norm": 0.4677465287869337,
      "learning_rate": 0.003,
      "loss": 4.2412,
      "step": 1919
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.42528920687498767,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 1920
    },
    {
      "epoch": 0.01921,
      "grad_norm": 0.4717244787464016,
      "learning_rate": 0.003,
      "loss": 4.2309,
      "step": 1921
    },
    {
      "epoch": 0.01922,
      "grad_norm": 0.4420806610942072,
      "learning_rate": 0.003,
      "loss": 4.2448,
      "step": 1922
    },
    {
      "epoch": 0.01923,
      "grad_norm": 0.42058067854047654,
      "learning_rate": 0.003,
      "loss": 4.2491,
      "step": 1923
    },
    {
      "epoch": 0.01924,
      "grad_norm": 0.3467314160541135,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 1924
    },
    {
      "epoch": 0.01925,
      "grad_norm": 0.3293466925660612,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 1925
    },
    {
      "epoch": 0.01926,
      "grad_norm": 0.41102145828379955,
      "learning_rate": 0.003,
      "loss": 4.2416,
      "step": 1926
    },
    {
      "epoch": 0.01927,
      "grad_norm": 0.720656359870687,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 1927
    },
    {
      "epoch": 0.01928,
      "grad_norm": 1.1844212503280394,
      "learning_rate": 0.003,
      "loss": 4.249,
      "step": 1928
    },
    {
      "epoch": 0.01929,
      "grad_norm": 0.9144474201054079,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 1929
    },
    {
      "epoch": 0.0193,
      "grad_norm": 0.8894910892570072,
      "learning_rate": 0.003,
      "loss": 4.2586,
      "step": 1930
    },
    {
      "epoch": 0.01931,
      "grad_norm": 1.0125767880809944,
      "learning_rate": 0.003,
      "loss": 4.2809,
      "step": 1931
    },
    {
      "epoch": 0.01932,
      "grad_norm": 0.6957841268471371,
      "learning_rate": 0.003,
      "loss": 4.2689,
      "step": 1932
    },
    {
      "epoch": 0.01933,
      "grad_norm": 0.967902005641644,
      "learning_rate": 0.003,
      "loss": 4.2719,
      "step": 1933
    },
    {
      "epoch": 0.01934,
      "grad_norm": 0.9288023682490801,
      "learning_rate": 0.003,
      "loss": 4.2872,
      "step": 1934
    },
    {
      "epoch": 0.01935,
      "grad_norm": 0.9414019337954325,
      "learning_rate": 0.003,
      "loss": 4.3063,
      "step": 1935
    },
    {
      "epoch": 0.01936,
      "grad_norm": 0.9694221827159286,
      "learning_rate": 0.003,
      "loss": 4.2962,
      "step": 1936
    },
    {
      "epoch": 0.01937,
      "grad_norm": 1.001938596118208,
      "learning_rate": 0.003,
      "loss": 4.3121,
      "step": 1937
    },
    {
      "epoch": 0.01938,
      "grad_norm": 1.0140385438058088,
      "learning_rate": 0.003,
      "loss": 4.3241,
      "step": 1938
    },
    {
      "epoch": 0.01939,
      "grad_norm": 1.2235452961348985,
      "learning_rate": 0.003,
      "loss": 4.3646,
      "step": 1939
    },
    {
      "epoch": 0.0194,
      "grad_norm": 0.8846695796998231,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1940
    },
    {
      "epoch": 0.01941,
      "grad_norm": 0.8286594640280003,
      "learning_rate": 0.003,
      "loss": 4.3066,
      "step": 1941
    },
    {
      "epoch": 0.01942,
      "grad_norm": 0.7123881249011045,
      "learning_rate": 0.003,
      "loss": 4.3226,
      "step": 1942
    },
    {
      "epoch": 0.01943,
      "grad_norm": 0.6089638608494183,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1943
    },
    {
      "epoch": 0.01944,
      "grad_norm": 0.5712540744847615,
      "learning_rate": 0.003,
      "loss": 4.3258,
      "step": 1944
    },
    {
      "epoch": 0.01945,
      "grad_norm": 0.5597808479773261,
      "learning_rate": 0.003,
      "loss": 4.2885,
      "step": 1945
    },
    {
      "epoch": 0.01946,
      "grad_norm": 0.5954089244059145,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 1946
    },
    {
      "epoch": 0.01947,
      "grad_norm": 0.5594072113001303,
      "learning_rate": 0.003,
      "loss": 4.2869,
      "step": 1947
    },
    {
      "epoch": 0.01948,
      "grad_norm": 0.47470802243769694,
      "learning_rate": 0.003,
      "loss": 4.3012,
      "step": 1948
    },
    {
      "epoch": 0.01949,
      "grad_norm": 0.5123315729723936,
      "learning_rate": 0.003,
      "loss": 4.2904,
      "step": 1949
    },
    {
      "epoch": 0.0195,
      "grad_norm": 0.5068647256374985,
      "learning_rate": 0.003,
      "loss": 4.2768,
      "step": 1950
    },
    {
      "epoch": 0.01951,
      "grad_norm": 0.4942238546854525,
      "learning_rate": 0.003,
      "loss": 4.2687,
      "step": 1951
    },
    {
      "epoch": 0.01952,
      "grad_norm": 0.4526189118950263,
      "learning_rate": 0.003,
      "loss": 4.271,
      "step": 1952
    },
    {
      "epoch": 0.01953,
      "grad_norm": 0.4370221049383596,
      "learning_rate": 0.003,
      "loss": 4.275,
      "step": 1953
    },
    {
      "epoch": 0.01954,
      "grad_norm": 0.46324265238758683,
      "learning_rate": 0.003,
      "loss": 4.2733,
      "step": 1954
    },
    {
      "epoch": 0.01955,
      "grad_norm": 0.5071666518202474,
      "learning_rate": 0.003,
      "loss": 4.2654,
      "step": 1955
    },
    {
      "epoch": 0.01956,
      "grad_norm": 0.6181690576282608,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1956
    },
    {
      "epoch": 0.01957,
      "grad_norm": 0.5796028080408675,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 1957
    },
    {
      "epoch": 0.01958,
      "grad_norm": 0.41315504837285055,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 1958
    },
    {
      "epoch": 0.01959,
      "grad_norm": 0.3341067438397588,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 1959
    },
    {
      "epoch": 0.0196,
      "grad_norm": 0.3891099796372788,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1960
    },
    {
      "epoch": 0.01961,
      "grad_norm": 0.43000625586768587,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1961
    },
    {
      "epoch": 0.01962,
      "grad_norm": 0.43925709381263844,
      "learning_rate": 0.003,
      "loss": 4.232,
      "step": 1962
    },
    {
      "epoch": 0.01963,
      "grad_norm": 0.42981173506186865,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 1963
    },
    {
      "epoch": 0.01964,
      "grad_norm": 0.4097996230629943,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 1964
    },
    {
      "epoch": 0.01965,
      "grad_norm": 0.3678222300164623,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 1965
    },
    {
      "epoch": 0.01966,
      "grad_norm": 0.3826622250643601,
      "learning_rate": 0.003,
      "loss": 4.2372,
      "step": 1966
    },
    {
      "epoch": 0.01967,
      "grad_norm": 0.40760983332927614,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 1967
    },
    {
      "epoch": 0.01968,
      "grad_norm": 0.47242767064200136,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 1968
    },
    {
      "epoch": 0.01969,
      "grad_norm": 0.588742886193704,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 1969
    },
    {
      "epoch": 0.0197,
      "grad_norm": 0.7240833755896967,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 1970
    },
    {
      "epoch": 0.01971,
      "grad_norm": 0.7681447765560055,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1971
    },
    {
      "epoch": 0.01972,
      "grad_norm": 0.6596049230522953,
      "learning_rate": 0.003,
      "loss": 4.2329,
      "step": 1972
    },
    {
      "epoch": 0.01973,
      "grad_norm": 0.5971490272128623,
      "learning_rate": 0.003,
      "loss": 4.2129,
      "step": 1973
    },
    {
      "epoch": 0.01974,
      "grad_norm": 0.5788210723943882,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 1974
    },
    {
      "epoch": 0.01975,
      "grad_norm": 0.5716517631988168,
      "learning_rate": 0.003,
      "loss": 4.2434,
      "step": 1975
    },
    {
      "epoch": 0.01976,
      "grad_norm": 0.5714647695261786,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 1976
    },
    {
      "epoch": 0.01977,
      "grad_norm": 0.4672772924879757,
      "learning_rate": 0.003,
      "loss": 4.2626,
      "step": 1977
    },
    {
      "epoch": 0.01978,
      "grad_norm": 0.4919465044055173,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1978
    },
    {
      "epoch": 0.01979,
      "grad_norm": 0.5361240818371313,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 1979
    },
    {
      "epoch": 0.0198,
      "grad_norm": 0.6323418711078611,
      "learning_rate": 0.003,
      "loss": 4.2334,
      "step": 1980
    },
    {
      "epoch": 0.01981,
      "grad_norm": 0.6568538387899918,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 1981
    },
    {
      "epoch": 0.01982,
      "grad_norm": 0.5202437330212393,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 1982
    },
    {
      "epoch": 0.01983,
      "grad_norm": 0.50161503607477,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 1983
    },
    {
      "epoch": 0.01984,
      "grad_norm": 0.6088039267450153,
      "learning_rate": 0.003,
      "loss": 4.2476,
      "step": 1984
    },
    {
      "epoch": 0.01985,
      "grad_norm": 0.8039703319398841,
      "learning_rate": 0.003,
      "loss": 4.247,
      "step": 1985
    },
    {
      "epoch": 0.01986,
      "grad_norm": 1.0279100079228762,
      "learning_rate": 0.003,
      "loss": 4.2896,
      "step": 1986
    },
    {
      "epoch": 0.01987,
      "grad_norm": 0.8414880380409095,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 1987
    },
    {
      "epoch": 0.01988,
      "grad_norm": 0.8182097035427517,
      "learning_rate": 0.003,
      "loss": 4.2565,
      "step": 1988
    },
    {
      "epoch": 0.01989,
      "grad_norm": 0.7584478804000142,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 1989
    },
    {
      "epoch": 0.0199,
      "grad_norm": 0.6298670596846694,
      "learning_rate": 0.003,
      "loss": 4.2542,
      "step": 1990
    },
    {
      "epoch": 0.01991,
      "grad_norm": 0.5918495849768016,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 1991
    },
    {
      "epoch": 0.01992,
      "grad_norm": 0.6341092964743213,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1992
    },
    {
      "epoch": 0.01993,
      "grad_norm": 0.690814419492087,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 1993
    },
    {
      "epoch": 0.01994,
      "grad_norm": 0.738327319887457,
      "learning_rate": 0.003,
      "loss": 4.2804,
      "step": 1994
    },
    {
      "epoch": 0.01995,
      "grad_norm": 0.9314757846548688,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1995
    },
    {
      "epoch": 0.01996,
      "grad_norm": 0.9849485592924706,
      "learning_rate": 0.003,
      "loss": 4.2595,
      "step": 1996
    },
    {
      "epoch": 0.01997,
      "grad_norm": 0.8581365774503161,
      "learning_rate": 0.003,
      "loss": 4.2889,
      "step": 1997
    },
    {
      "epoch": 0.01998,
      "grad_norm": 0.824906277718389,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1998
    },
    {
      "epoch": 0.01999,
      "grad_norm": 0.8642378679098298,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1999
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7257405855967545,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 2000
    },
    {
      "epoch": 0.02001,
      "grad_norm": 0.7164159344618267,
      "learning_rate": 0.003,
      "loss": 4.2892,
      "step": 2001
    },
    {
      "epoch": 0.02002,
      "grad_norm": 0.7067957991027757,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 2002
    },
    {
      "epoch": 0.02003,
      "grad_norm": 0.7144493226672485,
      "learning_rate": 0.003,
      "loss": 4.2802,
      "step": 2003
    },
    {
      "epoch": 0.02004,
      "grad_norm": 0.8392479210866416,
      "learning_rate": 0.003,
      "loss": 4.2749,
      "step": 2004
    },
    {
      "epoch": 0.02005,
      "grad_norm": 0.7896031315305061,
      "learning_rate": 0.003,
      "loss": 4.2792,
      "step": 2005
    },
    {
      "epoch": 0.02006,
      "grad_norm": 0.7812166363163485,
      "learning_rate": 0.003,
      "loss": 4.2791,
      "step": 2006
    },
    {
      "epoch": 0.02007,
      "grad_norm": 0.720616320355632,
      "learning_rate": 0.003,
      "loss": 4.2906,
      "step": 2007
    },
    {
      "epoch": 0.02008,
      "grad_norm": 0.7275037377798674,
      "learning_rate": 0.003,
      "loss": 4.2831,
      "step": 2008
    },
    {
      "epoch": 0.02009,
      "grad_norm": 0.8040540578216816,
      "learning_rate": 0.003,
      "loss": 4.2757,
      "step": 2009
    },
    {
      "epoch": 0.0201,
      "grad_norm": 0.8028591997627386,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 2010
    },
    {
      "epoch": 0.02011,
      "grad_norm": 0.7203630829267829,
      "learning_rate": 0.003,
      "loss": 4.2738,
      "step": 2011
    },
    {
      "epoch": 0.02012,
      "grad_norm": 0.6305312780193182,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 2012
    },
    {
      "epoch": 0.02013,
      "grad_norm": 0.4989290466950007,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 2013
    },
    {
      "epoch": 0.02014,
      "grad_norm": 0.5235253503408636,
      "learning_rate": 0.003,
      "loss": 4.2549,
      "step": 2014
    },
    {
      "epoch": 0.02015,
      "grad_norm": 0.49148432915624135,
      "learning_rate": 0.003,
      "loss": 4.2456,
      "step": 2015
    },
    {
      "epoch": 0.02016,
      "grad_norm": 0.5171965972543717,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 2016
    },
    {
      "epoch": 0.02017,
      "grad_norm": 0.48975410612827497,
      "learning_rate": 0.003,
      "loss": 4.2428,
      "step": 2017
    },
    {
      "epoch": 0.02018,
      "grad_norm": 0.532740553989744,
      "learning_rate": 0.003,
      "loss": 4.2458,
      "step": 2018
    },
    {
      "epoch": 0.02019,
      "grad_norm": 0.48750173027142363,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 2019
    },
    {
      "epoch": 0.0202,
      "grad_norm": 0.4089724033791691,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2020
    },
    {
      "epoch": 0.02021,
      "grad_norm": 0.4207487504513156,
      "learning_rate": 0.003,
      "loss": 4.265,
      "step": 2021
    },
    {
      "epoch": 0.02022,
      "grad_norm": 0.4057461757528402,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2022
    },
    {
      "epoch": 0.02023,
      "grad_norm": 0.3580229208354821,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2023
    },
    {
      "epoch": 0.02024,
      "grad_norm": 0.3567498277820771,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2024
    },
    {
      "epoch": 0.02025,
      "grad_norm": 0.43295354825834576,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 2025
    },
    {
      "epoch": 0.02026,
      "grad_norm": 0.4300502575103401,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2026
    },
    {
      "epoch": 0.02027,
      "grad_norm": 0.5507801143322275,
      "learning_rate": 0.003,
      "loss": 4.2429,
      "step": 2027
    },
    {
      "epoch": 0.02028,
      "grad_norm": 0.6603955599139062,
      "learning_rate": 0.003,
      "loss": 4.2319,
      "step": 2028
    },
    {
      "epoch": 0.02029,
      "grad_norm": 0.7462999560296353,
      "learning_rate": 0.003,
      "loss": 4.2391,
      "step": 2029
    },
    {
      "epoch": 0.0203,
      "grad_norm": 0.6312368220072189,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2030
    },
    {
      "epoch": 0.02031,
      "grad_norm": 0.5002880523213,
      "learning_rate": 0.003,
      "loss": 4.2308,
      "step": 2031
    },
    {
      "epoch": 0.02032,
      "grad_norm": 0.7101695088498386,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2032
    },
    {
      "epoch": 0.02033,
      "grad_norm": 0.7026230897172011,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 2033
    },
    {
      "epoch": 0.02034,
      "grad_norm": 0.4672671459141678,
      "learning_rate": 0.003,
      "loss": 4.2396,
      "step": 2034
    },
    {
      "epoch": 0.02035,
      "grad_norm": 0.5167562590378834,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2035
    },
    {
      "epoch": 0.02036,
      "grad_norm": 0.5204324719963792,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2036
    },
    {
      "epoch": 0.02037,
      "grad_norm": 0.44171452211516693,
      "learning_rate": 0.003,
      "loss": 4.2379,
      "step": 2037
    },
    {
      "epoch": 0.02038,
      "grad_norm": 0.443929190404615,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 2038
    },
    {
      "epoch": 0.02039,
      "grad_norm": 0.48540831581824334,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2039
    },
    {
      "epoch": 0.0204,
      "grad_norm": 0.5410457621233177,
      "learning_rate": 0.003,
      "loss": 4.236,
      "step": 2040
    },
    {
      "epoch": 0.02041,
      "grad_norm": 0.5544368623280378,
      "learning_rate": 0.003,
      "loss": 4.1991,
      "step": 2041
    },
    {
      "epoch": 0.02042,
      "grad_norm": 0.5894517852974022,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2042
    },
    {
      "epoch": 0.02043,
      "grad_norm": 0.6318863235101739,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 2043
    },
    {
      "epoch": 0.02044,
      "grad_norm": 0.6044374925441985,
      "learning_rate": 0.003,
      "loss": 4.2207,
      "step": 2044
    },
    {
      "epoch": 0.02045,
      "grad_norm": 0.5638648488316749,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 2045
    },
    {
      "epoch": 0.02046,
      "grad_norm": 0.514843677649273,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2046
    },
    {
      "epoch": 0.02047,
      "grad_norm": 0.4850628116632027,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 2047
    },
    {
      "epoch": 0.02048,
      "grad_norm": 0.5067500543661038,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2048
    },
    {
      "epoch": 0.02049,
      "grad_norm": 0.5005632559023494,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 2049
    },
    {
      "epoch": 0.0205,
      "grad_norm": 0.5338842758485439,
      "learning_rate": 0.003,
      "loss": 4.2016,
      "step": 2050
    },
    {
      "epoch": 0.02051,
      "grad_norm": 0.5673366342029961,
      "learning_rate": 0.003,
      "loss": 4.2056,
      "step": 2051
    },
    {
      "epoch": 0.02052,
      "grad_norm": 0.6208513233991159,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 2052
    },
    {
      "epoch": 0.02053,
      "grad_norm": 0.7925288751630323,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2053
    },
    {
      "epoch": 0.02054,
      "grad_norm": 0.9689478763760375,
      "learning_rate": 0.003,
      "loss": 4.2703,
      "step": 2054
    },
    {
      "epoch": 0.02055,
      "grad_norm": 0.9174532624746199,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2055
    },
    {
      "epoch": 0.02056,
      "grad_norm": 0.7759474219779441,
      "learning_rate": 0.003,
      "loss": 4.2289,
      "step": 2056
    },
    {
      "epoch": 0.02057,
      "grad_norm": 0.6855719086766776,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 2057
    },
    {
      "epoch": 0.02058,
      "grad_norm": 0.664344253711328,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2058
    },
    {
      "epoch": 0.02059,
      "grad_norm": 0.7126024175545171,
      "learning_rate": 0.003,
      "loss": 4.2205,
      "step": 2059
    },
    {
      "epoch": 0.0206,
      "grad_norm": 0.8304297596763678,
      "learning_rate": 0.003,
      "loss": 4.2739,
      "step": 2060
    },
    {
      "epoch": 0.02061,
      "grad_norm": 0.8055338565488124,
      "learning_rate": 0.003,
      "loss": 4.2502,
      "step": 2061
    },
    {
      "epoch": 0.02062,
      "grad_norm": 0.7656045427907143,
      "learning_rate": 0.003,
      "loss": 4.2474,
      "step": 2062
    },
    {
      "epoch": 0.02063,
      "grad_norm": 0.7423249019964357,
      "learning_rate": 0.003,
      "loss": 4.2458,
      "step": 2063
    },
    {
      "epoch": 0.02064,
      "grad_norm": 0.7236715566519402,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2064
    },
    {
      "epoch": 0.02065,
      "grad_norm": 0.7531415944255598,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2065
    },
    {
      "epoch": 0.02066,
      "grad_norm": 0.6867969137582566,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2066
    },
    {
      "epoch": 0.02067,
      "grad_norm": 0.5938985343647097,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 2067
    },
    {
      "epoch": 0.02068,
      "grad_norm": 0.4863087277096077,
      "learning_rate": 0.003,
      "loss": 4.2209,
      "step": 2068
    },
    {
      "epoch": 0.02069,
      "grad_norm": 0.4635605849081945,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2069
    },
    {
      "epoch": 0.0207,
      "grad_norm": 0.5457323036329618,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2070
    },
    {
      "epoch": 0.02071,
      "grad_norm": 0.5432685692905308,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2071
    },
    {
      "epoch": 0.02072,
      "grad_norm": 0.5685190700350947,
      "learning_rate": 0.003,
      "loss": 4.2519,
      "step": 2072
    },
    {
      "epoch": 0.02073,
      "grad_norm": 0.535137018993922,
      "learning_rate": 0.003,
      "loss": 4.2291,
      "step": 2073
    },
    {
      "epoch": 0.02074,
      "grad_norm": 0.5467639426283901,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2074
    },
    {
      "epoch": 0.02075,
      "grad_norm": 0.5451184946505037,
      "learning_rate": 0.003,
      "loss": 4.2644,
      "step": 2075
    },
    {
      "epoch": 0.02076,
      "grad_norm": 0.4600936162495081,
      "learning_rate": 0.003,
      "loss": 4.2518,
      "step": 2076
    },
    {
      "epoch": 0.02077,
      "grad_norm": 0.465650503497137,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2077
    },
    {
      "epoch": 0.02078,
      "grad_norm": 0.5411332311026555,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2078
    },
    {
      "epoch": 0.02079,
      "grad_norm": 0.5765467214387223,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 2079
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.5738914259466256,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2080
    },
    {
      "epoch": 0.02081,
      "grad_norm": 0.5810191862607845,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2081
    },
    {
      "epoch": 0.02082,
      "grad_norm": 0.5527464904961061,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2082
    },
    {
      "epoch": 0.02083,
      "grad_norm": 0.5331974469414995,
      "learning_rate": 0.003,
      "loss": 4.2351,
      "step": 2083
    },
    {
      "epoch": 0.02084,
      "grad_norm": 0.5880130787094308,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2084
    },
    {
      "epoch": 0.02085,
      "grad_norm": 0.6630695440600266,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2085
    },
    {
      "epoch": 0.02086,
      "grad_norm": 0.6095683840332357,
      "learning_rate": 0.003,
      "loss": 4.2148,
      "step": 2086
    },
    {
      "epoch": 0.02087,
      "grad_norm": 0.6289181823836659,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2087
    },
    {
      "epoch": 0.02088,
      "grad_norm": 0.6691073473732546,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 2088
    },
    {
      "epoch": 0.02089,
      "grad_norm": 0.757943428461077,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2089
    },
    {
      "epoch": 0.0209,
      "grad_norm": 0.6942072774820578,
      "learning_rate": 0.003,
      "loss": 4.2464,
      "step": 2090
    },
    {
      "epoch": 0.02091,
      "grad_norm": 0.5265412592049787,
      "learning_rate": 0.003,
      "loss": 4.224,
      "step": 2091
    },
    {
      "epoch": 0.02092,
      "grad_norm": 0.6385734634683451,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2092
    },
    {
      "epoch": 0.02093,
      "grad_norm": 0.7053440248117775,
      "learning_rate": 0.003,
      "loss": 4.1968,
      "step": 2093
    },
    {
      "epoch": 0.02094,
      "grad_norm": 0.696978044878893,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2094
    },
    {
      "epoch": 0.02095,
      "grad_norm": 0.6272978652521859,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2095
    },
    {
      "epoch": 0.02096,
      "grad_norm": 0.643764393951801,
      "learning_rate": 0.003,
      "loss": 4.2395,
      "step": 2096
    },
    {
      "epoch": 0.02097,
      "grad_norm": 0.6586650506136257,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2097
    },
    {
      "epoch": 0.02098,
      "grad_norm": 0.5374026111379894,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 2098
    },
    {
      "epoch": 0.02099,
      "grad_norm": 0.463394138236571,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 2099
    },
    {
      "epoch": 0.021,
      "grad_norm": 0.43247948811237297,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 2100
    },
    {
      "epoch": 0.02101,
      "grad_norm": 0.4029126824700864,
      "learning_rate": 0.003,
      "loss": 4.2421,
      "step": 2101
    },
    {
      "epoch": 0.02102,
      "grad_norm": 0.4076539308302002,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2102
    },
    {
      "epoch": 0.02103,
      "grad_norm": 0.48572239313264476,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2103
    },
    {
      "epoch": 0.02104,
      "grad_norm": 0.6011024498535701,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2104
    },
    {
      "epoch": 0.02105,
      "grad_norm": 0.7452127300170479,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 2105
    },
    {
      "epoch": 0.02106,
      "grad_norm": 0.8682450979969631,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2106
    },
    {
      "epoch": 0.02107,
      "grad_norm": 0.8835439612331999,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2107
    },
    {
      "epoch": 0.02108,
      "grad_norm": 0.7854943005617361,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2108
    },
    {
      "epoch": 0.02109,
      "grad_norm": 0.7880515723278634,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2109
    },
    {
      "epoch": 0.0211,
      "grad_norm": 0.7374506931044654,
      "learning_rate": 0.003,
      "loss": 4.234,
      "step": 2110
    },
    {
      "epoch": 0.02111,
      "grad_norm": 0.8090504896236194,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2111
    },
    {
      "epoch": 0.02112,
      "grad_norm": 0.7918025257999607,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 2112
    },
    {
      "epoch": 0.02113,
      "grad_norm": 0.8131758042516601,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 2113
    },
    {
      "epoch": 0.02114,
      "grad_norm": 0.7527860897593042,
      "learning_rate": 0.003,
      "loss": 4.2365,
      "step": 2114
    },
    {
      "epoch": 0.02115,
      "grad_norm": 0.7359409139929027,
      "learning_rate": 0.003,
      "loss": 4.231,
      "step": 2115
    },
    {
      "epoch": 0.02116,
      "grad_norm": 0.6808675729987482,
      "learning_rate": 0.003,
      "loss": 4.2544,
      "step": 2116
    },
    {
      "epoch": 0.02117,
      "grad_norm": 0.5755888828726318,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2117
    },
    {
      "epoch": 0.02118,
      "grad_norm": 0.5921868695179046,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2118
    },
    {
      "epoch": 0.02119,
      "grad_norm": 0.5455603705053529,
      "learning_rate": 0.003,
      "loss": 4.2459,
      "step": 2119
    },
    {
      "epoch": 0.0212,
      "grad_norm": 0.627867188843003,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 2120
    },
    {
      "epoch": 0.02121,
      "grad_norm": 0.611945450027123,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2121
    },
    {
      "epoch": 0.02122,
      "grad_norm": 0.568436827457235,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2122
    },
    {
      "epoch": 0.02123,
      "grad_norm": 0.5350506674860982,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2123
    },
    {
      "epoch": 0.02124,
      "grad_norm": 0.5325059896638067,
      "learning_rate": 0.003,
      "loss": 4.2254,
      "step": 2124
    },
    {
      "epoch": 0.02125,
      "grad_norm": 0.55145580695339,
      "learning_rate": 0.003,
      "loss": 4.2384,
      "step": 2125
    },
    {
      "epoch": 0.02126,
      "grad_norm": 0.6190295521827949,
      "learning_rate": 0.003,
      "loss": 4.2377,
      "step": 2126
    },
    {
      "epoch": 0.02127,
      "grad_norm": 0.5815814384311068,
      "learning_rate": 0.003,
      "loss": 4.244,
      "step": 2127
    },
    {
      "epoch": 0.02128,
      "grad_norm": 0.624207657173836,
      "learning_rate": 0.003,
      "loss": 4.2362,
      "step": 2128
    },
    {
      "epoch": 0.02129,
      "grad_norm": 0.637928608874603,
      "learning_rate": 0.003,
      "loss": 4.2364,
      "step": 2129
    },
    {
      "epoch": 0.0213,
      "grad_norm": 0.5372631019114236,
      "learning_rate": 0.003,
      "loss": 4.2069,
      "step": 2130
    },
    {
      "epoch": 0.02131,
      "grad_norm": 0.5117120142897393,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2131
    },
    {
      "epoch": 0.02132,
      "grad_norm": 0.5272503917214815,
      "learning_rate": 0.003,
      "loss": 4.2429,
      "step": 2132
    },
    {
      "epoch": 0.02133,
      "grad_norm": 0.4922684782001162,
      "learning_rate": 0.003,
      "loss": 4.2337,
      "step": 2133
    },
    {
      "epoch": 0.02134,
      "grad_norm": 0.4711594178288037,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2134
    },
    {
      "epoch": 0.02135,
      "grad_norm": 0.5354798117430423,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2135
    },
    {
      "epoch": 0.02136,
      "grad_norm": 0.5972846763679789,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2136
    },
    {
      "epoch": 0.02137,
      "grad_norm": 0.6875217013113815,
      "learning_rate": 0.003,
      "loss": 4.2468,
      "step": 2137
    },
    {
      "epoch": 0.02138,
      "grad_norm": 0.9636076155805002,
      "learning_rate": 0.003,
      "loss": 4.273,
      "step": 2138
    },
    {
      "epoch": 0.02139,
      "grad_norm": 1.148302508200364,
      "learning_rate": 0.003,
      "loss": 4.2769,
      "step": 2139
    },
    {
      "epoch": 0.0214,
      "grad_norm": 0.7025550992151242,
      "learning_rate": 0.003,
      "loss": 4.2515,
      "step": 2140
    },
    {
      "epoch": 0.02141,
      "grad_norm": 0.7381063691718428,
      "learning_rate": 0.003,
      "loss": 4.2408,
      "step": 2141
    },
    {
      "epoch": 0.02142,
      "grad_norm": 0.6164956740364308,
      "learning_rate": 0.003,
      "loss": 4.2223,
      "step": 2142
    },
    {
      "epoch": 0.02143,
      "grad_norm": 0.6846147414300063,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2143
    },
    {
      "epoch": 0.02144,
      "grad_norm": 0.7043565504624957,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2144
    },
    {
      "epoch": 0.02145,
      "grad_norm": 0.6306040325927637,
      "learning_rate": 0.003,
      "loss": 4.2262,
      "step": 2145
    },
    {
      "epoch": 0.02146,
      "grad_norm": 0.5782503991731893,
      "learning_rate": 0.003,
      "loss": 4.2372,
      "step": 2146
    },
    {
      "epoch": 0.02147,
      "grad_norm": 0.5354606170184357,
      "learning_rate": 0.003,
      "loss": 4.2548,
      "step": 2147
    },
    {
      "epoch": 0.02148,
      "grad_norm": 0.4707102689450845,
      "learning_rate": 0.003,
      "loss": 4.2629,
      "step": 2148
    },
    {
      "epoch": 0.02149,
      "grad_norm": 0.5098358773254507,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2149
    },
    {
      "epoch": 0.0215,
      "grad_norm": 0.5420555838700287,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 2150
    },
    {
      "epoch": 0.02151,
      "grad_norm": 0.5090710234825387,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2151
    },
    {
      "epoch": 0.02152,
      "grad_norm": 0.5210838533104809,
      "learning_rate": 0.003,
      "loss": 4.2153,
      "step": 2152
    },
    {
      "epoch": 0.02153,
      "grad_norm": 0.4582952267704145,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2153
    },
    {
      "epoch": 0.02154,
      "grad_norm": 0.5288143265950994,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2154
    },
    {
      "epoch": 0.02155,
      "grad_norm": 0.568233364630795,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 2155
    },
    {
      "epoch": 0.02156,
      "grad_norm": 0.6709041798606031,
      "learning_rate": 0.003,
      "loss": 4.2356,
      "step": 2156
    },
    {
      "epoch": 0.02157,
      "grad_norm": 0.837609334642183,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2157
    },
    {
      "epoch": 0.02158,
      "grad_norm": 0.9025483102003374,
      "learning_rate": 0.003,
      "loss": 4.2406,
      "step": 2158
    },
    {
      "epoch": 0.02159,
      "grad_norm": 0.6757453813747746,
      "learning_rate": 0.003,
      "loss": 4.2475,
      "step": 2159
    },
    {
      "epoch": 0.0216,
      "grad_norm": 0.6195925236969745,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2160
    },
    {
      "epoch": 0.02161,
      "grad_norm": 0.6541705747885649,
      "learning_rate": 0.003,
      "loss": 4.2346,
      "step": 2161
    },
    {
      "epoch": 0.02162,
      "grad_norm": 0.713506704868693,
      "learning_rate": 0.003,
      "loss": 4.2445,
      "step": 2162
    },
    {
      "epoch": 0.02163,
      "grad_norm": 0.7618153376621027,
      "learning_rate": 0.003,
      "loss": 4.2619,
      "step": 2163
    },
    {
      "epoch": 0.02164,
      "grad_norm": 0.7193438764197988,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2164
    },
    {
      "epoch": 0.02165,
      "grad_norm": 0.7614872081882262,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2165
    },
    {
      "epoch": 0.02166,
      "grad_norm": 0.7583750488187075,
      "learning_rate": 0.003,
      "loss": 4.2616,
      "step": 2166
    },
    {
      "epoch": 0.02167,
      "grad_norm": 0.5886983647176827,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2167
    },
    {
      "epoch": 0.02168,
      "grad_norm": 0.5673249719944714,
      "learning_rate": 0.003,
      "loss": 4.229,
      "step": 2168
    },
    {
      "epoch": 0.02169,
      "grad_norm": 0.539350720737554,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 2169
    },
    {
      "epoch": 0.0217,
      "grad_norm": 0.505175388578022,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2170
    },
    {
      "epoch": 0.02171,
      "grad_norm": 0.523933575801358,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2171
    },
    {
      "epoch": 0.02172,
      "grad_norm": 0.4755293785333703,
      "learning_rate": 0.003,
      "loss": 4.2128,
      "step": 2172
    },
    {
      "epoch": 0.02173,
      "grad_norm": 0.5360790560792152,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2173
    },
    {
      "epoch": 0.02174,
      "grad_norm": 0.5613507175665455,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2174
    },
    {
      "epoch": 0.02175,
      "grad_norm": 0.7273309266731635,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 2175
    },
    {
      "epoch": 0.02176,
      "grad_norm": 0.8055224057820037,
      "learning_rate": 0.003,
      "loss": 4.2357,
      "step": 2176
    },
    {
      "epoch": 0.02177,
      "grad_norm": 0.7934658981354097,
      "learning_rate": 0.003,
      "loss": 4.2405,
      "step": 2177
    },
    {
      "epoch": 0.02178,
      "grad_norm": 0.680464890242631,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2178
    },
    {
      "epoch": 0.02179,
      "grad_norm": 0.6249756747331623,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2179
    },
    {
      "epoch": 0.0218,
      "grad_norm": 0.6384620261389609,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 2180
    },
    {
      "epoch": 0.02181,
      "grad_norm": 0.6338413006913513,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2181
    },
    {
      "epoch": 0.02182,
      "grad_norm": 0.602201817766222,
      "learning_rate": 0.003,
      "loss": 4.2311,
      "step": 2182
    },
    {
      "epoch": 0.02183,
      "grad_norm": 0.6159529942047347,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2183
    },
    {
      "epoch": 0.02184,
      "grad_norm": 0.654749250757439,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2184
    },
    {
      "epoch": 0.02185,
      "grad_norm": 0.798418167230075,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2185
    },
    {
      "epoch": 0.02186,
      "grad_norm": 0.9118929371333987,
      "learning_rate": 0.003,
      "loss": 4.2486,
      "step": 2186
    },
    {
      "epoch": 0.02187,
      "grad_norm": 0.8282661348302262,
      "learning_rate": 0.003,
      "loss": 4.2454,
      "step": 2187
    },
    {
      "epoch": 0.02188,
      "grad_norm": 0.6832425004401769,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 2188
    },
    {
      "epoch": 0.02189,
      "grad_norm": 0.6446063652482951,
      "learning_rate": 0.003,
      "loss": 4.2521,
      "step": 2189
    },
    {
      "epoch": 0.0219,
      "grad_norm": 0.5887684645304796,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2190
    },
    {
      "epoch": 0.02191,
      "grad_norm": 0.5977971628739048,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 2191
    },
    {
      "epoch": 0.02192,
      "grad_norm": 0.4949212999134859,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2192
    },
    {
      "epoch": 0.02193,
      "grad_norm": 0.5003189031862624,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2193
    },
    {
      "epoch": 0.02194,
      "grad_norm": 0.48226309241618065,
      "learning_rate": 0.003,
      "loss": 4.2678,
      "step": 2194
    },
    {
      "epoch": 0.02195,
      "grad_norm": 0.544272590519486,
      "learning_rate": 0.003,
      "loss": 4.2332,
      "step": 2195
    },
    {
      "epoch": 0.02196,
      "grad_norm": 0.6071662200973025,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 2196
    },
    {
      "epoch": 0.02197,
      "grad_norm": 0.7366010094026804,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2197
    },
    {
      "epoch": 0.02198,
      "grad_norm": 0.7102801228364903,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2198
    },
    {
      "epoch": 0.02199,
      "grad_norm": 0.6007866089006326,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2199
    },
    {
      "epoch": 0.022,
      "grad_norm": 0.5670114788301772,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 2200
    },
    {
      "epoch": 0.02201,
      "grad_norm": 0.4763785124586338,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2201
    },
    {
      "epoch": 0.02202,
      "grad_norm": 0.42903876892090725,
      "learning_rate": 0.003,
      "loss": 4.2163,
      "step": 2202
    },
    {
      "epoch": 0.02203,
      "grad_norm": 0.4156329769757173,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 2203
    },
    {
      "epoch": 0.02204,
      "grad_norm": 0.4681197264286167,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2204
    },
    {
      "epoch": 0.02205,
      "grad_norm": 0.5059629089964722,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 2205
    },
    {
      "epoch": 0.02206,
      "grad_norm": 0.6309485655238869,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2206
    },
    {
      "epoch": 0.02207,
      "grad_norm": 0.6824022074223641,
      "learning_rate": 0.003,
      "loss": 4.2243,
      "step": 2207
    },
    {
      "epoch": 0.02208,
      "grad_norm": 0.6950242305361404,
      "learning_rate": 0.003,
      "loss": 4.2081,
      "step": 2208
    },
    {
      "epoch": 0.02209,
      "grad_norm": 0.8264786459594594,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 2209
    },
    {
      "epoch": 0.0221,
      "grad_norm": 0.8486387875780823,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2210
    },
    {
      "epoch": 0.02211,
      "grad_norm": 0.7761571300506404,
      "learning_rate": 0.003,
      "loss": 4.2269,
      "step": 2211
    },
    {
      "epoch": 0.02212,
      "grad_norm": 0.6379539901749441,
      "learning_rate": 0.003,
      "loss": 4.2233,
      "step": 2212
    },
    {
      "epoch": 0.02213,
      "grad_norm": 0.6089238962453585,
      "learning_rate": 0.003,
      "loss": 4.2296,
      "step": 2213
    },
    {
      "epoch": 0.02214,
      "grad_norm": 0.5796568814987876,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2214
    },
    {
      "epoch": 0.02215,
      "grad_norm": 0.5660272621079528,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2215
    },
    {
      "epoch": 0.02216,
      "grad_norm": 0.5885926435515716,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2216
    },
    {
      "epoch": 0.02217,
      "grad_norm": 0.5475660813804676,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2217
    },
    {
      "epoch": 0.02218,
      "grad_norm": 0.5260715025238017,
      "learning_rate": 0.003,
      "loss": 4.2304,
      "step": 2218
    },
    {
      "epoch": 0.02219,
      "grad_norm": 0.5716509549906074,
      "learning_rate": 0.003,
      "loss": 4.2216,
      "step": 2219
    },
    {
      "epoch": 0.0222,
      "grad_norm": 0.5917539957492576,
      "learning_rate": 0.003,
      "loss": 4.2195,
      "step": 2220
    },
    {
      "epoch": 0.02221,
      "grad_norm": 0.6044205892693139,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 2221
    },
    {
      "epoch": 0.02222,
      "grad_norm": 0.567760157124041,
      "learning_rate": 0.003,
      "loss": 4.2063,
      "step": 2222
    },
    {
      "epoch": 0.02223,
      "grad_norm": 0.577201891352313,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2223
    },
    {
      "epoch": 0.02224,
      "grad_norm": 0.5589539299230697,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2224
    },
    {
      "epoch": 0.02225,
      "grad_norm": 0.578872867820463,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2225
    },
    {
      "epoch": 0.02226,
      "grad_norm": 0.6018889202924489,
      "learning_rate": 0.003,
      "loss": 4.2223,
      "step": 2226
    },
    {
      "epoch": 0.02227,
      "grad_norm": 0.6659111341486127,
      "learning_rate": 0.003,
      "loss": 4.2192,
      "step": 2227
    },
    {
      "epoch": 0.02228,
      "grad_norm": 0.6307543898445819,
      "learning_rate": 0.003,
      "loss": 4.2071,
      "step": 2228
    },
    {
      "epoch": 0.02229,
      "grad_norm": 0.5993087464100946,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 2229
    },
    {
      "epoch": 0.0223,
      "grad_norm": 0.6286376964291409,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2230
    },
    {
      "epoch": 0.02231,
      "grad_norm": 0.872328678246551,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 2231
    },
    {
      "epoch": 0.02232,
      "grad_norm": 1.0461803323605487,
      "learning_rate": 0.003,
      "loss": 4.2634,
      "step": 2232
    },
    {
      "epoch": 0.02233,
      "grad_norm": 0.9329588220097066,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 2233
    },
    {
      "epoch": 0.02234,
      "grad_norm": 0.6760145633983702,
      "learning_rate": 0.003,
      "loss": 4.2457,
      "step": 2234
    },
    {
      "epoch": 0.02235,
      "grad_norm": 0.7832849984021282,
      "learning_rate": 0.003,
      "loss": 4.281,
      "step": 2235
    },
    {
      "epoch": 0.02236,
      "grad_norm": 0.6196461184379554,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 2236
    },
    {
      "epoch": 0.02237,
      "grad_norm": 0.5630794415245316,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2237
    },
    {
      "epoch": 0.02238,
      "grad_norm": 0.4338854754415404,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2238
    },
    {
      "epoch": 0.02239,
      "grad_norm": 0.43304313838731295,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2239
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.3849146263035166,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 2240
    },
    {
      "epoch": 0.02241,
      "grad_norm": 0.3700093297043444,
      "learning_rate": 0.003,
      "loss": 4.221,
      "step": 2241
    },
    {
      "epoch": 0.02242,
      "grad_norm": 0.3483834690743904,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2242
    },
    {
      "epoch": 0.02243,
      "grad_norm": 0.38069058438024445,
      "learning_rate": 0.003,
      "loss": 4.2335,
      "step": 2243
    },
    {
      "epoch": 0.02244,
      "grad_norm": 0.4503598461041079,
      "learning_rate": 0.003,
      "loss": 4.21,
      "step": 2244
    },
    {
      "epoch": 0.02245,
      "grad_norm": 0.5173040984777282,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2245
    },
    {
      "epoch": 0.02246,
      "grad_norm": 0.5743673128739117,
      "learning_rate": 0.003,
      "loss": 4.2315,
      "step": 2246
    },
    {
      "epoch": 0.02247,
      "grad_norm": 0.6564841670845484,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2247
    },
    {
      "epoch": 0.02248,
      "grad_norm": 0.7271803573068377,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2248
    },
    {
      "epoch": 0.02249,
      "grad_norm": 0.6357921170928627,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2249
    },
    {
      "epoch": 0.0225,
      "grad_norm": 0.5333359806181992,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2250
    },
    {
      "epoch": 0.02251,
      "grad_norm": 0.5481515163618663,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 2251
    },
    {
      "epoch": 0.02252,
      "grad_norm": 0.5849718573683582,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 2252
    },
    {
      "epoch": 0.02253,
      "grad_norm": 0.692221940613648,
      "learning_rate": 0.003,
      "loss": 4.2188,
      "step": 2253
    },
    {
      "epoch": 0.02254,
      "grad_norm": 0.8350743358894537,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 2254
    },
    {
      "epoch": 0.02255,
      "grad_norm": 0.9700169658063796,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2255
    },
    {
      "epoch": 0.02256,
      "grad_norm": 1.1849964912746291,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 2256
    },
    {
      "epoch": 0.02257,
      "grad_norm": 0.8019987212669306,
      "learning_rate": 0.003,
      "loss": 4.2432,
      "step": 2257
    },
    {
      "epoch": 0.02258,
      "grad_norm": 0.7178719029324007,
      "learning_rate": 0.003,
      "loss": 4.2513,
      "step": 2258
    },
    {
      "epoch": 0.02259,
      "grad_norm": 0.8182127992878601,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 2259
    },
    {
      "epoch": 0.0226,
      "grad_norm": 0.7294977278785716,
      "learning_rate": 0.003,
      "loss": 4.2695,
      "step": 2260
    },
    {
      "epoch": 0.02261,
      "grad_norm": 0.6869467420486108,
      "learning_rate": 0.003,
      "loss": 4.2439,
      "step": 2261
    },
    {
      "epoch": 0.02262,
      "grad_norm": 0.7258556366490507,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2262
    },
    {
      "epoch": 0.02263,
      "grad_norm": 0.6949107967496511,
      "learning_rate": 0.003,
      "loss": 4.2554,
      "step": 2263
    },
    {
      "epoch": 0.02264,
      "grad_norm": 0.6840843296451264,
      "learning_rate": 0.003,
      "loss": 4.2468,
      "step": 2264
    },
    {
      "epoch": 0.02265,
      "grad_norm": 0.6101088207573817,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 2265
    },
    {
      "epoch": 0.02266,
      "grad_norm": 0.6149463166497737,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2266
    },
    {
      "epoch": 0.02267,
      "grad_norm": 0.566032425336374,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 2267
    },
    {
      "epoch": 0.02268,
      "grad_norm": 0.5712592856836044,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2268
    },
    {
      "epoch": 0.02269,
      "grad_norm": 0.5122859378101323,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2269
    },
    {
      "epoch": 0.0227,
      "grad_norm": 0.4651490853981594,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2270
    },
    {
      "epoch": 0.02271,
      "grad_norm": 0.39776686394339583,
      "learning_rate": 0.003,
      "loss": 4.2286,
      "step": 2271
    },
    {
      "epoch": 0.02272,
      "grad_norm": 0.4302685557347805,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 2272
    },
    {
      "epoch": 0.02273,
      "grad_norm": 0.467593917531494,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2273
    },
    {
      "epoch": 0.02274,
      "grad_norm": 0.632433536625551,
      "learning_rate": 0.003,
      "loss": 4.2141,
      "step": 2274
    },
    {
      "epoch": 0.02275,
      "grad_norm": 0.8580139662336462,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2275
    },
    {
      "epoch": 0.02276,
      "grad_norm": 0.9516059646785548,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2276
    },
    {
      "epoch": 0.02277,
      "grad_norm": 0.7436294288419117,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2277
    },
    {
      "epoch": 0.02278,
      "grad_norm": 0.6418671737236206,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2278
    },
    {
      "epoch": 0.02279,
      "grad_norm": 0.7656816707126055,
      "learning_rate": 0.003,
      "loss": 4.2435,
      "step": 2279
    },
    {
      "epoch": 0.0228,
      "grad_norm": 0.6177672920210129,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2280
    },
    {
      "epoch": 0.02281,
      "grad_norm": 0.5895210519635931,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 2281
    },
    {
      "epoch": 0.02282,
      "grad_norm": 0.5868753488307538,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2282
    },
    {
      "epoch": 0.02283,
      "grad_norm": 0.6052691283316047,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2283
    },
    {
      "epoch": 0.02284,
      "grad_norm": 0.5832141161476754,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2284
    },
    {
      "epoch": 0.02285,
      "grad_norm": 0.5796401382132979,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 2285
    },
    {
      "epoch": 0.02286,
      "grad_norm": 0.5270223366823183,
      "learning_rate": 0.003,
      "loss": 4.2443,
      "step": 2286
    },
    {
      "epoch": 0.02287,
      "grad_norm": 0.4022883813311557,
      "learning_rate": 0.003,
      "loss": 4.2176,
      "step": 2287
    },
    {
      "epoch": 0.02288,
      "grad_norm": 0.4249187639605795,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 2288
    },
    {
      "epoch": 0.02289,
      "grad_norm": 0.39026571046663244,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2289
    },
    {
      "epoch": 0.0229,
      "grad_norm": 0.4142451551190753,
      "learning_rate": 0.003,
      "loss": 4.2147,
      "step": 2290
    },
    {
      "epoch": 0.02291,
      "grad_norm": 0.48765765895038643,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2291
    },
    {
      "epoch": 0.02292,
      "grad_norm": 0.5671990063966691,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 2292
    },
    {
      "epoch": 0.02293,
      "grad_norm": 0.5656764518255547,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2293
    },
    {
      "epoch": 0.02294,
      "grad_norm": 0.5275578505406908,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2294
    },
    {
      "epoch": 0.02295,
      "grad_norm": 0.48221866990126555,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 2295
    },
    {
      "epoch": 0.02296,
      "grad_norm": 0.5188277549634542,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2296
    },
    {
      "epoch": 0.02297,
      "grad_norm": 0.6417395019786397,
      "learning_rate": 0.003,
      "loss": 4.199,
      "step": 2297
    },
    {
      "epoch": 0.02298,
      "grad_norm": 0.7094858197213939,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2298
    },
    {
      "epoch": 0.02299,
      "grad_norm": 0.7715781893149263,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2299
    },
    {
      "epoch": 0.023,
      "grad_norm": 0.8183493893838668,
      "learning_rate": 0.003,
      "loss": 4.2495,
      "step": 2300
    },
    {
      "epoch": 0.02301,
      "grad_norm": 0.7727005595462422,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2301
    },
    {
      "epoch": 0.02302,
      "grad_norm": 0.789250865027278,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2302
    },
    {
      "epoch": 0.02303,
      "grad_norm": 0.8561833032989024,
      "learning_rate": 0.003,
      "loss": 4.2229,
      "step": 2303
    },
    {
      "epoch": 0.02304,
      "grad_norm": 0.7983476489527453,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2304
    },
    {
      "epoch": 0.02305,
      "grad_norm": 0.792691144527778,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2305
    },
    {
      "epoch": 0.02306,
      "grad_norm": 0.8333614803540381,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2306
    },
    {
      "epoch": 0.02307,
      "grad_norm": 0.8709733257410114,
      "learning_rate": 0.003,
      "loss": 4.248,
      "step": 2307
    },
    {
      "epoch": 0.02308,
      "grad_norm": 0.7840461913162227,
      "learning_rate": 0.003,
      "loss": 4.2376,
      "step": 2308
    },
    {
      "epoch": 0.02309,
      "grad_norm": 0.7160902424813279,
      "learning_rate": 0.003,
      "loss": 4.263,
      "step": 2309
    },
    {
      "epoch": 0.0231,
      "grad_norm": 0.7757918757353481,
      "learning_rate": 0.003,
      "loss": 4.2432,
      "step": 2310
    },
    {
      "epoch": 0.02311,
      "grad_norm": 0.8687865089693275,
      "learning_rate": 0.003,
      "loss": 4.2573,
      "step": 2311
    },
    {
      "epoch": 0.02312,
      "grad_norm": 0.7846302126542865,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2312
    },
    {
      "epoch": 0.02313,
      "grad_norm": 0.7297339955221905,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2313
    },
    {
      "epoch": 0.02314,
      "grad_norm": 0.6956047853880688,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2314
    },
    {
      "epoch": 0.02315,
      "grad_norm": 0.6937368414181051,
      "learning_rate": 0.003,
      "loss": 4.2413,
      "step": 2315
    },
    {
      "epoch": 0.02316,
      "grad_norm": 0.6709916211621512,
      "learning_rate": 0.003,
      "loss": 4.2522,
      "step": 2316
    },
    {
      "epoch": 0.02317,
      "grad_norm": 0.6213877076356003,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2317
    },
    {
      "epoch": 0.02318,
      "grad_norm": 0.6088693827583501,
      "learning_rate": 0.003,
      "loss": 4.2184,
      "step": 2318
    },
    {
      "epoch": 0.02319,
      "grad_norm": 0.5989733797969562,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 2319
    },
    {
      "epoch": 0.0232,
      "grad_norm": 0.55615386704494,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2320
    },
    {
      "epoch": 0.02321,
      "grad_norm": 0.5520477296154882,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2321
    },
    {
      "epoch": 0.02322,
      "grad_norm": 0.5223023663129024,
      "learning_rate": 0.003,
      "loss": 4.2284,
      "step": 2322
    },
    {
      "epoch": 0.02323,
      "grad_norm": 0.5716320780098701,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2323
    },
    {
      "epoch": 0.02324,
      "grad_norm": 0.6956733472532048,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 2324
    },
    {
      "epoch": 0.02325,
      "grad_norm": 0.6924168616855679,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2325
    },
    {
      "epoch": 0.02326,
      "grad_norm": 0.6921696722241616,
      "learning_rate": 0.003,
      "loss": 4.2476,
      "step": 2326
    },
    {
      "epoch": 0.02327,
      "grad_norm": 0.6351653773148184,
      "learning_rate": 0.003,
      "loss": 4.2319,
      "step": 2327
    },
    {
      "epoch": 0.02328,
      "grad_norm": 0.5655372914501792,
      "learning_rate": 0.003,
      "loss": 4.2261,
      "step": 2328
    },
    {
      "epoch": 0.02329,
      "grad_norm": 0.6342553615471473,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 2329
    },
    {
      "epoch": 0.0233,
      "grad_norm": 0.7154577148277532,
      "learning_rate": 0.003,
      "loss": 4.2152,
      "step": 2330
    },
    {
      "epoch": 0.02331,
      "grad_norm": 0.73894366097317,
      "learning_rate": 0.003,
      "loss": 4.2587,
      "step": 2331
    },
    {
      "epoch": 0.02332,
      "grad_norm": 0.7851778924150568,
      "learning_rate": 0.003,
      "loss": 4.2242,
      "step": 2332
    },
    {
      "epoch": 0.02333,
      "grad_norm": 0.7287502366417434,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2333
    },
    {
      "epoch": 0.02334,
      "grad_norm": 0.69094548866664,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 2334
    },
    {
      "epoch": 0.02335,
      "grad_norm": 0.6431810869476228,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2335
    },
    {
      "epoch": 0.02336,
      "grad_norm": 0.6507673975963663,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2336
    },
    {
      "epoch": 0.02337,
      "grad_norm": 0.7087163474933083,
      "learning_rate": 0.003,
      "loss": 4.2296,
      "step": 2337
    },
    {
      "epoch": 0.02338,
      "grad_norm": 0.6659743716890278,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2338
    },
    {
      "epoch": 0.02339,
      "grad_norm": 0.5625472175847084,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2339
    },
    {
      "epoch": 0.0234,
      "grad_norm": 0.536202676220793,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2340
    },
    {
      "epoch": 0.02341,
      "grad_norm": 0.4892866338203838,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2341
    },
    {
      "epoch": 0.02342,
      "grad_norm": 0.4796875190367091,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2342
    },
    {
      "epoch": 0.02343,
      "grad_norm": 0.43636850083944123,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 2343
    },
    {
      "epoch": 0.02344,
      "grad_norm": 0.4075473684449614,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2344
    },
    {
      "epoch": 0.02345,
      "grad_norm": 0.4126961162680678,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2345
    },
    {
      "epoch": 0.02346,
      "grad_norm": 0.4484646299481725,
      "learning_rate": 0.003,
      "loss": 4.1936,
      "step": 2346
    },
    {
      "epoch": 0.02347,
      "grad_norm": 0.4884767738301177,
      "learning_rate": 0.003,
      "loss": 4.2318,
      "step": 2347
    },
    {
      "epoch": 0.02348,
      "grad_norm": 0.4734903472149947,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2348
    },
    {
      "epoch": 0.02349,
      "grad_norm": 0.4515057387841677,
      "learning_rate": 0.003,
      "loss": 4.2087,
      "step": 2349
    },
    {
      "epoch": 0.0235,
      "grad_norm": 0.43769490231507474,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2350
    },
    {
      "epoch": 0.02351,
      "grad_norm": 0.3875892706722824,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2351
    },
    {
      "epoch": 0.02352,
      "grad_norm": 0.42012415359844946,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 2352
    },
    {
      "epoch": 0.02353,
      "grad_norm": 0.5800085038244271,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2353
    },
    {
      "epoch": 0.02354,
      "grad_norm": 0.9008192298721557,
      "learning_rate": 0.003,
      "loss": 4.2263,
      "step": 2354
    },
    {
      "epoch": 0.02355,
      "grad_norm": 1.174355012660842,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2355
    },
    {
      "epoch": 0.02356,
      "grad_norm": 0.6757502362960481,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2356
    },
    {
      "epoch": 0.02357,
      "grad_norm": 0.7784416987559685,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2357
    },
    {
      "epoch": 0.02358,
      "grad_norm": 0.8207171439845993,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2358
    },
    {
      "epoch": 0.02359,
      "grad_norm": 0.6974291909709516,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 2359
    },
    {
      "epoch": 0.0236,
      "grad_norm": 0.6185472104370391,
      "learning_rate": 0.003,
      "loss": 4.2471,
      "step": 2360
    },
    {
      "epoch": 0.02361,
      "grad_norm": 0.5773489611025617,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2361
    },
    {
      "epoch": 0.02362,
      "grad_norm": 0.5560165828585443,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 2362
    },
    {
      "epoch": 0.02363,
      "grad_norm": 0.5892532979179639,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2363
    },
    {
      "epoch": 0.02364,
      "grad_norm": 0.6748161886878429,
      "learning_rate": 0.003,
      "loss": 4.2153,
      "step": 2364
    },
    {
      "epoch": 0.02365,
      "grad_norm": 0.6598699309472811,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2365
    },
    {
      "epoch": 0.02366,
      "grad_norm": 0.6235317744814385,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2366
    },
    {
      "epoch": 0.02367,
      "grad_norm": 0.6532173738874076,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 2367
    },
    {
      "epoch": 0.02368,
      "grad_norm": 0.6882158454966492,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2368
    },
    {
      "epoch": 0.02369,
      "grad_norm": 0.7139133860202842,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2369
    },
    {
      "epoch": 0.0237,
      "grad_norm": 0.6776525799328785,
      "learning_rate": 0.003,
      "loss": 4.2355,
      "step": 2370
    },
    {
      "epoch": 0.02371,
      "grad_norm": 0.7002833727021253,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2371
    },
    {
      "epoch": 0.02372,
      "grad_norm": 0.7223141017308237,
      "learning_rate": 0.003,
      "loss": 4.2127,
      "step": 2372
    },
    {
      "epoch": 0.02373,
      "grad_norm": 0.7004144412962903,
      "learning_rate": 0.003,
      "loss": 4.2003,
      "step": 2373
    },
    {
      "epoch": 0.02374,
      "grad_norm": 0.6806743664027574,
      "learning_rate": 0.003,
      "loss": 4.2074,
      "step": 2374
    },
    {
      "epoch": 0.02375,
      "grad_norm": 0.6705132166888766,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2375
    },
    {
      "epoch": 0.02376,
      "grad_norm": 0.5005122224104306,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 2376
    },
    {
      "epoch": 0.02377,
      "grad_norm": 0.4877832394535443,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2377
    },
    {
      "epoch": 0.02378,
      "grad_norm": 0.5211637717940496,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2378
    },
    {
      "epoch": 0.02379,
      "grad_norm": 0.6384313580109909,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2379
    },
    {
      "epoch": 0.0238,
      "grad_norm": 0.7131026886322172,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 2380
    },
    {
      "epoch": 0.02381,
      "grad_norm": 0.6937779462786806,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2381
    },
    {
      "epoch": 0.02382,
      "grad_norm": 0.6337775600473754,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2382
    },
    {
      "epoch": 0.02383,
      "grad_norm": 0.5646754547957066,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2383
    },
    {
      "epoch": 0.02384,
      "grad_norm": 0.5471514877116583,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 2384
    },
    {
      "epoch": 0.02385,
      "grad_norm": 0.5791601386868904,
      "learning_rate": 0.003,
      "loss": 4.2135,
      "step": 2385
    },
    {
      "epoch": 0.02386,
      "grad_norm": 0.5976046883190522,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2386
    },
    {
      "epoch": 0.02387,
      "grad_norm": 0.6655482933999028,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2387
    },
    {
      "epoch": 0.02388,
      "grad_norm": 0.7334421734543064,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 2388
    },
    {
      "epoch": 0.02389,
      "grad_norm": 0.7475937838490452,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 2389
    },
    {
      "epoch": 0.0239,
      "grad_norm": 0.7091495046860523,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2390
    },
    {
      "epoch": 0.02391,
      "grad_norm": 0.6788986600308993,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2391
    },
    {
      "epoch": 0.02392,
      "grad_norm": 0.6156974740652746,
      "learning_rate": 0.003,
      "loss": 4.2308,
      "step": 2392
    },
    {
      "epoch": 0.02393,
      "grad_norm": 0.5941793885935496,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 2393
    },
    {
      "epoch": 0.02394,
      "grad_norm": 0.5863170046009283,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2394
    },
    {
      "epoch": 0.02395,
      "grad_norm": 0.7380191806299783,
      "learning_rate": 0.003,
      "loss": 4.2191,
      "step": 2395
    },
    {
      "epoch": 0.02396,
      "grad_norm": 0.8712169644578591,
      "learning_rate": 0.003,
      "loss": 4.2425,
      "step": 2396
    },
    {
      "epoch": 0.02397,
      "grad_norm": 0.8536583911176343,
      "learning_rate": 0.003,
      "loss": 4.2305,
      "step": 2397
    },
    {
      "epoch": 0.02398,
      "grad_norm": 0.7082106435478536,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2398
    },
    {
      "epoch": 0.02399,
      "grad_norm": 0.6784314246730624,
      "learning_rate": 0.003,
      "loss": 4.2124,
      "step": 2399
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.7409520225246221,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 2400
    },
    {
      "epoch": 0.02401,
      "grad_norm": 0.7471077758325675,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2401
    },
    {
      "epoch": 0.02402,
      "grad_norm": 0.7167664651670812,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2402
    },
    {
      "epoch": 0.02403,
      "grad_norm": 0.7133428609804063,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 2403
    },
    {
      "epoch": 0.02404,
      "grad_norm": 0.6914213864059761,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 2404
    },
    {
      "epoch": 0.02405,
      "grad_norm": 0.6223491766257139,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2405
    },
    {
      "epoch": 0.02406,
      "grad_norm": 0.6402462299592815,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2406
    },
    {
      "epoch": 0.02407,
      "grad_norm": 0.5600794609103489,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 2407
    },
    {
      "epoch": 0.02408,
      "grad_norm": 0.4916021474517548,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2408
    },
    {
      "epoch": 0.02409,
      "grad_norm": 0.46461844041782707,
      "learning_rate": 0.003,
      "loss": 4.2116,
      "step": 2409
    },
    {
      "epoch": 0.0241,
      "grad_norm": 0.5180903118600274,
      "learning_rate": 0.003,
      "loss": 4.2239,
      "step": 2410
    },
    {
      "epoch": 0.02411,
      "grad_norm": 0.6289211485922821,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 2411
    },
    {
      "epoch": 0.02412,
      "grad_norm": 0.625292309840426,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2412
    },
    {
      "epoch": 0.02413,
      "grad_norm": 0.5316162511007585,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 2413
    },
    {
      "epoch": 0.02414,
      "grad_norm": 0.569080239332035,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2414
    },
    {
      "epoch": 0.02415,
      "grad_norm": 0.6830500278720256,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2415
    },
    {
      "epoch": 0.02416,
      "grad_norm": 0.8230987152330054,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 2416
    },
    {
      "epoch": 0.02417,
      "grad_norm": 0.9992750696467811,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 2417
    },
    {
      "epoch": 0.02418,
      "grad_norm": 0.7172338223207045,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 2418
    },
    {
      "epoch": 0.02419,
      "grad_norm": 0.6066239591362272,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2419
    },
    {
      "epoch": 0.0242,
      "grad_norm": 0.6615008268071535,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2420
    },
    {
      "epoch": 0.02421,
      "grad_norm": 0.601398245236051,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 2421
    },
    {
      "epoch": 0.02422,
      "grad_norm": 0.5164536803360787,
      "learning_rate": 0.003,
      "loss": 4.2444,
      "step": 2422
    },
    {
      "epoch": 0.02423,
      "grad_norm": 0.47425853655922334,
      "learning_rate": 0.003,
      "loss": 4.2323,
      "step": 2423
    },
    {
      "epoch": 0.02424,
      "grad_norm": 0.45557836325278056,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 2424
    },
    {
      "epoch": 0.02425,
      "grad_norm": 0.37992471673130124,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 2425
    },
    {
      "epoch": 0.02426,
      "grad_norm": 0.4388918700897719,
      "learning_rate": 0.003,
      "loss": 4.2146,
      "step": 2426
    },
    {
      "epoch": 0.02427,
      "grad_norm": 0.47592644536510775,
      "learning_rate": 0.003,
      "loss": 4.2395,
      "step": 2427
    },
    {
      "epoch": 0.02428,
      "grad_norm": 0.41079892436338244,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2428
    },
    {
      "epoch": 0.02429,
      "grad_norm": 0.48266596555203656,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2429
    },
    {
      "epoch": 0.0243,
      "grad_norm": 0.58630499435022,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2430
    },
    {
      "epoch": 0.02431,
      "grad_norm": 0.6362348679805433,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2431
    },
    {
      "epoch": 0.02432,
      "grad_norm": 0.6409895194011197,
      "learning_rate": 0.003,
      "loss": 4.2237,
      "step": 2432
    },
    {
      "epoch": 0.02433,
      "grad_norm": 0.7045349857959661,
      "learning_rate": 0.003,
      "loss": 4.239,
      "step": 2433
    },
    {
      "epoch": 0.02434,
      "grad_norm": 0.6987807589267505,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2434
    },
    {
      "epoch": 0.02435,
      "grad_norm": 0.5878135110602212,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2435
    },
    {
      "epoch": 0.02436,
      "grad_norm": 0.5898177674736178,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2436
    },
    {
      "epoch": 0.02437,
      "grad_norm": 0.5929083773979196,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 2437
    },
    {
      "epoch": 0.02438,
      "grad_norm": 0.6434362451996457,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 2438
    },
    {
      "epoch": 0.02439,
      "grad_norm": 0.6611917410629563,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2439
    },
    {
      "epoch": 0.0244,
      "grad_norm": 0.7743759672777403,
      "learning_rate": 0.003,
      "loss": 4.2424,
      "step": 2440
    },
    {
      "epoch": 0.02441,
      "grad_norm": 0.7462269728851785,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2441
    },
    {
      "epoch": 0.02442,
      "grad_norm": 0.7608157833129767,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2442
    },
    {
      "epoch": 0.02443,
      "grad_norm": 0.79170341480151,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 2443
    },
    {
      "epoch": 0.02444,
      "grad_norm": 0.7939192302282532,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2444
    },
    {
      "epoch": 0.02445,
      "grad_norm": 0.7598676209330011,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2445
    },
    {
      "epoch": 0.02446,
      "grad_norm": 0.657657458880643,
      "learning_rate": 0.003,
      "loss": 4.2177,
      "step": 2446
    },
    {
      "epoch": 0.02447,
      "grad_norm": 0.636985404351659,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2447
    },
    {
      "epoch": 0.02448,
      "grad_norm": 0.6540065013234154,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 2448
    },
    {
      "epoch": 0.02449,
      "grad_norm": 0.6430617212308102,
      "learning_rate": 0.003,
      "loss": 4.218,
      "step": 2449
    },
    {
      "epoch": 0.0245,
      "grad_norm": 0.6859449218709958,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2450
    },
    {
      "epoch": 0.02451,
      "grad_norm": 0.6585394183553759,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2451
    },
    {
      "epoch": 0.02452,
      "grad_norm": 0.5944042022079538,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2452
    },
    {
      "epoch": 0.02453,
      "grad_norm": 0.6245280633667185,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 2453
    },
    {
      "epoch": 0.02454,
      "grad_norm": 0.6285302652257769,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 2454
    },
    {
      "epoch": 0.02455,
      "grad_norm": 0.630536065900029,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2455
    },
    {
      "epoch": 0.02456,
      "grad_norm": 0.6392454947407117,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 2456
    },
    {
      "epoch": 0.02457,
      "grad_norm": 0.732191714374005,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2457
    },
    {
      "epoch": 0.02458,
      "grad_norm": 0.8168261785432145,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 2458
    },
    {
      "epoch": 0.02459,
      "grad_norm": 0.8118895866121905,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2459
    },
    {
      "epoch": 0.0246,
      "grad_norm": 0.7597126257704254,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2460
    },
    {
      "epoch": 0.02461,
      "grad_norm": 0.7096048570719959,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2461
    },
    {
      "epoch": 0.02462,
      "grad_norm": 0.6412516044054245,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 2462
    },
    {
      "epoch": 0.02463,
      "grad_norm": 0.6158305280803775,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2463
    },
    {
      "epoch": 0.02464,
      "grad_norm": 0.5837936979418801,
      "learning_rate": 0.003,
      "loss": 4.2078,
      "step": 2464
    },
    {
      "epoch": 0.02465,
      "grad_norm": 0.5652391806413423,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2465
    },
    {
      "epoch": 0.02466,
      "grad_norm": 0.5314994263522052,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 2466
    },
    {
      "epoch": 0.02467,
      "grad_norm": 0.5492980443942803,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 2467
    },
    {
      "epoch": 0.02468,
      "grad_norm": 0.5476949286484541,
      "learning_rate": 0.003,
      "loss": 4.2248,
      "step": 2468
    },
    {
      "epoch": 0.02469,
      "grad_norm": 0.5739182126857109,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 2469
    },
    {
      "epoch": 0.0247,
      "grad_norm": 0.6455625601186472,
      "learning_rate": 0.003,
      "loss": 4.2003,
      "step": 2470
    },
    {
      "epoch": 0.02471,
      "grad_norm": 0.7500801989374446,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2471
    },
    {
      "epoch": 0.02472,
      "grad_norm": 0.8015350177741498,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 2472
    },
    {
      "epoch": 0.02473,
      "grad_norm": 0.7700030742386493,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 2473
    },
    {
      "epoch": 0.02474,
      "grad_norm": 0.6591968552189625,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 2474
    },
    {
      "epoch": 0.02475,
      "grad_norm": 0.5913092934763802,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2475
    },
    {
      "epoch": 0.02476,
      "grad_norm": 0.6828419322627938,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2476
    },
    {
      "epoch": 0.02477,
      "grad_norm": 0.7429757949041498,
      "learning_rate": 0.003,
      "loss": 4.2273,
      "step": 2477
    },
    {
      "epoch": 0.02478,
      "grad_norm": 0.7533856527859732,
      "learning_rate": 0.003,
      "loss": 4.2029,
      "step": 2478
    },
    {
      "epoch": 0.02479,
      "grad_norm": 0.6271813598380395,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 2479
    },
    {
      "epoch": 0.0248,
      "grad_norm": 0.7048163325838352,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 2480
    },
    {
      "epoch": 0.02481,
      "grad_norm": 0.7059455660386844,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2481
    },
    {
      "epoch": 0.02482,
      "grad_norm": 0.6095979307747832,
      "learning_rate": 0.003,
      "loss": 4.2232,
      "step": 2482
    },
    {
      "epoch": 0.02483,
      "grad_norm": 0.589007344890445,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 2483
    },
    {
      "epoch": 0.02484,
      "grad_norm": 0.5795798273023001,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2484
    },
    {
      "epoch": 0.02485,
      "grad_norm": 0.6558357821242041,
      "learning_rate": 0.003,
      "loss": 4.2264,
      "step": 2485
    },
    {
      "epoch": 0.02486,
      "grad_norm": 0.6694443853300054,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2486
    },
    {
      "epoch": 0.02487,
      "grad_norm": 0.7548583562826651,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2487
    },
    {
      "epoch": 0.02488,
      "grad_norm": 0.6086868759932925,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 2488
    },
    {
      "epoch": 0.02489,
      "grad_norm": 0.5543068870038478,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 2489
    },
    {
      "epoch": 0.0249,
      "grad_norm": 0.4695404599404173,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 2490
    },
    {
      "epoch": 0.02491,
      "grad_norm": 0.5098980691467636,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2491
    },
    {
      "epoch": 0.02492,
      "grad_norm": 0.5029414147729624,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2492
    },
    {
      "epoch": 0.02493,
      "grad_norm": 0.5644399016032551,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 2493
    },
    {
      "epoch": 0.02494,
      "grad_norm": 0.6367872115761151,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2494
    },
    {
      "epoch": 0.02495,
      "grad_norm": 0.6767787025658096,
      "learning_rate": 0.003,
      "loss": 4.2079,
      "step": 2495
    },
    {
      "epoch": 0.02496,
      "grad_norm": 0.696192347480972,
      "learning_rate": 0.003,
      "loss": 4.1973,
      "step": 2496
    },
    {
      "epoch": 0.02497,
      "grad_norm": 0.6551341280875559,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2497
    },
    {
      "epoch": 0.02498,
      "grad_norm": 0.6087269213019051,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 2498
    },
    {
      "epoch": 0.02499,
      "grad_norm": 0.6686291413156635,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2499
    },
    {
      "epoch": 0.025,
      "grad_norm": 0.6959767014574844,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2500
    },
    {
      "epoch": 0.02501,
      "grad_norm": 0.6984509411978452,
      "learning_rate": 0.003,
      "loss": 4.2249,
      "step": 2501
    },
    {
      "epoch": 0.02502,
      "grad_norm": 0.6296748558389454,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2502
    },
    {
      "epoch": 0.02503,
      "grad_norm": 0.7588933964884115,
      "learning_rate": 0.003,
      "loss": 4.2176,
      "step": 2503
    },
    {
      "epoch": 0.02504,
      "grad_norm": 0.7608780929496015,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2504
    },
    {
      "epoch": 0.02505,
      "grad_norm": 0.7626036518863434,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2505
    },
    {
      "epoch": 0.02506,
      "grad_norm": 0.8721628417499054,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2506
    },
    {
      "epoch": 0.02507,
      "grad_norm": 0.77178561857626,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2507
    },
    {
      "epoch": 0.02508,
      "grad_norm": 0.728729150203243,
      "learning_rate": 0.003,
      "loss": 4.2073,
      "step": 2508
    },
    {
      "epoch": 0.02509,
      "grad_norm": 0.6476865147041562,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2509
    },
    {
      "epoch": 0.0251,
      "grad_norm": 0.6920620401373876,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2510
    },
    {
      "epoch": 0.02511,
      "grad_norm": 0.7129030622439039,
      "learning_rate": 0.003,
      "loss": 4.2288,
      "step": 2511
    },
    {
      "epoch": 0.02512,
      "grad_norm": 0.7903195113153619,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2512
    },
    {
      "epoch": 0.02513,
      "grad_norm": 0.7730609450390953,
      "learning_rate": 0.003,
      "loss": 4.2181,
      "step": 2513
    },
    {
      "epoch": 0.02514,
      "grad_norm": 0.656826259281822,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 2514
    },
    {
      "epoch": 0.02515,
      "grad_norm": 0.6482134555064485,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2515
    },
    {
      "epoch": 0.02516,
      "grad_norm": 0.5967395863996815,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 2516
    },
    {
      "epoch": 0.02517,
      "grad_norm": 0.5007565780061344,
      "learning_rate": 0.003,
      "loss": 4.2015,
      "step": 2517
    },
    {
      "epoch": 0.02518,
      "grad_norm": 0.5217525966473281,
      "learning_rate": 0.003,
      "loss": 4.2114,
      "step": 2518
    },
    {
      "epoch": 0.02519,
      "grad_norm": 0.4784613905987089,
      "learning_rate": 0.003,
      "loss": 4.1871,
      "step": 2519
    },
    {
      "epoch": 0.0252,
      "grad_norm": 0.4743563932590768,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2520
    },
    {
      "epoch": 0.02521,
      "grad_norm": 0.487720505759532,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 2521
    },
    {
      "epoch": 0.02522,
      "grad_norm": 0.5322961687406275,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 2522
    },
    {
      "epoch": 0.02523,
      "grad_norm": 0.5314708118979667,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 2523
    },
    {
      "epoch": 0.02524,
      "grad_norm": 0.565643482510681,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2524
    },
    {
      "epoch": 0.02525,
      "grad_norm": 0.5980154086501233,
      "learning_rate": 0.003,
      "loss": 4.2204,
      "step": 2525
    },
    {
      "epoch": 0.02526,
      "grad_norm": 0.668292503382125,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 2526
    },
    {
      "epoch": 0.02527,
      "grad_norm": 0.8972899326232674,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2527
    },
    {
      "epoch": 0.02528,
      "grad_norm": 1.071209575597214,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 2528
    },
    {
      "epoch": 0.02529,
      "grad_norm": 0.7116312192642624,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2529
    },
    {
      "epoch": 0.0253,
      "grad_norm": 0.6832964815647001,
      "learning_rate": 0.003,
      "loss": 4.2324,
      "step": 2530
    },
    {
      "epoch": 0.02531,
      "grad_norm": 0.625455406806653,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 2531
    },
    {
      "epoch": 0.02532,
      "grad_norm": 0.5481673077994964,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2532
    },
    {
      "epoch": 0.02533,
      "grad_norm": 0.5897257835444514,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2533
    },
    {
      "epoch": 0.02534,
      "grad_norm": 0.5146548286556276,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2534
    },
    {
      "epoch": 0.02535,
      "grad_norm": 0.5177859418119969,
      "learning_rate": 0.003,
      "loss": 4.2022,
      "step": 2535
    },
    {
      "epoch": 0.02536,
      "grad_norm": 0.5298155181993902,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 2536
    },
    {
      "epoch": 0.02537,
      "grad_norm": 0.4935521410250174,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 2537
    },
    {
      "epoch": 0.02538,
      "grad_norm": 0.46468573263998625,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 2538
    },
    {
      "epoch": 0.02539,
      "grad_norm": 0.517571911001853,
      "learning_rate": 0.003,
      "loss": 4.2155,
      "step": 2539
    },
    {
      "epoch": 0.0254,
      "grad_norm": 0.5342760221554671,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2540
    },
    {
      "epoch": 0.02541,
      "grad_norm": 0.54814214731398,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2541
    },
    {
      "epoch": 0.02542,
      "grad_norm": 0.6094140763884728,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 2542
    },
    {
      "epoch": 0.02543,
      "grad_norm": 0.7209045415511117,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2543
    },
    {
      "epoch": 0.02544,
      "grad_norm": 0.8126054331647753,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 2544
    },
    {
      "epoch": 0.02545,
      "grad_norm": 0.7645479610935821,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2545
    },
    {
      "epoch": 0.02546,
      "grad_norm": 0.6438407031724117,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2546
    },
    {
      "epoch": 0.02547,
      "grad_norm": 0.6119848106712312,
      "learning_rate": 0.003,
      "loss": 4.2309,
      "step": 2547
    },
    {
      "epoch": 0.02548,
      "grad_norm": 0.5927448337585458,
      "learning_rate": 0.003,
      "loss": 4.2236,
      "step": 2548
    },
    {
      "epoch": 0.02549,
      "grad_norm": 0.8336977020901954,
      "learning_rate": 0.003,
      "loss": 4.2269,
      "step": 2549
    },
    {
      "epoch": 0.0255,
      "grad_norm": 1.0008812022371014,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2550
    },
    {
      "epoch": 0.02551,
      "grad_norm": 1.047497616869583,
      "learning_rate": 0.003,
      "loss": 4.2455,
      "step": 2551
    },
    {
      "epoch": 0.02552,
      "grad_norm": 0.9125531296176879,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2552
    },
    {
      "epoch": 0.02553,
      "grad_norm": 1.0094140787791404,
      "learning_rate": 0.003,
      "loss": 4.2917,
      "step": 2553
    },
    {
      "epoch": 0.02554,
      "grad_norm": 0.9401220726777249,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 2554
    },
    {
      "epoch": 0.02555,
      "grad_norm": 0.8810312726451306,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2555
    },
    {
      "epoch": 0.02556,
      "grad_norm": 0.8398872404556826,
      "learning_rate": 0.003,
      "loss": 4.242,
      "step": 2556
    },
    {
      "epoch": 0.02557,
      "grad_norm": 0.8674958106494866,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 2557
    },
    {
      "epoch": 0.02558,
      "grad_norm": 0.9359611753863145,
      "learning_rate": 0.003,
      "loss": 4.2763,
      "step": 2558
    },
    {
      "epoch": 0.02559,
      "grad_norm": 0.919728435180346,
      "learning_rate": 0.003,
      "loss": 4.2634,
      "step": 2559
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.9704240601933475,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2560
    },
    {
      "epoch": 0.02561,
      "grad_norm": 0.9645281717104208,
      "learning_rate": 0.003,
      "loss": 4.2533,
      "step": 2561
    },
    {
      "epoch": 0.02562,
      "grad_norm": 0.9553791325261367,
      "learning_rate": 0.003,
      "loss": 4.2754,
      "step": 2562
    },
    {
      "epoch": 0.02563,
      "grad_norm": 0.7877299682148988,
      "learning_rate": 0.003,
      "loss": 4.2515,
      "step": 2563
    },
    {
      "epoch": 0.02564,
      "grad_norm": 0.6960170044487899,
      "learning_rate": 0.003,
      "loss": 4.264,
      "step": 2564
    },
    {
      "epoch": 0.02565,
      "grad_norm": 0.5853819163555847,
      "learning_rate": 0.003,
      "loss": 4.2346,
      "step": 2565
    },
    {
      "epoch": 0.02566,
      "grad_norm": 0.56684551943134,
      "learning_rate": 0.003,
      "loss": 4.2543,
      "step": 2566
    },
    {
      "epoch": 0.02567,
      "grad_norm": 0.5518127543434307,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 2567
    },
    {
      "epoch": 0.02568,
      "grad_norm": 0.5196560828829251,
      "learning_rate": 0.003,
      "loss": 4.2402,
      "step": 2568
    },
    {
      "epoch": 0.02569,
      "grad_norm": 0.595356322869423,
      "learning_rate": 0.003,
      "loss": 4.2224,
      "step": 2569
    },
    {
      "epoch": 0.0257,
      "grad_norm": 0.7758881983100933,
      "learning_rate": 0.003,
      "loss": 4.2302,
      "step": 2570
    },
    {
      "epoch": 0.02571,
      "grad_norm": 0.9784633915798607,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 2571
    },
    {
      "epoch": 0.02572,
      "grad_norm": 0.9890948659534822,
      "learning_rate": 0.003,
      "loss": 4.2413,
      "step": 2572
    },
    {
      "epoch": 0.02573,
      "grad_norm": 0.7618440411400482,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2573
    },
    {
      "epoch": 0.02574,
      "grad_norm": 0.5829959121875294,
      "learning_rate": 0.003,
      "loss": 4.2662,
      "step": 2574
    },
    {
      "epoch": 0.02575,
      "grad_norm": 0.5754964401101073,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2575
    },
    {
      "epoch": 0.02576,
      "grad_norm": 0.5023263337601672,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2576
    },
    {
      "epoch": 0.02577,
      "grad_norm": 0.5094113404385299,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2577
    },
    {
      "epoch": 0.02578,
      "grad_norm": 0.45458875378333236,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2578
    },
    {
      "epoch": 0.02579,
      "grad_norm": 0.376766997653558,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 2579
    },
    {
      "epoch": 0.0258,
      "grad_norm": 0.3615714006582123,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 2580
    },
    {
      "epoch": 0.02581,
      "grad_norm": 0.34139864457507546,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2581
    },
    {
      "epoch": 0.02582,
      "grad_norm": 0.31484861485882354,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2582
    },
    {
      "epoch": 0.02583,
      "grad_norm": 0.27908396977569033,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 2583
    },
    {
      "epoch": 0.02584,
      "grad_norm": 0.28343226190675325,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2584
    },
    {
      "epoch": 0.02585,
      "grad_norm": 0.31066810836348263,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 2585
    },
    {
      "epoch": 0.02586,
      "grad_norm": 0.29167242806584853,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2586
    },
    {
      "epoch": 0.02587,
      "grad_norm": 0.3309461372612561,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 2587
    },
    {
      "epoch": 0.02588,
      "grad_norm": 0.3812190342327403,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2588
    },
    {
      "epoch": 0.02589,
      "grad_norm": 0.48608400124636386,
      "learning_rate": 0.003,
      "loss": 4.1958,
      "step": 2589
    },
    {
      "epoch": 0.0259,
      "grad_norm": 0.6708650835599492,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2590
    },
    {
      "epoch": 0.02591,
      "grad_norm": 0.7815243168293263,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 2591
    },
    {
      "epoch": 0.02592,
      "grad_norm": 0.6070733396030551,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2592
    },
    {
      "epoch": 0.02593,
      "grad_norm": 0.36755746327659883,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2593
    },
    {
      "epoch": 0.02594,
      "grad_norm": 0.5461931648813682,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2594
    },
    {
      "epoch": 0.02595,
      "grad_norm": 0.5548085387163001,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 2595
    },
    {
      "epoch": 0.02596,
      "grad_norm": 0.6279120258670815,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2596
    },
    {
      "epoch": 0.02597,
      "grad_norm": 0.6833879500904073,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 2597
    },
    {
      "epoch": 0.02598,
      "grad_norm": 0.7079000881755875,
      "learning_rate": 0.003,
      "loss": 4.2216,
      "step": 2598
    },
    {
      "epoch": 0.02599,
      "grad_norm": 0.6366436124583713,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2599
    },
    {
      "epoch": 0.026,
      "grad_norm": 0.5901740968221895,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 2600
    },
    {
      "epoch": 0.02601,
      "grad_norm": 0.5956938888246321,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2601
    },
    {
      "epoch": 0.02602,
      "grad_norm": 0.5493627512392831,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 2602
    },
    {
      "epoch": 0.02603,
      "grad_norm": 0.44261728447277765,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 2603
    },
    {
      "epoch": 0.02604,
      "grad_norm": 0.47710609408550947,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 2604
    },
    {
      "epoch": 0.02605,
      "grad_norm": 0.4918626291452229,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 2605
    },
    {
      "epoch": 0.02606,
      "grad_norm": 0.46913897415532124,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2606
    },
    {
      "epoch": 0.02607,
      "grad_norm": 0.48033538034271883,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 2607
    },
    {
      "epoch": 0.02608,
      "grad_norm": 0.5430081805575081,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 2608
    },
    {
      "epoch": 0.02609,
      "grad_norm": 0.7745086701981942,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2609
    },
    {
      "epoch": 0.0261,
      "grad_norm": 0.9079228266743159,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2610
    },
    {
      "epoch": 0.02611,
      "grad_norm": 0.8915671065001584,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 2611
    },
    {
      "epoch": 0.02612,
      "grad_norm": 0.8970335727218054,
      "learning_rate": 0.003,
      "loss": 4.2272,
      "step": 2612
    },
    {
      "epoch": 0.02613,
      "grad_norm": 0.7184841949590459,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 2613
    },
    {
      "epoch": 0.02614,
      "grad_norm": 0.6361221793660993,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 2614
    },
    {
      "epoch": 0.02615,
      "grad_norm": 0.6008095298009088,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 2615
    },
    {
      "epoch": 0.02616,
      "grad_norm": 0.5894242905465984,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2616
    },
    {
      "epoch": 0.02617,
      "grad_norm": 0.6436011542740302,
      "learning_rate": 0.003,
      "loss": 4.1808,
      "step": 2617
    },
    {
      "epoch": 0.02618,
      "grad_norm": 0.7283160674464684,
      "learning_rate": 0.003,
      "loss": 4.2126,
      "step": 2618
    },
    {
      "epoch": 0.02619,
      "grad_norm": 0.7948866659668812,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2619
    },
    {
      "epoch": 0.0262,
      "grad_norm": 0.7483085908476613,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2620
    },
    {
      "epoch": 0.02621,
      "grad_norm": 0.6674824981271792,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 2621
    },
    {
      "epoch": 0.02622,
      "grad_norm": 0.6038002286952583,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2622
    },
    {
      "epoch": 0.02623,
      "grad_norm": 0.6014737867483161,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2623
    },
    {
      "epoch": 0.02624,
      "grad_norm": 0.5903601826725474,
      "learning_rate": 0.003,
      "loss": 4.2493,
      "step": 2624
    },
    {
      "epoch": 0.02625,
      "grad_norm": 0.6662689883238633,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2625
    },
    {
      "epoch": 0.02626,
      "grad_norm": 0.6773686120638865,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 2626
    },
    {
      "epoch": 0.02627,
      "grad_norm": 0.6189409649870486,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 2627
    },
    {
      "epoch": 0.02628,
      "grad_norm": 0.6566341590011362,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2628
    },
    {
      "epoch": 0.02629,
      "grad_norm": 0.7499843369890263,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 2629
    },
    {
      "epoch": 0.0263,
      "grad_norm": 0.7612666434746392,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 2630
    },
    {
      "epoch": 0.02631,
      "grad_norm": 0.6535861726328742,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 2631
    },
    {
      "epoch": 0.02632,
      "grad_norm": 0.6394295181872582,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 2632
    },
    {
      "epoch": 0.02633,
      "grad_norm": 0.6261623421196629,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 2633
    },
    {
      "epoch": 0.02634,
      "grad_norm": 0.574244319715974,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2634
    },
    {
      "epoch": 0.02635,
      "grad_norm": 0.4907934619440069,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2635
    },
    {
      "epoch": 0.02636,
      "grad_norm": 0.47213288352020044,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2636
    },
    {
      "epoch": 0.02637,
      "grad_norm": 0.4457276996768415,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2637
    },
    {
      "epoch": 0.02638,
      "grad_norm": 0.45345762795109046,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2638
    },
    {
      "epoch": 0.02639,
      "grad_norm": 0.49627655516291913,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 2639
    },
    {
      "epoch": 0.0264,
      "grad_norm": 0.5785078427721256,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 2640
    },
    {
      "epoch": 0.02641,
      "grad_norm": 0.7855264178492048,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 2641
    },
    {
      "epoch": 0.02642,
      "grad_norm": 0.9564797199743369,
      "learning_rate": 0.003,
      "loss": 4.2106,
      "step": 2642
    },
    {
      "epoch": 0.02643,
      "grad_norm": 0.9009180286527748,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 2643
    },
    {
      "epoch": 0.02644,
      "grad_norm": 0.6806094030785705,
      "learning_rate": 0.003,
      "loss": 4.2066,
      "step": 2644
    },
    {
      "epoch": 0.02645,
      "grad_norm": 0.6950198239199242,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2645
    },
    {
      "epoch": 0.02646,
      "grad_norm": 0.6983266391512944,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2646
    },
    {
      "epoch": 0.02647,
      "grad_norm": 0.6601491893781873,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 2647
    },
    {
      "epoch": 0.02648,
      "grad_norm": 0.6355139962402977,
      "learning_rate": 0.003,
      "loss": 4.2257,
      "step": 2648
    },
    {
      "epoch": 0.02649,
      "grad_norm": 0.5820480039604559,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2649
    },
    {
      "epoch": 0.0265,
      "grad_norm": 0.5775761818558494,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 2650
    },
    {
      "epoch": 0.02651,
      "grad_norm": 0.5277676161314399,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2651
    },
    {
      "epoch": 0.02652,
      "grad_norm": 0.5297857391988389,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2652
    },
    {
      "epoch": 0.02653,
      "grad_norm": 0.5897074985274656,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 2653
    },
    {
      "epoch": 0.02654,
      "grad_norm": 0.6546317550221565,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 2654
    },
    {
      "epoch": 0.02655,
      "grad_norm": 0.6571917314744374,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 2655
    },
    {
      "epoch": 0.02656,
      "grad_norm": 0.6300742703920978,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 2656
    },
    {
      "epoch": 0.02657,
      "grad_norm": 0.6437986671993883,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 2657
    },
    {
      "epoch": 0.02658,
      "grad_norm": 0.6540624785623048,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2658
    },
    {
      "epoch": 0.02659,
      "grad_norm": 0.6833039431896774,
      "learning_rate": 0.003,
      "loss": 4.2004,
      "step": 2659
    },
    {
      "epoch": 0.0266,
      "grad_norm": 0.6560030968945707,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 2660
    },
    {
      "epoch": 0.02661,
      "grad_norm": 0.6045077056358651,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2661
    },
    {
      "epoch": 0.02662,
      "grad_norm": 0.5982743265464419,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 2662
    },
    {
      "epoch": 0.02663,
      "grad_norm": 0.6142468679426575,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 2663
    },
    {
      "epoch": 0.02664,
      "grad_norm": 0.5662553451652035,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 2664
    },
    {
      "epoch": 0.02665,
      "grad_norm": 0.5761075041061567,
      "learning_rate": 0.003,
      "loss": 4.2236,
      "step": 2665
    },
    {
      "epoch": 0.02666,
      "grad_norm": 0.6870114427260798,
      "learning_rate": 0.003,
      "loss": 4.2193,
      "step": 2666
    },
    {
      "epoch": 0.02667,
      "grad_norm": 0.8311453279031553,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 2667
    },
    {
      "epoch": 0.02668,
      "grad_norm": 0.9940869003642662,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 2668
    },
    {
      "epoch": 0.02669,
      "grad_norm": 0.9365809099626523,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2669
    },
    {
      "epoch": 0.0267,
      "grad_norm": 0.7814988387932911,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2670
    },
    {
      "epoch": 0.02671,
      "grad_norm": 0.7536334880750852,
      "learning_rate": 0.003,
      "loss": 4.2232,
      "step": 2671
    },
    {
      "epoch": 0.02672,
      "grad_norm": 0.6570195531624667,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2672
    },
    {
      "epoch": 0.02673,
      "grad_norm": 0.5512466402663184,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2673
    },
    {
      "epoch": 0.02674,
      "grad_norm": 0.61723011024151,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 2674
    },
    {
      "epoch": 0.02675,
      "grad_norm": 0.6694888220820355,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 2675
    },
    {
      "epoch": 0.02676,
      "grad_norm": 0.684903042237445,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2676
    },
    {
      "epoch": 0.02677,
      "grad_norm": 0.6916594775917959,
      "learning_rate": 0.003,
      "loss": 4.2254,
      "step": 2677
    },
    {
      "epoch": 0.02678,
      "grad_norm": 0.5888995948139836,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2678
    },
    {
      "epoch": 0.02679,
      "grad_norm": 0.46001488738668683,
      "learning_rate": 0.003,
      "loss": 4.1818,
      "step": 2679
    },
    {
      "epoch": 0.0268,
      "grad_norm": 0.48559488053080857,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 2680
    },
    {
      "epoch": 0.02681,
      "grad_norm": 0.5328577853550642,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2681
    },
    {
      "epoch": 0.02682,
      "grad_norm": 0.5478209027693539,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 2682
    },
    {
      "epoch": 0.02683,
      "grad_norm": 0.5231582728945893,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2683
    },
    {
      "epoch": 0.02684,
      "grad_norm": 0.5004578011633336,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2684
    },
    {
      "epoch": 0.02685,
      "grad_norm": 0.5339999052577957,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2685
    },
    {
      "epoch": 0.02686,
      "grad_norm": 0.5808031106775001,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2686
    },
    {
      "epoch": 0.02687,
      "grad_norm": 0.6474970996641412,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 2687
    },
    {
      "epoch": 0.02688,
      "grad_norm": 0.6071304394655341,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 2688
    },
    {
      "epoch": 0.02689,
      "grad_norm": 0.6303081168202617,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 2689
    },
    {
      "epoch": 0.0269,
      "grad_norm": 0.6875418523901222,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 2690
    },
    {
      "epoch": 0.02691,
      "grad_norm": 0.8385284997095487,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2691
    },
    {
      "epoch": 0.02692,
      "grad_norm": 0.9480846054518673,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2692
    },
    {
      "epoch": 0.02693,
      "grad_norm": 0.8071009083907058,
      "learning_rate": 0.003,
      "loss": 4.2239,
      "step": 2693
    },
    {
      "epoch": 0.02694,
      "grad_norm": 0.726325622879689,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 2694
    },
    {
      "epoch": 0.02695,
      "grad_norm": 0.6357501756622079,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2695
    },
    {
      "epoch": 0.02696,
      "grad_norm": 0.5914031503189017,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2696
    },
    {
      "epoch": 0.02697,
      "grad_norm": 0.6523831217047013,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2697
    },
    {
      "epoch": 0.02698,
      "grad_norm": 0.6028229475417466,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2698
    },
    {
      "epoch": 0.02699,
      "grad_norm": 0.5476388167406291,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2699
    },
    {
      "epoch": 0.027,
      "grad_norm": 0.4742225811072004,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 2700
    },
    {
      "epoch": 0.02701,
      "grad_norm": 0.4202221652460036,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 2701
    },
    {
      "epoch": 0.02702,
      "grad_norm": 0.43365355703010106,
      "learning_rate": 0.003,
      "loss": 4.2144,
      "step": 2702
    },
    {
      "epoch": 0.02703,
      "grad_norm": 0.4838782470885569,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2703
    },
    {
      "epoch": 0.02704,
      "grad_norm": 0.5383020066295563,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 2704
    },
    {
      "epoch": 0.02705,
      "grad_norm": 0.6989516882897651,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2705
    },
    {
      "epoch": 0.02706,
      "grad_norm": 0.7486144178237021,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 2706
    },
    {
      "epoch": 0.02707,
      "grad_norm": 0.7126817427692109,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 2707
    },
    {
      "epoch": 0.02708,
      "grad_norm": 0.6105439188174862,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2708
    },
    {
      "epoch": 0.02709,
      "grad_norm": 0.5700397686001112,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 2709
    },
    {
      "epoch": 0.0271,
      "grad_norm": 0.6208775729436938,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 2710
    },
    {
      "epoch": 0.02711,
      "grad_norm": 0.6014094255252589,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 2711
    },
    {
      "epoch": 0.02712,
      "grad_norm": 0.5261948655268764,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 2712
    },
    {
      "epoch": 0.02713,
      "grad_norm": 0.5550123847377345,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 2713
    },
    {
      "epoch": 0.02714,
      "grad_norm": 0.56736840403282,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 2714
    },
    {
      "epoch": 0.02715,
      "grad_norm": 0.6628029062916901,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2715
    },
    {
      "epoch": 0.02716,
      "grad_norm": 0.768979209194684,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 2716
    },
    {
      "epoch": 0.02717,
      "grad_norm": 0.7482884136468829,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 2717
    },
    {
      "epoch": 0.02718,
      "grad_norm": 0.8846409333789726,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2718
    },
    {
      "epoch": 0.02719,
      "grad_norm": 0.9429345703060683,
      "learning_rate": 0.003,
      "loss": 4.2045,
      "step": 2719
    },
    {
      "epoch": 0.0272,
      "grad_norm": 1.002947468833307,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2720
    },
    {
      "epoch": 0.02721,
      "grad_norm": 1.0695835852166422,
      "learning_rate": 0.003,
      "loss": 4.2286,
      "step": 2721
    },
    {
      "epoch": 0.02722,
      "grad_norm": 1.1056646889932085,
      "learning_rate": 0.003,
      "loss": 4.2273,
      "step": 2722
    },
    {
      "epoch": 0.02723,
      "grad_norm": 0.8267701861491207,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 2723
    },
    {
      "epoch": 0.02724,
      "grad_norm": 0.7367716189571131,
      "learning_rate": 0.003,
      "loss": 4.2292,
      "step": 2724
    },
    {
      "epoch": 0.02725,
      "grad_norm": 0.6044896083683174,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2725
    },
    {
      "epoch": 0.02726,
      "grad_norm": 0.6956030827952617,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 2726
    },
    {
      "epoch": 0.02727,
      "grad_norm": 0.7192323257722564,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2727
    },
    {
      "epoch": 0.02728,
      "grad_norm": 0.7783208693346495,
      "learning_rate": 0.003,
      "loss": 4.2088,
      "step": 2728
    },
    {
      "epoch": 0.02729,
      "grad_norm": 0.8318571752999154,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 2729
    },
    {
      "epoch": 0.0273,
      "grad_norm": 0.6686152556603595,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2730
    },
    {
      "epoch": 0.02731,
      "grad_norm": 0.6832226230920126,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 2731
    },
    {
      "epoch": 0.02732,
      "grad_norm": 0.6331150968875209,
      "learning_rate": 0.003,
      "loss": 4.1987,
      "step": 2732
    },
    {
      "epoch": 0.02733,
      "grad_norm": 0.5278467341667037,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2733
    },
    {
      "epoch": 0.02734,
      "grad_norm": 0.6138764744804129,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 2734
    },
    {
      "epoch": 0.02735,
      "grad_norm": 0.561739809597342,
      "learning_rate": 0.003,
      "loss": 4.2193,
      "step": 2735
    },
    {
      "epoch": 0.02736,
      "grad_norm": 0.5908346453733819,
      "learning_rate": 0.003,
      "loss": 4.1928,
      "step": 2736
    },
    {
      "epoch": 0.02737,
      "grad_norm": 0.6511959635659452,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 2737
    },
    {
      "epoch": 0.02738,
      "grad_norm": 0.758698109151286,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2738
    },
    {
      "epoch": 0.02739,
      "grad_norm": 0.6956474809291019,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2739
    },
    {
      "epoch": 0.0274,
      "grad_norm": 0.5058919640608095,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 2740
    },
    {
      "epoch": 0.02741,
      "grad_norm": 0.5563232333292056,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2741
    },
    {
      "epoch": 0.02742,
      "grad_norm": 0.5987655354363043,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 2742
    },
    {
      "epoch": 0.02743,
      "grad_norm": 0.5914696396737046,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2743
    },
    {
      "epoch": 0.02744,
      "grad_norm": 0.5630038713206789,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2744
    },
    {
      "epoch": 0.02745,
      "grad_norm": 0.587542894365965,
      "learning_rate": 0.003,
      "loss": 4.2027,
      "step": 2745
    },
    {
      "epoch": 0.02746,
      "grad_norm": 0.5934042861360144,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 2746
    },
    {
      "epoch": 0.02747,
      "grad_norm": 0.6474787518036716,
      "learning_rate": 0.003,
      "loss": 4.212,
      "step": 2747
    },
    {
      "epoch": 0.02748,
      "grad_norm": 0.8113625213803506,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2748
    },
    {
      "epoch": 0.02749,
      "grad_norm": 0.8229694069124209,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 2749
    },
    {
      "epoch": 0.0275,
      "grad_norm": 0.6793478072709392,
      "learning_rate": 0.003,
      "loss": 4.2077,
      "step": 2750
    },
    {
      "epoch": 0.02751,
      "grad_norm": 0.5656831593340759,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 2751
    },
    {
      "epoch": 0.02752,
      "grad_norm": 0.5267924527455705,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2752
    },
    {
      "epoch": 0.02753,
      "grad_norm": 0.502928202841188,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 2753
    },
    {
      "epoch": 0.02754,
      "grad_norm": 0.5044833881071361,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 2754
    },
    {
      "epoch": 0.02755,
      "grad_norm": 0.5267594677995416,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 2755
    },
    {
      "epoch": 0.02756,
      "grad_norm": 0.456921415827541,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 2756
    },
    {
      "epoch": 0.02757,
      "grad_norm": 0.3836455619810196,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 2757
    },
    {
      "epoch": 0.02758,
      "grad_norm": 0.34559498426529056,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 2758
    },
    {
      "epoch": 0.02759,
      "grad_norm": 0.34478709094038296,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 2759
    },
    {
      "epoch": 0.0276,
      "grad_norm": 0.4284232121155303,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 2760
    },
    {
      "epoch": 0.02761,
      "grad_norm": 0.5071377792443562,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2761
    },
    {
      "epoch": 0.02762,
      "grad_norm": 0.6171334423886827,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 2762
    },
    {
      "epoch": 0.02763,
      "grad_norm": 0.7563311828472695,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 2763
    },
    {
      "epoch": 0.02764,
      "grad_norm": 0.8185996809603724,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 2764
    },
    {
      "epoch": 0.02765,
      "grad_norm": 0.7325003643978839,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2765
    },
    {
      "epoch": 0.02766,
      "grad_norm": 0.6977141939643235,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 2766
    },
    {
      "epoch": 0.02767,
      "grad_norm": 0.7178457591084105,
      "learning_rate": 0.003,
      "loss": 4.2198,
      "step": 2767
    },
    {
      "epoch": 0.02768,
      "grad_norm": 0.6603543944468468,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 2768
    },
    {
      "epoch": 0.02769,
      "grad_norm": 0.5908808640128104,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2769
    },
    {
      "epoch": 0.0277,
      "grad_norm": 0.5584733753440203,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 2770
    },
    {
      "epoch": 0.02771,
      "grad_norm": 0.5513412132170109,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 2771
    },
    {
      "epoch": 0.02772,
      "grad_norm": 0.601007172067716,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 2772
    },
    {
      "epoch": 0.02773,
      "grad_norm": 0.6222859555759426,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 2773
    },
    {
      "epoch": 0.02774,
      "grad_norm": 0.6298455082538394,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 2774
    },
    {
      "epoch": 0.02775,
      "grad_norm": 0.6384186594957891,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 2775
    },
    {
      "epoch": 0.02776,
      "grad_norm": 0.7183697808219275,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 2776
    },
    {
      "epoch": 0.02777,
      "grad_norm": 0.6932591793310324,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 2777
    },
    {
      "epoch": 0.02778,
      "grad_norm": 0.6667113422946584,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 2778
    },
    {
      "epoch": 0.02779,
      "grad_norm": 0.7407795606572514,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 2779
    },
    {
      "epoch": 0.0278,
      "grad_norm": 0.7295794698337943,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 2780
    },
    {
      "epoch": 0.02781,
      "grad_norm": 0.6893914165082023,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 2781
    },
    {
      "epoch": 0.02782,
      "grad_norm": 0.6620017500188085,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 2782
    },
    {
      "epoch": 0.02783,
      "grad_norm": 0.6131468932443346,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 2783
    },
    {
      "epoch": 0.02784,
      "grad_norm": 0.5579371518365059,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 2784
    },
    {
      "epoch": 0.02785,
      "grad_norm": 0.49723848657236286,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 2785
    },
    {
      "epoch": 0.02786,
      "grad_norm": 0.5125002182675953,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 2786
    },
    {
      "epoch": 0.02787,
      "grad_norm": 0.4914565242864978,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 2787
    },
    {
      "epoch": 0.02788,
      "grad_norm": 0.526484975930702,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 2788
    },
    {
      "epoch": 0.02789,
      "grad_norm": 0.5321807051700945,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2789
    },
    {
      "epoch": 0.0279,
      "grad_norm": 0.5840495108099473,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 2790
    },
    {
      "epoch": 0.02791,
      "grad_norm": 0.9202391536169947,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 2791
    },
    {
      "epoch": 0.02792,
      "grad_norm": 1.2841478556987087,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 2792
    },
    {
      "epoch": 0.02793,
      "grad_norm": 0.6986972433822171,
      "learning_rate": 0.003,
      "loss": 4.2026,
      "step": 2793
    },
    {
      "epoch": 0.02794,
      "grad_norm": 0.7400669972097395,
      "learning_rate": 0.003,
      "loss": 4.2072,
      "step": 2794
    },
    {
      "epoch": 0.02795,
      "grad_norm": 0.7145362265408897,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2795
    },
    {
      "epoch": 0.02796,
      "grad_norm": 0.7481134935341638,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2796
    },
    {
      "epoch": 0.02797,
      "grad_norm": 0.8256336591556148,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2797
    },
    {
      "epoch": 0.02798,
      "grad_norm": 0.7297049193325166,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 2798
    },
    {
      "epoch": 0.02799,
      "grad_norm": 0.7233108511585719,
      "learning_rate": 0.003,
      "loss": 4.2063,
      "step": 2799
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.6743448137124989,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2800
    },
    {
      "epoch": 0.02801,
      "grad_norm": 0.5867417428513181,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2801
    },
    {
      "epoch": 0.02802,
      "grad_norm": 0.557083005494101,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 2802
    },
    {
      "epoch": 0.02803,
      "grad_norm": 0.5239629190819096,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2803
    },
    {
      "epoch": 0.02804,
      "grad_norm": 0.6122062528440474,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 2804
    },
    {
      "epoch": 0.02805,
      "grad_norm": 0.6123404104585692,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2805
    },
    {
      "epoch": 0.02806,
      "grad_norm": 0.5813003741795452,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2806
    },
    {
      "epoch": 0.02807,
      "grad_norm": 0.555658293190141,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 2807
    },
    {
      "epoch": 0.02808,
      "grad_norm": 0.4856792434285324,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 2808
    },
    {
      "epoch": 0.02809,
      "grad_norm": 0.468227745280838,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2809
    },
    {
      "epoch": 0.0281,
      "grad_norm": 0.5953915884658988,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 2810
    },
    {
      "epoch": 0.02811,
      "grad_norm": 0.9133221862875325,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 2811
    },
    {
      "epoch": 0.02812,
      "grad_norm": 1.275925834959494,
      "learning_rate": 0.003,
      "loss": 4.2304,
      "step": 2812
    },
    {
      "epoch": 0.02813,
      "grad_norm": 0.8483651283891129,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2813
    },
    {
      "epoch": 0.02814,
      "grad_norm": 0.7605354432635538,
      "learning_rate": 0.003,
      "loss": 4.202,
      "step": 2814
    },
    {
      "epoch": 0.02815,
      "grad_norm": 0.7567043943015741,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2815
    },
    {
      "epoch": 0.02816,
      "grad_norm": 0.7448943831936015,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2816
    },
    {
      "epoch": 0.02817,
      "grad_norm": 0.7400121216625585,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2817
    },
    {
      "epoch": 0.02818,
      "grad_norm": 0.7612379755026737,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2818
    },
    {
      "epoch": 0.02819,
      "grad_norm": 0.8916358831666468,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2819
    },
    {
      "epoch": 0.0282,
      "grad_norm": 0.9386399567045983,
      "learning_rate": 0.003,
      "loss": 4.2191,
      "step": 2820
    },
    {
      "epoch": 0.02821,
      "grad_norm": 0.7968339375438817,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2821
    },
    {
      "epoch": 0.02822,
      "grad_norm": 0.6707514048211725,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2822
    },
    {
      "epoch": 0.02823,
      "grad_norm": 0.5937548176700269,
      "learning_rate": 0.003,
      "loss": 4.2229,
      "step": 2823
    },
    {
      "epoch": 0.02824,
      "grad_norm": 0.6160995840345511,
      "learning_rate": 0.003,
      "loss": 4.2078,
      "step": 2824
    },
    {
      "epoch": 0.02825,
      "grad_norm": 0.5897186283971302,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2825
    },
    {
      "epoch": 0.02826,
      "grad_norm": 0.5674030456447229,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 2826
    },
    {
      "epoch": 0.02827,
      "grad_norm": 0.5509162115210677,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 2827
    },
    {
      "epoch": 0.02828,
      "grad_norm": 0.5443211688206793,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 2828
    },
    {
      "epoch": 0.02829,
      "grad_norm": 0.5032331740089974,
      "learning_rate": 0.003,
      "loss": 4.227,
      "step": 2829
    },
    {
      "epoch": 0.0283,
      "grad_norm": 0.5225637902550168,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 2830
    },
    {
      "epoch": 0.02831,
      "grad_norm": 0.5311323075927029,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 2831
    },
    {
      "epoch": 0.02832,
      "grad_norm": 0.5001952573238827,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2832
    },
    {
      "epoch": 0.02833,
      "grad_norm": 0.466662882413672,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 2833
    },
    {
      "epoch": 0.02834,
      "grad_norm": 0.4772870203488476,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 2834
    },
    {
      "epoch": 0.02835,
      "grad_norm": 0.5535871706800527,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 2835
    },
    {
      "epoch": 0.02836,
      "grad_norm": 0.7239867170041147,
      "learning_rate": 0.003,
      "loss": 4.2004,
      "step": 2836
    },
    {
      "epoch": 0.02837,
      "grad_norm": 0.7597045607127227,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 2837
    },
    {
      "epoch": 0.02838,
      "grad_norm": 0.7252587205241631,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 2838
    },
    {
      "epoch": 0.02839,
      "grad_norm": 0.8402252149672428,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 2839
    },
    {
      "epoch": 0.0284,
      "grad_norm": 0.8241726365065422,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 2840
    },
    {
      "epoch": 0.02841,
      "grad_norm": 0.8096094003742889,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 2841
    },
    {
      "epoch": 0.02842,
      "grad_norm": 0.7156086401419698,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 2842
    },
    {
      "epoch": 0.02843,
      "grad_norm": 0.6040106819890795,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 2843
    },
    {
      "epoch": 0.02844,
      "grad_norm": 0.5060645820894467,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2844
    },
    {
      "epoch": 0.02845,
      "grad_norm": 0.5041938926420962,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 2845
    },
    {
      "epoch": 0.02846,
      "grad_norm": 0.5025147774215044,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 2846
    },
    {
      "epoch": 0.02847,
      "grad_norm": 0.5625355648833802,
      "learning_rate": 0.003,
      "loss": 4.2236,
      "step": 2847
    },
    {
      "epoch": 0.02848,
      "grad_norm": 0.6048155911479377,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 2848
    },
    {
      "epoch": 0.02849,
      "grad_norm": 0.593616598386133,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 2849
    },
    {
      "epoch": 0.0285,
      "grad_norm": 0.5421469421547838,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 2850
    },
    {
      "epoch": 0.02851,
      "grad_norm": 0.5145785466987678,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2851
    },
    {
      "epoch": 0.02852,
      "grad_norm": 0.5364202493290657,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 2852
    },
    {
      "epoch": 0.02853,
      "grad_norm": 0.5092167845332397,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 2853
    },
    {
      "epoch": 0.02854,
      "grad_norm": 0.505959030155662,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 2854
    },
    {
      "epoch": 0.02855,
      "grad_norm": 0.48727747027775575,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 2855
    },
    {
      "epoch": 0.02856,
      "grad_norm": 0.5114663663985488,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2856
    },
    {
      "epoch": 0.02857,
      "grad_norm": 0.6029175042652142,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2857
    },
    {
      "epoch": 0.02858,
      "grad_norm": 0.811258279718228,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 2858
    },
    {
      "epoch": 0.02859,
      "grad_norm": 1.000140796502162,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 2859
    },
    {
      "epoch": 0.0286,
      "grad_norm": 1.2449496760884864,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2860
    },
    {
      "epoch": 0.02861,
      "grad_norm": 0.7665309921680512,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 2861
    },
    {
      "epoch": 0.02862,
      "grad_norm": 0.7010737782221613,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2862
    },
    {
      "epoch": 0.02863,
      "grad_norm": 0.6652925807236452,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2863
    },
    {
      "epoch": 0.02864,
      "grad_norm": 0.6660638892873599,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2864
    },
    {
      "epoch": 0.02865,
      "grad_norm": 0.7030991193559302,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2865
    },
    {
      "epoch": 0.02866,
      "grad_norm": 0.6597702895413415,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 2866
    },
    {
      "epoch": 0.02867,
      "grad_norm": 0.582862490658125,
      "learning_rate": 0.003,
      "loss": 4.1819,
      "step": 2867
    },
    {
      "epoch": 0.02868,
      "grad_norm": 0.5208311845539698,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2868
    },
    {
      "epoch": 0.02869,
      "grad_norm": 0.5269361881811672,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2869
    },
    {
      "epoch": 0.0287,
      "grad_norm": 0.49588966178431665,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2870
    },
    {
      "epoch": 0.02871,
      "grad_norm": 0.4427797615796878,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2871
    },
    {
      "epoch": 0.02872,
      "grad_norm": 0.46354755259563285,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2872
    },
    {
      "epoch": 0.02873,
      "grad_norm": 0.5003897365609394,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 2873
    },
    {
      "epoch": 0.02874,
      "grad_norm": 0.5095674852487324,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2874
    },
    {
      "epoch": 0.02875,
      "grad_norm": 0.5767552596221609,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 2875
    },
    {
      "epoch": 0.02876,
      "grad_norm": 0.7553686836410436,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 2876
    },
    {
      "epoch": 0.02877,
      "grad_norm": 0.927267465496802,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 2877
    },
    {
      "epoch": 0.02878,
      "grad_norm": 0.8275286861154829,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 2878
    },
    {
      "epoch": 0.02879,
      "grad_norm": 0.6254369579498115,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 2879
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.6608202030691483,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2880
    },
    {
      "epoch": 0.02881,
      "grad_norm": 0.6264026993054118,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2881
    },
    {
      "epoch": 0.02882,
      "grad_norm": 0.6456049764969926,
      "learning_rate": 0.003,
      "loss": 4.2046,
      "step": 2882
    },
    {
      "epoch": 0.02883,
      "grad_norm": 0.631140439713875,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2883
    },
    {
      "epoch": 0.02884,
      "grad_norm": 0.6167258778841849,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2884
    },
    {
      "epoch": 0.02885,
      "grad_norm": 0.6237538943365489,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 2885
    },
    {
      "epoch": 0.02886,
      "grad_norm": 0.577915341916395,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 2886
    },
    {
      "epoch": 0.02887,
      "grad_norm": 0.5869716498143316,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 2887
    },
    {
      "epoch": 0.02888,
      "grad_norm": 0.6367834629014674,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2888
    },
    {
      "epoch": 0.02889,
      "grad_norm": 0.6424828882281004,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 2889
    },
    {
      "epoch": 0.0289,
      "grad_norm": 0.7456946160039144,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2890
    },
    {
      "epoch": 0.02891,
      "grad_norm": 0.8611032962542791,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2891
    },
    {
      "epoch": 0.02892,
      "grad_norm": 1.0298003752084743,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2892
    },
    {
      "epoch": 0.02893,
      "grad_norm": 1.0054991154141122,
      "learning_rate": 0.003,
      "loss": 4.1976,
      "step": 2893
    },
    {
      "epoch": 0.02894,
      "grad_norm": 0.8964526887398983,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2894
    },
    {
      "epoch": 0.02895,
      "grad_norm": 0.786662499637371,
      "learning_rate": 0.003,
      "loss": 4.2494,
      "step": 2895
    },
    {
      "epoch": 0.02896,
      "grad_norm": 0.8924645045032006,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2896
    },
    {
      "epoch": 0.02897,
      "grad_norm": 0.9444560121857192,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2897
    },
    {
      "epoch": 0.02898,
      "grad_norm": 0.9041547595653124,
      "learning_rate": 0.003,
      "loss": 4.2475,
      "step": 2898
    },
    {
      "epoch": 0.02899,
      "grad_norm": 0.8479058583034529,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2899
    },
    {
      "epoch": 0.029,
      "grad_norm": 0.8224886244859558,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2900
    },
    {
      "epoch": 0.02901,
      "grad_norm": 0.8698134304184088,
      "learning_rate": 0.003,
      "loss": 4.2375,
      "step": 2901
    },
    {
      "epoch": 0.02902,
      "grad_norm": 0.8486827254456236,
      "learning_rate": 0.003,
      "loss": 4.2524,
      "step": 2902
    },
    {
      "epoch": 0.02903,
      "grad_norm": 0.7415128492690074,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2903
    },
    {
      "epoch": 0.02904,
      "grad_norm": 0.7083318227885205,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 2904
    },
    {
      "epoch": 0.02905,
      "grad_norm": 0.7648353967865369,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2905
    },
    {
      "epoch": 0.02906,
      "grad_norm": 0.7593644185333667,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 2906
    },
    {
      "epoch": 0.02907,
      "grad_norm": 0.7189838848454702,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 2907
    },
    {
      "epoch": 0.02908,
      "grad_norm": 0.7101023301849051,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2908
    },
    {
      "epoch": 0.02909,
      "grad_norm": 0.6273709094950615,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 2909
    },
    {
      "epoch": 0.0291,
      "grad_norm": 0.5304669195124173,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2910
    },
    {
      "epoch": 0.02911,
      "grad_norm": 0.509577660549234,
      "learning_rate": 0.003,
      "loss": 4.2021,
      "step": 2911
    },
    {
      "epoch": 0.02912,
      "grad_norm": 0.4255206741887875,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 2912
    },
    {
      "epoch": 0.02913,
      "grad_norm": 0.4561469328827923,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2913
    },
    {
      "epoch": 0.02914,
      "grad_norm": 0.42117739519724395,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 2914
    },
    {
      "epoch": 0.02915,
      "grad_norm": 0.4011573145594706,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 2915
    },
    {
      "epoch": 0.02916,
      "grad_norm": 0.42970374324262517,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 2916
    },
    {
      "epoch": 0.02917,
      "grad_norm": 0.33514332187348916,
      "learning_rate": 0.003,
      "loss": 4.2087,
      "step": 2917
    },
    {
      "epoch": 0.02918,
      "grad_norm": 0.34222241717451907,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 2918
    },
    {
      "epoch": 0.02919,
      "grad_norm": 0.36583286994162584,
      "learning_rate": 0.003,
      "loss": 4.2045,
      "step": 2919
    },
    {
      "epoch": 0.0292,
      "grad_norm": 0.4446133039291897,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 2920
    },
    {
      "epoch": 0.02921,
      "grad_norm": 0.6109713288069767,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 2921
    },
    {
      "epoch": 0.02922,
      "grad_norm": 0.9167859634758244,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 2922
    },
    {
      "epoch": 0.02923,
      "grad_norm": 1.009808447907654,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 2923
    },
    {
      "epoch": 0.02924,
      "grad_norm": 0.8524826722350543,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2924
    },
    {
      "epoch": 0.02925,
      "grad_norm": 0.8032284343679224,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2925
    },
    {
      "epoch": 0.02926,
      "grad_norm": 0.749692494912656,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2926
    },
    {
      "epoch": 0.02927,
      "grad_norm": 0.7088429283617566,
      "learning_rate": 0.003,
      "loss": 4.1975,
      "step": 2927
    },
    {
      "epoch": 0.02928,
      "grad_norm": 0.6042074975753539,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2928
    },
    {
      "epoch": 0.02929,
      "grad_norm": 0.563152730036225,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2929
    },
    {
      "epoch": 0.0293,
      "grad_norm": 0.44275110512836524,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 2930
    },
    {
      "epoch": 0.02931,
      "grad_norm": 0.4579944440403804,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2931
    },
    {
      "epoch": 0.02932,
      "grad_norm": 0.4515235240773822,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 2932
    },
    {
      "epoch": 0.02933,
      "grad_norm": 0.46401883936201904,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2933
    },
    {
      "epoch": 0.02934,
      "grad_norm": 0.4813031783733538,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 2934
    },
    {
      "epoch": 0.02935,
      "grad_norm": 0.4967851984899721,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 2935
    },
    {
      "epoch": 0.02936,
      "grad_norm": 0.5320015979752634,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 2936
    },
    {
      "epoch": 0.02937,
      "grad_norm": 0.5208542547284235,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 2937
    },
    {
      "epoch": 0.02938,
      "grad_norm": 0.4869610046256959,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 2938
    },
    {
      "epoch": 0.02939,
      "grad_norm": 0.43982195879012664,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 2939
    },
    {
      "epoch": 0.0294,
      "grad_norm": 0.43165943569371323,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2940
    },
    {
      "epoch": 0.02941,
      "grad_norm": 0.42317517860985626,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 2941
    },
    {
      "epoch": 0.02942,
      "grad_norm": 0.46023364048218307,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2942
    },
    {
      "epoch": 0.02943,
      "grad_norm": 0.5385981811058449,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 2943
    },
    {
      "epoch": 0.02944,
      "grad_norm": 0.7649280046904252,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2944
    },
    {
      "epoch": 0.02945,
      "grad_norm": 1.1539036490870935,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2945
    },
    {
      "epoch": 0.02946,
      "grad_norm": 0.9395984374200569,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 2946
    },
    {
      "epoch": 0.02947,
      "grad_norm": 0.8080599605170256,
      "learning_rate": 0.003,
      "loss": 4.224,
      "step": 2947
    },
    {
      "epoch": 0.02948,
      "grad_norm": 0.7434332174406536,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2948
    },
    {
      "epoch": 0.02949,
      "grad_norm": 0.7928578212224346,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 2949
    },
    {
      "epoch": 0.0295,
      "grad_norm": 0.8333301095683224,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2950
    },
    {
      "epoch": 0.02951,
      "grad_norm": 0.9482608719829599,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 2951
    },
    {
      "epoch": 0.02952,
      "grad_norm": 0.8328172653762101,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2952
    },
    {
      "epoch": 0.02953,
      "grad_norm": 0.8348300580576069,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2953
    },
    {
      "epoch": 0.02954,
      "grad_norm": 0.8191965718315765,
      "learning_rate": 0.003,
      "loss": 4.2499,
      "step": 2954
    },
    {
      "epoch": 0.02955,
      "grad_norm": 0.8592579474886709,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2955
    },
    {
      "epoch": 0.02956,
      "grad_norm": 0.768153733637731,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2956
    },
    {
      "epoch": 0.02957,
      "grad_norm": 0.619447070815721,
      "learning_rate": 0.003,
      "loss": 4.2157,
      "step": 2957
    },
    {
      "epoch": 0.02958,
      "grad_norm": 0.6117541520949576,
      "learning_rate": 0.003,
      "loss": 4.2284,
      "step": 2958
    },
    {
      "epoch": 0.02959,
      "grad_norm": 0.5765518020385798,
      "learning_rate": 0.003,
      "loss": 4.2146,
      "step": 2959
    },
    {
      "epoch": 0.0296,
      "grad_norm": 0.5472369901466232,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2960
    },
    {
      "epoch": 0.02961,
      "grad_norm": 0.5697675686509917,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 2961
    },
    {
      "epoch": 0.02962,
      "grad_norm": 0.5322939405778847,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 2962
    },
    {
      "epoch": 0.02963,
      "grad_norm": 0.4995062550237162,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 2963
    },
    {
      "epoch": 0.02964,
      "grad_norm": 0.48408735290928806,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 2964
    },
    {
      "epoch": 0.02965,
      "grad_norm": 0.49796931779139914,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 2965
    },
    {
      "epoch": 0.02966,
      "grad_norm": 0.49309681218010465,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 2966
    },
    {
      "epoch": 0.02967,
      "grad_norm": 0.46885663161103275,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2967
    },
    {
      "epoch": 0.02968,
      "grad_norm": 0.4307711779753192,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2968
    },
    {
      "epoch": 0.02969,
      "grad_norm": 0.38466337051238414,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 2969
    },
    {
      "epoch": 0.0297,
      "grad_norm": 0.46985591464325377,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 2970
    },
    {
      "epoch": 0.02971,
      "grad_norm": 0.6319423638686542,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 2971
    },
    {
      "epoch": 0.02972,
      "grad_norm": 0.8754437886538992,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 2972
    },
    {
      "epoch": 0.02973,
      "grad_norm": 1.020941685436833,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 2973
    },
    {
      "epoch": 0.02974,
      "grad_norm": 0.6824462354167666,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 2974
    },
    {
      "epoch": 0.02975,
      "grad_norm": 0.5456820041365307,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 2975
    },
    {
      "epoch": 0.02976,
      "grad_norm": 0.7254587394041584,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 2976
    },
    {
      "epoch": 0.02977,
      "grad_norm": 0.6896012196641018,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 2977
    },
    {
      "epoch": 0.02978,
      "grad_norm": 0.6446956558129,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 2978
    },
    {
      "epoch": 0.02979,
      "grad_norm": 0.5613291446841128,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 2979
    },
    {
      "epoch": 0.0298,
      "grad_norm": 0.578998122498601,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 2980
    },
    {
      "epoch": 0.02981,
      "grad_norm": 0.6195176572378017,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 2981
    },
    {
      "epoch": 0.02982,
      "grad_norm": 0.6240562576963727,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 2982
    },
    {
      "epoch": 0.02983,
      "grad_norm": 0.5968266153844792,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2983
    },
    {
      "epoch": 0.02984,
      "grad_norm": 0.5306795307358292,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 2984
    },
    {
      "epoch": 0.02985,
      "grad_norm": 0.5463323846035709,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2985
    },
    {
      "epoch": 0.02986,
      "grad_norm": 0.5421781401610759,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 2986
    },
    {
      "epoch": 0.02987,
      "grad_norm": 0.5293611287733503,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 2987
    },
    {
      "epoch": 0.02988,
      "grad_norm": 0.6408997854011281,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 2988
    },
    {
      "epoch": 0.02989,
      "grad_norm": 0.7426268122761913,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2989
    },
    {
      "epoch": 0.0299,
      "grad_norm": 0.7556899506663288,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 2990
    },
    {
      "epoch": 0.02991,
      "grad_norm": 0.6415270525884926,
      "learning_rate": 0.003,
      "loss": 4.1985,
      "step": 2991
    },
    {
      "epoch": 0.02992,
      "grad_norm": 0.6368634341681172,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2992
    },
    {
      "epoch": 0.02993,
      "grad_norm": 0.6410538934602643,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 2993
    },
    {
      "epoch": 0.02994,
      "grad_norm": 0.6332716830387952,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 2994
    },
    {
      "epoch": 0.02995,
      "grad_norm": 0.7865491161110946,
      "learning_rate": 0.003,
      "loss": 4.2175,
      "step": 2995
    },
    {
      "epoch": 0.02996,
      "grad_norm": 0.9230963051423904,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 2996
    },
    {
      "epoch": 0.02997,
      "grad_norm": 1.1769719664750078,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 2997
    },
    {
      "epoch": 0.02998,
      "grad_norm": 0.965735521679244,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2998
    },
    {
      "epoch": 0.02999,
      "grad_norm": 0.9025433565090871,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 2999
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9697259711524333,
      "learning_rate": 0.003,
      "loss": 4.2399,
      "step": 3000
    },
    {
      "epoch": 0.03001,
      "grad_norm": 0.9254213834179852,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3001
    },
    {
      "epoch": 0.03002,
      "grad_norm": 0.8893198223091714,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 3002
    },
    {
      "epoch": 0.03003,
      "grad_norm": 0.7192875473487559,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 3003
    },
    {
      "epoch": 0.03004,
      "grad_norm": 0.7886569353807684,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3004
    },
    {
      "epoch": 0.03005,
      "grad_norm": 0.8011423250434415,
      "learning_rate": 0.003,
      "loss": 4.2294,
      "step": 3005
    },
    {
      "epoch": 0.03006,
      "grad_norm": 0.7535404952002084,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 3006
    },
    {
      "epoch": 0.03007,
      "grad_norm": 0.7314424724357438,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 3007
    },
    {
      "epoch": 0.03008,
      "grad_norm": 0.6096876125818365,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 3008
    },
    {
      "epoch": 0.03009,
      "grad_norm": 0.5575234963619043,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3009
    },
    {
      "epoch": 0.0301,
      "grad_norm": 0.5598489300006476,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 3010
    },
    {
      "epoch": 0.03011,
      "grad_norm": 0.7310511801710661,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 3011
    },
    {
      "epoch": 0.03012,
      "grad_norm": 0.8536268304907085,
      "learning_rate": 0.003,
      "loss": 4.2002,
      "step": 3012
    },
    {
      "epoch": 0.03013,
      "grad_norm": 0.9271684945658873,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3013
    },
    {
      "epoch": 0.03014,
      "grad_norm": 0.748204257953396,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3014
    },
    {
      "epoch": 0.03015,
      "grad_norm": 0.7294374875175993,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 3015
    },
    {
      "epoch": 0.03016,
      "grad_norm": 0.8462639497588901,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 3016
    },
    {
      "epoch": 0.03017,
      "grad_norm": 0.6499456234415616,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 3017
    },
    {
      "epoch": 0.03018,
      "grad_norm": 0.5875121164816828,
      "learning_rate": 0.003,
      "loss": 4.1824,
      "step": 3018
    },
    {
      "epoch": 0.03019,
      "grad_norm": 0.47508344745103015,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3019
    },
    {
      "epoch": 0.0302,
      "grad_norm": 0.5142366251911704,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3020
    },
    {
      "epoch": 0.03021,
      "grad_norm": 0.5290255986656376,
      "learning_rate": 0.003,
      "loss": 4.2069,
      "step": 3021
    },
    {
      "epoch": 0.03022,
      "grad_norm": 0.5261392358925515,
      "learning_rate": 0.003,
      "loss": 4.2026,
      "step": 3022
    },
    {
      "epoch": 0.03023,
      "grad_norm": 0.5585956938981086,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 3023
    },
    {
      "epoch": 0.03024,
      "grad_norm": 0.6114817871124547,
      "learning_rate": 0.003,
      "loss": 4.2096,
      "step": 3024
    },
    {
      "epoch": 0.03025,
      "grad_norm": 0.6824857482246931,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3025
    },
    {
      "epoch": 0.03026,
      "grad_norm": 0.7984707065148595,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3026
    },
    {
      "epoch": 0.03027,
      "grad_norm": 0.7174251636707903,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3027
    },
    {
      "epoch": 0.03028,
      "grad_norm": 0.7106059576910744,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3028
    },
    {
      "epoch": 0.03029,
      "grad_norm": 0.6122719077257656,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 3029
    },
    {
      "epoch": 0.0303,
      "grad_norm": 0.5305954769529888,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3030
    },
    {
      "epoch": 0.03031,
      "grad_norm": 0.4997654568254282,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3031
    },
    {
      "epoch": 0.03032,
      "grad_norm": 0.4912614975804802,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3032
    },
    {
      "epoch": 0.03033,
      "grad_norm": 0.4920559211009805,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3033
    },
    {
      "epoch": 0.03034,
      "grad_norm": 0.5605701380054549,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 3034
    },
    {
      "epoch": 0.03035,
      "grad_norm": 0.6077583896543486,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 3035
    },
    {
      "epoch": 0.03036,
      "grad_norm": 0.7202855019986794,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3036
    },
    {
      "epoch": 0.03037,
      "grad_norm": 0.6575890137997805,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3037
    },
    {
      "epoch": 0.03038,
      "grad_norm": 0.46568815919813755,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 3038
    },
    {
      "epoch": 0.03039,
      "grad_norm": 0.49641007446862917,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 3039
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.559012034381986,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3040
    },
    {
      "epoch": 0.03041,
      "grad_norm": 0.5533954561872954,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 3041
    },
    {
      "epoch": 0.03042,
      "grad_norm": 0.47365032322165673,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 3042
    },
    {
      "epoch": 0.03043,
      "grad_norm": 0.4920139603559505,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3043
    },
    {
      "epoch": 0.03044,
      "grad_norm": 0.48267895485751067,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 3044
    },
    {
      "epoch": 0.03045,
      "grad_norm": 0.40892249484891385,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3045
    },
    {
      "epoch": 0.03046,
      "grad_norm": 0.5179956198309833,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3046
    },
    {
      "epoch": 0.03047,
      "grad_norm": 0.6458691485842255,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3047
    },
    {
      "epoch": 0.03048,
      "grad_norm": 0.805749623290173,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3048
    },
    {
      "epoch": 0.03049,
      "grad_norm": 0.8776696886409071,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3049
    },
    {
      "epoch": 0.0305,
      "grad_norm": 0.8822405330269477,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 3050
    },
    {
      "epoch": 0.03051,
      "grad_norm": 0.9277197792519675,
      "learning_rate": 0.003,
      "loss": 4.1966,
      "step": 3051
    },
    {
      "epoch": 0.03052,
      "grad_norm": 0.9287607223949175,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3052
    },
    {
      "epoch": 0.03053,
      "grad_norm": 0.8621618117255709,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 3053
    },
    {
      "epoch": 0.03054,
      "grad_norm": 0.9192475292904252,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 3054
    },
    {
      "epoch": 0.03055,
      "grad_norm": 0.9250509957208142,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 3055
    },
    {
      "epoch": 0.03056,
      "grad_norm": 0.9134243567876251,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 3056
    },
    {
      "epoch": 0.03057,
      "grad_norm": 1.0483835943699353,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 3057
    },
    {
      "epoch": 0.03058,
      "grad_norm": 1.0909797627134183,
      "learning_rate": 0.003,
      "loss": 4.2072,
      "step": 3058
    },
    {
      "epoch": 0.03059,
      "grad_norm": 1.0929367898362812,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 3059
    },
    {
      "epoch": 0.0306,
      "grad_norm": 0.9499317872652826,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 3060
    },
    {
      "epoch": 0.03061,
      "grad_norm": 0.7530643446613368,
      "learning_rate": 0.003,
      "loss": 4.227,
      "step": 3061
    },
    {
      "epoch": 0.03062,
      "grad_norm": 0.7822441045416462,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 3062
    },
    {
      "epoch": 0.03063,
      "grad_norm": 0.7991934606488392,
      "learning_rate": 0.003,
      "loss": 4.1973,
      "step": 3063
    },
    {
      "epoch": 0.03064,
      "grad_norm": 0.7614903963432643,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 3064
    },
    {
      "epoch": 0.03065,
      "grad_norm": 0.6047999858157871,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3065
    },
    {
      "epoch": 0.03066,
      "grad_norm": 0.6254938393897825,
      "learning_rate": 0.003,
      "loss": 4.231,
      "step": 3066
    },
    {
      "epoch": 0.03067,
      "grad_norm": 0.5333801571885867,
      "learning_rate": 0.003,
      "loss": 4.2275,
      "step": 3067
    },
    {
      "epoch": 0.03068,
      "grad_norm": 0.47159284158763254,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 3068
    },
    {
      "epoch": 0.03069,
      "grad_norm": 0.4095985602684375,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3069
    },
    {
      "epoch": 0.0307,
      "grad_norm": 0.4380075531913516,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 3070
    },
    {
      "epoch": 0.03071,
      "grad_norm": 0.520842979900822,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 3071
    },
    {
      "epoch": 0.03072,
      "grad_norm": 0.6276268407397474,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3072
    },
    {
      "epoch": 0.03073,
      "grad_norm": 0.7317184572201801,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3073
    },
    {
      "epoch": 0.03074,
      "grad_norm": 0.7701795105904193,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3074
    },
    {
      "epoch": 0.03075,
      "grad_norm": 0.6957977102957552,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3075
    },
    {
      "epoch": 0.03076,
      "grad_norm": 0.6144911834647919,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 3076
    },
    {
      "epoch": 0.03077,
      "grad_norm": 0.692008332614251,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 3077
    },
    {
      "epoch": 0.03078,
      "grad_norm": 0.6946737312829024,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 3078
    },
    {
      "epoch": 0.03079,
      "grad_norm": 0.7552084198206422,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3079
    },
    {
      "epoch": 0.0308,
      "grad_norm": 0.7025348126305976,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 3080
    },
    {
      "epoch": 0.03081,
      "grad_norm": 0.5947816851076059,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3081
    },
    {
      "epoch": 0.03082,
      "grad_norm": 0.6508903175127416,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3082
    },
    {
      "epoch": 0.03083,
      "grad_norm": 0.8364672171654759,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3083
    },
    {
      "epoch": 0.03084,
      "grad_norm": 0.7270205647692827,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3084
    },
    {
      "epoch": 0.03085,
      "grad_norm": 0.5251957082188036,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 3085
    },
    {
      "epoch": 0.03086,
      "grad_norm": 0.4928794972094713,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 3086
    },
    {
      "epoch": 0.03087,
      "grad_norm": 0.4574857190743631,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3087
    },
    {
      "epoch": 0.03088,
      "grad_norm": 0.48967300167720096,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 3088
    },
    {
      "epoch": 0.03089,
      "grad_norm": 0.5020327161233649,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3089
    },
    {
      "epoch": 0.0309,
      "grad_norm": 0.5318564623713369,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3090
    },
    {
      "epoch": 0.03091,
      "grad_norm": 0.5321720354076895,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 3091
    },
    {
      "epoch": 0.03092,
      "grad_norm": 0.45641360749571647,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 3092
    },
    {
      "epoch": 0.03093,
      "grad_norm": 0.40991320622723604,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3093
    },
    {
      "epoch": 0.03094,
      "grad_norm": 0.4792246990497778,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3094
    },
    {
      "epoch": 0.03095,
      "grad_norm": 0.5559515338310668,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3095
    },
    {
      "epoch": 0.03096,
      "grad_norm": 0.7462971856723287,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 3096
    },
    {
      "epoch": 0.03097,
      "grad_norm": 0.9002817455009458,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 3097
    },
    {
      "epoch": 0.03098,
      "grad_norm": 0.8351003992845788,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3098
    },
    {
      "epoch": 0.03099,
      "grad_norm": 0.5935585890716526,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3099
    },
    {
      "epoch": 0.031,
      "grad_norm": 0.6353446062973269,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3100
    },
    {
      "epoch": 0.03101,
      "grad_norm": 0.7592462024173295,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3101
    },
    {
      "epoch": 0.03102,
      "grad_norm": 0.8288461234391342,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3102
    },
    {
      "epoch": 0.03103,
      "grad_norm": 0.8186091723896449,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3103
    },
    {
      "epoch": 0.03104,
      "grad_norm": 0.730911001668181,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3104
    },
    {
      "epoch": 0.03105,
      "grad_norm": 0.6266459445253275,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 3105
    },
    {
      "epoch": 0.03106,
      "grad_norm": 0.6545747432475193,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3106
    },
    {
      "epoch": 0.03107,
      "grad_norm": 0.5231672698914397,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 3107
    },
    {
      "epoch": 0.03108,
      "grad_norm": 0.5734245951904136,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 3108
    },
    {
      "epoch": 0.03109,
      "grad_norm": 0.5288754437005536,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3109
    },
    {
      "epoch": 0.0311,
      "grad_norm": 0.5320691683351324,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 3110
    },
    {
      "epoch": 0.03111,
      "grad_norm": 0.5370642425053924,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3111
    },
    {
      "epoch": 0.03112,
      "grad_norm": 0.49890497665091543,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3112
    },
    {
      "epoch": 0.03113,
      "grad_norm": 0.4336869241892021,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3113
    },
    {
      "epoch": 0.03114,
      "grad_norm": 0.45513880225763054,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3114
    },
    {
      "epoch": 0.03115,
      "grad_norm": 0.5541487651615646,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3115
    },
    {
      "epoch": 0.03116,
      "grad_norm": 0.6085416142601979,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3116
    },
    {
      "epoch": 0.03117,
      "grad_norm": 0.6289436131363786,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3117
    },
    {
      "epoch": 0.03118,
      "grad_norm": 0.7648349803443671,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3118
    },
    {
      "epoch": 0.03119,
      "grad_norm": 0.9905397386097399,
      "learning_rate": 0.003,
      "loss": 4.1975,
      "step": 3119
    },
    {
      "epoch": 0.0312,
      "grad_norm": 1.002027828062209,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 3120
    },
    {
      "epoch": 0.03121,
      "grad_norm": 0.9945235603188975,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3121
    },
    {
      "epoch": 0.03122,
      "grad_norm": 0.9749277139432609,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 3122
    },
    {
      "epoch": 0.03123,
      "grad_norm": 0.9176347284109458,
      "learning_rate": 0.003,
      "loss": 4.2123,
      "step": 3123
    },
    {
      "epoch": 0.03124,
      "grad_norm": 0.8222379595933585,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3124
    },
    {
      "epoch": 0.03125,
      "grad_norm": 0.8029594745648053,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 3125
    },
    {
      "epoch": 0.03126,
      "grad_norm": 0.7973924670105094,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 3126
    },
    {
      "epoch": 0.03127,
      "grad_norm": 0.8183339868293253,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 3127
    },
    {
      "epoch": 0.03128,
      "grad_norm": 0.8564648965676082,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 3128
    },
    {
      "epoch": 0.03129,
      "grad_norm": 0.6856535056536724,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 3129
    },
    {
      "epoch": 0.0313,
      "grad_norm": 0.6641053792145878,
      "learning_rate": 0.003,
      "loss": 4.2086,
      "step": 3130
    },
    {
      "epoch": 0.03131,
      "grad_norm": 0.5479229508159585,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 3131
    },
    {
      "epoch": 0.03132,
      "grad_norm": 0.5311614836536821,
      "learning_rate": 0.003,
      "loss": 4.2024,
      "step": 3132
    },
    {
      "epoch": 0.03133,
      "grad_norm": 0.5105588102398402,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 3133
    },
    {
      "epoch": 0.03134,
      "grad_norm": 0.5009888970015204,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3134
    },
    {
      "epoch": 0.03135,
      "grad_norm": 0.4297510332483796,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3135
    },
    {
      "epoch": 0.03136,
      "grad_norm": 0.4354586739325799,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3136
    },
    {
      "epoch": 0.03137,
      "grad_norm": 0.5083671646646589,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3137
    },
    {
      "epoch": 0.03138,
      "grad_norm": 0.6199429454274028,
      "learning_rate": 0.003,
      "loss": 4.1701,
      "step": 3138
    },
    {
      "epoch": 0.03139,
      "grad_norm": 0.7271551302730402,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3139
    },
    {
      "epoch": 0.0314,
      "grad_norm": 0.7358854576257201,
      "learning_rate": 0.003,
      "loss": 4.1788,
      "step": 3140
    },
    {
      "epoch": 0.03141,
      "grad_norm": 0.6326154627482451,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 3141
    },
    {
      "epoch": 0.03142,
      "grad_norm": 0.5588364977660311,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 3142
    },
    {
      "epoch": 0.03143,
      "grad_norm": 0.6311591083308525,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3143
    },
    {
      "epoch": 0.03144,
      "grad_norm": 0.5392546760146013,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 3144
    },
    {
      "epoch": 0.03145,
      "grad_norm": 0.4557888424936214,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3145
    },
    {
      "epoch": 0.03146,
      "grad_norm": 0.4216370024975091,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 3146
    },
    {
      "epoch": 0.03147,
      "grad_norm": 0.4295618903624941,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3147
    },
    {
      "epoch": 0.03148,
      "grad_norm": 0.4097670324789773,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3148
    },
    {
      "epoch": 0.03149,
      "grad_norm": 0.4802416636246548,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 3149
    },
    {
      "epoch": 0.0315,
      "grad_norm": 0.46441368907739394,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 3150
    },
    {
      "epoch": 0.03151,
      "grad_norm": 0.44059093730228843,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3151
    },
    {
      "epoch": 0.03152,
      "grad_norm": 0.4172894456146658,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3152
    },
    {
      "epoch": 0.03153,
      "grad_norm": 0.42959521977879306,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3153
    },
    {
      "epoch": 0.03154,
      "grad_norm": 0.43398804524875734,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3154
    },
    {
      "epoch": 0.03155,
      "grad_norm": 0.4562270589856803,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3155
    },
    {
      "epoch": 0.03156,
      "grad_norm": 0.5126486371492089,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 3156
    },
    {
      "epoch": 0.03157,
      "grad_norm": 0.6526844959193616,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3157
    },
    {
      "epoch": 0.03158,
      "grad_norm": 0.9480089169547891,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3158
    },
    {
      "epoch": 0.03159,
      "grad_norm": 1.3561889811332835,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 3159
    },
    {
      "epoch": 0.0316,
      "grad_norm": 0.5982846825437411,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 3160
    },
    {
      "epoch": 0.03161,
      "grad_norm": 0.8732729212340313,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3161
    },
    {
      "epoch": 0.03162,
      "grad_norm": 0.886847324325859,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 3162
    },
    {
      "epoch": 0.03163,
      "grad_norm": 0.99270442281753,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 3163
    },
    {
      "epoch": 0.03164,
      "grad_norm": 0.9630909842912381,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 3164
    },
    {
      "epoch": 0.03165,
      "grad_norm": 0.7916297793185595,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3165
    },
    {
      "epoch": 0.03166,
      "grad_norm": 0.6926990051674062,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 3166
    },
    {
      "epoch": 0.03167,
      "grad_norm": 0.681279202527116,
      "learning_rate": 0.003,
      "loss": 4.1779,
      "step": 3167
    },
    {
      "epoch": 0.03168,
      "grad_norm": 0.6889533564228932,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 3168
    },
    {
      "epoch": 0.03169,
      "grad_norm": 0.8396344102094035,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 3169
    },
    {
      "epoch": 0.0317,
      "grad_norm": 0.9428653644577029,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 3170
    },
    {
      "epoch": 0.03171,
      "grad_norm": 0.691282163039264,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 3171
    },
    {
      "epoch": 0.03172,
      "grad_norm": 0.665795792414971,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 3172
    },
    {
      "epoch": 0.03173,
      "grad_norm": 0.6751127918552098,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3173
    },
    {
      "epoch": 0.03174,
      "grad_norm": 0.6066754986654972,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 3174
    },
    {
      "epoch": 0.03175,
      "grad_norm": 0.5400765945883416,
      "learning_rate": 0.003,
      "loss": 4.2066,
      "step": 3175
    },
    {
      "epoch": 0.03176,
      "grad_norm": 0.5542770572580099,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3176
    },
    {
      "epoch": 0.03177,
      "grad_norm": 0.5967671493496074,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3177
    },
    {
      "epoch": 0.03178,
      "grad_norm": 0.6105572005396745,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 3178
    },
    {
      "epoch": 0.03179,
      "grad_norm": 0.6783275770388425,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 3179
    },
    {
      "epoch": 0.0318,
      "grad_norm": 0.7073200859467578,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3180
    },
    {
      "epoch": 0.03181,
      "grad_norm": 0.6783381680497546,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 3181
    },
    {
      "epoch": 0.03182,
      "grad_norm": 0.7230568807336094,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 3182
    },
    {
      "epoch": 0.03183,
      "grad_norm": 0.873578952331532,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 3183
    },
    {
      "epoch": 0.03184,
      "grad_norm": 0.9905329983068216,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 3184
    },
    {
      "epoch": 0.03185,
      "grad_norm": 1.2552538351706124,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 3185
    },
    {
      "epoch": 0.03186,
      "grad_norm": 0.6881365341663627,
      "learning_rate": 0.003,
      "loss": 4.1977,
      "step": 3186
    },
    {
      "epoch": 0.03187,
      "grad_norm": 0.6603777519922737,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3187
    },
    {
      "epoch": 0.03188,
      "grad_norm": 0.6114536474568962,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3188
    },
    {
      "epoch": 0.03189,
      "grad_norm": 0.7652970454563929,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3189
    },
    {
      "epoch": 0.0319,
      "grad_norm": 0.7756244045621472,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3190
    },
    {
      "epoch": 0.03191,
      "grad_norm": 0.7586188953077919,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3191
    },
    {
      "epoch": 0.03192,
      "grad_norm": 0.7248898112293884,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 3192
    },
    {
      "epoch": 0.03193,
      "grad_norm": 0.6819049982170161,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 3193
    },
    {
      "epoch": 0.03194,
      "grad_norm": 0.6384912160022395,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 3194
    },
    {
      "epoch": 0.03195,
      "grad_norm": 0.6207563594114237,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3195
    },
    {
      "epoch": 0.03196,
      "grad_norm": 0.5716998964191464,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3196
    },
    {
      "epoch": 0.03197,
      "grad_norm": 0.5492769888070425,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 3197
    },
    {
      "epoch": 0.03198,
      "grad_norm": 0.5349886855188608,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3198
    },
    {
      "epoch": 0.03199,
      "grad_norm": 0.4722551144791865,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3199
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.5241213036464709,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3200
    },
    {
      "epoch": 0.03201,
      "grad_norm": 0.5556232177618735,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 3201
    },
    {
      "epoch": 0.03202,
      "grad_norm": 0.5951717317344422,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3202
    },
    {
      "epoch": 0.03203,
      "grad_norm": 0.5874723863194432,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 3203
    },
    {
      "epoch": 0.03204,
      "grad_norm": 0.5662458323617429,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3204
    },
    {
      "epoch": 0.03205,
      "grad_norm": 0.55236769595308,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3205
    },
    {
      "epoch": 0.03206,
      "grad_norm": 0.5082397875263351,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3206
    },
    {
      "epoch": 0.03207,
      "grad_norm": 0.4452387164637836,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 3207
    },
    {
      "epoch": 0.03208,
      "grad_norm": 0.48293819644672603,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3208
    },
    {
      "epoch": 0.03209,
      "grad_norm": 0.4498162894819178,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3209
    },
    {
      "epoch": 0.0321,
      "grad_norm": 0.5708600401920843,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3210
    },
    {
      "epoch": 0.03211,
      "grad_norm": 0.6897782500954599,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3211
    },
    {
      "epoch": 0.03212,
      "grad_norm": 0.8577311234564894,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3212
    },
    {
      "epoch": 0.03213,
      "grad_norm": 0.7479816094202391,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3213
    },
    {
      "epoch": 0.03214,
      "grad_norm": 0.5575837603075381,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 3214
    },
    {
      "epoch": 0.03215,
      "grad_norm": 0.619113370679725,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 3215
    },
    {
      "epoch": 0.03216,
      "grad_norm": 0.6857581625178918,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3216
    },
    {
      "epoch": 0.03217,
      "grad_norm": 0.6339316692312089,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 3217
    },
    {
      "epoch": 0.03218,
      "grad_norm": 0.5361569097117066,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3218
    },
    {
      "epoch": 0.03219,
      "grad_norm": 0.5146188024497156,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 3219
    },
    {
      "epoch": 0.0322,
      "grad_norm": 0.5580750782179962,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3220
    },
    {
      "epoch": 0.03221,
      "grad_norm": 0.5127806683543551,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3221
    },
    {
      "epoch": 0.03222,
      "grad_norm": 0.5372160864602477,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3222
    },
    {
      "epoch": 0.03223,
      "grad_norm": 0.5923510679532299,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3223
    },
    {
      "epoch": 0.03224,
      "grad_norm": 0.6861592854273826,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 3224
    },
    {
      "epoch": 0.03225,
      "grad_norm": 0.7582363764178276,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3225
    },
    {
      "epoch": 0.03226,
      "grad_norm": 0.8207513579274669,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 3226
    },
    {
      "epoch": 0.03227,
      "grad_norm": 0.7077066475989446,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3227
    },
    {
      "epoch": 0.03228,
      "grad_norm": 0.6232151026364198,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3228
    },
    {
      "epoch": 0.03229,
      "grad_norm": 0.6228354375566014,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 3229
    },
    {
      "epoch": 0.0323,
      "grad_norm": 0.6011100180442941,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3230
    },
    {
      "epoch": 0.03231,
      "grad_norm": 0.6278235480602589,
      "learning_rate": 0.003,
      "loss": 4.1733,
      "step": 3231
    },
    {
      "epoch": 0.03232,
      "grad_norm": 0.5999968897563468,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3232
    },
    {
      "epoch": 0.03233,
      "grad_norm": 0.5467043906444932,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3233
    },
    {
      "epoch": 0.03234,
      "grad_norm": 0.616637901979197,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 3234
    },
    {
      "epoch": 0.03235,
      "grad_norm": 0.7932914435950059,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3235
    },
    {
      "epoch": 0.03236,
      "grad_norm": 0.9734805326028607,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3236
    },
    {
      "epoch": 0.03237,
      "grad_norm": 0.9035113763708699,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3237
    },
    {
      "epoch": 0.03238,
      "grad_norm": 0.8124977646963972,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3238
    },
    {
      "epoch": 0.03239,
      "grad_norm": 0.7169306328435582,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 3239
    },
    {
      "epoch": 0.0324,
      "grad_norm": 0.7507848511489641,
      "learning_rate": 0.003,
      "loss": 4.1984,
      "step": 3240
    },
    {
      "epoch": 0.03241,
      "grad_norm": 0.871055197405415,
      "learning_rate": 0.003,
      "loss": 4.1945,
      "step": 3241
    },
    {
      "epoch": 0.03242,
      "grad_norm": 0.9726257649238742,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3242
    },
    {
      "epoch": 0.03243,
      "grad_norm": 0.8829139791894937,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3243
    },
    {
      "epoch": 0.03244,
      "grad_norm": 0.9226257984088151,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 3244
    },
    {
      "epoch": 0.03245,
      "grad_norm": 1.0602033368371822,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 3245
    },
    {
      "epoch": 0.03246,
      "grad_norm": 0.8832790843056474,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 3246
    },
    {
      "epoch": 0.03247,
      "grad_norm": 0.8993256456137283,
      "learning_rate": 0.003,
      "loss": 4.2129,
      "step": 3247
    },
    {
      "epoch": 0.03248,
      "grad_norm": 0.9121935412385299,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 3248
    },
    {
      "epoch": 0.03249,
      "grad_norm": 0.8613725910612215,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3249
    },
    {
      "epoch": 0.0325,
      "grad_norm": 0.8167292628339079,
      "learning_rate": 0.003,
      "loss": 4.2315,
      "step": 3250
    },
    {
      "epoch": 0.03251,
      "grad_norm": 0.7863361659710474,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 3251
    },
    {
      "epoch": 0.03252,
      "grad_norm": 0.7493875681769112,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3252
    },
    {
      "epoch": 0.03253,
      "grad_norm": 0.7679048561396902,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 3253
    },
    {
      "epoch": 0.03254,
      "grad_norm": 0.7654672625645944,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 3254
    },
    {
      "epoch": 0.03255,
      "grad_norm": 0.7464808737181244,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 3255
    },
    {
      "epoch": 0.03256,
      "grad_norm": 0.6526585542637006,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 3256
    },
    {
      "epoch": 0.03257,
      "grad_norm": 0.6001937061564249,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 3257
    },
    {
      "epoch": 0.03258,
      "grad_norm": 0.5472158831270236,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 3258
    },
    {
      "epoch": 0.03259,
      "grad_norm": 0.4785164815267161,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 3259
    },
    {
      "epoch": 0.0326,
      "grad_norm": 0.38861603840814574,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3260
    },
    {
      "epoch": 0.03261,
      "grad_norm": 0.41654558334327835,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3261
    },
    {
      "epoch": 0.03262,
      "grad_norm": 0.41147900352780037,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3262
    },
    {
      "epoch": 0.03263,
      "grad_norm": 0.4340386920329771,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 3263
    },
    {
      "epoch": 0.03264,
      "grad_norm": 0.47093951437519954,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3264
    },
    {
      "epoch": 0.03265,
      "grad_norm": 0.5195090423674352,
      "learning_rate": 0.003,
      "loss": 4.184,
      "step": 3265
    },
    {
      "epoch": 0.03266,
      "grad_norm": 0.5846656667385064,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 3266
    },
    {
      "epoch": 0.03267,
      "grad_norm": 0.6621919529113969,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 3267
    },
    {
      "epoch": 0.03268,
      "grad_norm": 0.6864096228211959,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 3268
    },
    {
      "epoch": 0.03269,
      "grad_norm": 0.5308237484050747,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3269
    },
    {
      "epoch": 0.0327,
      "grad_norm": 0.44850448788276254,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3270
    },
    {
      "epoch": 0.03271,
      "grad_norm": 0.4693563167350624,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3271
    },
    {
      "epoch": 0.03272,
      "grad_norm": 0.5323713025056388,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3272
    },
    {
      "epoch": 0.03273,
      "grad_norm": 0.6692459000105278,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3273
    },
    {
      "epoch": 0.03274,
      "grad_norm": 0.7924729751268105,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3274
    },
    {
      "epoch": 0.03275,
      "grad_norm": 0.8832852537291822,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 3275
    },
    {
      "epoch": 0.03276,
      "grad_norm": 0.9440382055401524,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3276
    },
    {
      "epoch": 0.03277,
      "grad_norm": 0.9157723514845009,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 3277
    },
    {
      "epoch": 0.03278,
      "grad_norm": 0.7757959986300896,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3278
    },
    {
      "epoch": 0.03279,
      "grad_norm": 0.7052884741699046,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3279
    },
    {
      "epoch": 0.0328,
      "grad_norm": 0.6748931568117461,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3280
    },
    {
      "epoch": 0.03281,
      "grad_norm": 0.5834976642550973,
      "learning_rate": 0.003,
      "loss": 4.1984,
      "step": 3281
    },
    {
      "epoch": 0.03282,
      "grad_norm": 0.5830462450862388,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3282
    },
    {
      "epoch": 0.03283,
      "grad_norm": 0.6046371763595715,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3283
    },
    {
      "epoch": 0.03284,
      "grad_norm": 0.498692543866341,
      "learning_rate": 0.003,
      "loss": 4.188,
      "step": 3284
    },
    {
      "epoch": 0.03285,
      "grad_norm": 0.5300083365332573,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 3285
    },
    {
      "epoch": 0.03286,
      "grad_norm": 0.5302483285107112,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3286
    },
    {
      "epoch": 0.03287,
      "grad_norm": 0.5280212789702404,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3287
    },
    {
      "epoch": 0.03288,
      "grad_norm": 0.49328823308562514,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3288
    },
    {
      "epoch": 0.03289,
      "grad_norm": 0.5955758847359914,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3289
    },
    {
      "epoch": 0.0329,
      "grad_norm": 0.7318966219724384,
      "learning_rate": 0.003,
      "loss": 4.1954,
      "step": 3290
    },
    {
      "epoch": 0.03291,
      "grad_norm": 0.8969015331966617,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3291
    },
    {
      "epoch": 0.03292,
      "grad_norm": 0.8171747463806796,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3292
    },
    {
      "epoch": 0.03293,
      "grad_norm": 0.6454312253277896,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3293
    },
    {
      "epoch": 0.03294,
      "grad_norm": 0.5476436730454269,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3294
    },
    {
      "epoch": 0.03295,
      "grad_norm": 0.6343237379540597,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3295
    },
    {
      "epoch": 0.03296,
      "grad_norm": 0.6111699411729721,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 3296
    },
    {
      "epoch": 0.03297,
      "grad_norm": 0.5298030355791016,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3297
    },
    {
      "epoch": 0.03298,
      "grad_norm": 0.4836663999032712,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3298
    },
    {
      "epoch": 0.03299,
      "grad_norm": 0.4701496219943231,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3299
    },
    {
      "epoch": 0.033,
      "grad_norm": 0.4623026228792837,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 3300
    },
    {
      "epoch": 0.03301,
      "grad_norm": 0.43500908939154875,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 3301
    },
    {
      "epoch": 0.03302,
      "grad_norm": 0.4287201358461857,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3302
    },
    {
      "epoch": 0.03303,
      "grad_norm": 0.46641640405345786,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3303
    },
    {
      "epoch": 0.03304,
      "grad_norm": 0.588941013591916,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3304
    },
    {
      "epoch": 0.03305,
      "grad_norm": 0.7079568978470296,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3305
    },
    {
      "epoch": 0.03306,
      "grad_norm": 0.829985756720969,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 3306
    },
    {
      "epoch": 0.03307,
      "grad_norm": 0.8462386242609186,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3307
    },
    {
      "epoch": 0.03308,
      "grad_norm": 0.8293495815638529,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3308
    },
    {
      "epoch": 0.03309,
      "grad_norm": 0.8456212329927598,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3309
    },
    {
      "epoch": 0.0331,
      "grad_norm": 0.8565848835132709,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3310
    },
    {
      "epoch": 0.03311,
      "grad_norm": 0.9394898981807235,
      "learning_rate": 0.003,
      "loss": 4.1969,
      "step": 3311
    },
    {
      "epoch": 0.03312,
      "grad_norm": 1.0401594073450542,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3312
    },
    {
      "epoch": 0.03313,
      "grad_norm": 0.8840161283334239,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 3313
    },
    {
      "epoch": 0.03314,
      "grad_norm": 0.8560525559559692,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3314
    },
    {
      "epoch": 0.03315,
      "grad_norm": 0.7954589166824607,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 3315
    },
    {
      "epoch": 0.03316,
      "grad_norm": 0.7407877779230415,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 3316
    },
    {
      "epoch": 0.03317,
      "grad_norm": 0.7593343697168953,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 3317
    },
    {
      "epoch": 0.03318,
      "grad_norm": 0.7588615866933531,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3318
    },
    {
      "epoch": 0.03319,
      "grad_norm": 0.6406451388828956,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3319
    },
    {
      "epoch": 0.0332,
      "grad_norm": 0.6857052599814105,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3320
    },
    {
      "epoch": 0.03321,
      "grad_norm": 0.6857879313241626,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 3321
    },
    {
      "epoch": 0.03322,
      "grad_norm": 0.7686016779789973,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3322
    },
    {
      "epoch": 0.03323,
      "grad_norm": 0.8269613306774392,
      "learning_rate": 0.003,
      "loss": 4.2053,
      "step": 3323
    },
    {
      "epoch": 0.03324,
      "grad_norm": 0.9528423775782789,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3324
    },
    {
      "epoch": 0.03325,
      "grad_norm": 1.0794820717536844,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 3325
    },
    {
      "epoch": 0.03326,
      "grad_norm": 0.8445799040478842,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3326
    },
    {
      "epoch": 0.03327,
      "grad_norm": 0.6629463030927865,
      "learning_rate": 0.003,
      "loss": 4.1918,
      "step": 3327
    },
    {
      "epoch": 0.03328,
      "grad_norm": 0.6487519835718909,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 3328
    },
    {
      "epoch": 0.03329,
      "grad_norm": 0.6298798087889773,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3329
    },
    {
      "epoch": 0.0333,
      "grad_norm": 0.6333191541087022,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3330
    },
    {
      "epoch": 0.03331,
      "grad_norm": 0.7837692756055505,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3331
    },
    {
      "epoch": 0.03332,
      "grad_norm": 0.9187938715930933,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3332
    },
    {
      "epoch": 0.03333,
      "grad_norm": 0.7284609287843756,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3333
    },
    {
      "epoch": 0.03334,
      "grad_norm": 0.5950364415684378,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3334
    },
    {
      "epoch": 0.03335,
      "grad_norm": 0.5204758750774462,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3335
    },
    {
      "epoch": 0.03336,
      "grad_norm": 0.5061539602167332,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 3336
    },
    {
      "epoch": 0.03337,
      "grad_norm": 0.42913500260913456,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3337
    },
    {
      "epoch": 0.03338,
      "grad_norm": 0.4415664602358343,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3338
    },
    {
      "epoch": 0.03339,
      "grad_norm": 0.4739540929710416,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3339
    },
    {
      "epoch": 0.0334,
      "grad_norm": 0.4177466598922738,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3340
    },
    {
      "epoch": 0.03341,
      "grad_norm": 0.4450986126917478,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 3341
    },
    {
      "epoch": 0.03342,
      "grad_norm": 0.4857573129812489,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 3342
    },
    {
      "epoch": 0.03343,
      "grad_norm": 0.6478603105495,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 3343
    },
    {
      "epoch": 0.03344,
      "grad_norm": 0.7654176175893398,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3344
    },
    {
      "epoch": 0.03345,
      "grad_norm": 0.708719161890371,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3345
    },
    {
      "epoch": 0.03346,
      "grad_norm": 0.5692838217453585,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 3346
    },
    {
      "epoch": 0.03347,
      "grad_norm": 0.5899959723847616,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3347
    },
    {
      "epoch": 0.03348,
      "grad_norm": 0.6486978345976652,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3348
    },
    {
      "epoch": 0.03349,
      "grad_norm": 0.5576273675617567,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3349
    },
    {
      "epoch": 0.0335,
      "grad_norm": 0.537004248184913,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 3350
    },
    {
      "epoch": 0.03351,
      "grad_norm": 0.5567929131713881,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 3351
    },
    {
      "epoch": 0.03352,
      "grad_norm": 0.6714613195190436,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3352
    },
    {
      "epoch": 0.03353,
      "grad_norm": 0.7428392470438228,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3353
    },
    {
      "epoch": 0.03354,
      "grad_norm": 0.709523442230735,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 3354
    },
    {
      "epoch": 0.03355,
      "grad_norm": 0.5743763408378283,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3355
    },
    {
      "epoch": 0.03356,
      "grad_norm": 0.5372193296353462,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3356
    },
    {
      "epoch": 0.03357,
      "grad_norm": 0.6152109328121239,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3357
    },
    {
      "epoch": 0.03358,
      "grad_norm": 0.693520738126006,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3358
    },
    {
      "epoch": 0.03359,
      "grad_norm": 0.6800317331746373,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3359
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.6462929043552973,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3360
    },
    {
      "epoch": 0.03361,
      "grad_norm": 0.6413289136062978,
      "learning_rate": 0.003,
      "loss": 4.1788,
      "step": 3361
    },
    {
      "epoch": 0.03362,
      "grad_norm": 0.6955328156949359,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 3362
    },
    {
      "epoch": 0.03363,
      "grad_norm": 0.6235123487755582,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 3363
    },
    {
      "epoch": 0.03364,
      "grad_norm": 0.602598837086171,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 3364
    },
    {
      "epoch": 0.03365,
      "grad_norm": 0.7356364448082628,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 3365
    },
    {
      "epoch": 0.03366,
      "grad_norm": 0.8244656072335483,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 3366
    },
    {
      "epoch": 0.03367,
      "grad_norm": 0.7703618126132621,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3367
    },
    {
      "epoch": 0.03368,
      "grad_norm": 0.6946923139107304,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 3368
    },
    {
      "epoch": 0.03369,
      "grad_norm": 0.7107703988620854,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3369
    },
    {
      "epoch": 0.0337,
      "grad_norm": 0.6762732803354303,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3370
    },
    {
      "epoch": 0.03371,
      "grad_norm": 0.6545021306919707,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 3371
    },
    {
      "epoch": 0.03372,
      "grad_norm": 0.6538700683175451,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3372
    },
    {
      "epoch": 0.03373,
      "grad_norm": 0.601220154679101,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3373
    },
    {
      "epoch": 0.03374,
      "grad_norm": 0.5930617819105674,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3374
    },
    {
      "epoch": 0.03375,
      "grad_norm": 0.5268626720580842,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 3375
    },
    {
      "epoch": 0.03376,
      "grad_norm": 0.5347783123792315,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 3376
    },
    {
      "epoch": 0.03377,
      "grad_norm": 0.5216348207352511,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3377
    },
    {
      "epoch": 0.03378,
      "grad_norm": 0.6194694687851028,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3378
    },
    {
      "epoch": 0.03379,
      "grad_norm": 0.6549669142309517,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3379
    },
    {
      "epoch": 0.0338,
      "grad_norm": 0.6860555079944507,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3380
    },
    {
      "epoch": 0.03381,
      "grad_norm": 0.7654947303377718,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3381
    },
    {
      "epoch": 0.03382,
      "grad_norm": 0.8842593404054855,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3382
    },
    {
      "epoch": 0.03383,
      "grad_norm": 1.2792086110225658,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3383
    },
    {
      "epoch": 0.03384,
      "grad_norm": 0.9270742666385557,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3384
    },
    {
      "epoch": 0.03385,
      "grad_norm": 0.7871637830554715,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3385
    },
    {
      "epoch": 0.03386,
      "grad_norm": 0.8239773523698274,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 3386
    },
    {
      "epoch": 0.03387,
      "grad_norm": 0.73835170938435,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3387
    },
    {
      "epoch": 0.03388,
      "grad_norm": 0.6929973646655294,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3388
    },
    {
      "epoch": 0.03389,
      "grad_norm": 0.7227682380021057,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 3389
    },
    {
      "epoch": 0.0339,
      "grad_norm": 0.6143798247462846,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3390
    },
    {
      "epoch": 0.03391,
      "grad_norm": 0.6734624195180631,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3391
    },
    {
      "epoch": 0.03392,
      "grad_norm": 0.6594794409943178,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3392
    },
    {
      "epoch": 0.03393,
      "grad_norm": 0.5959392487968519,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 3393
    },
    {
      "epoch": 0.03394,
      "grad_norm": 0.5142376929909485,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3394
    },
    {
      "epoch": 0.03395,
      "grad_norm": 0.48066315678667376,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3395
    },
    {
      "epoch": 0.03396,
      "grad_norm": 0.4969135823877688,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 3396
    },
    {
      "epoch": 0.03397,
      "grad_norm": 0.4793333077073278,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 3397
    },
    {
      "epoch": 0.03398,
      "grad_norm": 0.5618846612168108,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 3398
    },
    {
      "epoch": 0.03399,
      "grad_norm": 0.598645896556551,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 3399
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.5909466070203917,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3400
    },
    {
      "epoch": 0.03401,
      "grad_norm": 0.5544973914148591,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3401
    },
    {
      "epoch": 0.03402,
      "grad_norm": 0.5799704954663493,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3402
    },
    {
      "epoch": 0.03403,
      "grad_norm": 0.6679893543628809,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3403
    },
    {
      "epoch": 0.03404,
      "grad_norm": 0.7464230169510123,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 3404
    },
    {
      "epoch": 0.03405,
      "grad_norm": 0.7894447498122126,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3405
    },
    {
      "epoch": 0.03406,
      "grad_norm": 0.7021432861635766,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 3406
    },
    {
      "epoch": 0.03407,
      "grad_norm": 0.5878306021412367,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 3407
    },
    {
      "epoch": 0.03408,
      "grad_norm": 0.5913015706762682,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 3408
    },
    {
      "epoch": 0.03409,
      "grad_norm": 0.6312028321034342,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 3409
    },
    {
      "epoch": 0.0341,
      "grad_norm": 0.7898832429117774,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3410
    },
    {
      "epoch": 0.03411,
      "grad_norm": 0.7714877743902544,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3411
    },
    {
      "epoch": 0.03412,
      "grad_norm": 0.7513990558104037,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 3412
    },
    {
      "epoch": 0.03413,
      "grad_norm": 0.7461988498435763,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3413
    },
    {
      "epoch": 0.03414,
      "grad_norm": 0.7030092576304908,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 3414
    },
    {
      "epoch": 0.03415,
      "grad_norm": 0.7996163589913181,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 3415
    },
    {
      "epoch": 0.03416,
      "grad_norm": 0.9173276805644571,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 3416
    },
    {
      "epoch": 0.03417,
      "grad_norm": 1.1614777308174031,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 3417
    },
    {
      "epoch": 0.03418,
      "grad_norm": 1.0497658143246895,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3418
    },
    {
      "epoch": 0.03419,
      "grad_norm": 0.8865365686182854,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3419
    },
    {
      "epoch": 0.0342,
      "grad_norm": 0.9104853531188979,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 3420
    },
    {
      "epoch": 0.03421,
      "grad_norm": 0.8728209224738288,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 3421
    },
    {
      "epoch": 0.03422,
      "grad_norm": 0.8151487908159633,
      "learning_rate": 0.003,
      "loss": 4.2121,
      "step": 3422
    },
    {
      "epoch": 0.03423,
      "grad_norm": 0.7760561564619395,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3423
    },
    {
      "epoch": 0.03424,
      "grad_norm": 0.7902274791310419,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 3424
    },
    {
      "epoch": 0.03425,
      "grad_norm": 0.7840677839560138,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3425
    },
    {
      "epoch": 0.03426,
      "grad_norm": 0.7200842223622085,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3426
    },
    {
      "epoch": 0.03427,
      "grad_norm": 0.6907951008112484,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3427
    },
    {
      "epoch": 0.03428,
      "grad_norm": 0.6706145072117141,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3428
    },
    {
      "epoch": 0.03429,
      "grad_norm": 0.5775498037003282,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3429
    },
    {
      "epoch": 0.0343,
      "grad_norm": 0.5280875253924556,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3430
    },
    {
      "epoch": 0.03431,
      "grad_norm": 0.5133378996898088,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3431
    },
    {
      "epoch": 0.03432,
      "grad_norm": 0.5402448334590585,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3432
    },
    {
      "epoch": 0.03433,
      "grad_norm": 0.5171387300004384,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 3433
    },
    {
      "epoch": 0.03434,
      "grad_norm": 0.5654372536507138,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3434
    },
    {
      "epoch": 0.03435,
      "grad_norm": 0.7195487294501742,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3435
    },
    {
      "epoch": 0.03436,
      "grad_norm": 0.9836128063160017,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3436
    },
    {
      "epoch": 0.03437,
      "grad_norm": 1.000819383038808,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3437
    },
    {
      "epoch": 0.03438,
      "grad_norm": 0.7833210304414927,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3438
    },
    {
      "epoch": 0.03439,
      "grad_norm": 0.7155950508827565,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 3439
    },
    {
      "epoch": 0.0344,
      "grad_norm": 0.7239664621860699,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3440
    },
    {
      "epoch": 0.03441,
      "grad_norm": 0.6825035000283939,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3441
    },
    {
      "epoch": 0.03442,
      "grad_norm": 0.6099118062197426,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 3442
    },
    {
      "epoch": 0.03443,
      "grad_norm": 0.6931617154820389,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3443
    },
    {
      "epoch": 0.03444,
      "grad_norm": 0.6606283519642352,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3444
    },
    {
      "epoch": 0.03445,
      "grad_norm": 0.5738857319007685,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 3445
    },
    {
      "epoch": 0.03446,
      "grad_norm": 0.6247920672226246,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 3446
    },
    {
      "epoch": 0.03447,
      "grad_norm": 0.6141551380791748,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3447
    },
    {
      "epoch": 0.03448,
      "grad_norm": 0.5719627595458622,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 3448
    },
    {
      "epoch": 0.03449,
      "grad_norm": 0.5136603871940236,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 3449
    },
    {
      "epoch": 0.0345,
      "grad_norm": 0.4556692204413723,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3450
    },
    {
      "epoch": 0.03451,
      "grad_norm": 0.43785806673883965,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 3451
    },
    {
      "epoch": 0.03452,
      "grad_norm": 0.4135943023058251,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3452
    },
    {
      "epoch": 0.03453,
      "grad_norm": 0.41685265470403127,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 3453
    },
    {
      "epoch": 0.03454,
      "grad_norm": 0.4541366407006024,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 3454
    },
    {
      "epoch": 0.03455,
      "grad_norm": 0.4820485430446041,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 3455
    },
    {
      "epoch": 0.03456,
      "grad_norm": 0.6070175235924494,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3456
    },
    {
      "epoch": 0.03457,
      "grad_norm": 0.9155671860985577,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 3457
    },
    {
      "epoch": 0.03458,
      "grad_norm": 1.0743568532560335,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3458
    },
    {
      "epoch": 0.03459,
      "grad_norm": 0.6911554531998789,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 3459
    },
    {
      "epoch": 0.0346,
      "grad_norm": 0.7160032080825989,
      "learning_rate": 0.003,
      "loss": 4.1932,
      "step": 3460
    },
    {
      "epoch": 0.03461,
      "grad_norm": 0.7205146036072163,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 3461
    },
    {
      "epoch": 0.03462,
      "grad_norm": 0.6026232163742171,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 3462
    },
    {
      "epoch": 0.03463,
      "grad_norm": 0.6794227213204947,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 3463
    },
    {
      "epoch": 0.03464,
      "grad_norm": 0.6456036736035401,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3464
    },
    {
      "epoch": 0.03465,
      "grad_norm": 0.7130971772716332,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3465
    },
    {
      "epoch": 0.03466,
      "grad_norm": 0.6881411959714172,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3466
    },
    {
      "epoch": 0.03467,
      "grad_norm": 0.5261069185968785,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3467
    },
    {
      "epoch": 0.03468,
      "grad_norm": 0.6133848295254514,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3468
    },
    {
      "epoch": 0.03469,
      "grad_norm": 0.5746409041252338,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3469
    },
    {
      "epoch": 0.0347,
      "grad_norm": 0.509910411545196,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3470
    },
    {
      "epoch": 0.03471,
      "grad_norm": 0.5549185257359567,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 3471
    },
    {
      "epoch": 0.03472,
      "grad_norm": 0.5919654155995967,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3472
    },
    {
      "epoch": 0.03473,
      "grad_norm": 0.6922934316382823,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3473
    },
    {
      "epoch": 0.03474,
      "grad_norm": 0.664824640285904,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 3474
    },
    {
      "epoch": 0.03475,
      "grad_norm": 0.690088727777654,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 3475
    },
    {
      "epoch": 0.03476,
      "grad_norm": 0.6776855408026875,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3476
    },
    {
      "epoch": 0.03477,
      "grad_norm": 0.7001953445833428,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3477
    },
    {
      "epoch": 0.03478,
      "grad_norm": 0.7620329811614004,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3478
    },
    {
      "epoch": 0.03479,
      "grad_norm": 0.8981370425674162,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 3479
    },
    {
      "epoch": 0.0348,
      "grad_norm": 0.8250170448327679,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3480
    },
    {
      "epoch": 0.03481,
      "grad_norm": 0.6849653124584689,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3481
    },
    {
      "epoch": 0.03482,
      "grad_norm": 0.6955143755202231,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3482
    },
    {
      "epoch": 0.03483,
      "grad_norm": 0.8314937740860378,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3483
    },
    {
      "epoch": 0.03484,
      "grad_norm": 0.8591376385119152,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 3484
    },
    {
      "epoch": 0.03485,
      "grad_norm": 0.908594692668108,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3485
    },
    {
      "epoch": 0.03486,
      "grad_norm": 0.9172194797930587,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3486
    },
    {
      "epoch": 0.03487,
      "grad_norm": 0.8921568436215902,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3487
    },
    {
      "epoch": 0.03488,
      "grad_norm": 0.7601873754760123,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3488
    },
    {
      "epoch": 0.03489,
      "grad_norm": 0.7787148687779002,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 3489
    },
    {
      "epoch": 0.0349,
      "grad_norm": 0.7649701226102734,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3490
    },
    {
      "epoch": 0.03491,
      "grad_norm": 0.5441805689029632,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3491
    },
    {
      "epoch": 0.03492,
      "grad_norm": 0.5662780979613814,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3492
    },
    {
      "epoch": 0.03493,
      "grad_norm": 0.5236480373981434,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3493
    },
    {
      "epoch": 0.03494,
      "grad_norm": 0.5399778596820843,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3494
    },
    {
      "epoch": 0.03495,
      "grad_norm": 0.5540768413769496,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3495
    },
    {
      "epoch": 0.03496,
      "grad_norm": 0.5933746135701535,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3496
    },
    {
      "epoch": 0.03497,
      "grad_norm": 0.5852489998748386,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3497
    },
    {
      "epoch": 0.03498,
      "grad_norm": 0.6599448025327912,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3498
    },
    {
      "epoch": 0.03499,
      "grad_norm": 0.7172393903833442,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3499
    },
    {
      "epoch": 0.035,
      "grad_norm": 0.6405157363259029,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 3500
    },
    {
      "epoch": 0.03501,
      "grad_norm": 0.5654687699259198,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3501
    },
    {
      "epoch": 0.03502,
      "grad_norm": 0.618591430417172,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3502
    },
    {
      "epoch": 0.03503,
      "grad_norm": 0.6943996447233877,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3503
    },
    {
      "epoch": 0.03504,
      "grad_norm": 0.6621001515018434,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3504
    },
    {
      "epoch": 0.03505,
      "grad_norm": 0.597184679883689,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3505
    },
    {
      "epoch": 0.03506,
      "grad_norm": 0.5659263992573133,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3506
    },
    {
      "epoch": 0.03507,
      "grad_norm": 0.5952738202818753,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3507
    },
    {
      "epoch": 0.03508,
      "grad_norm": 0.6319710224532459,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 3508
    },
    {
      "epoch": 0.03509,
      "grad_norm": 0.7003635970154289,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 3509
    },
    {
      "epoch": 0.0351,
      "grad_norm": 0.7544632618493929,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 3510
    },
    {
      "epoch": 0.03511,
      "grad_norm": 0.8665700214441323,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3511
    },
    {
      "epoch": 0.03512,
      "grad_norm": 0.8564897161621264,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 3512
    },
    {
      "epoch": 0.03513,
      "grad_norm": 0.9197981119921014,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3513
    },
    {
      "epoch": 0.03514,
      "grad_norm": 0.8492203049153929,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3514
    },
    {
      "epoch": 0.03515,
      "grad_norm": 0.8901010567034684,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 3515
    },
    {
      "epoch": 0.03516,
      "grad_norm": 0.9283878054698729,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 3516
    },
    {
      "epoch": 0.03517,
      "grad_norm": 0.8804785262543259,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3517
    },
    {
      "epoch": 0.03518,
      "grad_norm": 0.8785403040867484,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3518
    },
    {
      "epoch": 0.03519,
      "grad_norm": 0.8144080502479469,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 3519
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.7871723040057216,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 3520
    },
    {
      "epoch": 0.03521,
      "grad_norm": 0.7368932966297566,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3521
    },
    {
      "epoch": 0.03522,
      "grad_norm": 0.7353338197865396,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3522
    },
    {
      "epoch": 0.03523,
      "grad_norm": 0.80714477911056,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 3523
    },
    {
      "epoch": 0.03524,
      "grad_norm": 0.8183318273894814,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3524
    },
    {
      "epoch": 0.03525,
      "grad_norm": 0.7083620836915047,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3525
    },
    {
      "epoch": 0.03526,
      "grad_norm": 0.6358752706280991,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3526
    },
    {
      "epoch": 0.03527,
      "grad_norm": 0.5882454641700064,
      "learning_rate": 0.003,
      "loss": 4.1866,
      "step": 3527
    },
    {
      "epoch": 0.03528,
      "grad_norm": 0.5418834609571274,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3528
    },
    {
      "epoch": 0.03529,
      "grad_norm": 0.4938060985233679,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3529
    },
    {
      "epoch": 0.0353,
      "grad_norm": 0.47222674272043563,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3530
    },
    {
      "epoch": 0.03531,
      "grad_norm": 0.4494800604692719,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3531
    },
    {
      "epoch": 0.03532,
      "grad_norm": 0.46025033508749047,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3532
    },
    {
      "epoch": 0.03533,
      "grad_norm": 0.5287514582627705,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3533
    },
    {
      "epoch": 0.03534,
      "grad_norm": 0.6330388929499089,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3534
    },
    {
      "epoch": 0.03535,
      "grad_norm": 0.6032059148264879,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 3535
    },
    {
      "epoch": 0.03536,
      "grad_norm": 0.5769078485745863,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 3536
    },
    {
      "epoch": 0.03537,
      "grad_norm": 0.7065282154758016,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3537
    },
    {
      "epoch": 0.03538,
      "grad_norm": 0.8671025316072222,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 3538
    },
    {
      "epoch": 0.03539,
      "grad_norm": 1.0158613655750308,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3539
    },
    {
      "epoch": 0.0354,
      "grad_norm": 0.8533960614991879,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 3540
    },
    {
      "epoch": 0.03541,
      "grad_norm": 0.6659419541207255,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3541
    },
    {
      "epoch": 0.03542,
      "grad_norm": 0.6774420684048681,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3542
    },
    {
      "epoch": 0.03543,
      "grad_norm": 0.7402573585555693,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3543
    },
    {
      "epoch": 0.03544,
      "grad_norm": 0.6226751107190542,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3544
    },
    {
      "epoch": 0.03545,
      "grad_norm": 0.5532949073684433,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 3545
    },
    {
      "epoch": 0.03546,
      "grad_norm": 0.521811868333153,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3546
    },
    {
      "epoch": 0.03547,
      "grad_norm": 0.5542126267573285,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3547
    },
    {
      "epoch": 0.03548,
      "grad_norm": 0.5294342446885085,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 3548
    },
    {
      "epoch": 0.03549,
      "grad_norm": 0.46845119607442226,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3549
    },
    {
      "epoch": 0.0355,
      "grad_norm": 0.4173574369972876,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 3550
    },
    {
      "epoch": 0.03551,
      "grad_norm": 0.5597688720149216,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3551
    },
    {
      "epoch": 0.03552,
      "grad_norm": 0.684279124843178,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 3552
    },
    {
      "epoch": 0.03553,
      "grad_norm": 0.7906598471557775,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 3553
    },
    {
      "epoch": 0.03554,
      "grad_norm": 0.8477558409372302,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3554
    },
    {
      "epoch": 0.03555,
      "grad_norm": 0.9130757811703537,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3555
    },
    {
      "epoch": 0.03556,
      "grad_norm": 1.0233508386779784,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3556
    },
    {
      "epoch": 0.03557,
      "grad_norm": 0.9577495816046034,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3557
    },
    {
      "epoch": 0.03558,
      "grad_norm": 0.6971645674236855,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 3558
    },
    {
      "epoch": 0.03559,
      "grad_norm": 0.615773531816,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3559
    },
    {
      "epoch": 0.0356,
      "grad_norm": 0.7565427067348032,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3560
    },
    {
      "epoch": 0.03561,
      "grad_norm": 0.7654646543935898,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 3561
    },
    {
      "epoch": 0.03562,
      "grad_norm": 0.7171787225466916,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3562
    },
    {
      "epoch": 0.03563,
      "grad_norm": 0.7202637052044578,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 3563
    },
    {
      "epoch": 0.03564,
      "grad_norm": 0.7813914185176697,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3564
    },
    {
      "epoch": 0.03565,
      "grad_norm": 0.9681384729900618,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3565
    },
    {
      "epoch": 0.03566,
      "grad_norm": 0.9815276969782409,
      "learning_rate": 0.003,
      "loss": 4.1871,
      "step": 3566
    },
    {
      "epoch": 0.03567,
      "grad_norm": 0.8545575756116207,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3567
    },
    {
      "epoch": 0.03568,
      "grad_norm": 0.8579697593867708,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3568
    },
    {
      "epoch": 0.03569,
      "grad_norm": 0.9295006292416876,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 3569
    },
    {
      "epoch": 0.0357,
      "grad_norm": 0.7824959494204987,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3570
    },
    {
      "epoch": 0.03571,
      "grad_norm": 0.715056502621953,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3571
    },
    {
      "epoch": 0.03572,
      "grad_norm": 0.7939274654308445,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 3572
    },
    {
      "epoch": 0.03573,
      "grad_norm": 0.8990540619234125,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 3573
    },
    {
      "epoch": 0.03574,
      "grad_norm": 0.8924167007734459,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 3574
    },
    {
      "epoch": 0.03575,
      "grad_norm": 0.9633999021454877,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3575
    },
    {
      "epoch": 0.03576,
      "grad_norm": 0.8337435293260566,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3576
    },
    {
      "epoch": 0.03577,
      "grad_norm": 0.7990789046598609,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 3577
    },
    {
      "epoch": 0.03578,
      "grad_norm": 0.740247280194808,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 3578
    },
    {
      "epoch": 0.03579,
      "grad_norm": 0.736535666256918,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 3579
    },
    {
      "epoch": 0.0358,
      "grad_norm": 0.7123148002792392,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3580
    },
    {
      "epoch": 0.03581,
      "grad_norm": 0.6711995328778474,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3581
    },
    {
      "epoch": 0.03582,
      "grad_norm": 0.6405574599918185,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 3582
    },
    {
      "epoch": 0.03583,
      "grad_norm": 0.6115022182938825,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3583
    },
    {
      "epoch": 0.03584,
      "grad_norm": 0.5561430180920142,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 3584
    },
    {
      "epoch": 0.03585,
      "grad_norm": 0.5150442741371403,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 3585
    },
    {
      "epoch": 0.03586,
      "grad_norm": 0.5324475951916865,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3586
    },
    {
      "epoch": 0.03587,
      "grad_norm": 0.5865627283177408,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3587
    },
    {
      "epoch": 0.03588,
      "grad_norm": 0.6404975493032687,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3588
    },
    {
      "epoch": 0.03589,
      "grad_norm": 0.6198477170373378,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3589
    },
    {
      "epoch": 0.0359,
      "grad_norm": 0.5436091514667869,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 3590
    },
    {
      "epoch": 0.03591,
      "grad_norm": 0.5153318776417137,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 3591
    },
    {
      "epoch": 0.03592,
      "grad_norm": 0.4789482382988687,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 3592
    },
    {
      "epoch": 0.03593,
      "grad_norm": 0.4332591255641176,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3593
    },
    {
      "epoch": 0.03594,
      "grad_norm": 0.46006005882861767,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 3594
    },
    {
      "epoch": 0.03595,
      "grad_norm": 0.39445716900409733,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 3595
    },
    {
      "epoch": 0.03596,
      "grad_norm": 0.40621407877255555,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3596
    },
    {
      "epoch": 0.03597,
      "grad_norm": 0.4360669588971265,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 3597
    },
    {
      "epoch": 0.03598,
      "grad_norm": 0.5144534194928903,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3598
    },
    {
      "epoch": 0.03599,
      "grad_norm": 0.7248031737778203,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3599
    },
    {
      "epoch": 0.036,
      "grad_norm": 1.0800621032582747,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 3600
    },
    {
      "epoch": 0.03601,
      "grad_norm": 0.9508478687848839,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3601
    },
    {
      "epoch": 0.03602,
      "grad_norm": 0.6514600121121226,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 3602
    },
    {
      "epoch": 0.03603,
      "grad_norm": 0.6278285308162193,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 3603
    },
    {
      "epoch": 0.03604,
      "grad_norm": 0.7950592611481373,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3604
    },
    {
      "epoch": 0.03605,
      "grad_norm": 0.7151902116041485,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3605
    },
    {
      "epoch": 0.03606,
      "grad_norm": 0.7086662453671662,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3606
    },
    {
      "epoch": 0.03607,
      "grad_norm": 0.7175966521512804,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3607
    },
    {
      "epoch": 0.03608,
      "grad_norm": 0.587764679138397,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 3608
    },
    {
      "epoch": 0.03609,
      "grad_norm": 0.5305480603629071,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3609
    },
    {
      "epoch": 0.0361,
      "grad_norm": 0.5522309324109884,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3610
    },
    {
      "epoch": 0.03611,
      "grad_norm": 0.5828315338943078,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 3611
    },
    {
      "epoch": 0.03612,
      "grad_norm": 0.6300633453025702,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 3612
    },
    {
      "epoch": 0.03613,
      "grad_norm": 0.617268364161966,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3613
    },
    {
      "epoch": 0.03614,
      "grad_norm": 0.6309109479392245,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 3614
    },
    {
      "epoch": 0.03615,
      "grad_norm": 0.570782892253401,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3615
    },
    {
      "epoch": 0.03616,
      "grad_norm": 0.5541049972487797,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3616
    },
    {
      "epoch": 0.03617,
      "grad_norm": 0.5951507982053302,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3617
    },
    {
      "epoch": 0.03618,
      "grad_norm": 0.7159646588024825,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3618
    },
    {
      "epoch": 0.03619,
      "grad_norm": 0.8176290428711182,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 3619
    },
    {
      "epoch": 0.0362,
      "grad_norm": 0.920159490234485,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3620
    },
    {
      "epoch": 0.03621,
      "grad_norm": 0.9049277558974237,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3621
    },
    {
      "epoch": 0.03622,
      "grad_norm": 0.9134672000988132,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 3622
    },
    {
      "epoch": 0.03623,
      "grad_norm": 0.8632217222934753,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 3623
    },
    {
      "epoch": 0.03624,
      "grad_norm": 0.7748608966158578,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 3624
    },
    {
      "epoch": 0.03625,
      "grad_norm": 0.6549902845991534,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3625
    },
    {
      "epoch": 0.03626,
      "grad_norm": 0.6025866731625074,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3626
    },
    {
      "epoch": 0.03627,
      "grad_norm": 0.5718452823869131,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 3627
    },
    {
      "epoch": 0.03628,
      "grad_norm": 0.5380938313028744,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3628
    },
    {
      "epoch": 0.03629,
      "grad_norm": 0.5051723060670195,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3629
    },
    {
      "epoch": 0.0363,
      "grad_norm": 0.5404489026019382,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 3630
    },
    {
      "epoch": 0.03631,
      "grad_norm": 0.6007935793023943,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 3631
    },
    {
      "epoch": 0.03632,
      "grad_norm": 0.620386838070519,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 3632
    },
    {
      "epoch": 0.03633,
      "grad_norm": 0.7326091251339881,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3633
    },
    {
      "epoch": 0.03634,
      "grad_norm": 0.9520374375008709,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3634
    },
    {
      "epoch": 0.03635,
      "grad_norm": 0.9435644392052507,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 3635
    },
    {
      "epoch": 0.03636,
      "grad_norm": 0.8034652176528725,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3636
    },
    {
      "epoch": 0.03637,
      "grad_norm": 0.8215010352273984,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3637
    },
    {
      "epoch": 0.03638,
      "grad_norm": 0.7729628364890975,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3638
    },
    {
      "epoch": 0.03639,
      "grad_norm": 0.7939041200619633,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3639
    },
    {
      "epoch": 0.0364,
      "grad_norm": 0.7624978886682702,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3640
    },
    {
      "epoch": 0.03641,
      "grad_norm": 0.6972108651120591,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 3641
    },
    {
      "epoch": 0.03642,
      "grad_norm": 0.686374864648811,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 3642
    },
    {
      "epoch": 0.03643,
      "grad_norm": 0.6994446209399746,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3643
    },
    {
      "epoch": 0.03644,
      "grad_norm": 0.7789189969606384,
      "learning_rate": 0.003,
      "loss": 4.1819,
      "step": 3644
    },
    {
      "epoch": 0.03645,
      "grad_norm": 0.7984886139835612,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3645
    },
    {
      "epoch": 0.03646,
      "grad_norm": 0.6181112574802159,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3646
    },
    {
      "epoch": 0.03647,
      "grad_norm": 0.6061654420727669,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3647
    },
    {
      "epoch": 0.03648,
      "grad_norm": 0.6696338465160281,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3648
    },
    {
      "epoch": 0.03649,
      "grad_norm": 0.7331012822671169,
      "learning_rate": 0.003,
      "loss": 4.1808,
      "step": 3649
    },
    {
      "epoch": 0.0365,
      "grad_norm": 0.7037525814274137,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3650
    },
    {
      "epoch": 0.03651,
      "grad_norm": 0.692119661540781,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 3651
    },
    {
      "epoch": 0.03652,
      "grad_norm": 0.6604395418921041,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3652
    },
    {
      "epoch": 0.03653,
      "grad_norm": 0.6346433017565251,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 3653
    },
    {
      "epoch": 0.03654,
      "grad_norm": 0.6759345560485471,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3654
    },
    {
      "epoch": 0.03655,
      "grad_norm": 0.7374777892156636,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 3655
    },
    {
      "epoch": 0.03656,
      "grad_norm": 0.8895381978397107,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3656
    },
    {
      "epoch": 0.03657,
      "grad_norm": 1.0700577195843572,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3657
    },
    {
      "epoch": 0.03658,
      "grad_norm": 0.8922825561167493,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 3658
    },
    {
      "epoch": 0.03659,
      "grad_norm": 0.754621627580339,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3659
    },
    {
      "epoch": 0.0366,
      "grad_norm": 0.8002619699354713,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 3660
    },
    {
      "epoch": 0.03661,
      "grad_norm": 0.7711130288745789,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3661
    },
    {
      "epoch": 0.03662,
      "grad_norm": 0.7607029556463581,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 3662
    },
    {
      "epoch": 0.03663,
      "grad_norm": 0.7440287835939471,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 3663
    },
    {
      "epoch": 0.03664,
      "grad_norm": 0.563780432136041,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3664
    },
    {
      "epoch": 0.03665,
      "grad_norm": 0.4357399894939763,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3665
    },
    {
      "epoch": 0.03666,
      "grad_norm": 0.45515608243379135,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 3666
    },
    {
      "epoch": 0.03667,
      "grad_norm": 0.4706495224710868,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 3667
    },
    {
      "epoch": 0.03668,
      "grad_norm": 0.4175208514628951,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 3668
    },
    {
      "epoch": 0.03669,
      "grad_norm": 0.3785566363290307,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 3669
    },
    {
      "epoch": 0.0367,
      "grad_norm": 0.3437728850428848,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3670
    },
    {
      "epoch": 0.03671,
      "grad_norm": 0.3820679778361713,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 3671
    },
    {
      "epoch": 0.03672,
      "grad_norm": 0.44348210068257454,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3672
    },
    {
      "epoch": 0.03673,
      "grad_norm": 0.5490971493740642,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3673
    },
    {
      "epoch": 0.03674,
      "grad_norm": 0.6223490419700332,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 3674
    },
    {
      "epoch": 0.03675,
      "grad_norm": 0.6342035142985044,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3675
    },
    {
      "epoch": 0.03676,
      "grad_norm": 0.631553220636773,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 3676
    },
    {
      "epoch": 0.03677,
      "grad_norm": 0.8208223460663321,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 3677
    },
    {
      "epoch": 0.03678,
      "grad_norm": 1.057395404894455,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 3678
    },
    {
      "epoch": 0.03679,
      "grad_norm": 1.059489837092536,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3679
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.9648807335139974,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3680
    },
    {
      "epoch": 0.03681,
      "grad_norm": 0.9082948485919852,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 3681
    },
    {
      "epoch": 0.03682,
      "grad_norm": 0.7608012026223391,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3682
    },
    {
      "epoch": 0.03683,
      "grad_norm": 0.7012632382999905,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3683
    },
    {
      "epoch": 0.03684,
      "grad_norm": 0.7244234234962599,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3684
    },
    {
      "epoch": 0.03685,
      "grad_norm": 0.6796055439571254,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3685
    },
    {
      "epoch": 0.03686,
      "grad_norm": 0.6397461073626532,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3686
    },
    {
      "epoch": 0.03687,
      "grad_norm": 0.6837549545590554,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 3687
    },
    {
      "epoch": 0.03688,
      "grad_norm": 0.612454393388502,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 3688
    },
    {
      "epoch": 0.03689,
      "grad_norm": 0.6808423570337984,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3689
    },
    {
      "epoch": 0.0369,
      "grad_norm": 0.6866647703326533,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 3690
    },
    {
      "epoch": 0.03691,
      "grad_norm": 0.7296388488733391,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3691
    },
    {
      "epoch": 0.03692,
      "grad_norm": 0.8733542369985744,
      "learning_rate": 0.003,
      "loss": 4.2056,
      "step": 3692
    },
    {
      "epoch": 0.03693,
      "grad_norm": 0.9120312999635036,
      "learning_rate": 0.003,
      "loss": 4.1787,
      "step": 3693
    },
    {
      "epoch": 0.03694,
      "grad_norm": 0.7928228320045024,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3694
    },
    {
      "epoch": 0.03695,
      "grad_norm": 0.7574597489784519,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3695
    },
    {
      "epoch": 0.03696,
      "grad_norm": 0.7264595127076895,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3696
    },
    {
      "epoch": 0.03697,
      "grad_norm": 0.797502557212387,
      "learning_rate": 0.003,
      "loss": 4.1553,
      "step": 3697
    },
    {
      "epoch": 0.03698,
      "grad_norm": 0.8741564039842601,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 3698
    },
    {
      "epoch": 0.03699,
      "grad_norm": 0.8855946379470947,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 3699
    },
    {
      "epoch": 0.037,
      "grad_norm": 0.8586025114326333,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3700
    },
    {
      "epoch": 0.03701,
      "grad_norm": 0.8930614814476187,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3701
    },
    {
      "epoch": 0.03702,
      "grad_norm": 0.8802482281802025,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3702
    },
    {
      "epoch": 0.03703,
      "grad_norm": 0.7549215853060027,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3703
    },
    {
      "epoch": 0.03704,
      "grad_norm": 0.6295591556831198,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3704
    },
    {
      "epoch": 0.03705,
      "grad_norm": 0.5580114287621604,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3705
    },
    {
      "epoch": 0.03706,
      "grad_norm": 0.493748778061454,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3706
    },
    {
      "epoch": 0.03707,
      "grad_norm": 0.501411513908697,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3707
    },
    {
      "epoch": 0.03708,
      "grad_norm": 0.46976615957109447,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3708
    },
    {
      "epoch": 0.03709,
      "grad_norm": 0.4882450257172406,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 3709
    },
    {
      "epoch": 0.0371,
      "grad_norm": 0.5609237339392169,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3710
    },
    {
      "epoch": 0.03711,
      "grad_norm": 0.577744009339436,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3711
    },
    {
      "epoch": 0.03712,
      "grad_norm": 0.5617641082054192,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3712
    },
    {
      "epoch": 0.03713,
      "grad_norm": 0.6129339279774879,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 3713
    },
    {
      "epoch": 0.03714,
      "grad_norm": 0.761778934710984,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3714
    },
    {
      "epoch": 0.03715,
      "grad_norm": 0.936702433405622,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3715
    },
    {
      "epoch": 0.03716,
      "grad_norm": 1.0004110813808647,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3716
    },
    {
      "epoch": 0.03717,
      "grad_norm": 0.9246847467387567,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3717
    },
    {
      "epoch": 0.03718,
      "grad_norm": 0.8130004478592973,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 3718
    },
    {
      "epoch": 0.03719,
      "grad_norm": 0.8110626454843745,
      "learning_rate": 0.003,
      "loss": 4.1701,
      "step": 3719
    },
    {
      "epoch": 0.0372,
      "grad_norm": 0.6478944328491517,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3720
    },
    {
      "epoch": 0.03721,
      "grad_norm": 0.6036054997551548,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 3721
    },
    {
      "epoch": 0.03722,
      "grad_norm": 0.47934112751684016,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3722
    },
    {
      "epoch": 0.03723,
      "grad_norm": 0.5465870050613278,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 3723
    },
    {
      "epoch": 0.03724,
      "grad_norm": 0.5794138576951311,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3724
    },
    {
      "epoch": 0.03725,
      "grad_norm": 0.5728374193198114,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3725
    },
    {
      "epoch": 0.03726,
      "grad_norm": 0.5999625053130218,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3726
    },
    {
      "epoch": 0.03727,
      "grad_norm": 0.6205981266168359,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3727
    },
    {
      "epoch": 0.03728,
      "grad_norm": 0.6360363945769689,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 3728
    },
    {
      "epoch": 0.03729,
      "grad_norm": 0.5842605492528835,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3729
    },
    {
      "epoch": 0.0373,
      "grad_norm": 0.5328025691146133,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3730
    },
    {
      "epoch": 0.03731,
      "grad_norm": 0.53824919717479,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3731
    },
    {
      "epoch": 0.03732,
      "grad_norm": 0.6816149389539496,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3732
    },
    {
      "epoch": 0.03733,
      "grad_norm": 0.7649227254747324,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3733
    },
    {
      "epoch": 0.03734,
      "grad_norm": 0.723858182349595,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3734
    },
    {
      "epoch": 0.03735,
      "grad_norm": 0.7221215059978716,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3735
    },
    {
      "epoch": 0.03736,
      "grad_norm": 0.7400835788486658,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3736
    },
    {
      "epoch": 0.03737,
      "grad_norm": 0.6969381819145593,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 3737
    },
    {
      "epoch": 0.03738,
      "grad_norm": 0.6127048568575166,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3738
    },
    {
      "epoch": 0.03739,
      "grad_norm": 0.6440493642792959,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 3739
    },
    {
      "epoch": 0.0374,
      "grad_norm": 0.72519391026854,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3740
    },
    {
      "epoch": 0.03741,
      "grad_norm": 0.7715264840802699,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 3741
    },
    {
      "epoch": 0.03742,
      "grad_norm": 0.6234960786729838,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 3742
    },
    {
      "epoch": 0.03743,
      "grad_norm": 0.529614255393143,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 3743
    },
    {
      "epoch": 0.03744,
      "grad_norm": 0.5395949119845926,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3744
    },
    {
      "epoch": 0.03745,
      "grad_norm": 0.5717733035088027,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 3745
    },
    {
      "epoch": 0.03746,
      "grad_norm": 0.7148827950766982,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3746
    },
    {
      "epoch": 0.03747,
      "grad_norm": 0.9619906254460838,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3747
    },
    {
      "epoch": 0.03748,
      "grad_norm": 0.9454260095382823,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3748
    },
    {
      "epoch": 0.03749,
      "grad_norm": 0.8505540045662432,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 3749
    },
    {
      "epoch": 0.0375,
      "grad_norm": 0.8001368316453729,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3750
    },
    {
      "epoch": 0.03751,
      "grad_norm": 0.6787039450042418,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 3751
    },
    {
      "epoch": 0.03752,
      "grad_norm": 0.6675510373317245,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3752
    },
    {
      "epoch": 0.03753,
      "grad_norm": 0.659564153036602,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3753
    },
    {
      "epoch": 0.03754,
      "grad_norm": 0.5879492269067329,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3754
    },
    {
      "epoch": 0.03755,
      "grad_norm": 0.5251202407537541,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3755
    },
    {
      "epoch": 0.03756,
      "grad_norm": 0.47935868138717536,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 3756
    },
    {
      "epoch": 0.03757,
      "grad_norm": 0.5841333632592701,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3757
    },
    {
      "epoch": 0.03758,
      "grad_norm": 0.7521988739330352,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 3758
    },
    {
      "epoch": 0.03759,
      "grad_norm": 0.8503529306111015,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 3759
    },
    {
      "epoch": 0.0376,
      "grad_norm": 0.8653069459339167,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 3760
    },
    {
      "epoch": 0.03761,
      "grad_norm": 0.8195705554984235,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 3761
    },
    {
      "epoch": 0.03762,
      "grad_norm": 0.7776430458932468,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 3762
    },
    {
      "epoch": 0.03763,
      "grad_norm": 0.8090187425891137,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 3763
    },
    {
      "epoch": 0.03764,
      "grad_norm": 0.7299629395412605,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3764
    },
    {
      "epoch": 0.03765,
      "grad_norm": 0.7220941046527883,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 3765
    },
    {
      "epoch": 0.03766,
      "grad_norm": 0.7938574946684543,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3766
    },
    {
      "epoch": 0.03767,
      "grad_norm": 0.848179339225032,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 3767
    },
    {
      "epoch": 0.03768,
      "grad_norm": 0.8273555270559375,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3768
    },
    {
      "epoch": 0.03769,
      "grad_norm": 0.7189795085544652,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 3769
    },
    {
      "epoch": 0.0377,
      "grad_norm": 0.6877419998361242,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 3770
    },
    {
      "epoch": 0.03771,
      "grad_norm": 0.7347886642206836,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3771
    },
    {
      "epoch": 0.03772,
      "grad_norm": 0.7369268379376811,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3772
    },
    {
      "epoch": 0.03773,
      "grad_norm": 0.7648153892095982,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3773
    },
    {
      "epoch": 0.03774,
      "grad_norm": 0.689330468097358,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 3774
    },
    {
      "epoch": 0.03775,
      "grad_norm": 0.7165358486039451,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3775
    },
    {
      "epoch": 0.03776,
      "grad_norm": 0.7489974209205834,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3776
    },
    {
      "epoch": 0.03777,
      "grad_norm": 0.8249125931936281,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3777
    },
    {
      "epoch": 0.03778,
      "grad_norm": 0.7978997683857884,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 3778
    },
    {
      "epoch": 0.03779,
      "grad_norm": 0.7763812864712815,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3779
    },
    {
      "epoch": 0.0378,
      "grad_norm": 0.6854145264424516,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3780
    },
    {
      "epoch": 0.03781,
      "grad_norm": 0.7017290642180594,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3781
    },
    {
      "epoch": 0.03782,
      "grad_norm": 0.7499872472851988,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3782
    },
    {
      "epoch": 0.03783,
      "grad_norm": 0.7066696741848768,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3783
    },
    {
      "epoch": 0.03784,
      "grad_norm": 0.6293823118533233,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3784
    },
    {
      "epoch": 0.03785,
      "grad_norm": 0.5600002308125822,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3785
    },
    {
      "epoch": 0.03786,
      "grad_norm": 0.4985251419936004,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3786
    },
    {
      "epoch": 0.03787,
      "grad_norm": 0.4682470683406906,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3787
    },
    {
      "epoch": 0.03788,
      "grad_norm": 0.507866310001821,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3788
    },
    {
      "epoch": 0.03789,
      "grad_norm": 0.5463073884456529,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 3789
    },
    {
      "epoch": 0.0379,
      "grad_norm": 0.6172159531979761,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 3790
    },
    {
      "epoch": 0.03791,
      "grad_norm": 0.7402780853790922,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 3791
    },
    {
      "epoch": 0.03792,
      "grad_norm": 0.9096314434511164,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3792
    },
    {
      "epoch": 0.03793,
      "grad_norm": 0.8009156787875767,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 3793
    },
    {
      "epoch": 0.03794,
      "grad_norm": 0.6007769119418681,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3794
    },
    {
      "epoch": 0.03795,
      "grad_norm": 0.6248284046940696,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3795
    },
    {
      "epoch": 0.03796,
      "grad_norm": 0.6313675517193291,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 3796
    },
    {
      "epoch": 0.03797,
      "grad_norm": 0.6260974457476847,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 3797
    },
    {
      "epoch": 0.03798,
      "grad_norm": 0.6329676643978727,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 3798
    },
    {
      "epoch": 0.03799,
      "grad_norm": 0.6413791763016119,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3799
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.5746316704038654,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3800
    },
    {
      "epoch": 0.03801,
      "grad_norm": 0.5630475643545394,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3801
    },
    {
      "epoch": 0.03802,
      "grad_norm": 0.5962762680426152,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 3802
    },
    {
      "epoch": 0.03803,
      "grad_norm": 0.7181192675393114,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3803
    },
    {
      "epoch": 0.03804,
      "grad_norm": 0.7862836747313122,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3804
    },
    {
      "epoch": 0.03805,
      "grad_norm": 0.8710510112491459,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3805
    },
    {
      "epoch": 0.03806,
      "grad_norm": 0.8521659658182485,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3806
    },
    {
      "epoch": 0.03807,
      "grad_norm": 0.7446144627159563,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3807
    },
    {
      "epoch": 0.03808,
      "grad_norm": 0.7136381828235824,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3808
    },
    {
      "epoch": 0.03809,
      "grad_norm": 0.6617555150784667,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3809
    },
    {
      "epoch": 0.0381,
      "grad_norm": 0.5589720929569325,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 3810
    },
    {
      "epoch": 0.03811,
      "grad_norm": 0.513976956124665,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 3811
    },
    {
      "epoch": 0.03812,
      "grad_norm": 0.5341053812083173,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3812
    },
    {
      "epoch": 0.03813,
      "grad_norm": 0.5180551827098935,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 3813
    },
    {
      "epoch": 0.03814,
      "grad_norm": 0.5664522386981432,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3814
    },
    {
      "epoch": 0.03815,
      "grad_norm": 0.7610361605531731,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 3815
    },
    {
      "epoch": 0.03816,
      "grad_norm": 0.9587327701936716,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3816
    },
    {
      "epoch": 0.03817,
      "grad_norm": 0.935563437309735,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 3817
    },
    {
      "epoch": 0.03818,
      "grad_norm": 1.0060426088546508,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 3818
    },
    {
      "epoch": 0.03819,
      "grad_norm": 0.9721837543200701,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3819
    },
    {
      "epoch": 0.0382,
      "grad_norm": 1.1304312225449777,
      "learning_rate": 0.003,
      "loss": 4.2256,
      "step": 3820
    },
    {
      "epoch": 0.03821,
      "grad_norm": 1.0177206332843154,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 3821
    },
    {
      "epoch": 0.03822,
      "grad_norm": 1.031220537341774,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 3822
    },
    {
      "epoch": 0.03823,
      "grad_norm": 1.0388061376025717,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3823
    },
    {
      "epoch": 0.03824,
      "grad_norm": 0.9677858739885562,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 3824
    },
    {
      "epoch": 0.03825,
      "grad_norm": 0.937987879509307,
      "learning_rate": 0.003,
      "loss": 4.2046,
      "step": 3825
    },
    {
      "epoch": 0.03826,
      "grad_norm": 0.8230886831468718,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3826
    },
    {
      "epoch": 0.03827,
      "grad_norm": 0.6935375745622523,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 3827
    },
    {
      "epoch": 0.03828,
      "grad_norm": 0.6856721015265079,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 3828
    },
    {
      "epoch": 0.03829,
      "grad_norm": 0.6394378471082416,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3829
    },
    {
      "epoch": 0.0383,
      "grad_norm": 0.6570235063756633,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3830
    },
    {
      "epoch": 0.03831,
      "grad_norm": 0.6505218073754993,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 3831
    },
    {
      "epoch": 0.03832,
      "grad_norm": 0.5403168994861741,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3832
    },
    {
      "epoch": 0.03833,
      "grad_norm": 0.42095483529999317,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3833
    },
    {
      "epoch": 0.03834,
      "grad_norm": 0.40483366089412304,
      "learning_rate": 0.003,
      "loss": 4.1845,
      "step": 3834
    },
    {
      "epoch": 0.03835,
      "grad_norm": 0.4630812031805051,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3835
    },
    {
      "epoch": 0.03836,
      "grad_norm": 0.4841208007987568,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3836
    },
    {
      "epoch": 0.03837,
      "grad_norm": 0.536907199381203,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3837
    },
    {
      "epoch": 0.03838,
      "grad_norm": 0.6117335045256618,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3838
    },
    {
      "epoch": 0.03839,
      "grad_norm": 0.6780085079827656,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3839
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.697829286285506,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3840
    },
    {
      "epoch": 0.03841,
      "grad_norm": 0.8027464809531286,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3841
    },
    {
      "epoch": 0.03842,
      "grad_norm": 0.8177694533436255,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 3842
    },
    {
      "epoch": 0.03843,
      "grad_norm": 0.6989170877767756,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 3843
    },
    {
      "epoch": 0.03844,
      "grad_norm": 0.6383070406944257,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3844
    },
    {
      "epoch": 0.03845,
      "grad_norm": 0.6201943361574275,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3845
    },
    {
      "epoch": 0.03846,
      "grad_norm": 0.6699207633477972,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 3846
    },
    {
      "epoch": 0.03847,
      "grad_norm": 0.5790976587711703,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3847
    },
    {
      "epoch": 0.03848,
      "grad_norm": 0.5911075701092868,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 3848
    },
    {
      "epoch": 0.03849,
      "grad_norm": 0.6969380862395563,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3849
    },
    {
      "epoch": 0.0385,
      "grad_norm": 0.7972901632396154,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3850
    },
    {
      "epoch": 0.03851,
      "grad_norm": 0.8999501436569268,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 3851
    },
    {
      "epoch": 0.03852,
      "grad_norm": 0.9359674147442539,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3852
    },
    {
      "epoch": 0.03853,
      "grad_norm": 0.7466141042924882,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3853
    },
    {
      "epoch": 0.03854,
      "grad_norm": 0.5969895365980001,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3854
    },
    {
      "epoch": 0.03855,
      "grad_norm": 0.5270288768727529,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 3855
    },
    {
      "epoch": 0.03856,
      "grad_norm": 0.5338061114857245,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 3856
    },
    {
      "epoch": 0.03857,
      "grad_norm": 0.5294295191115548,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3857
    },
    {
      "epoch": 0.03858,
      "grad_norm": 0.5523522268399486,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 3858
    },
    {
      "epoch": 0.03859,
      "grad_norm": 0.6152187330517869,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3859
    },
    {
      "epoch": 0.0386,
      "grad_norm": 0.6117532516023865,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 3860
    },
    {
      "epoch": 0.03861,
      "grad_norm": 0.5524498434352159,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3861
    },
    {
      "epoch": 0.03862,
      "grad_norm": 0.5236517072626808,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 3862
    },
    {
      "epoch": 0.03863,
      "grad_norm": 0.6248759021675342,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 3863
    },
    {
      "epoch": 0.03864,
      "grad_norm": 0.679464652102799,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3864
    },
    {
      "epoch": 0.03865,
      "grad_norm": 0.6453393413543939,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3865
    },
    {
      "epoch": 0.03866,
      "grad_norm": 0.6101015046125351,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 3866
    },
    {
      "epoch": 0.03867,
      "grad_norm": 0.5167462630710293,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 3867
    },
    {
      "epoch": 0.03868,
      "grad_norm": 0.501446932546633,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3868
    },
    {
      "epoch": 0.03869,
      "grad_norm": 0.4709811100238584,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 3869
    },
    {
      "epoch": 0.0387,
      "grad_norm": 0.4875337874718785,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 3870
    },
    {
      "epoch": 0.03871,
      "grad_norm": 0.5681317313929587,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 3871
    },
    {
      "epoch": 0.03872,
      "grad_norm": 0.6978195894520031,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 3872
    },
    {
      "epoch": 0.03873,
      "grad_norm": 0.9026562899987328,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3873
    },
    {
      "epoch": 0.03874,
      "grad_norm": 1.2643110286573773,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 3874
    },
    {
      "epoch": 0.03875,
      "grad_norm": 0.8522074037692399,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3875
    },
    {
      "epoch": 0.03876,
      "grad_norm": 0.7221351700519376,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 3876
    },
    {
      "epoch": 0.03877,
      "grad_norm": 0.6790217877458674,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3877
    },
    {
      "epoch": 0.03878,
      "grad_norm": 0.6442745541470235,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 3878
    },
    {
      "epoch": 0.03879,
      "grad_norm": 0.5773572680715529,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3879
    },
    {
      "epoch": 0.0388,
      "grad_norm": 0.6630378193548411,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 3880
    },
    {
      "epoch": 0.03881,
      "grad_norm": 0.6679169810032245,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3881
    },
    {
      "epoch": 0.03882,
      "grad_norm": 0.7019796842116967,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3882
    },
    {
      "epoch": 0.03883,
      "grad_norm": 0.6656806453760354,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 3883
    },
    {
      "epoch": 0.03884,
      "grad_norm": 0.7319329418461972,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 3884
    },
    {
      "epoch": 0.03885,
      "grad_norm": 0.9033013251391556,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3885
    },
    {
      "epoch": 0.03886,
      "grad_norm": 1.0950277457821216,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 3886
    },
    {
      "epoch": 0.03887,
      "grad_norm": 0.7794992948946918,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3887
    },
    {
      "epoch": 0.03888,
      "grad_norm": 0.6776556880519633,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 3888
    },
    {
      "epoch": 0.03889,
      "grad_norm": 0.6740634088352366,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3889
    },
    {
      "epoch": 0.0389,
      "grad_norm": 0.6800323734503214,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 3890
    },
    {
      "epoch": 0.03891,
      "grad_norm": 0.6548487525930513,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 3891
    },
    {
      "epoch": 0.03892,
      "grad_norm": 0.6949410167297108,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3892
    },
    {
      "epoch": 0.03893,
      "grad_norm": 0.71600442881932,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 3893
    },
    {
      "epoch": 0.03894,
      "grad_norm": 0.6445397470488516,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3894
    },
    {
      "epoch": 0.03895,
      "grad_norm": 0.5602823203406133,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 3895
    },
    {
      "epoch": 0.03896,
      "grad_norm": 0.5429794704774388,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 3896
    },
    {
      "epoch": 0.03897,
      "grad_norm": 0.5936531994862636,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 3897
    },
    {
      "epoch": 0.03898,
      "grad_norm": 0.6856826306823404,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3898
    },
    {
      "epoch": 0.03899,
      "grad_norm": 0.7274726655026192,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3899
    },
    {
      "epoch": 0.039,
      "grad_norm": 0.8037659881555415,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 3900
    },
    {
      "epoch": 0.03901,
      "grad_norm": 0.8316671306971751,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3901
    },
    {
      "epoch": 0.03902,
      "grad_norm": 0.8611987009716957,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 3902
    },
    {
      "epoch": 0.03903,
      "grad_norm": 0.8344071211818236,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 3903
    },
    {
      "epoch": 0.03904,
      "grad_norm": 0.7880927476353365,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 3904
    },
    {
      "epoch": 0.03905,
      "grad_norm": 0.778724013223126,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 3905
    },
    {
      "epoch": 0.03906,
      "grad_norm": 0.7715175718675301,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3906
    },
    {
      "epoch": 0.03907,
      "grad_norm": 0.7877862833527057,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3907
    },
    {
      "epoch": 0.03908,
      "grad_norm": 0.5993067619626092,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 3908
    },
    {
      "epoch": 0.03909,
      "grad_norm": 0.5889198426858406,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3909
    },
    {
      "epoch": 0.0391,
      "grad_norm": 0.5523722340939393,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3910
    },
    {
      "epoch": 0.03911,
      "grad_norm": 0.5127903685980498,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 3911
    },
    {
      "epoch": 0.03912,
      "grad_norm": 0.4766305366831587,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 3912
    },
    {
      "epoch": 0.03913,
      "grad_norm": 0.5086608596079191,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 3913
    },
    {
      "epoch": 0.03914,
      "grad_norm": 0.5215056781723886,
      "learning_rate": 0.003,
      "loss": 4.1622,
      "step": 3914
    },
    {
      "epoch": 0.03915,
      "grad_norm": 0.5669876961115498,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 3915
    },
    {
      "epoch": 0.03916,
      "grad_norm": 0.5924249172540584,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3916
    },
    {
      "epoch": 0.03917,
      "grad_norm": 0.6024923400756873,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3917
    },
    {
      "epoch": 0.03918,
      "grad_norm": 0.7101951940455182,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 3918
    },
    {
      "epoch": 0.03919,
      "grad_norm": 0.8265326178992785,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 3919
    },
    {
      "epoch": 0.0392,
      "grad_norm": 1.076222106498144,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 3920
    },
    {
      "epoch": 0.03921,
      "grad_norm": 0.9195914439109347,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3921
    },
    {
      "epoch": 0.03922,
      "grad_norm": 0.8170632240304666,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3922
    },
    {
      "epoch": 0.03923,
      "grad_norm": 0.8296100524137636,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 3923
    },
    {
      "epoch": 0.03924,
      "grad_norm": 0.8264396450686449,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3924
    },
    {
      "epoch": 0.03925,
      "grad_norm": 0.8983164449879958,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3925
    },
    {
      "epoch": 0.03926,
      "grad_norm": 0.8746883993886628,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3926
    },
    {
      "epoch": 0.03927,
      "grad_norm": 0.8997462460768232,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3927
    },
    {
      "epoch": 0.03928,
      "grad_norm": 0.8314963835673442,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 3928
    },
    {
      "epoch": 0.03929,
      "grad_norm": 0.7199296155552709,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 3929
    },
    {
      "epoch": 0.0393,
      "grad_norm": 0.6856199549997152,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3930
    },
    {
      "epoch": 0.03931,
      "grad_norm": 0.6802907106791054,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3931
    },
    {
      "epoch": 0.03932,
      "grad_norm": 0.7056782066262727,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3932
    },
    {
      "epoch": 0.03933,
      "grad_norm": 0.84224228414277,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 3933
    },
    {
      "epoch": 0.03934,
      "grad_norm": 0.9417246071568263,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3934
    },
    {
      "epoch": 0.03935,
      "grad_norm": 0.8605169810225617,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3935
    },
    {
      "epoch": 0.03936,
      "grad_norm": 0.8668689820297688,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 3936
    },
    {
      "epoch": 0.03937,
      "grad_norm": 0.7302292343482224,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3937
    },
    {
      "epoch": 0.03938,
      "grad_norm": 0.7852118199812202,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3938
    },
    {
      "epoch": 0.03939,
      "grad_norm": 0.7372234256187263,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 3939
    },
    {
      "epoch": 0.0394,
      "grad_norm": 0.7135965416181232,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3940
    },
    {
      "epoch": 0.03941,
      "grad_norm": 0.598564224938674,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3941
    },
    {
      "epoch": 0.03942,
      "grad_norm": 0.5266569159659855,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3942
    },
    {
      "epoch": 0.03943,
      "grad_norm": 0.5409156386273862,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3943
    },
    {
      "epoch": 0.03944,
      "grad_norm": 0.604265732288973,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3944
    },
    {
      "epoch": 0.03945,
      "grad_norm": 0.5831656289599461,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 3945
    },
    {
      "epoch": 0.03946,
      "grad_norm": 0.5502377902249579,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3946
    },
    {
      "epoch": 0.03947,
      "grad_norm": 0.48330198086471926,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3947
    },
    {
      "epoch": 0.03948,
      "grad_norm": 0.45335380074512316,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3948
    },
    {
      "epoch": 0.03949,
      "grad_norm": 0.4365298076816215,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 3949
    },
    {
      "epoch": 0.0395,
      "grad_norm": 0.4330234892790934,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3950
    },
    {
      "epoch": 0.03951,
      "grad_norm": 0.4549033232539336,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 3951
    },
    {
      "epoch": 0.03952,
      "grad_norm": 0.5088568380699875,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3952
    },
    {
      "epoch": 0.03953,
      "grad_norm": 0.611258865255683,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 3953
    },
    {
      "epoch": 0.03954,
      "grad_norm": 0.6882585765157053,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 3954
    },
    {
      "epoch": 0.03955,
      "grad_norm": 0.7656685556557281,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3955
    },
    {
      "epoch": 0.03956,
      "grad_norm": 0.8113542097479195,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3956
    },
    {
      "epoch": 0.03957,
      "grad_norm": 0.8632342447207374,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3957
    },
    {
      "epoch": 0.03958,
      "grad_norm": 1.0300805032856515,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 3958
    },
    {
      "epoch": 0.03959,
      "grad_norm": 1.4201099525549408,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3959
    },
    {
      "epoch": 0.0396,
      "grad_norm": 0.8108000226279894,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3960
    },
    {
      "epoch": 0.03961,
      "grad_norm": 0.8230830778410857,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 3961
    },
    {
      "epoch": 0.03962,
      "grad_norm": 0.8784449509664598,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3962
    },
    {
      "epoch": 0.03963,
      "grad_norm": 0.8465039181944972,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3963
    },
    {
      "epoch": 0.03964,
      "grad_norm": 0.8689576663713054,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3964
    },
    {
      "epoch": 0.03965,
      "grad_norm": 0.8789965519139235,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3965
    },
    {
      "epoch": 0.03966,
      "grad_norm": 0.7504804588349563,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 3966
    },
    {
      "epoch": 0.03967,
      "grad_norm": 0.7521340457571327,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3967
    },
    {
      "epoch": 0.03968,
      "grad_norm": 0.6566639153668963,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 3968
    },
    {
      "epoch": 0.03969,
      "grad_norm": 0.6193517682452371,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3969
    },
    {
      "epoch": 0.0397,
      "grad_norm": 0.618307384827639,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 3970
    },
    {
      "epoch": 0.03971,
      "grad_norm": 0.5807398202018539,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 3971
    },
    {
      "epoch": 0.03972,
      "grad_norm": 0.5003114971259446,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3972
    },
    {
      "epoch": 0.03973,
      "grad_norm": 0.5672839085406842,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3973
    },
    {
      "epoch": 0.03974,
      "grad_norm": 0.6148540153542106,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 3974
    },
    {
      "epoch": 0.03975,
      "grad_norm": 0.6299273778345961,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 3975
    },
    {
      "epoch": 0.03976,
      "grad_norm": 0.7124442131113389,
      "learning_rate": 0.003,
      "loss": 4.1845,
      "step": 3976
    },
    {
      "epoch": 0.03977,
      "grad_norm": 0.8025542381027738,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3977
    },
    {
      "epoch": 0.03978,
      "grad_norm": 0.8588749028025513,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3978
    },
    {
      "epoch": 0.03979,
      "grad_norm": 0.8573131443233668,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3979
    },
    {
      "epoch": 0.0398,
      "grad_norm": 0.99280645435225,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3980
    },
    {
      "epoch": 0.03981,
      "grad_norm": 1.1460073770386754,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3981
    },
    {
      "epoch": 0.03982,
      "grad_norm": 0.9873321030261758,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3982
    },
    {
      "epoch": 0.03983,
      "grad_norm": 0.8416549998631022,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3983
    },
    {
      "epoch": 0.03984,
      "grad_norm": 0.7402406551683491,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3984
    },
    {
      "epoch": 0.03985,
      "grad_norm": 0.7030279267262828,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3985
    },
    {
      "epoch": 0.03986,
      "grad_norm": 0.6894694286447128,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3986
    },
    {
      "epoch": 0.03987,
      "grad_norm": 0.7314973515669282,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3987
    },
    {
      "epoch": 0.03988,
      "grad_norm": 0.8590107763209545,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3988
    },
    {
      "epoch": 0.03989,
      "grad_norm": 1.074747879568787,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 3989
    },
    {
      "epoch": 0.0399,
      "grad_norm": 0.8434989382900636,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3990
    },
    {
      "epoch": 0.03991,
      "grad_norm": 0.7206979857545632,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 3991
    },
    {
      "epoch": 0.03992,
      "grad_norm": 0.8037268570780255,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3992
    },
    {
      "epoch": 0.03993,
      "grad_norm": 0.7620905271249203,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3993
    },
    {
      "epoch": 0.03994,
      "grad_norm": 0.738025311420094,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3994
    },
    {
      "epoch": 0.03995,
      "grad_norm": 0.6172466009583261,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 3995
    },
    {
      "epoch": 0.03996,
      "grad_norm": 0.452402581761887,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3996
    },
    {
      "epoch": 0.03997,
      "grad_norm": 0.48368404250480046,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3997
    },
    {
      "epoch": 0.03998,
      "grad_norm": 0.47207135976156783,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3998
    },
    {
      "epoch": 0.03999,
      "grad_norm": 0.502083976437197,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3999
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4504866573185321,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 4000
    },
    {
      "epoch": 0.04001,
      "grad_norm": 0.37734509945831174,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4001
    },
    {
      "epoch": 0.04002,
      "grad_norm": 0.4339763243340729,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4002
    },
    {
      "epoch": 0.04003,
      "grad_norm": 0.5726849542049914,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 4003
    },
    {
      "epoch": 0.04004,
      "grad_norm": 0.71841237008688,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 4004
    },
    {
      "epoch": 0.04005,
      "grad_norm": 0.9930905212295545,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 4005
    },
    {
      "epoch": 0.04006,
      "grad_norm": 0.9991984295961828,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4006
    },
    {
      "epoch": 0.04007,
      "grad_norm": 0.7547073099000725,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 4007
    },
    {
      "epoch": 0.04008,
      "grad_norm": 0.8454739420896483,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4008
    },
    {
      "epoch": 0.04009,
      "grad_norm": 0.9067851236139133,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4009
    },
    {
      "epoch": 0.0401,
      "grad_norm": 0.8062732196382164,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4010
    },
    {
      "epoch": 0.04011,
      "grad_norm": 0.6938007348622179,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4011
    },
    {
      "epoch": 0.04012,
      "grad_norm": 0.7757979776927771,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 4012
    },
    {
      "epoch": 0.04013,
      "grad_norm": 0.7089158793355179,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4013
    },
    {
      "epoch": 0.04014,
      "grad_norm": 0.6111952836402943,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 4014
    },
    {
      "epoch": 0.04015,
      "grad_norm": 0.5804891668774357,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4015
    },
    {
      "epoch": 0.04016,
      "grad_norm": 0.5604585117010126,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 4016
    },
    {
      "epoch": 0.04017,
      "grad_norm": 0.5830753494163682,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 4017
    },
    {
      "epoch": 0.04018,
      "grad_norm": 0.52884054520163,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4018
    },
    {
      "epoch": 0.04019,
      "grad_norm": 0.5435284400458248,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 4019
    },
    {
      "epoch": 0.0402,
      "grad_norm": 0.46054232620433116,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4020
    },
    {
      "epoch": 0.04021,
      "grad_norm": 0.5140849815317029,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 4021
    },
    {
      "epoch": 0.04022,
      "grad_norm": 0.6311629918528258,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4022
    },
    {
      "epoch": 0.04023,
      "grad_norm": 0.7137869361747575,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 4023
    },
    {
      "epoch": 0.04024,
      "grad_norm": 0.738420470053228,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 4024
    },
    {
      "epoch": 0.04025,
      "grad_norm": 0.6521094783928104,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4025
    },
    {
      "epoch": 0.04026,
      "grad_norm": 0.5809109807556073,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4026
    },
    {
      "epoch": 0.04027,
      "grad_norm": 0.6439840472927373,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 4027
    },
    {
      "epoch": 0.04028,
      "grad_norm": 0.6879545278017376,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4028
    },
    {
      "epoch": 0.04029,
      "grad_norm": 0.7939138849392092,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4029
    },
    {
      "epoch": 0.0403,
      "grad_norm": 0.7119398099333982,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4030
    },
    {
      "epoch": 0.04031,
      "grad_norm": 0.6966756671526879,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4031
    },
    {
      "epoch": 0.04032,
      "grad_norm": 0.6596032753240082,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 4032
    },
    {
      "epoch": 0.04033,
      "grad_norm": 0.6274155855378644,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4033
    },
    {
      "epoch": 0.04034,
      "grad_norm": 0.6095955743702651,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 4034
    },
    {
      "epoch": 0.04035,
      "grad_norm": 0.555350198090057,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4035
    },
    {
      "epoch": 0.04036,
      "grad_norm": 0.5197769667331041,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 4036
    },
    {
      "epoch": 0.04037,
      "grad_norm": 0.5661546926366932,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 4037
    },
    {
      "epoch": 0.04038,
      "grad_norm": 0.7699720543941366,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4038
    },
    {
      "epoch": 0.04039,
      "grad_norm": 1.210094587806928,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4039
    },
    {
      "epoch": 0.0404,
      "grad_norm": 0.868075390136909,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 4040
    },
    {
      "epoch": 0.04041,
      "grad_norm": 0.7416181269518949,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 4041
    },
    {
      "epoch": 0.04042,
      "grad_norm": 0.6648042947771692,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 4042
    },
    {
      "epoch": 0.04043,
      "grad_norm": 0.6609860437382262,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4043
    },
    {
      "epoch": 0.04044,
      "grad_norm": 0.7121985413545872,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4044
    },
    {
      "epoch": 0.04045,
      "grad_norm": 0.7259672418788464,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 4045
    },
    {
      "epoch": 0.04046,
      "grad_norm": 0.6676245217285697,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4046
    },
    {
      "epoch": 0.04047,
      "grad_norm": 0.6576773790552027,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 4047
    },
    {
      "epoch": 0.04048,
      "grad_norm": 0.7504199241468649,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 4048
    },
    {
      "epoch": 0.04049,
      "grad_norm": 0.6934067855311765,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 4049
    },
    {
      "epoch": 0.0405,
      "grad_norm": 0.6792487502012533,
      "learning_rate": 0.003,
      "loss": 4.1721,
      "step": 4050
    },
    {
      "epoch": 0.04051,
      "grad_norm": 0.6504046520169164,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4051
    },
    {
      "epoch": 0.04052,
      "grad_norm": 0.5049122912594802,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 4052
    },
    {
      "epoch": 0.04053,
      "grad_norm": 0.5305812046390571,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 4053
    },
    {
      "epoch": 0.04054,
      "grad_norm": 0.595657788891312,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4054
    },
    {
      "epoch": 0.04055,
      "grad_norm": 0.8174417297511951,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 4055
    },
    {
      "epoch": 0.04056,
      "grad_norm": 1.2164744618417813,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4056
    },
    {
      "epoch": 0.04057,
      "grad_norm": 0.7248345125453411,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4057
    },
    {
      "epoch": 0.04058,
      "grad_norm": 0.6128740383918565,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 4058
    },
    {
      "epoch": 0.04059,
      "grad_norm": 0.7656551573931919,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 4059
    },
    {
      "epoch": 0.0406,
      "grad_norm": 0.7644654016064842,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 4060
    },
    {
      "epoch": 0.04061,
      "grad_norm": 0.7754515611223848,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4061
    },
    {
      "epoch": 0.04062,
      "grad_norm": 0.8608274358738601,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4062
    },
    {
      "epoch": 0.04063,
      "grad_norm": 0.9067607362608292,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4063
    },
    {
      "epoch": 0.04064,
      "grad_norm": 1.0529705240815344,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 4064
    },
    {
      "epoch": 0.04065,
      "grad_norm": 1.0102927624010265,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 4065
    },
    {
      "epoch": 0.04066,
      "grad_norm": 0.9478175395069827,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 4066
    },
    {
      "epoch": 0.04067,
      "grad_norm": 0.8098599255455219,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4067
    },
    {
      "epoch": 0.04068,
      "grad_norm": 0.724209064256899,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 4068
    },
    {
      "epoch": 0.04069,
      "grad_norm": 0.6809092940169613,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 4069
    },
    {
      "epoch": 0.0407,
      "grad_norm": 0.8132916778007174,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 4070
    },
    {
      "epoch": 0.04071,
      "grad_norm": 1.041265005432495,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4071
    },
    {
      "epoch": 0.04072,
      "grad_norm": 1.0457271292201864,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4072
    },
    {
      "epoch": 0.04073,
      "grad_norm": 1.0651363452233251,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 4073
    },
    {
      "epoch": 0.04074,
      "grad_norm": 0.7177727168488233,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 4074
    },
    {
      "epoch": 0.04075,
      "grad_norm": 0.6005805584962535,
      "learning_rate": 0.003,
      "loss": 4.1746,
      "step": 4075
    },
    {
      "epoch": 0.04076,
      "grad_norm": 0.5914694996561694,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 4076
    },
    {
      "epoch": 0.04077,
      "grad_norm": 0.6911502018139879,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 4077
    },
    {
      "epoch": 0.04078,
      "grad_norm": 0.7912939945570531,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 4078
    },
    {
      "epoch": 0.04079,
      "grad_norm": 0.8570836696530718,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 4079
    },
    {
      "epoch": 0.0408,
      "grad_norm": 0.7585471415380697,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4080
    },
    {
      "epoch": 0.04081,
      "grad_norm": 0.7520683150191194,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4081
    },
    {
      "epoch": 0.04082,
      "grad_norm": 0.6537470491867149,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 4082
    },
    {
      "epoch": 0.04083,
      "grad_norm": 0.6723574711008242,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 4083
    },
    {
      "epoch": 0.04084,
      "grad_norm": 0.6506573739916557,
      "learning_rate": 0.003,
      "loss": 4.1816,
      "step": 4084
    },
    {
      "epoch": 0.04085,
      "grad_norm": 0.7215838325103069,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 4085
    },
    {
      "epoch": 0.04086,
      "grad_norm": 0.7304080342645308,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 4086
    },
    {
      "epoch": 0.04087,
      "grad_norm": 0.736672307736761,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 4087
    },
    {
      "epoch": 0.04088,
      "grad_norm": 0.7327696397562437,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4088
    },
    {
      "epoch": 0.04089,
      "grad_norm": 0.737838253705245,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4089
    },
    {
      "epoch": 0.0409,
      "grad_norm": 0.6618217098522069,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4090
    },
    {
      "epoch": 0.04091,
      "grad_norm": 0.6534380155154119,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 4091
    },
    {
      "epoch": 0.04092,
      "grad_norm": 0.6174715229139612,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 4092
    },
    {
      "epoch": 0.04093,
      "grad_norm": 0.6189167119848286,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 4093
    },
    {
      "epoch": 0.04094,
      "grad_norm": 0.6641703981374822,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4094
    },
    {
      "epoch": 0.04095,
      "grad_norm": 0.6395123879631477,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 4095
    },
    {
      "epoch": 0.04096,
      "grad_norm": 0.6349994704357759,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4096
    },
    {
      "epoch": 0.04097,
      "grad_norm": 0.6463455807108017,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 4097
    },
    {
      "epoch": 0.04098,
      "grad_norm": 0.6932323673060099,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4098
    },
    {
      "epoch": 0.04099,
      "grad_norm": 0.8807834378700455,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 4099
    },
    {
      "epoch": 0.041,
      "grad_norm": 1.010325661821192,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 4100
    },
    {
      "epoch": 0.04101,
      "grad_norm": 0.8888939381607318,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4101
    },
    {
      "epoch": 0.04102,
      "grad_norm": 0.7433180771066499,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4102
    },
    {
      "epoch": 0.04103,
      "grad_norm": 0.6923500471319527,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4103
    },
    {
      "epoch": 0.04104,
      "grad_norm": 0.6523846830610261,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4104
    },
    {
      "epoch": 0.04105,
      "grad_norm": 0.6124945853172781,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 4105
    },
    {
      "epoch": 0.04106,
      "grad_norm": 0.6180742314657086,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 4106
    },
    {
      "epoch": 0.04107,
      "grad_norm": 0.7675574080126237,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 4107
    },
    {
      "epoch": 0.04108,
      "grad_norm": 0.7972279438911031,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 4108
    },
    {
      "epoch": 0.04109,
      "grad_norm": 0.7338110856400634,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4109
    },
    {
      "epoch": 0.0411,
      "grad_norm": 0.5620305965378466,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4110
    },
    {
      "epoch": 0.04111,
      "grad_norm": 0.5981690023054039,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 4111
    },
    {
      "epoch": 0.04112,
      "grad_norm": 0.5672857968992558,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4112
    },
    {
      "epoch": 0.04113,
      "grad_norm": 0.6312145325378008,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 4113
    },
    {
      "epoch": 0.04114,
      "grad_norm": 0.6035265736834228,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4114
    },
    {
      "epoch": 0.04115,
      "grad_norm": 0.5962258124728763,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 4115
    },
    {
      "epoch": 0.04116,
      "grad_norm": 0.5290947291389287,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4116
    },
    {
      "epoch": 0.04117,
      "grad_norm": 0.475360712634243,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 4117
    },
    {
      "epoch": 0.04118,
      "grad_norm": 0.48139980415634037,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4118
    },
    {
      "epoch": 0.04119,
      "grad_norm": 0.4768773023811471,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4119
    },
    {
      "epoch": 0.0412,
      "grad_norm": 0.4715918252478233,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 4120
    },
    {
      "epoch": 0.04121,
      "grad_norm": 0.4631389077981398,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 4121
    },
    {
      "epoch": 0.04122,
      "grad_norm": 0.5695292166099397,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4122
    },
    {
      "epoch": 0.04123,
      "grad_norm": 0.6344096447219111,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4123
    },
    {
      "epoch": 0.04124,
      "grad_norm": 0.7628425133873309,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4124
    },
    {
      "epoch": 0.04125,
      "grad_norm": 1.0834853187583475,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 4125
    },
    {
      "epoch": 0.04126,
      "grad_norm": 1.108357219795881,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4126
    },
    {
      "epoch": 0.04127,
      "grad_norm": 0.7334164794838763,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4127
    },
    {
      "epoch": 0.04128,
      "grad_norm": 0.7315856737953997,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 4128
    },
    {
      "epoch": 0.04129,
      "grad_norm": 0.67750855822388,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4129
    },
    {
      "epoch": 0.0413,
      "grad_norm": 0.6477790148509263,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4130
    },
    {
      "epoch": 0.04131,
      "grad_norm": 0.6976553932039135,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4131
    },
    {
      "epoch": 0.04132,
      "grad_norm": 0.7575720136801286,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4132
    },
    {
      "epoch": 0.04133,
      "grad_norm": 0.7834309464101223,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4133
    },
    {
      "epoch": 0.04134,
      "grad_norm": 0.8769385664533508,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4134
    },
    {
      "epoch": 0.04135,
      "grad_norm": 0.9288719850360465,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4135
    },
    {
      "epoch": 0.04136,
      "grad_norm": 0.8613590746245083,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 4136
    },
    {
      "epoch": 0.04137,
      "grad_norm": 0.7914311956529176,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 4137
    },
    {
      "epoch": 0.04138,
      "grad_norm": 0.8323094727140243,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4138
    },
    {
      "epoch": 0.04139,
      "grad_norm": 0.7498292112429189,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 4139
    },
    {
      "epoch": 0.0414,
      "grad_norm": 0.7470547424534141,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4140
    },
    {
      "epoch": 0.04141,
      "grad_norm": 0.7086395261007737,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4141
    },
    {
      "epoch": 0.04142,
      "grad_norm": 0.7475027864042157,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4142
    },
    {
      "epoch": 0.04143,
      "grad_norm": 0.7096941472876405,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4143
    },
    {
      "epoch": 0.04144,
      "grad_norm": 0.6807921123663281,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 4144
    },
    {
      "epoch": 0.04145,
      "grad_norm": 0.6209916627342668,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4145
    },
    {
      "epoch": 0.04146,
      "grad_norm": 0.6603935131450228,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4146
    },
    {
      "epoch": 0.04147,
      "grad_norm": 0.766665555184346,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 4147
    },
    {
      "epoch": 0.04148,
      "grad_norm": 0.8232977544301274,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4148
    },
    {
      "epoch": 0.04149,
      "grad_norm": 0.9384903955246353,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4149
    },
    {
      "epoch": 0.0415,
      "grad_norm": 0.8526053994314199,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4150
    },
    {
      "epoch": 0.04151,
      "grad_norm": 0.703764014227338,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 4151
    },
    {
      "epoch": 0.04152,
      "grad_norm": 0.5883825477373993,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4152
    },
    {
      "epoch": 0.04153,
      "grad_norm": 0.5178464710197032,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 4153
    },
    {
      "epoch": 0.04154,
      "grad_norm": 0.4801954114695538,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4154
    },
    {
      "epoch": 0.04155,
      "grad_norm": 0.4883851788041572,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4155
    },
    {
      "epoch": 0.04156,
      "grad_norm": 0.5305154970213135,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4156
    },
    {
      "epoch": 0.04157,
      "grad_norm": 0.5969580348221766,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4157
    },
    {
      "epoch": 0.04158,
      "grad_norm": 0.604419085621543,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 4158
    },
    {
      "epoch": 0.04159,
      "grad_norm": 0.5952666616236362,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4159
    },
    {
      "epoch": 0.0416,
      "grad_norm": 0.6015748305688609,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4160
    },
    {
      "epoch": 0.04161,
      "grad_norm": 0.6509334448144796,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 4161
    },
    {
      "epoch": 0.04162,
      "grad_norm": 0.8275491529133575,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 4162
    },
    {
      "epoch": 0.04163,
      "grad_norm": 0.9190060913193079,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 4163
    },
    {
      "epoch": 0.04164,
      "grad_norm": 1.0804296755917129,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 4164
    },
    {
      "epoch": 0.04165,
      "grad_norm": 0.8918375998135468,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4165
    },
    {
      "epoch": 0.04166,
      "grad_norm": 0.6444987754007603,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4166
    },
    {
      "epoch": 0.04167,
      "grad_norm": 0.5987188953434583,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 4167
    },
    {
      "epoch": 0.04168,
      "grad_norm": 0.6931415065499815,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 4168
    },
    {
      "epoch": 0.04169,
      "grad_norm": 0.7939284298013517,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 4169
    },
    {
      "epoch": 0.0417,
      "grad_norm": 0.8929923791952147,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4170
    },
    {
      "epoch": 0.04171,
      "grad_norm": 0.8713733021965931,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4171
    },
    {
      "epoch": 0.04172,
      "grad_norm": 0.741654776741016,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 4172
    },
    {
      "epoch": 0.04173,
      "grad_norm": 0.6869389765405233,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 4173
    },
    {
      "epoch": 0.04174,
      "grad_norm": 0.7104276340717113,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4174
    },
    {
      "epoch": 0.04175,
      "grad_norm": 0.7652108394533633,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 4175
    },
    {
      "epoch": 0.04176,
      "grad_norm": 0.6595942295349072,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4176
    },
    {
      "epoch": 0.04177,
      "grad_norm": 0.6050614217333806,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4177
    },
    {
      "epoch": 0.04178,
      "grad_norm": 0.6649317208659106,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4178
    },
    {
      "epoch": 0.04179,
      "grad_norm": 0.7497130033497713,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4179
    },
    {
      "epoch": 0.0418,
      "grad_norm": 0.8135721504617128,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4180
    },
    {
      "epoch": 0.04181,
      "grad_norm": 0.8382943348914027,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4181
    },
    {
      "epoch": 0.04182,
      "grad_norm": 0.7581608595797077,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4182
    },
    {
      "epoch": 0.04183,
      "grad_norm": 0.7367386623848073,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4183
    },
    {
      "epoch": 0.04184,
      "grad_norm": 0.8223299343686243,
      "learning_rate": 0.003,
      "loss": 4.2008,
      "step": 4184
    },
    {
      "epoch": 0.04185,
      "grad_norm": 0.8253315995103551,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 4185
    },
    {
      "epoch": 0.04186,
      "grad_norm": 1.021967899357659,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 4186
    },
    {
      "epoch": 0.04187,
      "grad_norm": 1.1194144390778462,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 4187
    },
    {
      "epoch": 0.04188,
      "grad_norm": 0.8663014575987595,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4188
    },
    {
      "epoch": 0.04189,
      "grad_norm": 0.7730217042128824,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4189
    },
    {
      "epoch": 0.0419,
      "grad_norm": 0.718425867412397,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4190
    },
    {
      "epoch": 0.04191,
      "grad_norm": 0.634536414286615,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 4191
    },
    {
      "epoch": 0.04192,
      "grad_norm": 0.6074939161177713,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4192
    },
    {
      "epoch": 0.04193,
      "grad_norm": 0.5595906619172795,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4193
    },
    {
      "epoch": 0.04194,
      "grad_norm": 0.591321468676565,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4194
    },
    {
      "epoch": 0.04195,
      "grad_norm": 0.6331925592194563,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 4195
    },
    {
      "epoch": 0.04196,
      "grad_norm": 0.66777685730357,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 4196
    },
    {
      "epoch": 0.04197,
      "grad_norm": 0.6581215996860487,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 4197
    },
    {
      "epoch": 0.04198,
      "grad_norm": 0.5775968736275588,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4198
    },
    {
      "epoch": 0.04199,
      "grad_norm": 0.5115173192319249,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4199
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.4907326781591073,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4200
    },
    {
      "epoch": 0.04201,
      "grad_norm": 0.4340051119862648,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 4201
    },
    {
      "epoch": 0.04202,
      "grad_norm": 0.4270978633731861,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 4202
    },
    {
      "epoch": 0.04203,
      "grad_norm": 0.5415535220350182,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 4203
    },
    {
      "epoch": 0.04204,
      "grad_norm": 0.6653136591047163,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 4204
    },
    {
      "epoch": 0.04205,
      "grad_norm": 0.9015195834615578,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4205
    },
    {
      "epoch": 0.04206,
      "grad_norm": 0.9511163331115945,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 4206
    },
    {
      "epoch": 0.04207,
      "grad_norm": 0.8183175873095769,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4207
    },
    {
      "epoch": 0.04208,
      "grad_norm": 0.7983775516444301,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4208
    },
    {
      "epoch": 0.04209,
      "grad_norm": 0.9111818683711085,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 4209
    },
    {
      "epoch": 0.0421,
      "grad_norm": 0.8760572861926779,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 4210
    },
    {
      "epoch": 0.04211,
      "grad_norm": 0.8385383847029609,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4211
    },
    {
      "epoch": 0.04212,
      "grad_norm": 0.8523572548315141,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4212
    },
    {
      "epoch": 0.04213,
      "grad_norm": 0.87521517308783,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4213
    },
    {
      "epoch": 0.04214,
      "grad_norm": 0.8765783386936129,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 4214
    },
    {
      "epoch": 0.04215,
      "grad_norm": 0.797769330073817,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4215
    },
    {
      "epoch": 0.04216,
      "grad_norm": 0.7247822776339875,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4216
    },
    {
      "epoch": 0.04217,
      "grad_norm": 0.7261327052126046,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4217
    },
    {
      "epoch": 0.04218,
      "grad_norm": 0.7319178674334204,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 4218
    },
    {
      "epoch": 0.04219,
      "grad_norm": 0.7694832016892729,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 4219
    },
    {
      "epoch": 0.0422,
      "grad_norm": 0.757141250932244,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 4220
    },
    {
      "epoch": 0.04221,
      "grad_norm": 0.7290165220800948,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4221
    },
    {
      "epoch": 0.04222,
      "grad_norm": 0.6346702624845955,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 4222
    },
    {
      "epoch": 0.04223,
      "grad_norm": 0.583770301965496,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 4223
    },
    {
      "epoch": 0.04224,
      "grad_norm": 0.5350148270680584,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 4224
    },
    {
      "epoch": 0.04225,
      "grad_norm": 0.5316639979135299,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 4225
    },
    {
      "epoch": 0.04226,
      "grad_norm": 0.5763902675717133,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4226
    },
    {
      "epoch": 0.04227,
      "grad_norm": 0.6468360436042053,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4227
    },
    {
      "epoch": 0.04228,
      "grad_norm": 0.7289273143868911,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4228
    },
    {
      "epoch": 0.04229,
      "grad_norm": 0.759894491275707,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 4229
    },
    {
      "epoch": 0.0423,
      "grad_norm": 0.7516690067272278,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4230
    },
    {
      "epoch": 0.04231,
      "grad_norm": 0.6070127843457722,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4231
    },
    {
      "epoch": 0.04232,
      "grad_norm": 0.6987145523188838,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4232
    },
    {
      "epoch": 0.04233,
      "grad_norm": 0.9287852741458625,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 4233
    },
    {
      "epoch": 0.04234,
      "grad_norm": 0.9918383708905326,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4234
    },
    {
      "epoch": 0.04235,
      "grad_norm": 0.9367701682265075,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4235
    },
    {
      "epoch": 0.04236,
      "grad_norm": 0.8703490177121154,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4236
    },
    {
      "epoch": 0.04237,
      "grad_norm": 0.8156234434712465,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 4237
    },
    {
      "epoch": 0.04238,
      "grad_norm": 0.6835933384976534,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 4238
    },
    {
      "epoch": 0.04239,
      "grad_norm": 0.6391989756025438,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4239
    },
    {
      "epoch": 0.0424,
      "grad_norm": 0.6447631067586589,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4240
    },
    {
      "epoch": 0.04241,
      "grad_norm": 0.6715155999130873,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4241
    },
    {
      "epoch": 0.04242,
      "grad_norm": 0.6273748467552226,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4242
    },
    {
      "epoch": 0.04243,
      "grad_norm": 0.6467254126434867,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4243
    },
    {
      "epoch": 0.04244,
      "grad_norm": 0.6955126933633631,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4244
    },
    {
      "epoch": 0.04245,
      "grad_norm": 0.681078617510812,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 4245
    },
    {
      "epoch": 0.04246,
      "grad_norm": 0.69651922194907,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4246
    },
    {
      "epoch": 0.04247,
      "grad_norm": 0.7659042143848519,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4247
    },
    {
      "epoch": 0.04248,
      "grad_norm": 0.8045443211607587,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4248
    },
    {
      "epoch": 0.04249,
      "grad_norm": 0.7807725187988103,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 4249
    },
    {
      "epoch": 0.0425,
      "grad_norm": 0.797629821833922,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 4250
    },
    {
      "epoch": 0.04251,
      "grad_norm": 0.8773737541427695,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 4251
    },
    {
      "epoch": 0.04252,
      "grad_norm": 0.8764481339534494,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 4252
    },
    {
      "epoch": 0.04253,
      "grad_norm": 0.771295115347259,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4253
    },
    {
      "epoch": 0.04254,
      "grad_norm": 0.6812529052304258,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4254
    },
    {
      "epoch": 0.04255,
      "grad_norm": 0.6528883858102312,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4255
    },
    {
      "epoch": 0.04256,
      "grad_norm": 0.6360597468005796,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4256
    },
    {
      "epoch": 0.04257,
      "grad_norm": 0.6045194570542003,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4257
    },
    {
      "epoch": 0.04258,
      "grad_norm": 0.6237209269525407,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 4258
    },
    {
      "epoch": 0.04259,
      "grad_norm": 0.6534065774304211,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 4259
    },
    {
      "epoch": 0.0426,
      "grad_norm": 0.7228160196304746,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4260
    },
    {
      "epoch": 0.04261,
      "grad_norm": 0.7225504963671482,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4261
    },
    {
      "epoch": 0.04262,
      "grad_norm": 0.6030677337778022,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 4262
    },
    {
      "epoch": 0.04263,
      "grad_norm": 0.5576016271704423,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4263
    },
    {
      "epoch": 0.04264,
      "grad_norm": 0.4622794122851678,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4264
    },
    {
      "epoch": 0.04265,
      "grad_norm": 0.4498062625739869,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 4265
    },
    {
      "epoch": 0.04266,
      "grad_norm": 0.4663874509947672,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4266
    },
    {
      "epoch": 0.04267,
      "grad_norm": 0.5819381640130065,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 4267
    },
    {
      "epoch": 0.04268,
      "grad_norm": 0.7240543464190408,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4268
    },
    {
      "epoch": 0.04269,
      "grad_norm": 0.7869636361035491,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4269
    },
    {
      "epoch": 0.0427,
      "grad_norm": 0.7316564623442252,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 4270
    },
    {
      "epoch": 0.04271,
      "grad_norm": 0.6962340655904857,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 4271
    },
    {
      "epoch": 0.04272,
      "grad_norm": 0.7468024354946826,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4272
    },
    {
      "epoch": 0.04273,
      "grad_norm": 0.8315685932461937,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 4273
    },
    {
      "epoch": 0.04274,
      "grad_norm": 0.7971355124213763,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 4274
    },
    {
      "epoch": 0.04275,
      "grad_norm": 0.7711074822549306,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4275
    },
    {
      "epoch": 0.04276,
      "grad_norm": 0.8849517437250696,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4276
    },
    {
      "epoch": 0.04277,
      "grad_norm": 1.058466724741216,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 4277
    },
    {
      "epoch": 0.04278,
      "grad_norm": 0.9806877039749324,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 4278
    },
    {
      "epoch": 0.04279,
      "grad_norm": 1.1506116450442327,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4279
    },
    {
      "epoch": 0.0428,
      "grad_norm": 1.0111616301127737,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 4280
    },
    {
      "epoch": 0.04281,
      "grad_norm": 1.1240409097888207,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 4281
    },
    {
      "epoch": 0.04282,
      "grad_norm": 0.9707964829866726,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 4282
    },
    {
      "epoch": 0.04283,
      "grad_norm": 0.8952580683408602,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 4283
    },
    {
      "epoch": 0.04284,
      "grad_norm": 0.8037590999213429,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4284
    },
    {
      "epoch": 0.04285,
      "grad_norm": 0.7180846206853434,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 4285
    },
    {
      "epoch": 0.04286,
      "grad_norm": 0.6600031955008437,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 4286
    },
    {
      "epoch": 0.04287,
      "grad_norm": 0.624643362166525,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 4287
    },
    {
      "epoch": 0.04288,
      "grad_norm": 0.7314395975343404,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4288
    },
    {
      "epoch": 0.04289,
      "grad_norm": 0.7552436651554467,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 4289
    },
    {
      "epoch": 0.0429,
      "grad_norm": 0.7444307165904086,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4290
    },
    {
      "epoch": 0.04291,
      "grad_norm": 0.6792525045452005,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4291
    },
    {
      "epoch": 0.04292,
      "grad_norm": 0.6366634706551982,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 4292
    },
    {
      "epoch": 0.04293,
      "grad_norm": 0.5177321350541024,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 4293
    },
    {
      "epoch": 0.04294,
      "grad_norm": 0.5199647077622895,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 4294
    },
    {
      "epoch": 0.04295,
      "grad_norm": 0.533363434433449,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 4295
    },
    {
      "epoch": 0.04296,
      "grad_norm": 0.4910620913342609,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 4296
    },
    {
      "epoch": 0.04297,
      "grad_norm": 0.4528594955305011,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 4297
    },
    {
      "epoch": 0.04298,
      "grad_norm": 0.5508851288276945,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4298
    },
    {
      "epoch": 0.04299,
      "grad_norm": 0.6926359286804635,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4299
    },
    {
      "epoch": 0.043,
      "grad_norm": 0.776299147583224,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4300
    },
    {
      "epoch": 0.04301,
      "grad_norm": 0.7543517253530487,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4301
    },
    {
      "epoch": 0.04302,
      "grad_norm": 0.7190955364079332,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4302
    },
    {
      "epoch": 0.04303,
      "grad_norm": 0.7273212030383343,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4303
    },
    {
      "epoch": 0.04304,
      "grad_norm": 0.8042177427504058,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4304
    },
    {
      "epoch": 0.04305,
      "grad_norm": 0.889845680686584,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4305
    },
    {
      "epoch": 0.04306,
      "grad_norm": 0.8019734206580067,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 4306
    },
    {
      "epoch": 0.04307,
      "grad_norm": 0.7599014266097025,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4307
    },
    {
      "epoch": 0.04308,
      "grad_norm": 0.6928021109826253,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4308
    },
    {
      "epoch": 0.04309,
      "grad_norm": 0.6623360644735573,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4309
    },
    {
      "epoch": 0.0431,
      "grad_norm": 0.643977742791814,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4310
    },
    {
      "epoch": 0.04311,
      "grad_norm": 0.6752224622600257,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4311
    },
    {
      "epoch": 0.04312,
      "grad_norm": 0.685683319654323,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 4312
    },
    {
      "epoch": 0.04313,
      "grad_norm": 0.6828332575779398,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 4313
    },
    {
      "epoch": 0.04314,
      "grad_norm": 0.7207034799508291,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4314
    },
    {
      "epoch": 0.04315,
      "grad_norm": 0.6920628089309946,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 4315
    },
    {
      "epoch": 0.04316,
      "grad_norm": 0.6697894292684422,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4316
    },
    {
      "epoch": 0.04317,
      "grad_norm": 0.6529885706954907,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4317
    },
    {
      "epoch": 0.04318,
      "grad_norm": 0.6780113889358966,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 4318
    },
    {
      "epoch": 0.04319,
      "grad_norm": 0.6696493184651962,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 4319
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.6379331339678891,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4320
    },
    {
      "epoch": 0.04321,
      "grad_norm": 0.713977041740171,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4321
    },
    {
      "epoch": 0.04322,
      "grad_norm": 0.9061136199776986,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 4322
    },
    {
      "epoch": 0.04323,
      "grad_norm": 1.0087052417142888,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4323
    },
    {
      "epoch": 0.04324,
      "grad_norm": 0.9066972139100388,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4324
    },
    {
      "epoch": 0.04325,
      "grad_norm": 0.829210403933647,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 4325
    },
    {
      "epoch": 0.04326,
      "grad_norm": 0.766217873896832,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 4326
    },
    {
      "epoch": 0.04327,
      "grad_norm": 0.66336507203185,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4327
    },
    {
      "epoch": 0.04328,
      "grad_norm": 0.6687205522581232,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4328
    },
    {
      "epoch": 0.04329,
      "grad_norm": 0.6982687394228022,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 4329
    },
    {
      "epoch": 0.0433,
      "grad_norm": 0.8226633234518121,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 4330
    },
    {
      "epoch": 0.04331,
      "grad_norm": 0.7328602014834829,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4331
    },
    {
      "epoch": 0.04332,
      "grad_norm": 0.7041041791657667,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 4332
    },
    {
      "epoch": 0.04333,
      "grad_norm": 0.7435632839058932,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4333
    },
    {
      "epoch": 0.04334,
      "grad_norm": 0.8773446956801049,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4334
    },
    {
      "epoch": 0.04335,
      "grad_norm": 0.8446265063365133,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4335
    },
    {
      "epoch": 0.04336,
      "grad_norm": 0.7068555070364958,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 4336
    },
    {
      "epoch": 0.04337,
      "grad_norm": 0.7529497673494582,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4337
    },
    {
      "epoch": 0.04338,
      "grad_norm": 0.8822495743753999,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 4338
    },
    {
      "epoch": 0.04339,
      "grad_norm": 0.8362014809442191,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4339
    },
    {
      "epoch": 0.0434,
      "grad_norm": 0.8580345532471818,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 4340
    },
    {
      "epoch": 0.04341,
      "grad_norm": 0.8559919642238094,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4341
    },
    {
      "epoch": 0.04342,
      "grad_norm": 0.9879150305605277,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4342
    },
    {
      "epoch": 0.04343,
      "grad_norm": 1.0224948991525673,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 4343
    },
    {
      "epoch": 0.04344,
      "grad_norm": 0.8797661268558509,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4344
    },
    {
      "epoch": 0.04345,
      "grad_norm": 0.7071974800056289,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 4345
    },
    {
      "epoch": 0.04346,
      "grad_norm": 0.7231453989893986,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4346
    },
    {
      "epoch": 0.04347,
      "grad_norm": 0.7413522380132354,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4347
    },
    {
      "epoch": 0.04348,
      "grad_norm": 0.861089019789159,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4348
    },
    {
      "epoch": 0.04349,
      "grad_norm": 0.8929649634389639,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 4349
    },
    {
      "epoch": 0.0435,
      "grad_norm": 0.8422594079232042,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 4350
    },
    {
      "epoch": 0.04351,
      "grad_norm": 0.6766722178647709,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4351
    },
    {
      "epoch": 0.04352,
      "grad_norm": 0.6065148592107631,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 4352
    },
    {
      "epoch": 0.04353,
      "grad_norm": 0.650665279490602,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4353
    },
    {
      "epoch": 0.04354,
      "grad_norm": 0.6886947880787938,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 4354
    },
    {
      "epoch": 0.04355,
      "grad_norm": 0.671956570402557,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 4355
    },
    {
      "epoch": 0.04356,
      "grad_norm": 0.6051867008044065,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4356
    },
    {
      "epoch": 0.04357,
      "grad_norm": 0.565831993323353,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4357
    },
    {
      "epoch": 0.04358,
      "grad_norm": 0.6223890122466902,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 4358
    },
    {
      "epoch": 0.04359,
      "grad_norm": 0.5817182346909322,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4359
    },
    {
      "epoch": 0.0436,
      "grad_norm": 0.5416327077206258,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4360
    },
    {
      "epoch": 0.04361,
      "grad_norm": 0.5828369400063332,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 4361
    },
    {
      "epoch": 0.04362,
      "grad_norm": 0.6659513919032584,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4362
    },
    {
      "epoch": 0.04363,
      "grad_norm": 0.6690881007982171,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4363
    },
    {
      "epoch": 0.04364,
      "grad_norm": 0.6338233307172799,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4364
    },
    {
      "epoch": 0.04365,
      "grad_norm": 0.6477441011384779,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4365
    },
    {
      "epoch": 0.04366,
      "grad_norm": 0.7001290336621013,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 4366
    },
    {
      "epoch": 0.04367,
      "grad_norm": 0.7750910475232337,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4367
    },
    {
      "epoch": 0.04368,
      "grad_norm": 0.653538990880571,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 4368
    },
    {
      "epoch": 0.04369,
      "grad_norm": 0.5454509649351917,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 4369
    },
    {
      "epoch": 0.0437,
      "grad_norm": 0.587005709751278,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4370
    },
    {
      "epoch": 0.04371,
      "grad_norm": 0.6982746587986959,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4371
    },
    {
      "epoch": 0.04372,
      "grad_norm": 0.7399237590808896,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 4372
    },
    {
      "epoch": 0.04373,
      "grad_norm": 0.7016341160191462,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 4373
    },
    {
      "epoch": 0.04374,
      "grad_norm": 0.7430013647684137,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 4374
    },
    {
      "epoch": 0.04375,
      "grad_norm": 0.7446061707027156,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 4375
    },
    {
      "epoch": 0.04376,
      "grad_norm": 0.7746810507611076,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4376
    },
    {
      "epoch": 0.04377,
      "grad_norm": 0.9323940396007363,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 4377
    },
    {
      "epoch": 0.04378,
      "grad_norm": 1.139045470645915,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 4378
    },
    {
      "epoch": 0.04379,
      "grad_norm": 0.9525647154578588,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 4379
    },
    {
      "epoch": 0.0438,
      "grad_norm": 0.8243260327364462,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4380
    },
    {
      "epoch": 0.04381,
      "grad_norm": 0.6831625467049471,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4381
    },
    {
      "epoch": 0.04382,
      "grad_norm": 0.612690019661821,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4382
    },
    {
      "epoch": 0.04383,
      "grad_norm": 0.540088011425151,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4383
    },
    {
      "epoch": 0.04384,
      "grad_norm": 0.48067565504109466,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4384
    },
    {
      "epoch": 0.04385,
      "grad_norm": 0.4940838283566669,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4385
    },
    {
      "epoch": 0.04386,
      "grad_norm": 0.47625894049094,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4386
    },
    {
      "epoch": 0.04387,
      "grad_norm": 0.48055396677108986,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 4387
    },
    {
      "epoch": 0.04388,
      "grad_norm": 0.4652346965489437,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 4388
    },
    {
      "epoch": 0.04389,
      "grad_norm": 0.6054008250326102,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 4389
    },
    {
      "epoch": 0.0439,
      "grad_norm": 0.8224358615264853,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 4390
    },
    {
      "epoch": 0.04391,
      "grad_norm": 0.9124868764503055,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4391
    },
    {
      "epoch": 0.04392,
      "grad_norm": 0.9806682632795687,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4392
    },
    {
      "epoch": 0.04393,
      "grad_norm": 0.9973385060246203,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4393
    },
    {
      "epoch": 0.04394,
      "grad_norm": 0.9949589217331475,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 4394
    },
    {
      "epoch": 0.04395,
      "grad_norm": 0.9709607301881599,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4395
    },
    {
      "epoch": 0.04396,
      "grad_norm": 0.9645527804872946,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 4396
    },
    {
      "epoch": 0.04397,
      "grad_norm": 0.9816350655638709,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4397
    },
    {
      "epoch": 0.04398,
      "grad_norm": 1.0455773944250841,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4398
    },
    {
      "epoch": 0.04399,
      "grad_norm": 0.9119841908639416,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 4399
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.9349641099143394,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 4400
    },
    {
      "epoch": 0.04401,
      "grad_norm": 1.012943202195574,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 4401
    },
    {
      "epoch": 0.04402,
      "grad_norm": 0.9301815653214077,
      "learning_rate": 0.003,
      "loss": 4.1887,
      "step": 4402
    },
    {
      "epoch": 0.04403,
      "grad_norm": 0.8751459284087095,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4403
    },
    {
      "epoch": 0.04404,
      "grad_norm": 0.8722167037599222,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 4404
    },
    {
      "epoch": 0.04405,
      "grad_norm": 0.8424088244833199,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4405
    },
    {
      "epoch": 0.04406,
      "grad_norm": 0.8029456116125556,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 4406
    },
    {
      "epoch": 0.04407,
      "grad_norm": 0.7600665803647753,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 4407
    },
    {
      "epoch": 0.04408,
      "grad_norm": 0.7143442872514836,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4408
    },
    {
      "epoch": 0.04409,
      "grad_norm": 0.7965192349032346,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 4409
    },
    {
      "epoch": 0.0441,
      "grad_norm": 0.8792702370392786,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4410
    },
    {
      "epoch": 0.04411,
      "grad_norm": 1.0243973709917655,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4411
    },
    {
      "epoch": 0.04412,
      "grad_norm": 0.8145959259792708,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4412
    },
    {
      "epoch": 0.04413,
      "grad_norm": 0.57578344484875,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4413
    },
    {
      "epoch": 0.04414,
      "grad_norm": 0.5763746372907456,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 4414
    },
    {
      "epoch": 0.04415,
      "grad_norm": 0.4849406384890711,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 4415
    },
    {
      "epoch": 0.04416,
      "grad_norm": 0.5645793903941705,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4416
    },
    {
      "epoch": 0.04417,
      "grad_norm": 0.561088827608562,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 4417
    },
    {
      "epoch": 0.04418,
      "grad_norm": 0.5718569607650594,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 4418
    },
    {
      "epoch": 0.04419,
      "grad_norm": 0.502350939750905,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4419
    },
    {
      "epoch": 0.0442,
      "grad_norm": 0.45571038809991676,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 4420
    },
    {
      "epoch": 0.04421,
      "grad_norm": 0.42375826081335205,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 4421
    },
    {
      "epoch": 0.04422,
      "grad_norm": 0.42008823401459217,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4422
    },
    {
      "epoch": 0.04423,
      "grad_norm": 0.46008487233587403,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4423
    },
    {
      "epoch": 0.04424,
      "grad_norm": 0.5469701885775844,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 4424
    },
    {
      "epoch": 0.04425,
      "grad_norm": 0.6192720902784608,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 4425
    },
    {
      "epoch": 0.04426,
      "grad_norm": 0.6752662810050423,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4426
    },
    {
      "epoch": 0.04427,
      "grad_norm": 0.8141717149045535,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4427
    },
    {
      "epoch": 0.04428,
      "grad_norm": 1.1138698669039535,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4428
    },
    {
      "epoch": 0.04429,
      "grad_norm": 1.1125457214220589,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 4429
    },
    {
      "epoch": 0.0443,
      "grad_norm": 0.7372479542562949,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4430
    },
    {
      "epoch": 0.04431,
      "grad_norm": 0.7590662782504862,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 4431
    },
    {
      "epoch": 0.04432,
      "grad_norm": 0.7851056797042649,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 4432
    },
    {
      "epoch": 0.04433,
      "grad_norm": 0.6565615710180471,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 4433
    },
    {
      "epoch": 0.04434,
      "grad_norm": 0.5918951930492853,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4434
    },
    {
      "epoch": 0.04435,
      "grad_norm": 0.5712466192614544,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4435
    },
    {
      "epoch": 0.04436,
      "grad_norm": 0.6540737676285047,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 4436
    },
    {
      "epoch": 0.04437,
      "grad_norm": 0.6512769646169722,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4437
    },
    {
      "epoch": 0.04438,
      "grad_norm": 0.6313619524940465,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 4438
    },
    {
      "epoch": 0.04439,
      "grad_norm": 0.752951040748533,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4439
    },
    {
      "epoch": 0.0444,
      "grad_norm": 0.7444628766487162,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 4440
    },
    {
      "epoch": 0.04441,
      "grad_norm": 0.7171166547645912,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4441
    },
    {
      "epoch": 0.04442,
      "grad_norm": 0.7004680425501666,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4442
    },
    {
      "epoch": 0.04443,
      "grad_norm": 0.699693309924618,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 4443
    },
    {
      "epoch": 0.04444,
      "grad_norm": 0.605352884343741,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 4444
    },
    {
      "epoch": 0.04445,
      "grad_norm": 0.6910385491365711,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4445
    },
    {
      "epoch": 0.04446,
      "grad_norm": 0.6750353497967526,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 4446
    },
    {
      "epoch": 0.04447,
      "grad_norm": 0.5704187456661106,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4447
    },
    {
      "epoch": 0.04448,
      "grad_norm": 0.5831627691900045,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 4448
    },
    {
      "epoch": 0.04449,
      "grad_norm": 0.7383838513937211,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4449
    },
    {
      "epoch": 0.0445,
      "grad_norm": 0.9275667982571733,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4450
    },
    {
      "epoch": 0.04451,
      "grad_norm": 1.1278443077545208,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 4451
    },
    {
      "epoch": 0.04452,
      "grad_norm": 0.7200631704178471,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 4452
    },
    {
      "epoch": 0.04453,
      "grad_norm": 0.7319095347479152,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 4453
    },
    {
      "epoch": 0.04454,
      "grad_norm": 0.828360672866726,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4454
    },
    {
      "epoch": 0.04455,
      "grad_norm": 0.8149109397250537,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 4455
    },
    {
      "epoch": 0.04456,
      "grad_norm": 0.7949979530915787,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4456
    },
    {
      "epoch": 0.04457,
      "grad_norm": 0.7815335465443298,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4457
    },
    {
      "epoch": 0.04458,
      "grad_norm": 0.6667090111498273,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4458
    },
    {
      "epoch": 0.04459,
      "grad_norm": 0.6478410064314194,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4459
    },
    {
      "epoch": 0.0446,
      "grad_norm": 0.692693711021832,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 4460
    },
    {
      "epoch": 0.04461,
      "grad_norm": 0.7661349867430685,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 4461
    },
    {
      "epoch": 0.04462,
      "grad_norm": 0.9456365590075492,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4462
    },
    {
      "epoch": 0.04463,
      "grad_norm": 0.9219959129201606,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4463
    },
    {
      "epoch": 0.04464,
      "grad_norm": 0.8148900268740181,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4464
    },
    {
      "epoch": 0.04465,
      "grad_norm": 0.8174238600511685,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 4465
    },
    {
      "epoch": 0.04466,
      "grad_norm": 0.6910175481107252,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4466
    },
    {
      "epoch": 0.04467,
      "grad_norm": 0.8503966818321036,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4467
    },
    {
      "epoch": 0.04468,
      "grad_norm": 0.8010356993406599,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4468
    },
    {
      "epoch": 0.04469,
      "grad_norm": 0.7388705798425241,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 4469
    },
    {
      "epoch": 0.0447,
      "grad_norm": 0.776572618481807,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4470
    },
    {
      "epoch": 0.04471,
      "grad_norm": 0.8915085141537737,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 4471
    },
    {
      "epoch": 0.04472,
      "grad_norm": 0.9200136447240024,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4472
    },
    {
      "epoch": 0.04473,
      "grad_norm": 0.8156139485901094,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4473
    },
    {
      "epoch": 0.04474,
      "grad_norm": 0.7239763646047743,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4474
    },
    {
      "epoch": 0.04475,
      "grad_norm": 0.6835215989709738,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 4475
    },
    {
      "epoch": 0.04476,
      "grad_norm": 0.6497813876295598,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 4476
    },
    {
      "epoch": 0.04477,
      "grad_norm": 0.647414838457496,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 4477
    },
    {
      "epoch": 0.04478,
      "grad_norm": 0.559137538294316,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4478
    },
    {
      "epoch": 0.04479,
      "grad_norm": 0.5859013586768095,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 4479
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.5435317461744096,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 4480
    },
    {
      "epoch": 0.04481,
      "grad_norm": 0.6011946337549432,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 4481
    },
    {
      "epoch": 0.04482,
      "grad_norm": 0.7806327420943877,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4482
    },
    {
      "epoch": 0.04483,
      "grad_norm": 1.0587276903826628,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4483
    },
    {
      "epoch": 0.04484,
      "grad_norm": 0.9760427349795051,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 4484
    },
    {
      "epoch": 0.04485,
      "grad_norm": 0.7488828747485865,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4485
    },
    {
      "epoch": 0.04486,
      "grad_norm": 0.8084521903070014,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 4486
    },
    {
      "epoch": 0.04487,
      "grad_norm": 0.9117696333780938,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4487
    },
    {
      "epoch": 0.04488,
      "grad_norm": 0.8996969646129346,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 4488
    },
    {
      "epoch": 0.04489,
      "grad_norm": 0.8641616901342983,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4489
    },
    {
      "epoch": 0.0449,
      "grad_norm": 0.7280388583966139,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4490
    },
    {
      "epoch": 0.04491,
      "grad_norm": 0.6733842898509934,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4491
    },
    {
      "epoch": 0.04492,
      "grad_norm": 0.6313608339838067,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4492
    },
    {
      "epoch": 0.04493,
      "grad_norm": 0.6192244819984872,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 4493
    },
    {
      "epoch": 0.04494,
      "grad_norm": 0.5488813855675564,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4494
    },
    {
      "epoch": 0.04495,
      "grad_norm": 0.4779659159856146,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4495
    },
    {
      "epoch": 0.04496,
      "grad_norm": 0.5060747754385665,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4496
    },
    {
      "epoch": 0.04497,
      "grad_norm": 0.5808549525919214,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4497
    },
    {
      "epoch": 0.04498,
      "grad_norm": 0.6694153212993217,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4498
    },
    {
      "epoch": 0.04499,
      "grad_norm": 0.8090682417638635,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 4499
    },
    {
      "epoch": 0.045,
      "grad_norm": 0.9223841526186204,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4500
    },
    {
      "epoch": 0.04501,
      "grad_norm": 0.8349758444195646,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4501
    },
    {
      "epoch": 0.04502,
      "grad_norm": 0.7254525437975357,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 4502
    },
    {
      "epoch": 0.04503,
      "grad_norm": 0.649056521260754,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4503
    },
    {
      "epoch": 0.04504,
      "grad_norm": 0.713115140118379,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 4504
    },
    {
      "epoch": 0.04505,
      "grad_norm": 0.6117909603348726,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 4505
    },
    {
      "epoch": 0.04506,
      "grad_norm": 0.6145202957029064,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4506
    },
    {
      "epoch": 0.04507,
      "grad_norm": 0.5247412068715119,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 4507
    },
    {
      "epoch": 0.04508,
      "grad_norm": 0.5806017179220667,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4508
    },
    {
      "epoch": 0.04509,
      "grad_norm": 0.5825330876210861,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4509
    },
    {
      "epoch": 0.0451,
      "grad_norm": 0.7871833251091375,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4510
    },
    {
      "epoch": 0.04511,
      "grad_norm": 0.8750417888525687,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4511
    },
    {
      "epoch": 0.04512,
      "grad_norm": 0.8564995340864695,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 4512
    },
    {
      "epoch": 0.04513,
      "grad_norm": 0.7843001776115731,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4513
    },
    {
      "epoch": 0.04514,
      "grad_norm": 0.6796657929720622,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4514
    },
    {
      "epoch": 0.04515,
      "grad_norm": 0.6410320420606614,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4515
    },
    {
      "epoch": 0.04516,
      "grad_norm": 0.6710348380128827,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 4516
    },
    {
      "epoch": 0.04517,
      "grad_norm": 0.8318718390934561,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 4517
    },
    {
      "epoch": 0.04518,
      "grad_norm": 0.895874624608404,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4518
    },
    {
      "epoch": 0.04519,
      "grad_norm": 0.9529303615004293,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4519
    },
    {
      "epoch": 0.0452,
      "grad_norm": 1.00692610224058,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4520
    },
    {
      "epoch": 0.04521,
      "grad_norm": 1.0366622726930343,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 4521
    },
    {
      "epoch": 0.04522,
      "grad_norm": 1.0505298007829553,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 4522
    },
    {
      "epoch": 0.04523,
      "grad_norm": 1.0189524405337644,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4523
    },
    {
      "epoch": 0.04524,
      "grad_norm": 1.030832446298102,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4524
    },
    {
      "epoch": 0.04525,
      "grad_norm": 1.0054730825905025,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4525
    },
    {
      "epoch": 0.04526,
      "grad_norm": 0.9357043959907038,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 4526
    },
    {
      "epoch": 0.04527,
      "grad_norm": 0.880967833594335,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4527
    },
    {
      "epoch": 0.04528,
      "grad_norm": 0.95913766957737,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4528
    },
    {
      "epoch": 0.04529,
      "grad_norm": 0.7989651014070877,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4529
    },
    {
      "epoch": 0.0453,
      "grad_norm": 0.7380961288194754,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4530
    },
    {
      "epoch": 0.04531,
      "grad_norm": 0.7085687602336856,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4531
    },
    {
      "epoch": 0.04532,
      "grad_norm": 0.6574885202110717,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 4532
    },
    {
      "epoch": 0.04533,
      "grad_norm": 0.6519010842998008,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 4533
    },
    {
      "epoch": 0.04534,
      "grad_norm": 0.6093622128244879,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4534
    },
    {
      "epoch": 0.04535,
      "grad_norm": 0.5837584644000219,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 4535
    },
    {
      "epoch": 0.04536,
      "grad_norm": 0.6897731390503988,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 4536
    },
    {
      "epoch": 0.04537,
      "grad_norm": 0.6949801331902556,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 4537
    },
    {
      "epoch": 0.04538,
      "grad_norm": 0.6776975735023881,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4538
    },
    {
      "epoch": 0.04539,
      "grad_norm": 0.678568300988415,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 4539
    },
    {
      "epoch": 0.0454,
      "grad_norm": 0.6603018556737656,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 4540
    },
    {
      "epoch": 0.04541,
      "grad_norm": 0.6133835370391497,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4541
    },
    {
      "epoch": 0.04542,
      "grad_norm": 0.635535341053592,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 4542
    },
    {
      "epoch": 0.04543,
      "grad_norm": 0.668577889488378,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4543
    },
    {
      "epoch": 0.04544,
      "grad_norm": 0.7427455997801305,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4544
    },
    {
      "epoch": 0.04545,
      "grad_norm": 0.8639923142163211,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4545
    },
    {
      "epoch": 0.04546,
      "grad_norm": 0.6895806692659665,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 4546
    },
    {
      "epoch": 0.04547,
      "grad_norm": 0.49153824435946125,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4547
    },
    {
      "epoch": 0.04548,
      "grad_norm": 0.5516451722786672,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 4548
    },
    {
      "epoch": 0.04549,
      "grad_norm": 0.6029789439713806,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4549
    },
    {
      "epoch": 0.0455,
      "grad_norm": 0.66622853988702,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4550
    },
    {
      "epoch": 0.04551,
      "grad_norm": 0.7238008534924777,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4551
    },
    {
      "epoch": 0.04552,
      "grad_norm": 0.7824263722798367,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 4552
    },
    {
      "epoch": 0.04553,
      "grad_norm": 0.7388703999593268,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4553
    },
    {
      "epoch": 0.04554,
      "grad_norm": 0.6913086022347059,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4554
    },
    {
      "epoch": 0.04555,
      "grad_norm": 0.6442867508319275,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 4555
    },
    {
      "epoch": 0.04556,
      "grad_norm": 0.7238849740450599,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4556
    },
    {
      "epoch": 0.04557,
      "grad_norm": 0.7976327742193281,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 4557
    },
    {
      "epoch": 0.04558,
      "grad_norm": 0.8032287746582655,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 4558
    },
    {
      "epoch": 0.04559,
      "grad_norm": 0.784841139322348,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4559
    },
    {
      "epoch": 0.0456,
      "grad_norm": 0.8095586358578389,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4560
    },
    {
      "epoch": 0.04561,
      "grad_norm": 0.6748297662204855,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 4561
    },
    {
      "epoch": 0.04562,
      "grad_norm": 0.6987048066021282,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 4562
    },
    {
      "epoch": 0.04563,
      "grad_norm": 0.772254646470346,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4563
    },
    {
      "epoch": 0.04564,
      "grad_norm": 0.8035846762926936,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4564
    },
    {
      "epoch": 0.04565,
      "grad_norm": 0.7913541854291436,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 4565
    },
    {
      "epoch": 0.04566,
      "grad_norm": 0.7821636491006997,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4566
    },
    {
      "epoch": 0.04567,
      "grad_norm": 0.6863252350497664,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4567
    },
    {
      "epoch": 0.04568,
      "grad_norm": 0.6211213056274318,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 4568
    },
    {
      "epoch": 0.04569,
      "grad_norm": 0.7600485305292163,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4569
    },
    {
      "epoch": 0.0457,
      "grad_norm": 0.7935297140012192,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4570
    },
    {
      "epoch": 0.04571,
      "grad_norm": 0.8707627695603387,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 4571
    },
    {
      "epoch": 0.04572,
      "grad_norm": 0.9504850878398086,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4572
    },
    {
      "epoch": 0.04573,
      "grad_norm": 0.9772978151032115,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 4573
    },
    {
      "epoch": 0.04574,
      "grad_norm": 0.8349860867562747,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4574
    },
    {
      "epoch": 0.04575,
      "grad_norm": 0.7275515356936376,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4575
    },
    {
      "epoch": 0.04576,
      "grad_norm": 0.7453152360629928,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4576
    },
    {
      "epoch": 0.04577,
      "grad_norm": 0.731228336021725,
      "learning_rate": 0.003,
      "loss": 4.1686,
      "step": 4577
    },
    {
      "epoch": 0.04578,
      "grad_norm": 0.7565651611671647,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4578
    },
    {
      "epoch": 0.04579,
      "grad_norm": 0.9198271859494617,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 4579
    },
    {
      "epoch": 0.0458,
      "grad_norm": 1.0574643718931676,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 4580
    },
    {
      "epoch": 0.04581,
      "grad_norm": 0.9388725861321914,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 4581
    },
    {
      "epoch": 0.04582,
      "grad_norm": 0.7599199575014808,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 4582
    },
    {
      "epoch": 0.04583,
      "grad_norm": 0.6683292234471547,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4583
    },
    {
      "epoch": 0.04584,
      "grad_norm": 0.7512487692406998,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4584
    },
    {
      "epoch": 0.04585,
      "grad_norm": 0.7081640921869536,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4585
    },
    {
      "epoch": 0.04586,
      "grad_norm": 0.7168079051261622,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4586
    },
    {
      "epoch": 0.04587,
      "grad_norm": 0.814216089662895,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4587
    },
    {
      "epoch": 0.04588,
      "grad_norm": 0.8322807065043963,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4588
    },
    {
      "epoch": 0.04589,
      "grad_norm": 0.8173670039992074,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 4589
    },
    {
      "epoch": 0.0459,
      "grad_norm": 0.7002687692328883,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4590
    },
    {
      "epoch": 0.04591,
      "grad_norm": 0.667313250694922,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 4591
    },
    {
      "epoch": 0.04592,
      "grad_norm": 0.7721193873262023,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 4592
    },
    {
      "epoch": 0.04593,
      "grad_norm": 0.719951932975662,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4593
    },
    {
      "epoch": 0.04594,
      "grad_norm": 0.624806723533564,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4594
    },
    {
      "epoch": 0.04595,
      "grad_norm": 0.5756945318000921,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 4595
    },
    {
      "epoch": 0.04596,
      "grad_norm": 0.5657179032543712,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 4596
    },
    {
      "epoch": 0.04597,
      "grad_norm": 0.6256040629924149,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4597
    },
    {
      "epoch": 0.04598,
      "grad_norm": 0.6013110035097162,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 4598
    },
    {
      "epoch": 0.04599,
      "grad_norm": 0.7296067063170582,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 4599
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.9209598285204096,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4600
    },
    {
      "epoch": 0.04601,
      "grad_norm": 1.0160633328517232,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4601
    },
    {
      "epoch": 0.04602,
      "grad_norm": 0.8169968744441843,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4602
    },
    {
      "epoch": 0.04603,
      "grad_norm": 0.659664339286615,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4603
    },
    {
      "epoch": 0.04604,
      "grad_norm": 0.6592904260797935,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4604
    },
    {
      "epoch": 0.04605,
      "grad_norm": 0.6689967757886678,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 4605
    },
    {
      "epoch": 0.04606,
      "grad_norm": 0.6839466982971314,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 4606
    },
    {
      "epoch": 0.04607,
      "grad_norm": 0.6080788219356916,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 4607
    },
    {
      "epoch": 0.04608,
      "grad_norm": 0.6229335033564855,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4608
    },
    {
      "epoch": 0.04609,
      "grad_norm": 0.6110178676005829,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4609
    },
    {
      "epoch": 0.0461,
      "grad_norm": 0.6009057478520481,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 4610
    },
    {
      "epoch": 0.04611,
      "grad_norm": 0.667830986428792,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 4611
    },
    {
      "epoch": 0.04612,
      "grad_norm": 0.8308783400085262,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4612
    },
    {
      "epoch": 0.04613,
      "grad_norm": 0.9122587432173699,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4613
    },
    {
      "epoch": 0.04614,
      "grad_norm": 1.0201054294863492,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 4614
    },
    {
      "epoch": 0.04615,
      "grad_norm": 0.7968952064740754,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4615
    },
    {
      "epoch": 0.04616,
      "grad_norm": 0.7040440287863108,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4616
    },
    {
      "epoch": 0.04617,
      "grad_norm": 0.650743607835788,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4617
    },
    {
      "epoch": 0.04618,
      "grad_norm": 0.6876364900587607,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4618
    },
    {
      "epoch": 0.04619,
      "grad_norm": 0.674403145056888,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 4619
    },
    {
      "epoch": 0.0462,
      "grad_norm": 0.6278438951559039,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4620
    },
    {
      "epoch": 0.04621,
      "grad_norm": 0.5882186858354869,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4621
    },
    {
      "epoch": 0.04622,
      "grad_norm": 0.6093666817375198,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4622
    },
    {
      "epoch": 0.04623,
      "grad_norm": 0.5680931823986678,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 4623
    },
    {
      "epoch": 0.04624,
      "grad_norm": 0.6168812041909036,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 4624
    },
    {
      "epoch": 0.04625,
      "grad_norm": 0.6071577011256134,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4625
    },
    {
      "epoch": 0.04626,
      "grad_norm": 0.5293202352744564,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 4626
    },
    {
      "epoch": 0.04627,
      "grad_norm": 0.5193967115713368,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4627
    },
    {
      "epoch": 0.04628,
      "grad_norm": 0.6344196546552844,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4628
    },
    {
      "epoch": 0.04629,
      "grad_norm": 0.6660491057377373,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4629
    },
    {
      "epoch": 0.0463,
      "grad_norm": 0.6585553354499009,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4630
    },
    {
      "epoch": 0.04631,
      "grad_norm": 0.6065008696674108,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 4631
    },
    {
      "epoch": 0.04632,
      "grad_norm": 0.5310090972041924,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 4632
    },
    {
      "epoch": 0.04633,
      "grad_norm": 0.6357968891584513,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 4633
    },
    {
      "epoch": 0.04634,
      "grad_norm": 0.941578980575187,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4634
    },
    {
      "epoch": 0.04635,
      "grad_norm": 1.3572160948425378,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 4635
    },
    {
      "epoch": 0.04636,
      "grad_norm": 0.6013442724378576,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4636
    },
    {
      "epoch": 0.04637,
      "grad_norm": 0.6555645652247263,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4637
    },
    {
      "epoch": 0.04638,
      "grad_norm": 0.7328409108901823,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 4638
    },
    {
      "epoch": 0.04639,
      "grad_norm": 0.6867315466218923,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 4639
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.78928501351717,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4640
    },
    {
      "epoch": 0.04641,
      "grad_norm": 0.8455497454514354,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 4641
    },
    {
      "epoch": 0.04642,
      "grad_norm": 0.8465515736905618,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4642
    },
    {
      "epoch": 0.04643,
      "grad_norm": 0.7201267511516922,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 4643
    },
    {
      "epoch": 0.04644,
      "grad_norm": 0.7720786991059095,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 4644
    },
    {
      "epoch": 0.04645,
      "grad_norm": 0.7658630542557853,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 4645
    },
    {
      "epoch": 0.04646,
      "grad_norm": 0.8677913304110078,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 4646
    },
    {
      "epoch": 0.04647,
      "grad_norm": 0.8914361615284006,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4647
    },
    {
      "epoch": 0.04648,
      "grad_norm": 0.8603752804558079,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4648
    },
    {
      "epoch": 0.04649,
      "grad_norm": 0.8183443337243556,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 4649
    },
    {
      "epoch": 0.0465,
      "grad_norm": 0.6592577702353686,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 4650
    },
    {
      "epoch": 0.04651,
      "grad_norm": 0.6033831169642737,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 4651
    },
    {
      "epoch": 0.04652,
      "grad_norm": 0.6694785473055216,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 4652
    },
    {
      "epoch": 0.04653,
      "grad_norm": 0.6797455964877727,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4653
    },
    {
      "epoch": 0.04654,
      "grad_norm": 0.7400041353958778,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4654
    },
    {
      "epoch": 0.04655,
      "grad_norm": 0.884024246484862,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 4655
    },
    {
      "epoch": 0.04656,
      "grad_norm": 0.9724046681595542,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4656
    },
    {
      "epoch": 0.04657,
      "grad_norm": 0.9321801110747265,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 4657
    },
    {
      "epoch": 0.04658,
      "grad_norm": 0.7977247035986766,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 4658
    },
    {
      "epoch": 0.04659,
      "grad_norm": 0.7837947068767307,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4659
    },
    {
      "epoch": 0.0466,
      "grad_norm": 0.8489578033894326,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4660
    },
    {
      "epoch": 0.04661,
      "grad_norm": 0.9070979393786719,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4661
    },
    {
      "epoch": 0.04662,
      "grad_norm": 0.9271278255074006,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4662
    },
    {
      "epoch": 0.04663,
      "grad_norm": 0.9626508931970053,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4663
    },
    {
      "epoch": 0.04664,
      "grad_norm": 1.112988241634731,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4664
    },
    {
      "epoch": 0.04665,
      "grad_norm": 0.8921340954370226,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 4665
    },
    {
      "epoch": 0.04666,
      "grad_norm": 0.7837443287711118,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4666
    },
    {
      "epoch": 0.04667,
      "grad_norm": 0.7219249423509705,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4667
    },
    {
      "epoch": 0.04668,
      "grad_norm": 0.7307342741354349,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 4668
    },
    {
      "epoch": 0.04669,
      "grad_norm": 0.7401471208519357,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 4669
    },
    {
      "epoch": 0.0467,
      "grad_norm": 0.7077495687150961,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 4670
    },
    {
      "epoch": 0.04671,
      "grad_norm": 0.7086785987292082,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4671
    },
    {
      "epoch": 0.04672,
      "grad_norm": 0.7271653898870789,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4672
    },
    {
      "epoch": 0.04673,
      "grad_norm": 0.7152687743613511,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4673
    },
    {
      "epoch": 0.04674,
      "grad_norm": 0.8387550898694738,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4674
    },
    {
      "epoch": 0.04675,
      "grad_norm": 0.9264998756521228,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 4675
    },
    {
      "epoch": 0.04676,
      "grad_norm": 0.9760003719176258,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4676
    },
    {
      "epoch": 0.04677,
      "grad_norm": 0.9605322775687504,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4677
    },
    {
      "epoch": 0.04678,
      "grad_norm": 0.760672310236161,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 4678
    },
    {
      "epoch": 0.04679,
      "grad_norm": 0.6288734540073927,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4679
    },
    {
      "epoch": 0.0468,
      "grad_norm": 0.5340420649165649,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4680
    },
    {
      "epoch": 0.04681,
      "grad_norm": 0.5383218654638344,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 4681
    },
    {
      "epoch": 0.04682,
      "grad_norm": 0.5921170754815444,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4682
    },
    {
      "epoch": 0.04683,
      "grad_norm": 0.5986112675201656,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4683
    },
    {
      "epoch": 0.04684,
      "grad_norm": 0.6515018805380663,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4684
    },
    {
      "epoch": 0.04685,
      "grad_norm": 0.6225025584760357,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4685
    },
    {
      "epoch": 0.04686,
      "grad_norm": 0.5748567786413258,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4686
    },
    {
      "epoch": 0.04687,
      "grad_norm": 0.597224880471858,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4687
    },
    {
      "epoch": 0.04688,
      "grad_norm": 0.7033215388348866,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4688
    },
    {
      "epoch": 0.04689,
      "grad_norm": 0.8369783564473339,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 4689
    },
    {
      "epoch": 0.0469,
      "grad_norm": 0.8637378056248713,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4690
    },
    {
      "epoch": 0.04691,
      "grad_norm": 0.9863994238127417,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4691
    },
    {
      "epoch": 0.04692,
      "grad_norm": 1.1588055062272775,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 4692
    },
    {
      "epoch": 0.04693,
      "grad_norm": 1.0154258289459517,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4693
    },
    {
      "epoch": 0.04694,
      "grad_norm": 0.8542851528225356,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4694
    },
    {
      "epoch": 0.04695,
      "grad_norm": 0.7936283873091949,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 4695
    },
    {
      "epoch": 0.04696,
      "grad_norm": 0.8706455027386264,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 4696
    },
    {
      "epoch": 0.04697,
      "grad_norm": 0.888025520306328,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4697
    },
    {
      "epoch": 0.04698,
      "grad_norm": 0.8024507884497375,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4698
    },
    {
      "epoch": 0.04699,
      "grad_norm": 0.6716482593675086,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4699
    },
    {
      "epoch": 0.047,
      "grad_norm": 0.6875782963509716,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4700
    },
    {
      "epoch": 0.04701,
      "grad_norm": 0.6233215143134673,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4701
    },
    {
      "epoch": 0.04702,
      "grad_norm": 0.5392566756700929,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4702
    },
    {
      "epoch": 0.04703,
      "grad_norm": 0.5917569266603563,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 4703
    },
    {
      "epoch": 0.04704,
      "grad_norm": 0.6011918134456015,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 4704
    },
    {
      "epoch": 0.04705,
      "grad_norm": 0.6280725769151366,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 4705
    },
    {
      "epoch": 0.04706,
      "grad_norm": 0.6696710280228617,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 4706
    },
    {
      "epoch": 0.04707,
      "grad_norm": 0.5908607329438127,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4707
    },
    {
      "epoch": 0.04708,
      "grad_norm": 0.5153382206986533,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 4708
    },
    {
      "epoch": 0.04709,
      "grad_norm": 0.513601992478386,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4709
    },
    {
      "epoch": 0.0471,
      "grad_norm": 0.6490521053147635,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4710
    },
    {
      "epoch": 0.04711,
      "grad_norm": 0.7794691577627103,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4711
    },
    {
      "epoch": 0.04712,
      "grad_norm": 0.7688497592240825,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4712
    },
    {
      "epoch": 0.04713,
      "grad_norm": 0.8439539645333411,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 4713
    },
    {
      "epoch": 0.04714,
      "grad_norm": 0.9146059884037103,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4714
    },
    {
      "epoch": 0.04715,
      "grad_norm": 0.8646027137635436,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4715
    },
    {
      "epoch": 0.04716,
      "grad_norm": 0.7274583804499275,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4716
    },
    {
      "epoch": 0.04717,
      "grad_norm": 0.683321783957918,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 4717
    },
    {
      "epoch": 0.04718,
      "grad_norm": 0.7716759413978014,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 4718
    },
    {
      "epoch": 0.04719,
      "grad_norm": 0.79629647999161,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4719
    },
    {
      "epoch": 0.0472,
      "grad_norm": 0.7050760628166665,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4720
    },
    {
      "epoch": 0.04721,
      "grad_norm": 0.6252610323631939,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 4721
    },
    {
      "epoch": 0.04722,
      "grad_norm": 0.6123095727231639,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4722
    },
    {
      "epoch": 0.04723,
      "grad_norm": 0.6308650251833371,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 4723
    },
    {
      "epoch": 0.04724,
      "grad_norm": 0.6995282704649718,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 4724
    },
    {
      "epoch": 0.04725,
      "grad_norm": 0.822004537328782,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 4725
    },
    {
      "epoch": 0.04726,
      "grad_norm": 0.9108754250302765,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 4726
    },
    {
      "epoch": 0.04727,
      "grad_norm": 0.9359931741571359,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 4727
    },
    {
      "epoch": 0.04728,
      "grad_norm": 0.8319271568838779,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 4728
    },
    {
      "epoch": 0.04729,
      "grad_norm": 0.7272073389712985,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4729
    },
    {
      "epoch": 0.0473,
      "grad_norm": 0.6527677653973615,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4730
    },
    {
      "epoch": 0.04731,
      "grad_norm": 0.5930657999039424,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4731
    },
    {
      "epoch": 0.04732,
      "grad_norm": 0.6549784460880107,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4732
    },
    {
      "epoch": 0.04733,
      "grad_norm": 0.8327578035736742,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4733
    },
    {
      "epoch": 0.04734,
      "grad_norm": 1.104638767629771,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4734
    },
    {
      "epoch": 0.04735,
      "grad_norm": 0.9388734582276314,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 4735
    },
    {
      "epoch": 0.04736,
      "grad_norm": 0.7892905763980204,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4736
    },
    {
      "epoch": 0.04737,
      "grad_norm": 0.8500450443730917,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 4737
    },
    {
      "epoch": 0.04738,
      "grad_norm": 0.9289248020340674,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4738
    },
    {
      "epoch": 0.04739,
      "grad_norm": 0.955981339694237,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 4739
    },
    {
      "epoch": 0.0474,
      "grad_norm": 0.9041807476389423,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4740
    },
    {
      "epoch": 0.04741,
      "grad_norm": 0.8492001903774913,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4741
    },
    {
      "epoch": 0.04742,
      "grad_norm": 0.7853348835474101,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4742
    },
    {
      "epoch": 0.04743,
      "grad_norm": 0.7795682667464989,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4743
    },
    {
      "epoch": 0.04744,
      "grad_norm": 0.7267419747954841,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 4744
    },
    {
      "epoch": 0.04745,
      "grad_norm": 0.7262044399015473,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4745
    },
    {
      "epoch": 0.04746,
      "grad_norm": 0.7299893869109523,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4746
    },
    {
      "epoch": 0.04747,
      "grad_norm": 0.806239055464935,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4747
    },
    {
      "epoch": 0.04748,
      "grad_norm": 0.6789998274727187,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4748
    },
    {
      "epoch": 0.04749,
      "grad_norm": 0.54858705364636,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 4749
    },
    {
      "epoch": 0.0475,
      "grad_norm": 0.5060470689311919,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4750
    },
    {
      "epoch": 0.04751,
      "grad_norm": 0.5080665649296767,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 4751
    },
    {
      "epoch": 0.04752,
      "grad_norm": 0.5597535217525249,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4752
    },
    {
      "epoch": 0.04753,
      "grad_norm": 0.5815761039735459,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4753
    },
    {
      "epoch": 0.04754,
      "grad_norm": 0.6895819778653248,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4754
    },
    {
      "epoch": 0.04755,
      "grad_norm": 0.9251374027849657,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 4755
    },
    {
      "epoch": 0.04756,
      "grad_norm": 1.1438681733923501,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4756
    },
    {
      "epoch": 0.04757,
      "grad_norm": 0.7788109201469021,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4757
    },
    {
      "epoch": 0.04758,
      "grad_norm": 0.6624956821204955,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 4758
    },
    {
      "epoch": 0.04759,
      "grad_norm": 0.7037195304943054,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4759
    },
    {
      "epoch": 0.0476,
      "grad_norm": 0.7411798494302422,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 4760
    },
    {
      "epoch": 0.04761,
      "grad_norm": 0.7054700847471357,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4761
    },
    {
      "epoch": 0.04762,
      "grad_norm": 0.6226720435093628,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 4762
    },
    {
      "epoch": 0.04763,
      "grad_norm": 0.6886529154194588,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 4763
    },
    {
      "epoch": 0.04764,
      "grad_norm": 0.6992577861087502,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 4764
    },
    {
      "epoch": 0.04765,
      "grad_norm": 0.6613811777977939,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4765
    },
    {
      "epoch": 0.04766,
      "grad_norm": 0.5851808207831782,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 4766
    },
    {
      "epoch": 0.04767,
      "grad_norm": 0.6679898130461849,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4767
    },
    {
      "epoch": 0.04768,
      "grad_norm": 0.7821493077256183,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 4768
    },
    {
      "epoch": 0.04769,
      "grad_norm": 0.8355601404081668,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 4769
    },
    {
      "epoch": 0.0477,
      "grad_norm": 0.8479665704653694,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 4770
    },
    {
      "epoch": 0.04771,
      "grad_norm": 0.8866254860935738,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4771
    },
    {
      "epoch": 0.04772,
      "grad_norm": 0.915742333919775,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4772
    },
    {
      "epoch": 0.04773,
      "grad_norm": 0.754051301920501,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4773
    },
    {
      "epoch": 0.04774,
      "grad_norm": 0.7232939078296818,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4774
    },
    {
      "epoch": 0.04775,
      "grad_norm": 0.8169270534707214,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4775
    },
    {
      "epoch": 0.04776,
      "grad_norm": 0.8573605280685423,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4776
    },
    {
      "epoch": 0.04777,
      "grad_norm": 0.9120618578003061,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 4777
    },
    {
      "epoch": 0.04778,
      "grad_norm": 0.9392445159690339,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4778
    },
    {
      "epoch": 0.04779,
      "grad_norm": 0.8671768686651342,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4779
    },
    {
      "epoch": 0.0478,
      "grad_norm": 0.8025049205614178,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4780
    },
    {
      "epoch": 0.04781,
      "grad_norm": 0.7254367126713761,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4781
    },
    {
      "epoch": 0.04782,
      "grad_norm": 0.6859662827550447,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4782
    },
    {
      "epoch": 0.04783,
      "grad_norm": 0.6967231284903036,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4783
    },
    {
      "epoch": 0.04784,
      "grad_norm": 0.7980703901010124,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4784
    },
    {
      "epoch": 0.04785,
      "grad_norm": 0.8254970324649596,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4785
    },
    {
      "epoch": 0.04786,
      "grad_norm": 0.8777922852113546,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4786
    },
    {
      "epoch": 0.04787,
      "grad_norm": 0.7491389893707523,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4787
    },
    {
      "epoch": 0.04788,
      "grad_norm": 0.6503832601570363,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 4788
    },
    {
      "epoch": 0.04789,
      "grad_norm": 0.7031951667657491,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 4789
    },
    {
      "epoch": 0.0479,
      "grad_norm": 0.678065080300084,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 4790
    },
    {
      "epoch": 0.04791,
      "grad_norm": 0.523098323506981,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4791
    },
    {
      "epoch": 0.04792,
      "grad_norm": 0.5260803142583377,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4792
    },
    {
      "epoch": 0.04793,
      "grad_norm": 0.5215649768374346,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 4793
    },
    {
      "epoch": 0.04794,
      "grad_norm": 0.46968713617443264,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 4794
    },
    {
      "epoch": 0.04795,
      "grad_norm": 0.5247439754384676,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4795
    },
    {
      "epoch": 0.04796,
      "grad_norm": 0.6933674582411209,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 4796
    },
    {
      "epoch": 0.04797,
      "grad_norm": 0.9807791236957965,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4797
    },
    {
      "epoch": 0.04798,
      "grad_norm": 1.1741198222289444,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 4798
    },
    {
      "epoch": 0.04799,
      "grad_norm": 0.6525704641811733,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 4799
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.8089064094414999,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4800
    },
    {
      "epoch": 0.04801,
      "grad_norm": 0.870503514300495,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4801
    },
    {
      "epoch": 0.04802,
      "grad_norm": 0.6635652157598942,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 4802
    },
    {
      "epoch": 0.04803,
      "grad_norm": 0.6200303139785031,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 4803
    },
    {
      "epoch": 0.04804,
      "grad_norm": 0.762754226201963,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4804
    },
    {
      "epoch": 0.04805,
      "grad_norm": 0.7861473887229704,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4805
    },
    {
      "epoch": 0.04806,
      "grad_norm": 0.7696022521328222,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 4806
    },
    {
      "epoch": 0.04807,
      "grad_norm": 0.7570918480886352,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4807
    },
    {
      "epoch": 0.04808,
      "grad_norm": 0.8705313373191277,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 4808
    },
    {
      "epoch": 0.04809,
      "grad_norm": 0.8255463817975629,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 4809
    },
    {
      "epoch": 0.0481,
      "grad_norm": 0.8537469900625964,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4810
    },
    {
      "epoch": 0.04811,
      "grad_norm": 0.8483723617783365,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4811
    },
    {
      "epoch": 0.04812,
      "grad_norm": 0.7645031119405934,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4812
    },
    {
      "epoch": 0.04813,
      "grad_norm": 0.8283116512854986,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 4813
    },
    {
      "epoch": 0.04814,
      "grad_norm": 0.8641578230891004,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 4814
    },
    {
      "epoch": 0.04815,
      "grad_norm": 0.8919893073764681,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 4815
    },
    {
      "epoch": 0.04816,
      "grad_norm": 0.8648426828228101,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4816
    },
    {
      "epoch": 0.04817,
      "grad_norm": 0.9477238526245142,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 4817
    },
    {
      "epoch": 0.04818,
      "grad_norm": 0.9252788072968092,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4818
    },
    {
      "epoch": 0.04819,
      "grad_norm": 0.8797819035450328,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4819
    },
    {
      "epoch": 0.0482,
      "grad_norm": 0.9035562726868283,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4820
    },
    {
      "epoch": 0.04821,
      "grad_norm": 0.858921817091334,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 4821
    },
    {
      "epoch": 0.04822,
      "grad_norm": 0.6583705277011013,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4822
    },
    {
      "epoch": 0.04823,
      "grad_norm": 0.5779379743608074,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4823
    },
    {
      "epoch": 0.04824,
      "grad_norm": 0.5505773694357159,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 4824
    },
    {
      "epoch": 0.04825,
      "grad_norm": 0.5013668074041548,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4825
    },
    {
      "epoch": 0.04826,
      "grad_norm": 0.5240704697199119,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4826
    },
    {
      "epoch": 0.04827,
      "grad_norm": 0.533629158460227,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4827
    },
    {
      "epoch": 0.04828,
      "grad_norm": 0.5992841702695002,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 4828
    },
    {
      "epoch": 0.04829,
      "grad_norm": 0.6683602415515053,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 4829
    },
    {
      "epoch": 0.0483,
      "grad_norm": 0.7131645514243471,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4830
    },
    {
      "epoch": 0.04831,
      "grad_norm": 0.7918054904389603,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 4831
    },
    {
      "epoch": 0.04832,
      "grad_norm": 0.8874223110243422,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4832
    },
    {
      "epoch": 0.04833,
      "grad_norm": 0.9718339603744532,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4833
    },
    {
      "epoch": 0.04834,
      "grad_norm": 0.9758671382016597,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 4834
    },
    {
      "epoch": 0.04835,
      "grad_norm": 0.9485480338404471,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 4835
    },
    {
      "epoch": 0.04836,
      "grad_norm": 1.0708614312540052,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4836
    },
    {
      "epoch": 0.04837,
      "grad_norm": 1.005625524837446,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4837
    },
    {
      "epoch": 0.04838,
      "grad_norm": 1.142957704221482,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 4838
    },
    {
      "epoch": 0.04839,
      "grad_norm": 0.8876209726930161,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 4839
    },
    {
      "epoch": 0.0484,
      "grad_norm": 1.0000157052113117,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4840
    },
    {
      "epoch": 0.04841,
      "grad_norm": 0.8947816300534793,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 4841
    },
    {
      "epoch": 0.04842,
      "grad_norm": 1.0306766066880264,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 4842
    },
    {
      "epoch": 0.04843,
      "grad_norm": 1.1670348326264735,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 4843
    },
    {
      "epoch": 0.04844,
      "grad_norm": 0.8379009760125189,
      "learning_rate": 0.003,
      "loss": 4.1746,
      "step": 4844
    },
    {
      "epoch": 0.04845,
      "grad_norm": 0.9239064010183352,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4845
    },
    {
      "epoch": 0.04846,
      "grad_norm": 0.9084764134640881,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 4846
    },
    {
      "epoch": 0.04847,
      "grad_norm": 0.7704730780996573,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4847
    },
    {
      "epoch": 0.04848,
      "grad_norm": 0.8040182191975439,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 4848
    },
    {
      "epoch": 0.04849,
      "grad_norm": 0.9074519148966281,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 4849
    },
    {
      "epoch": 0.0485,
      "grad_norm": 1.1771056378102565,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 4850
    },
    {
      "epoch": 0.04851,
      "grad_norm": 0.9654993158924701,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 4851
    },
    {
      "epoch": 0.04852,
      "grad_norm": 0.7654220286623955,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 4852
    },
    {
      "epoch": 0.04853,
      "grad_norm": 0.7291058591496458,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4853
    },
    {
      "epoch": 0.04854,
      "grad_norm": 0.6997930550604129,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4854
    },
    {
      "epoch": 0.04855,
      "grad_norm": 0.8234663642795009,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 4855
    },
    {
      "epoch": 0.04856,
      "grad_norm": 0.7798398187689213,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 4856
    },
    {
      "epoch": 0.04857,
      "grad_norm": 0.6411382707024734,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 4857
    },
    {
      "epoch": 0.04858,
      "grad_norm": 0.6914930361794667,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4858
    },
    {
      "epoch": 0.04859,
      "grad_norm": 0.742724632667153,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 4859
    },
    {
      "epoch": 0.0486,
      "grad_norm": 0.6805430980474091,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4860
    },
    {
      "epoch": 0.04861,
      "grad_norm": 0.538868342282314,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4861
    },
    {
      "epoch": 0.04862,
      "grad_norm": 0.4415725147391537,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4862
    },
    {
      "epoch": 0.04863,
      "grad_norm": 0.46517280056665583,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 4863
    },
    {
      "epoch": 0.04864,
      "grad_norm": 0.44851841399681913,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4864
    },
    {
      "epoch": 0.04865,
      "grad_norm": 0.4755135869431016,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4865
    },
    {
      "epoch": 0.04866,
      "grad_norm": 0.449801980415272,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4866
    },
    {
      "epoch": 0.04867,
      "grad_norm": 0.41976113144986854,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 4867
    },
    {
      "epoch": 0.04868,
      "grad_norm": 0.41274576265380564,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4868
    },
    {
      "epoch": 0.04869,
      "grad_norm": 0.4120235509377836,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4869
    },
    {
      "epoch": 0.0487,
      "grad_norm": 0.45999567656300233,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 4870
    },
    {
      "epoch": 0.04871,
      "grad_norm": 0.5762534303017752,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4871
    },
    {
      "epoch": 0.04872,
      "grad_norm": 0.6778749879598042,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 4872
    },
    {
      "epoch": 0.04873,
      "grad_norm": 0.7636826849696239,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4873
    },
    {
      "epoch": 0.04874,
      "grad_norm": 0.7169955577952278,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4874
    },
    {
      "epoch": 0.04875,
      "grad_norm": 0.5747097191102749,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 4875
    },
    {
      "epoch": 0.04876,
      "grad_norm": 0.5097483392386485,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 4876
    },
    {
      "epoch": 0.04877,
      "grad_norm": 0.5291097317099629,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 4877
    },
    {
      "epoch": 0.04878,
      "grad_norm": 0.5314236989459756,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 4878
    },
    {
      "epoch": 0.04879,
      "grad_norm": 0.5936650310790298,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 4879
    },
    {
      "epoch": 0.0488,
      "grad_norm": 0.7538105145753773,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 4880
    },
    {
      "epoch": 0.04881,
      "grad_norm": 0.9419961284154782,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4881
    },
    {
      "epoch": 0.04882,
      "grad_norm": 1.125784565290687,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 4882
    },
    {
      "epoch": 0.04883,
      "grad_norm": 0.7937742064689157,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 4883
    },
    {
      "epoch": 0.04884,
      "grad_norm": 0.7276138590729601,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4884
    },
    {
      "epoch": 0.04885,
      "grad_norm": 0.7635121030741637,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4885
    },
    {
      "epoch": 0.04886,
      "grad_norm": 0.8525933214692812,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 4886
    },
    {
      "epoch": 0.04887,
      "grad_norm": 0.8979433474373131,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4887
    },
    {
      "epoch": 0.04888,
      "grad_norm": 0.9257981316717354,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4888
    },
    {
      "epoch": 0.04889,
      "grad_norm": 0.8410005218082275,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4889
    },
    {
      "epoch": 0.0489,
      "grad_norm": 0.8463344038585867,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 4890
    },
    {
      "epoch": 0.04891,
      "grad_norm": 0.7647404294541085,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4891
    },
    {
      "epoch": 0.04892,
      "grad_norm": 0.7679081300937539,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4892
    },
    {
      "epoch": 0.04893,
      "grad_norm": 0.8575058455749437,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4893
    },
    {
      "epoch": 0.04894,
      "grad_norm": 0.8645404495029662,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4894
    },
    {
      "epoch": 0.04895,
      "grad_norm": 0.8334125833446672,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4895
    },
    {
      "epoch": 0.04896,
      "grad_norm": 0.7851358889893844,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4896
    },
    {
      "epoch": 0.04897,
      "grad_norm": 0.7536500770250976,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4897
    },
    {
      "epoch": 0.04898,
      "grad_norm": 0.7125975228161553,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4898
    },
    {
      "epoch": 0.04899,
      "grad_norm": 0.6723493590903667,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 4899
    },
    {
      "epoch": 0.049,
      "grad_norm": 0.5994822751068893,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4900
    },
    {
      "epoch": 0.04901,
      "grad_norm": 0.562371614276483,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4901
    },
    {
      "epoch": 0.04902,
      "grad_norm": 0.6664145016393446,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4902
    },
    {
      "epoch": 0.04903,
      "grad_norm": 0.6848367937927413,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4903
    },
    {
      "epoch": 0.04904,
      "grad_norm": 0.6584184589527915,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4904
    },
    {
      "epoch": 0.04905,
      "grad_norm": 0.6238214939939397,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 4905
    },
    {
      "epoch": 0.04906,
      "grad_norm": 0.6975587494661175,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4906
    },
    {
      "epoch": 0.04907,
      "grad_norm": 0.8941666785365164,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 4907
    },
    {
      "epoch": 0.04908,
      "grad_norm": 1.081340571949139,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 4908
    },
    {
      "epoch": 0.04909,
      "grad_norm": 0.8257080561116473,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4909
    },
    {
      "epoch": 0.0491,
      "grad_norm": 0.6982931611612875,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4910
    },
    {
      "epoch": 0.04911,
      "grad_norm": 0.8003274963133196,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4911
    },
    {
      "epoch": 0.04912,
      "grad_norm": 0.8407576542296316,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 4912
    },
    {
      "epoch": 0.04913,
      "grad_norm": 0.9633535310499528,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 4913
    },
    {
      "epoch": 0.04914,
      "grad_norm": 0.9685795549695555,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4914
    },
    {
      "epoch": 0.04915,
      "grad_norm": 0.8915068602810806,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4915
    },
    {
      "epoch": 0.04916,
      "grad_norm": 0.7915988128949956,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 4916
    },
    {
      "epoch": 0.04917,
      "grad_norm": 0.7634607572088334,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 4917
    },
    {
      "epoch": 0.04918,
      "grad_norm": 0.7786335105527836,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4918
    },
    {
      "epoch": 0.04919,
      "grad_norm": 0.8773870158083756,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4919
    },
    {
      "epoch": 0.0492,
      "grad_norm": 0.7845152066567127,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4920
    },
    {
      "epoch": 0.04921,
      "grad_norm": 0.8250729750583956,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 4921
    },
    {
      "epoch": 0.04922,
      "grad_norm": 0.8194188230604832,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4922
    },
    {
      "epoch": 0.04923,
      "grad_norm": 0.8678442267593626,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4923
    },
    {
      "epoch": 0.04924,
      "grad_norm": 0.9322111328758901,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 4924
    },
    {
      "epoch": 0.04925,
      "grad_norm": 0.883656356593312,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4925
    },
    {
      "epoch": 0.04926,
      "grad_norm": 0.7834220179620971,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4926
    },
    {
      "epoch": 0.04927,
      "grad_norm": 0.648822263811632,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4927
    },
    {
      "epoch": 0.04928,
      "grad_norm": 0.6606153717082371,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 4928
    },
    {
      "epoch": 0.04929,
      "grad_norm": 0.6448549657492914,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4929
    },
    {
      "epoch": 0.0493,
      "grad_norm": 0.6537446324241543,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4930
    },
    {
      "epoch": 0.04931,
      "grad_norm": 0.6782203397244866,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 4931
    },
    {
      "epoch": 0.04932,
      "grad_norm": 0.7378598466407571,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4932
    },
    {
      "epoch": 0.04933,
      "grad_norm": 0.7395809589970013,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4933
    },
    {
      "epoch": 0.04934,
      "grad_norm": 0.7658041842655043,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 4934
    },
    {
      "epoch": 0.04935,
      "grad_norm": 0.9019517516492989,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 4935
    },
    {
      "epoch": 0.04936,
      "grad_norm": 1.1441200899996462,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4936
    },
    {
      "epoch": 0.04937,
      "grad_norm": 0.8788107449320854,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4937
    },
    {
      "epoch": 0.04938,
      "grad_norm": 0.8007894624899063,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 4938
    },
    {
      "epoch": 0.04939,
      "grad_norm": 0.789503524635859,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 4939
    },
    {
      "epoch": 0.0494,
      "grad_norm": 0.787654716377685,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4940
    },
    {
      "epoch": 0.04941,
      "grad_norm": 0.6688134872012756,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4941
    },
    {
      "epoch": 0.04942,
      "grad_norm": 0.5978385156801125,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4942
    },
    {
      "epoch": 0.04943,
      "grad_norm": 0.6124698730462602,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4943
    },
    {
      "epoch": 0.04944,
      "grad_norm": 0.5591922471734772,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 4944
    },
    {
      "epoch": 0.04945,
      "grad_norm": 0.4939591571209979,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4945
    },
    {
      "epoch": 0.04946,
      "grad_norm": 0.49333816406820896,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 4946
    },
    {
      "epoch": 0.04947,
      "grad_norm": 0.5127970104844136,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 4947
    },
    {
      "epoch": 0.04948,
      "grad_norm": 0.5804249788539954,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4948
    },
    {
      "epoch": 0.04949,
      "grad_norm": 0.6106490006275131,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4949
    },
    {
      "epoch": 0.0495,
      "grad_norm": 0.7326284913442733,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 4950
    },
    {
      "epoch": 0.04951,
      "grad_norm": 0.8806738854393903,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 4951
    },
    {
      "epoch": 0.04952,
      "grad_norm": 0.8907989929621909,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4952
    },
    {
      "epoch": 0.04953,
      "grad_norm": 0.6504370122838272,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 4953
    },
    {
      "epoch": 0.04954,
      "grad_norm": 0.5399865076183303,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 4954
    },
    {
      "epoch": 0.04955,
      "grad_norm": 0.6426116557249149,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 4955
    },
    {
      "epoch": 0.04956,
      "grad_norm": 0.6565119107623256,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4956
    },
    {
      "epoch": 0.04957,
      "grad_norm": 0.6722789026797903,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 4957
    },
    {
      "epoch": 0.04958,
      "grad_norm": 0.5890068452063758,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4958
    },
    {
      "epoch": 0.04959,
      "grad_norm": 0.5878861479683556,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4959
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.6509946110169393,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 4960
    },
    {
      "epoch": 0.04961,
      "grad_norm": 0.6156850509647105,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 4961
    },
    {
      "epoch": 0.04962,
      "grad_norm": 0.6497590683615507,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 4962
    },
    {
      "epoch": 0.04963,
      "grad_norm": 0.6412215380917496,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4963
    },
    {
      "epoch": 0.04964,
      "grad_norm": 0.6061448543738891,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4964
    },
    {
      "epoch": 0.04965,
      "grad_norm": 0.6123953168115248,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4965
    },
    {
      "epoch": 0.04966,
      "grad_norm": 0.8037505368816966,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 4966
    },
    {
      "epoch": 0.04967,
      "grad_norm": 0.9549949754561539,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 4967
    },
    {
      "epoch": 0.04968,
      "grad_norm": 1.1281595822055885,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4968
    },
    {
      "epoch": 0.04969,
      "grad_norm": 0.8652927562240527,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4969
    },
    {
      "epoch": 0.0497,
      "grad_norm": 0.7881031556981761,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4970
    },
    {
      "epoch": 0.04971,
      "grad_norm": 0.8417863445996929,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 4971
    },
    {
      "epoch": 0.04972,
      "grad_norm": 0.8493800636053089,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 4972
    },
    {
      "epoch": 0.04973,
      "grad_norm": 0.9639644128846697,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 4973
    },
    {
      "epoch": 0.04974,
      "grad_norm": 1.0979045911236074,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 4974
    },
    {
      "epoch": 0.04975,
      "grad_norm": 1.1499829958321686,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4975
    },
    {
      "epoch": 0.04976,
      "grad_norm": 0.896257641249768,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4976
    },
    {
      "epoch": 0.04977,
      "grad_norm": 0.8176607909447694,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 4977
    },
    {
      "epoch": 0.04978,
      "grad_norm": 0.8167430782012225,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4978
    },
    {
      "epoch": 0.04979,
      "grad_norm": 0.8298370431983914,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4979
    },
    {
      "epoch": 0.0498,
      "grad_norm": 0.8918442192162982,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 4980
    },
    {
      "epoch": 0.04981,
      "grad_norm": 0.8954883580668984,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 4981
    },
    {
      "epoch": 0.04982,
      "grad_norm": 0.8872182858380784,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 4982
    },
    {
      "epoch": 0.04983,
      "grad_norm": 0.7800382368496617,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 4983
    },
    {
      "epoch": 0.04984,
      "grad_norm": 0.7656679202701089,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 4984
    },
    {
      "epoch": 0.04985,
      "grad_norm": 0.8692356760179402,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4985
    },
    {
      "epoch": 0.04986,
      "grad_norm": 0.8658640091365618,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 4986
    },
    {
      "epoch": 0.04987,
      "grad_norm": 0.9418717422739423,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4987
    },
    {
      "epoch": 0.04988,
      "grad_norm": 0.9193482062490026,
      "learning_rate": 0.003,
      "loss": 4.1628,
      "step": 4988
    },
    {
      "epoch": 0.04989,
      "grad_norm": 0.8335538614629532,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 4989
    },
    {
      "epoch": 0.0499,
      "grad_norm": 0.7458856658289911,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4990
    },
    {
      "epoch": 0.04991,
      "grad_norm": 0.6549431595291397,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4991
    },
    {
      "epoch": 0.04992,
      "grad_norm": 0.6708966834866974,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 4992
    },
    {
      "epoch": 0.04993,
      "grad_norm": 0.6367637070694386,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 4993
    },
    {
      "epoch": 0.04994,
      "grad_norm": 0.6415299490019786,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4994
    },
    {
      "epoch": 0.04995,
      "grad_norm": 0.7355718950683062,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4995
    },
    {
      "epoch": 0.04996,
      "grad_norm": 0.8795955713743572,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 4996
    },
    {
      "epoch": 0.04997,
      "grad_norm": 1.1069671993764165,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4997
    },
    {
      "epoch": 0.04998,
      "grad_norm": 0.9565748099055139,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4998
    },
    {
      "epoch": 0.04999,
      "grad_norm": 0.8361592261029797,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 4999
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9200109034528712,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 5000
    },
    {
      "epoch": 0.05001,
      "grad_norm": 0.9617359528015215,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 5001
    },
    {
      "epoch": 0.05002,
      "grad_norm": 0.8052353535299214,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5002
    },
    {
      "epoch": 0.05003,
      "grad_norm": 0.7285922794371577,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 5003
    },
    {
      "epoch": 0.05004,
      "grad_norm": 0.6735195112013481,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5004
    },
    {
      "epoch": 0.05005,
      "grad_norm": 0.6049497214440789,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 5005
    },
    {
      "epoch": 0.05006,
      "grad_norm": 0.6066071944555775,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5006
    },
    {
      "epoch": 0.05007,
      "grad_norm": 0.558024137943532,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 5007
    },
    {
      "epoch": 0.05008,
      "grad_norm": 0.5395388306678914,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 5008
    },
    {
      "epoch": 0.05009,
      "grad_norm": 0.5556098595038315,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5009
    },
    {
      "epoch": 0.0501,
      "grad_norm": 0.6653438296094251,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 5010
    },
    {
      "epoch": 0.05011,
      "grad_norm": 0.77449202414198,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 5011
    },
    {
      "epoch": 0.05012,
      "grad_norm": 0.7658330704952865,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5012
    },
    {
      "epoch": 0.05013,
      "grad_norm": 0.9103652169172048,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 5013
    },
    {
      "epoch": 0.05014,
      "grad_norm": 1.0759287643443909,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5014
    },
    {
      "epoch": 0.05015,
      "grad_norm": 0.9691643650194389,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 5015
    },
    {
      "epoch": 0.05016,
      "grad_norm": 0.9073305570987539,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 5016
    },
    {
      "epoch": 0.05017,
      "grad_norm": 0.8073149334354928,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 5017
    },
    {
      "epoch": 0.05018,
      "grad_norm": 0.7017737986112383,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 5018
    },
    {
      "epoch": 0.05019,
      "grad_norm": 0.6777526168401347,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 5019
    },
    {
      "epoch": 0.0502,
      "grad_norm": 0.7096093854788539,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5020
    },
    {
      "epoch": 0.05021,
      "grad_norm": 0.6174117502687014,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5021
    },
    {
      "epoch": 0.05022,
      "grad_norm": 0.6289052553001543,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 5022
    },
    {
      "epoch": 0.05023,
      "grad_norm": 0.6439915077819784,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 5023
    },
    {
      "epoch": 0.05024,
      "grad_norm": 0.6395721391541763,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 5024
    },
    {
      "epoch": 0.05025,
      "grad_norm": 0.6213156408060325,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5025
    },
    {
      "epoch": 0.05026,
      "grad_norm": 0.6019908233874888,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5026
    },
    {
      "epoch": 0.05027,
      "grad_norm": 0.7041501406597449,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 5027
    },
    {
      "epoch": 0.05028,
      "grad_norm": 0.8719881236969269,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5028
    },
    {
      "epoch": 0.05029,
      "grad_norm": 0.9033846221678056,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5029
    },
    {
      "epoch": 0.0503,
      "grad_norm": 0.8943098923163394,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 5030
    },
    {
      "epoch": 0.05031,
      "grad_norm": 0.8280905971408706,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5031
    },
    {
      "epoch": 0.05032,
      "grad_norm": 0.8573742108993896,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 5032
    },
    {
      "epoch": 0.05033,
      "grad_norm": 0.8521216692297204,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5033
    },
    {
      "epoch": 0.05034,
      "grad_norm": 0.8336742938862856,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5034
    },
    {
      "epoch": 0.05035,
      "grad_norm": 0.7155132757115704,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5035
    },
    {
      "epoch": 0.05036,
      "grad_norm": 0.7077255442757248,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 5036
    },
    {
      "epoch": 0.05037,
      "grad_norm": 0.6598592587060891,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 5037
    },
    {
      "epoch": 0.05038,
      "grad_norm": 0.6985786753992844,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 5038
    },
    {
      "epoch": 0.05039,
      "grad_norm": 0.7840225900778709,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 5039
    },
    {
      "epoch": 0.0504,
      "grad_norm": 0.8187206569784481,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5040
    },
    {
      "epoch": 0.05041,
      "grad_norm": 0.8117416189384354,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5041
    },
    {
      "epoch": 0.05042,
      "grad_norm": 0.701867915130458,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5042
    },
    {
      "epoch": 0.05043,
      "grad_norm": 0.6612953236447207,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 5043
    },
    {
      "epoch": 0.05044,
      "grad_norm": 0.6173987969701078,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 5044
    },
    {
      "epoch": 0.05045,
      "grad_norm": 0.5892878279295536,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 5045
    },
    {
      "epoch": 0.05046,
      "grad_norm": 0.6842586991400519,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 5046
    },
    {
      "epoch": 0.05047,
      "grad_norm": 0.7460792551860184,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5047
    },
    {
      "epoch": 0.05048,
      "grad_norm": 0.7111178192864371,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 5048
    },
    {
      "epoch": 0.05049,
      "grad_norm": 0.646577206489084,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 5049
    },
    {
      "epoch": 0.0505,
      "grad_norm": 0.7133657320392227,
      "learning_rate": 0.003,
      "loss": 4.1772,
      "step": 5050
    },
    {
      "epoch": 0.05051,
      "grad_norm": 0.8879083064557955,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5051
    },
    {
      "epoch": 0.05052,
      "grad_norm": 1.108020136402945,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 5052
    },
    {
      "epoch": 0.05053,
      "grad_norm": 1.124376607679852,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 5053
    },
    {
      "epoch": 0.05054,
      "grad_norm": 0.8613572741956879,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 5054
    },
    {
      "epoch": 0.05055,
      "grad_norm": 0.9007503555201047,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 5055
    },
    {
      "epoch": 0.05056,
      "grad_norm": 0.9020326994115863,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5056
    },
    {
      "epoch": 0.05057,
      "grad_norm": 0.9194629842951593,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 5057
    },
    {
      "epoch": 0.05058,
      "grad_norm": 0.8058592428411899,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 5058
    },
    {
      "epoch": 0.05059,
      "grad_norm": 0.705098811845147,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 5059
    },
    {
      "epoch": 0.0506,
      "grad_norm": 0.6716192520354848,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 5060
    },
    {
      "epoch": 0.05061,
      "grad_norm": 0.7122976459366105,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 5061
    },
    {
      "epoch": 0.05062,
      "grad_norm": 0.8458397806323464,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5062
    },
    {
      "epoch": 0.05063,
      "grad_norm": 0.8509332439877665,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 5063
    },
    {
      "epoch": 0.05064,
      "grad_norm": 0.8628499996665681,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 5064
    },
    {
      "epoch": 0.05065,
      "grad_norm": 0.7651349441929993,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5065
    },
    {
      "epoch": 0.05066,
      "grad_norm": 0.6441373019881442,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 5066
    },
    {
      "epoch": 0.05067,
      "grad_norm": 0.6162135983622039,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5067
    },
    {
      "epoch": 0.05068,
      "grad_norm": 0.5839874533063438,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 5068
    },
    {
      "epoch": 0.05069,
      "grad_norm": 0.602540955663345,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5069
    },
    {
      "epoch": 0.0507,
      "grad_norm": 0.6829292685811817,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5070
    },
    {
      "epoch": 0.05071,
      "grad_norm": 0.9040694995935689,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 5071
    },
    {
      "epoch": 0.05072,
      "grad_norm": 1.1363622378094762,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 5072
    },
    {
      "epoch": 0.05073,
      "grad_norm": 0.6884937706217313,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 5073
    },
    {
      "epoch": 0.05074,
      "grad_norm": 0.6473518308046033,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 5074
    },
    {
      "epoch": 0.05075,
      "grad_norm": 0.7765044046650167,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5075
    },
    {
      "epoch": 0.05076,
      "grad_norm": 0.8329788470025239,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 5076
    },
    {
      "epoch": 0.05077,
      "grad_norm": 0.799881313047519,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 5077
    },
    {
      "epoch": 0.05078,
      "grad_norm": 0.7389949572586678,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5078
    },
    {
      "epoch": 0.05079,
      "grad_norm": 0.7330687471577605,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 5079
    },
    {
      "epoch": 0.0508,
      "grad_norm": 0.7072208630775181,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 5080
    },
    {
      "epoch": 0.05081,
      "grad_norm": 0.698914398304387,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 5081
    },
    {
      "epoch": 0.05082,
      "grad_norm": 0.6278523534587991,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 5082
    },
    {
      "epoch": 0.05083,
      "grad_norm": 0.5773478170755377,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 5083
    },
    {
      "epoch": 0.05084,
      "grad_norm": 0.4995734535669917,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 5084
    },
    {
      "epoch": 0.05085,
      "grad_norm": 0.503192190879057,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 5085
    },
    {
      "epoch": 0.05086,
      "grad_norm": 0.4849750033335867,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 5086
    },
    {
      "epoch": 0.05087,
      "grad_norm": 0.5401918808221647,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 5087
    },
    {
      "epoch": 0.05088,
      "grad_norm": 0.5431450812753981,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5088
    },
    {
      "epoch": 0.05089,
      "grad_norm": 0.5955708352853926,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5089
    },
    {
      "epoch": 0.0509,
      "grad_norm": 0.6944142097249647,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5090
    },
    {
      "epoch": 0.05091,
      "grad_norm": 0.7093523328553718,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 5091
    },
    {
      "epoch": 0.05092,
      "grad_norm": 0.711564968018302,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 5092
    },
    {
      "epoch": 0.05093,
      "grad_norm": 0.7293388159366392,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 5093
    },
    {
      "epoch": 0.05094,
      "grad_norm": 0.7789366794435865,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 5094
    },
    {
      "epoch": 0.05095,
      "grad_norm": 0.8943150130903914,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5095
    },
    {
      "epoch": 0.05096,
      "grad_norm": 1.2235914801717511,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5096
    },
    {
      "epoch": 0.05097,
      "grad_norm": 0.8627777574244729,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 5097
    },
    {
      "epoch": 0.05098,
      "grad_norm": 0.7848585017002957,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5098
    },
    {
      "epoch": 0.05099,
      "grad_norm": 0.7684212373548842,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 5099
    },
    {
      "epoch": 0.051,
      "grad_norm": 0.7191076738646075,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 5100
    },
    {
      "epoch": 0.05101,
      "grad_norm": 0.7450408749289968,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 5101
    },
    {
      "epoch": 0.05102,
      "grad_norm": 0.7788766932633848,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 5102
    },
    {
      "epoch": 0.05103,
      "grad_norm": 0.7398214547540392,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 5103
    },
    {
      "epoch": 0.05104,
      "grad_norm": 0.6558233797113744,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 5104
    },
    {
      "epoch": 0.05105,
      "grad_norm": 0.6381200092203351,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 5105
    },
    {
      "epoch": 0.05106,
      "grad_norm": 0.6828637852428295,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 5106
    },
    {
      "epoch": 0.05107,
      "grad_norm": 0.8597716654991152,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5107
    },
    {
      "epoch": 0.05108,
      "grad_norm": 1.2317482679596312,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 5108
    },
    {
      "epoch": 0.05109,
      "grad_norm": 0.8665302105620079,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5109
    },
    {
      "epoch": 0.0511,
      "grad_norm": 0.7969460956461537,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 5110
    },
    {
      "epoch": 0.05111,
      "grad_norm": 0.8007840700796149,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 5111
    },
    {
      "epoch": 0.05112,
      "grad_norm": 0.7681571978199957,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5112
    },
    {
      "epoch": 0.05113,
      "grad_norm": 0.6903861079813164,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 5113
    },
    {
      "epoch": 0.05114,
      "grad_norm": 0.7692348540414872,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 5114
    },
    {
      "epoch": 0.05115,
      "grad_norm": 0.7538155044071809,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 5115
    },
    {
      "epoch": 0.05116,
      "grad_norm": 0.7449617507426948,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 5116
    },
    {
      "epoch": 0.05117,
      "grad_norm": 0.8510379134985516,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5117
    },
    {
      "epoch": 0.05118,
      "grad_norm": 1.005098903517839,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 5118
    },
    {
      "epoch": 0.05119,
      "grad_norm": 1.0115097346662045,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5119
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.9811018804142781,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5120
    },
    {
      "epoch": 0.05121,
      "grad_norm": 0.9389084532957305,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5121
    },
    {
      "epoch": 0.05122,
      "grad_norm": 0.9691145521379205,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 5122
    },
    {
      "epoch": 0.05123,
      "grad_norm": 1.0049713917715095,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 5123
    },
    {
      "epoch": 0.05124,
      "grad_norm": 1.0419381303567035,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 5124
    },
    {
      "epoch": 0.05125,
      "grad_norm": 1.0446044411819475,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 5125
    },
    {
      "epoch": 0.05126,
      "grad_norm": 0.9019262378070061,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5126
    },
    {
      "epoch": 0.05127,
      "grad_norm": 0.8105987689505292,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 5127
    },
    {
      "epoch": 0.05128,
      "grad_norm": 0.7905970287718124,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 5128
    },
    {
      "epoch": 0.05129,
      "grad_norm": 0.7417838905848333,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5129
    },
    {
      "epoch": 0.0513,
      "grad_norm": 0.7900814171820785,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 5130
    },
    {
      "epoch": 0.05131,
      "grad_norm": 0.9133150183860361,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 5131
    },
    {
      "epoch": 0.05132,
      "grad_norm": 1.095265924476809,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 5132
    },
    {
      "epoch": 0.05133,
      "grad_norm": 0.9471945091609965,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 5133
    },
    {
      "epoch": 0.05134,
      "grad_norm": 0.9244905428462661,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 5134
    },
    {
      "epoch": 0.05135,
      "grad_norm": 0.8334274421733712,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5135
    },
    {
      "epoch": 0.05136,
      "grad_norm": 0.709427559475487,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5136
    },
    {
      "epoch": 0.05137,
      "grad_norm": 0.6442219169362268,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 5137
    },
    {
      "epoch": 0.05138,
      "grad_norm": 0.6467988163207677,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5138
    },
    {
      "epoch": 0.05139,
      "grad_norm": 0.733478794767403,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5139
    },
    {
      "epoch": 0.0514,
      "grad_norm": 0.7297546736273143,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5140
    },
    {
      "epoch": 0.05141,
      "grad_norm": 0.6341213802128426,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 5141
    },
    {
      "epoch": 0.05142,
      "grad_norm": 0.5776293822275287,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5142
    },
    {
      "epoch": 0.05143,
      "grad_norm": 0.5773363793450792,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5143
    },
    {
      "epoch": 0.05144,
      "grad_norm": 0.5526645311406644,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 5144
    },
    {
      "epoch": 0.05145,
      "grad_norm": 0.4890054978586333,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 5145
    },
    {
      "epoch": 0.05146,
      "grad_norm": 0.4190024922610593,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 5146
    },
    {
      "epoch": 0.05147,
      "grad_norm": 0.44043856065420567,
      "learning_rate": 0.003,
      "loss": 4.1214,
      "step": 5147
    },
    {
      "epoch": 0.05148,
      "grad_norm": 0.45582477546885664,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 5148
    },
    {
      "epoch": 0.05149,
      "grad_norm": 0.5300055959379171,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 5149
    },
    {
      "epoch": 0.0515,
      "grad_norm": 0.5337764008665007,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5150
    },
    {
      "epoch": 0.05151,
      "grad_norm": 0.4777932939886135,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 5151
    },
    {
      "epoch": 0.05152,
      "grad_norm": 0.4822263308806615,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 5152
    },
    {
      "epoch": 0.05153,
      "grad_norm": 0.5781114808382272,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 5153
    },
    {
      "epoch": 0.05154,
      "grad_norm": 0.7109250842959455,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 5154
    },
    {
      "epoch": 0.05155,
      "grad_norm": 0.9418503203747088,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5155
    },
    {
      "epoch": 0.05156,
      "grad_norm": 1.2647418011213238,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 5156
    },
    {
      "epoch": 0.05157,
      "grad_norm": 0.7308469531008109,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 5157
    },
    {
      "epoch": 0.05158,
      "grad_norm": 0.7875229543444077,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5158
    },
    {
      "epoch": 0.05159,
      "grad_norm": 0.8140638394753378,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 5159
    },
    {
      "epoch": 0.0516,
      "grad_norm": 0.8244642815731201,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5160
    },
    {
      "epoch": 0.05161,
      "grad_norm": 0.8540456002198428,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 5161
    },
    {
      "epoch": 0.05162,
      "grad_norm": 0.8746074474591011,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5162
    },
    {
      "epoch": 0.05163,
      "grad_norm": 0.8737342817314073,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 5163
    },
    {
      "epoch": 0.05164,
      "grad_norm": 1.0627737585477899,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5164
    },
    {
      "epoch": 0.05165,
      "grad_norm": 0.9694679199622456,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 5165
    },
    {
      "epoch": 0.05166,
      "grad_norm": 0.9344677027950545,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5166
    },
    {
      "epoch": 0.05167,
      "grad_norm": 0.7891486782935601,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 5167
    },
    {
      "epoch": 0.05168,
      "grad_norm": 0.7729700436459497,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 5168
    },
    {
      "epoch": 0.05169,
      "grad_norm": 0.7768744996829114,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5169
    },
    {
      "epoch": 0.0517,
      "grad_norm": 0.9172514868358831,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 5170
    },
    {
      "epoch": 0.05171,
      "grad_norm": 0.987191990102733,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5171
    },
    {
      "epoch": 0.05172,
      "grad_norm": 1.1074558271172996,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 5172
    },
    {
      "epoch": 0.05173,
      "grad_norm": 0.9112090797136994,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 5173
    },
    {
      "epoch": 0.05174,
      "grad_norm": 0.8508573888938128,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 5174
    },
    {
      "epoch": 0.05175,
      "grad_norm": 0.9226457815827247,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 5175
    },
    {
      "epoch": 0.05176,
      "grad_norm": 0.9401989711162229,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5176
    },
    {
      "epoch": 0.05177,
      "grad_norm": 1.1169814516735364,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5177
    },
    {
      "epoch": 0.05178,
      "grad_norm": 0.8757415219342446,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5178
    },
    {
      "epoch": 0.05179,
      "grad_norm": 0.8953548399429396,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 5179
    },
    {
      "epoch": 0.0518,
      "grad_norm": 0.8270061517274871,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 5180
    },
    {
      "epoch": 0.05181,
      "grad_norm": 0.7996635692475278,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 5181
    },
    {
      "epoch": 0.05182,
      "grad_norm": 0.788819919508774,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 5182
    },
    {
      "epoch": 0.05183,
      "grad_norm": 0.7520155289113659,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5183
    },
    {
      "epoch": 0.05184,
      "grad_norm": 0.7236732746268649,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 5184
    },
    {
      "epoch": 0.05185,
      "grad_norm": 0.638115349783052,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 5185
    },
    {
      "epoch": 0.05186,
      "grad_norm": 0.5723057160378794,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 5186
    },
    {
      "epoch": 0.05187,
      "grad_norm": 0.6462132298250771,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 5187
    },
    {
      "epoch": 0.05188,
      "grad_norm": 0.7054920468123554,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 5188
    },
    {
      "epoch": 0.05189,
      "grad_norm": 0.8179909749537473,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 5189
    },
    {
      "epoch": 0.0519,
      "grad_norm": 0.9666616633136137,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 5190
    },
    {
      "epoch": 0.05191,
      "grad_norm": 0.8948010215701201,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 5191
    },
    {
      "epoch": 0.05192,
      "grad_norm": 0.7272065263730562,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 5192
    },
    {
      "epoch": 0.05193,
      "grad_norm": 0.7326420198310921,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5193
    },
    {
      "epoch": 0.05194,
      "grad_norm": 0.7400163904092779,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 5194
    },
    {
      "epoch": 0.05195,
      "grad_norm": 0.6229710454294611,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 5195
    },
    {
      "epoch": 0.05196,
      "grad_norm": 0.5732082487024617,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 5196
    },
    {
      "epoch": 0.05197,
      "grad_norm": 0.5437048621681219,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5197
    },
    {
      "epoch": 0.05198,
      "grad_norm": 0.5106768440457569,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 5198
    },
    {
      "epoch": 0.05199,
      "grad_norm": 0.46246539222179384,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 5199
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.48176692281015543,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5200
    },
    {
      "epoch": 0.05201,
      "grad_norm": 0.4828183154586405,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 5201
    },
    {
      "epoch": 0.05202,
      "grad_norm": 0.5257398649757856,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 5202
    },
    {
      "epoch": 0.05203,
      "grad_norm": 0.5163132111756201,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 5203
    },
    {
      "epoch": 0.05204,
      "grad_norm": 0.47004258100640817,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5204
    },
    {
      "epoch": 0.05205,
      "grad_norm": 0.5080321067769532,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5205
    },
    {
      "epoch": 0.05206,
      "grad_norm": 0.6713429412709093,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5206
    },
    {
      "epoch": 0.05207,
      "grad_norm": 0.9187866373124713,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 5207
    },
    {
      "epoch": 0.05208,
      "grad_norm": 1.2677728854398744,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5208
    },
    {
      "epoch": 0.05209,
      "grad_norm": 0.630859673605965,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 5209
    },
    {
      "epoch": 0.0521,
      "grad_norm": 0.7350336486110594,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 5210
    },
    {
      "epoch": 0.05211,
      "grad_norm": 0.8202119917312024,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 5211
    },
    {
      "epoch": 0.05212,
      "grad_norm": 0.6430476198919388,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 5212
    },
    {
      "epoch": 0.05213,
      "grad_norm": 0.6006713399050833,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5213
    },
    {
      "epoch": 0.05214,
      "grad_norm": 0.6412056440427675,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 5214
    },
    {
      "epoch": 0.05215,
      "grad_norm": 0.7406745452916796,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 5215
    },
    {
      "epoch": 0.05216,
      "grad_norm": 0.7338642035681033,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5216
    },
    {
      "epoch": 0.05217,
      "grad_norm": 0.7548188627814898,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5217
    },
    {
      "epoch": 0.05218,
      "grad_norm": 0.7790433320975734,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 5218
    },
    {
      "epoch": 0.05219,
      "grad_norm": 0.921132206986461,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5219
    },
    {
      "epoch": 0.0522,
      "grad_norm": 1.0167153212263549,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 5220
    },
    {
      "epoch": 0.05221,
      "grad_norm": 1.0121309018306264,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5221
    },
    {
      "epoch": 0.05222,
      "grad_norm": 0.907628952065032,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 5222
    },
    {
      "epoch": 0.05223,
      "grad_norm": 0.919506193257904,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 5223
    },
    {
      "epoch": 0.05224,
      "grad_norm": 1.1506026821508868,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 5224
    },
    {
      "epoch": 0.05225,
      "grad_norm": 0.899823660266816,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 5225
    },
    {
      "epoch": 0.05226,
      "grad_norm": 0.8527078083046193,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 5226
    },
    {
      "epoch": 0.05227,
      "grad_norm": 0.7079246341060081,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 5227
    },
    {
      "epoch": 0.05228,
      "grad_norm": 0.8044831433064553,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 5228
    },
    {
      "epoch": 0.05229,
      "grad_norm": 0.8720881208580172,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 5229
    },
    {
      "epoch": 0.0523,
      "grad_norm": 0.8795412540846314,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 5230
    },
    {
      "epoch": 0.05231,
      "grad_norm": 0.7917348425475053,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 5231
    },
    {
      "epoch": 0.05232,
      "grad_norm": 0.752556657478349,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 5232
    },
    {
      "epoch": 0.05233,
      "grad_norm": 0.6869072276302872,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5233
    },
    {
      "epoch": 0.05234,
      "grad_norm": 0.7324907875751323,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 5234
    },
    {
      "epoch": 0.05235,
      "grad_norm": 0.721046302164716,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 5235
    },
    {
      "epoch": 0.05236,
      "grad_norm": 0.575714162268126,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 5236
    },
    {
      "epoch": 0.05237,
      "grad_norm": 0.516817288215746,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 5237
    },
    {
      "epoch": 0.05238,
      "grad_norm": 0.6049328985205596,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 5238
    },
    {
      "epoch": 0.05239,
      "grad_norm": 0.645411189054774,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 5239
    },
    {
      "epoch": 0.0524,
      "grad_norm": 0.7516501855794593,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5240
    },
    {
      "epoch": 0.05241,
      "grad_norm": 0.8580499313970857,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 5241
    },
    {
      "epoch": 0.05242,
      "grad_norm": 0.8683371010488704,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 5242
    },
    {
      "epoch": 0.05243,
      "grad_norm": 1.0371281140802466,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 5243
    },
    {
      "epoch": 0.05244,
      "grad_norm": 1.0230759917025969,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5244
    },
    {
      "epoch": 0.05245,
      "grad_norm": 0.8780400373348358,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 5245
    },
    {
      "epoch": 0.05246,
      "grad_norm": 0.8613363719642216,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5246
    },
    {
      "epoch": 0.05247,
      "grad_norm": 0.7710466126666472,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5247
    },
    {
      "epoch": 0.05248,
      "grad_norm": 0.7486554824013956,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5248
    },
    {
      "epoch": 0.05249,
      "grad_norm": 0.7704795977105405,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 5249
    },
    {
      "epoch": 0.0525,
      "grad_norm": 0.7893461765166555,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 5250
    },
    {
      "epoch": 0.05251,
      "grad_norm": 0.7489202376303653,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 5251
    },
    {
      "epoch": 0.05252,
      "grad_norm": 0.6609520253229195,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 5252
    },
    {
      "epoch": 0.05253,
      "grad_norm": 0.719719165362111,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 5253
    },
    {
      "epoch": 0.05254,
      "grad_norm": 0.6856836855692663,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5254
    },
    {
      "epoch": 0.05255,
      "grad_norm": 0.6331973623044866,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 5255
    },
    {
      "epoch": 0.05256,
      "grad_norm": 0.6209166464347697,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5256
    },
    {
      "epoch": 0.05257,
      "grad_norm": 0.6703957390397629,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5257
    },
    {
      "epoch": 0.05258,
      "grad_norm": 0.8005691943001225,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5258
    },
    {
      "epoch": 0.05259,
      "grad_norm": 1.178008398076746,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5259
    },
    {
      "epoch": 0.0526,
      "grad_norm": 0.9009691862241965,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 5260
    },
    {
      "epoch": 0.05261,
      "grad_norm": 0.636909529940799,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 5261
    },
    {
      "epoch": 0.05262,
      "grad_norm": 0.7583190817879178,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 5262
    },
    {
      "epoch": 0.05263,
      "grad_norm": 0.7940859742603288,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 5263
    },
    {
      "epoch": 0.05264,
      "grad_norm": 0.7387297524848515,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 5264
    },
    {
      "epoch": 0.05265,
      "grad_norm": 0.6381909927316334,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 5265
    },
    {
      "epoch": 0.05266,
      "grad_norm": 0.6211413466674444,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 5266
    },
    {
      "epoch": 0.05267,
      "grad_norm": 0.6304792137219847,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 5267
    },
    {
      "epoch": 0.05268,
      "grad_norm": 0.7489765141892385,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 5268
    },
    {
      "epoch": 0.05269,
      "grad_norm": 0.7861495121827866,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 5269
    },
    {
      "epoch": 0.0527,
      "grad_norm": 0.6974038599114732,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 5270
    },
    {
      "epoch": 0.05271,
      "grad_norm": 0.664530980238474,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 5271
    },
    {
      "epoch": 0.05272,
      "grad_norm": 0.7520422410793401,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5272
    },
    {
      "epoch": 0.05273,
      "grad_norm": 0.7793196930797887,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 5273
    },
    {
      "epoch": 0.05274,
      "grad_norm": 0.754818417244099,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 5274
    },
    {
      "epoch": 0.05275,
      "grad_norm": 0.6242567287914901,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 5275
    },
    {
      "epoch": 0.05276,
      "grad_norm": 0.6899528691289879,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 5276
    },
    {
      "epoch": 0.05277,
      "grad_norm": 0.7574326268248618,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 5277
    },
    {
      "epoch": 0.05278,
      "grad_norm": 0.8507468220046873,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5278
    },
    {
      "epoch": 0.05279,
      "grad_norm": 1.0353055112068175,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5279
    },
    {
      "epoch": 0.0528,
      "grad_norm": 0.9033296737516481,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5280
    },
    {
      "epoch": 0.05281,
      "grad_norm": 0.7963839547721746,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5281
    },
    {
      "epoch": 0.05282,
      "grad_norm": 0.7877787577000847,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 5282
    },
    {
      "epoch": 0.05283,
      "grad_norm": 0.7846974861191656,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 5283
    },
    {
      "epoch": 0.05284,
      "grad_norm": 0.7458279107143788,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 5284
    },
    {
      "epoch": 0.05285,
      "grad_norm": 0.7229333553728315,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 5285
    },
    {
      "epoch": 0.05286,
      "grad_norm": 0.7184374854077129,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 5286
    },
    {
      "epoch": 0.05287,
      "grad_norm": 0.8297178227719375,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5287
    },
    {
      "epoch": 0.05288,
      "grad_norm": 0.7621123902466413,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5288
    },
    {
      "epoch": 0.05289,
      "grad_norm": 0.6339905585597592,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 5289
    },
    {
      "epoch": 0.0529,
      "grad_norm": 0.6327911322808994,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5290
    },
    {
      "epoch": 0.05291,
      "grad_norm": 0.6026020302477917,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 5291
    },
    {
      "epoch": 0.05292,
      "grad_norm": 0.6437340105854593,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5292
    },
    {
      "epoch": 0.05293,
      "grad_norm": 0.6314109214308004,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5293
    },
    {
      "epoch": 0.05294,
      "grad_norm": 0.695308432779079,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 5294
    },
    {
      "epoch": 0.05295,
      "grad_norm": 0.7787212860512488,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5295
    },
    {
      "epoch": 0.05296,
      "grad_norm": 0.8622976689690274,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 5296
    },
    {
      "epoch": 0.05297,
      "grad_norm": 1.0166098716112735,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5297
    },
    {
      "epoch": 0.05298,
      "grad_norm": 1.0869357162009616,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 5298
    },
    {
      "epoch": 0.05299,
      "grad_norm": 0.885010326633525,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 5299
    },
    {
      "epoch": 0.053,
      "grad_norm": 0.9615808194178252,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 5300
    },
    {
      "epoch": 0.05301,
      "grad_norm": 0.938352188140003,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 5301
    },
    {
      "epoch": 0.05302,
      "grad_norm": 1.0040646683623418,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 5302
    },
    {
      "epoch": 0.05303,
      "grad_norm": 0.8200261070086993,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 5303
    },
    {
      "epoch": 0.05304,
      "grad_norm": 0.8306683481132952,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 5304
    },
    {
      "epoch": 0.05305,
      "grad_norm": 0.835451021401934,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5305
    },
    {
      "epoch": 0.05306,
      "grad_norm": 0.999299771292447,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 5306
    },
    {
      "epoch": 0.05307,
      "grad_norm": 1.1575629558742315,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 5307
    },
    {
      "epoch": 0.05308,
      "grad_norm": 0.8821432895719332,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 5308
    },
    {
      "epoch": 0.05309,
      "grad_norm": 0.7478897292069714,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 5309
    },
    {
      "epoch": 0.0531,
      "grad_norm": 0.6757983881370974,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5310
    },
    {
      "epoch": 0.05311,
      "grad_norm": 0.734265220677749,
      "learning_rate": 0.003,
      "loss": 4.165,
      "step": 5311
    },
    {
      "epoch": 0.05312,
      "grad_norm": 0.7403158021925053,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5312
    },
    {
      "epoch": 0.05313,
      "grad_norm": 0.7001445193969814,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 5313
    },
    {
      "epoch": 0.05314,
      "grad_norm": 0.6690291331175298,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5314
    },
    {
      "epoch": 0.05315,
      "grad_norm": 0.692987823884034,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5315
    },
    {
      "epoch": 0.05316,
      "grad_norm": 0.6767422015060799,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 5316
    },
    {
      "epoch": 0.05317,
      "grad_norm": 0.6288130150337862,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 5317
    },
    {
      "epoch": 0.05318,
      "grad_norm": 0.6126535331811985,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 5318
    },
    {
      "epoch": 0.05319,
      "grad_norm": 0.6921182533464885,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 5319
    },
    {
      "epoch": 0.0532,
      "grad_norm": 0.8654384133463597,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 5320
    },
    {
      "epoch": 0.05321,
      "grad_norm": 1.33574080453646,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5321
    },
    {
      "epoch": 0.05322,
      "grad_norm": 0.7268715200875471,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 5322
    },
    {
      "epoch": 0.05323,
      "grad_norm": 0.8318376363678193,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 5323
    },
    {
      "epoch": 0.05324,
      "grad_norm": 0.9201296394107168,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 5324
    },
    {
      "epoch": 0.05325,
      "grad_norm": 0.9283907989830872,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 5325
    },
    {
      "epoch": 0.05326,
      "grad_norm": 0.9049955824604045,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 5326
    },
    {
      "epoch": 0.05327,
      "grad_norm": 0.8645534067837618,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 5327
    },
    {
      "epoch": 0.05328,
      "grad_norm": 0.7530167384055072,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 5328
    },
    {
      "epoch": 0.05329,
      "grad_norm": 0.7611151412670385,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5329
    },
    {
      "epoch": 0.0533,
      "grad_norm": 0.6706563948860642,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5330
    },
    {
      "epoch": 0.05331,
      "grad_norm": 0.7287872383125151,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 5331
    },
    {
      "epoch": 0.05332,
      "grad_norm": 0.6990347485461705,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5332
    },
    {
      "epoch": 0.05333,
      "grad_norm": 0.7153797786309389,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5333
    },
    {
      "epoch": 0.05334,
      "grad_norm": 0.7135824620276469,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 5334
    },
    {
      "epoch": 0.05335,
      "grad_norm": 0.7329026729817923,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 5335
    },
    {
      "epoch": 0.05336,
      "grad_norm": 0.7489264419035738,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 5336
    },
    {
      "epoch": 0.05337,
      "grad_norm": 0.7598169235752914,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 5337
    },
    {
      "epoch": 0.05338,
      "grad_norm": 0.7398262818382618,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5338
    },
    {
      "epoch": 0.05339,
      "grad_norm": 0.7503534133517454,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5339
    },
    {
      "epoch": 0.0534,
      "grad_norm": 0.729974918134158,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 5340
    },
    {
      "epoch": 0.05341,
      "grad_norm": 0.6588879077709251,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 5341
    },
    {
      "epoch": 0.05342,
      "grad_norm": 0.7419685883521477,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 5342
    },
    {
      "epoch": 0.05343,
      "grad_norm": 0.7435550390936541,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 5343
    },
    {
      "epoch": 0.05344,
      "grad_norm": 0.7863852626600986,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 5344
    },
    {
      "epoch": 0.05345,
      "grad_norm": 0.7422447513853418,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 5345
    },
    {
      "epoch": 0.05346,
      "grad_norm": 0.6629198259774158,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5346
    },
    {
      "epoch": 0.05347,
      "grad_norm": 0.6664361566776935,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 5347
    },
    {
      "epoch": 0.05348,
      "grad_norm": 0.7259069460111056,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 5348
    },
    {
      "epoch": 0.05349,
      "grad_norm": 0.8960329268103234,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 5349
    },
    {
      "epoch": 0.0535,
      "grad_norm": 1.2972339587908424,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 5350
    },
    {
      "epoch": 0.05351,
      "grad_norm": 0.8742838225263962,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 5351
    },
    {
      "epoch": 0.05352,
      "grad_norm": 0.7189234033594293,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5352
    },
    {
      "epoch": 0.05353,
      "grad_norm": 0.8249722357305902,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5353
    },
    {
      "epoch": 0.05354,
      "grad_norm": 0.9093171506347467,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5354
    },
    {
      "epoch": 0.05355,
      "grad_norm": 0.9642691428952453,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5355
    },
    {
      "epoch": 0.05356,
      "grad_norm": 1.0666591729077117,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 5356
    },
    {
      "epoch": 0.05357,
      "grad_norm": 1.1481513734933346,
      "learning_rate": 0.003,
      "loss": 4.1499,
      "step": 5357
    },
    {
      "epoch": 0.05358,
      "grad_norm": 0.8272666359467848,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 5358
    },
    {
      "epoch": 0.05359,
      "grad_norm": 0.7172771084344366,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 5359
    },
    {
      "epoch": 0.0536,
      "grad_norm": 0.6651106652374006,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 5360
    },
    {
      "epoch": 0.05361,
      "grad_norm": 0.7099023505414537,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5361
    },
    {
      "epoch": 0.05362,
      "grad_norm": 0.6814935799785445,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5362
    },
    {
      "epoch": 0.05363,
      "grad_norm": 0.7116843146393923,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 5363
    },
    {
      "epoch": 0.05364,
      "grad_norm": 0.8468291440238759,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5364
    },
    {
      "epoch": 0.05365,
      "grad_norm": 0.9232590055754448,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 5365
    },
    {
      "epoch": 0.05366,
      "grad_norm": 0.9207798580705393,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5366
    },
    {
      "epoch": 0.05367,
      "grad_norm": 0.903400216800342,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 5367
    },
    {
      "epoch": 0.05368,
      "grad_norm": 0.8164520073516316,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 5368
    },
    {
      "epoch": 0.05369,
      "grad_norm": 0.7772994418518063,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 5369
    },
    {
      "epoch": 0.0537,
      "grad_norm": 0.880991172920623,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 5370
    },
    {
      "epoch": 0.05371,
      "grad_norm": 0.8578371500269605,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 5371
    },
    {
      "epoch": 0.05372,
      "grad_norm": 0.8581893520081385,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5372
    },
    {
      "epoch": 0.05373,
      "grad_norm": 0.981605807782507,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 5373
    },
    {
      "epoch": 0.05374,
      "grad_norm": 1.0221056617839879,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 5374
    },
    {
      "epoch": 0.05375,
      "grad_norm": 0.8349259297399914,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5375
    },
    {
      "epoch": 0.05376,
      "grad_norm": 0.6337577803423684,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5376
    },
    {
      "epoch": 0.05377,
      "grad_norm": 0.6889832338024495,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5377
    },
    {
      "epoch": 0.05378,
      "grad_norm": 0.7812861326156888,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 5378
    },
    {
      "epoch": 0.05379,
      "grad_norm": 0.8471373099098546,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 5379
    },
    {
      "epoch": 0.0538,
      "grad_norm": 0.8235164446325877,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 5380
    },
    {
      "epoch": 0.05381,
      "grad_norm": 0.7191229707868476,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5381
    },
    {
      "epoch": 0.05382,
      "grad_norm": 0.6499844207771936,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 5382
    },
    {
      "epoch": 0.05383,
      "grad_norm": 0.6556993226066228,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 5383
    },
    {
      "epoch": 0.05384,
      "grad_norm": 0.6563361402381379,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 5384
    },
    {
      "epoch": 0.05385,
      "grad_norm": 0.7347199963990706,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 5385
    },
    {
      "epoch": 0.05386,
      "grad_norm": 0.7386855978909174,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5386
    },
    {
      "epoch": 0.05387,
      "grad_norm": 0.6950422799141344,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 5387
    },
    {
      "epoch": 0.05388,
      "grad_norm": 0.5813469350124647,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5388
    },
    {
      "epoch": 0.05389,
      "grad_norm": 0.5663689684535403,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 5389
    },
    {
      "epoch": 0.0539,
      "grad_norm": 0.5465150881145437,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 5390
    },
    {
      "epoch": 0.05391,
      "grad_norm": 0.6133981789080953,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 5391
    },
    {
      "epoch": 0.05392,
      "grad_norm": 0.6736582173461655,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 5392
    },
    {
      "epoch": 0.05393,
      "grad_norm": 0.7301761943101791,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 5393
    },
    {
      "epoch": 0.05394,
      "grad_norm": 0.8165543355372945,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 5394
    },
    {
      "epoch": 0.05395,
      "grad_norm": 0.9718965866957342,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 5395
    },
    {
      "epoch": 0.05396,
      "grad_norm": 1.187963468121871,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 5396
    },
    {
      "epoch": 0.05397,
      "grad_norm": 0.8173506639861493,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5397
    },
    {
      "epoch": 0.05398,
      "grad_norm": 0.8358938356109366,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5398
    },
    {
      "epoch": 0.05399,
      "grad_norm": 1.0290513982103986,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 5399
    },
    {
      "epoch": 0.054,
      "grad_norm": 1.080858286517883,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 5400
    },
    {
      "epoch": 0.05401,
      "grad_norm": 0.9077257276299444,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 5401
    },
    {
      "epoch": 0.05402,
      "grad_norm": 1.0072208959923346,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 5402
    },
    {
      "epoch": 0.05403,
      "grad_norm": 1.018927241348422,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 5403
    },
    {
      "epoch": 0.05404,
      "grad_norm": 0.869754384415822,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 5404
    },
    {
      "epoch": 0.05405,
      "grad_norm": 0.8797782462066771,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 5405
    },
    {
      "epoch": 0.05406,
      "grad_norm": 0.9001689861982061,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 5406
    },
    {
      "epoch": 0.05407,
      "grad_norm": 0.8502046414935157,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 5407
    },
    {
      "epoch": 0.05408,
      "grad_norm": 0.7664799629174679,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 5408
    },
    {
      "epoch": 0.05409,
      "grad_norm": 0.6908041565692945,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 5409
    },
    {
      "epoch": 0.0541,
      "grad_norm": 0.6610468895485156,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5410
    },
    {
      "epoch": 0.05411,
      "grad_norm": 0.7119400907145296,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 5411
    },
    {
      "epoch": 0.05412,
      "grad_norm": 0.663756122102936,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 5412
    },
    {
      "epoch": 0.05413,
      "grad_norm": 0.6934560877059226,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 5413
    },
    {
      "epoch": 0.05414,
      "grad_norm": 0.7346678872710226,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 5414
    },
    {
      "epoch": 0.05415,
      "grad_norm": 0.8484600669855001,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 5415
    },
    {
      "epoch": 0.05416,
      "grad_norm": 1.0151357887958885,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 5416
    },
    {
      "epoch": 0.05417,
      "grad_norm": 1.0367007805029973,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 5417
    },
    {
      "epoch": 0.05418,
      "grad_norm": 0.8644340293592367,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 5418
    },
    {
      "epoch": 0.05419,
      "grad_norm": 0.7954788907717502,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5419
    },
    {
      "epoch": 0.0542,
      "grad_norm": 0.6744403315171975,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5420
    },
    {
      "epoch": 0.05421,
      "grad_norm": 0.6512161506363956,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 5421
    },
    {
      "epoch": 0.05422,
      "grad_norm": 0.6012962342683661,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 5422
    },
    {
      "epoch": 0.05423,
      "grad_norm": 0.5672068462884734,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 5423
    },
    {
      "epoch": 0.05424,
      "grad_norm": 0.5505478147926042,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 5424
    },
    {
      "epoch": 0.05425,
      "grad_norm": 0.5656768972103682,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5425
    },
    {
      "epoch": 0.05426,
      "grad_norm": 0.6139416554795246,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 5426
    },
    {
      "epoch": 0.05427,
      "grad_norm": 0.7235910686302464,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 5427
    },
    {
      "epoch": 0.05428,
      "grad_norm": 0.9324553754288041,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 5428
    },
    {
      "epoch": 0.05429,
      "grad_norm": 1.240086702831149,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5429
    },
    {
      "epoch": 0.0543,
      "grad_norm": 0.6358810736408764,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 5430
    },
    {
      "epoch": 0.05431,
      "grad_norm": 0.6446879509210102,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5431
    },
    {
      "epoch": 0.05432,
      "grad_norm": 0.7579426004521167,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 5432
    },
    {
      "epoch": 0.05433,
      "grad_norm": 0.7180614229941668,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 5433
    },
    {
      "epoch": 0.05434,
      "grad_norm": 0.7525090079920899,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 5434
    },
    {
      "epoch": 0.05435,
      "grad_norm": 0.7330233888201235,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 5435
    },
    {
      "epoch": 0.05436,
      "grad_norm": 0.6353200125329312,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5436
    },
    {
      "epoch": 0.05437,
      "grad_norm": 0.5773159842467662,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5437
    },
    {
      "epoch": 0.05438,
      "grad_norm": 0.5441102130695965,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 5438
    },
    {
      "epoch": 0.05439,
      "grad_norm": 0.5835331287048283,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 5439
    },
    {
      "epoch": 0.0544,
      "grad_norm": 0.60640408988456,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 5440
    },
    {
      "epoch": 0.05441,
      "grad_norm": 0.6241847646948565,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5441
    },
    {
      "epoch": 0.05442,
      "grad_norm": 0.6464314803977715,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 5442
    },
    {
      "epoch": 0.05443,
      "grad_norm": 0.7145704106013125,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5443
    },
    {
      "epoch": 0.05444,
      "grad_norm": 0.9202361132826254,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 5444
    },
    {
      "epoch": 0.05445,
      "grad_norm": 1.0938082156986777,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 5445
    },
    {
      "epoch": 0.05446,
      "grad_norm": 0.938114872647745,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 5446
    },
    {
      "epoch": 0.05447,
      "grad_norm": 0.7770727627970194,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 5447
    },
    {
      "epoch": 0.05448,
      "grad_norm": 0.9145478575365268,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5448
    },
    {
      "epoch": 0.05449,
      "grad_norm": 1.2469972170687962,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 5449
    },
    {
      "epoch": 0.0545,
      "grad_norm": 0.9093855660849631,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 5450
    },
    {
      "epoch": 0.05451,
      "grad_norm": 0.8857139983218779,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 5451
    },
    {
      "epoch": 0.05452,
      "grad_norm": 0.8900426362781307,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 5452
    },
    {
      "epoch": 0.05453,
      "grad_norm": 0.8375394525311214,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5453
    },
    {
      "epoch": 0.05454,
      "grad_norm": 0.8515818685266928,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5454
    },
    {
      "epoch": 0.05455,
      "grad_norm": 0.8260003014827662,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 5455
    },
    {
      "epoch": 0.05456,
      "grad_norm": 0.7287425178770975,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5456
    },
    {
      "epoch": 0.05457,
      "grad_norm": 0.7579041524211054,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 5457
    },
    {
      "epoch": 0.05458,
      "grad_norm": 0.8277850916477079,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 5458
    },
    {
      "epoch": 0.05459,
      "grad_norm": 0.9693114941982469,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 5459
    },
    {
      "epoch": 0.0546,
      "grad_norm": 0.9765459017034847,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 5460
    },
    {
      "epoch": 0.05461,
      "grad_norm": 0.791633262589639,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 5461
    },
    {
      "epoch": 0.05462,
      "grad_norm": 0.8490495779181783,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 5462
    },
    {
      "epoch": 0.05463,
      "grad_norm": 0.9783106702125903,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5463
    },
    {
      "epoch": 0.05464,
      "grad_norm": 1.0312992529126999,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 5464
    },
    {
      "epoch": 0.05465,
      "grad_norm": 1.0898859521438415,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 5465
    },
    {
      "epoch": 0.05466,
      "grad_norm": 0.9011074511913453,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 5466
    },
    {
      "epoch": 0.05467,
      "grad_norm": 0.7944273383073479,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 5467
    },
    {
      "epoch": 0.05468,
      "grad_norm": 0.7557868646316354,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 5468
    },
    {
      "epoch": 0.05469,
      "grad_norm": 0.7157700961434887,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 5469
    },
    {
      "epoch": 0.0547,
      "grad_norm": 0.768865679961248,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5470
    },
    {
      "epoch": 0.05471,
      "grad_norm": 0.7253564868288621,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5471
    },
    {
      "epoch": 0.05472,
      "grad_norm": 0.6484270659754396,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 5472
    },
    {
      "epoch": 0.05473,
      "grad_norm": 0.6149159552987197,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 5473
    },
    {
      "epoch": 0.05474,
      "grad_norm": 0.6394638714309846,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 5474
    },
    {
      "epoch": 0.05475,
      "grad_norm": 0.7019738393808744,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 5475
    },
    {
      "epoch": 0.05476,
      "grad_norm": 0.7528679854178247,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 5476
    },
    {
      "epoch": 0.05477,
      "grad_norm": 0.8641692686702116,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 5477
    },
    {
      "epoch": 0.05478,
      "grad_norm": 0.9329968095336455,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5478
    },
    {
      "epoch": 0.05479,
      "grad_norm": 0.8530617715262148,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 5479
    },
    {
      "epoch": 0.0548,
      "grad_norm": 0.6851831025467184,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5480
    },
    {
      "epoch": 0.05481,
      "grad_norm": 0.6300236876868771,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 5481
    },
    {
      "epoch": 0.05482,
      "grad_norm": 0.6745238952273943,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 5482
    },
    {
      "epoch": 0.05483,
      "grad_norm": 0.6653313106296262,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 5483
    },
    {
      "epoch": 0.05484,
      "grad_norm": 0.7093577864302207,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 5484
    },
    {
      "epoch": 0.05485,
      "grad_norm": 0.7814477121541693,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5485
    },
    {
      "epoch": 0.05486,
      "grad_norm": 0.8390976033521745,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 5486
    },
    {
      "epoch": 0.05487,
      "grad_norm": 0.761138743567861,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 5487
    },
    {
      "epoch": 0.05488,
      "grad_norm": 0.7296061446756823,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 5488
    },
    {
      "epoch": 0.05489,
      "grad_norm": 0.7768186211594449,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 5489
    },
    {
      "epoch": 0.0549,
      "grad_norm": 0.6724947950772555,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 5490
    },
    {
      "epoch": 0.05491,
      "grad_norm": 0.6345273171856477,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 5491
    },
    {
      "epoch": 0.05492,
      "grad_norm": 0.6187943049842857,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 5492
    },
    {
      "epoch": 0.05493,
      "grad_norm": 0.736687820563599,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 5493
    },
    {
      "epoch": 0.05494,
      "grad_norm": 0.7510302372633021,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 5494
    },
    {
      "epoch": 0.05495,
      "grad_norm": 0.8712240659486696,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 5495
    },
    {
      "epoch": 0.05496,
      "grad_norm": 1.0341625719582008,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 5496
    },
    {
      "epoch": 0.05497,
      "grad_norm": 1.1451338915999956,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 5497
    },
    {
      "epoch": 0.05498,
      "grad_norm": 0.8431680462491885,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5498
    },
    {
      "epoch": 0.05499,
      "grad_norm": 0.816700771440371,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 5499
    },
    {
      "epoch": 0.055,
      "grad_norm": 0.799679407985695,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 5500
    },
    {
      "epoch": 0.05501,
      "grad_norm": 0.9037921014823048,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5501
    },
    {
      "epoch": 0.05502,
      "grad_norm": 1.0586237652595547,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 5502
    },
    {
      "epoch": 0.05503,
      "grad_norm": 1.031551443780613,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 5503
    },
    {
      "epoch": 0.05504,
      "grad_norm": 0.9058140828678377,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 5504
    },
    {
      "epoch": 0.05505,
      "grad_norm": 0.8558402590801321,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 5505
    },
    {
      "epoch": 0.05506,
      "grad_norm": 0.9182865122687004,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 5506
    },
    {
      "epoch": 0.05507,
      "grad_norm": 0.7933354119287489,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 5507
    },
    {
      "epoch": 0.05508,
      "grad_norm": 0.7272589705323323,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 5508
    },
    {
      "epoch": 0.05509,
      "grad_norm": 0.7161156271163123,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5509
    },
    {
      "epoch": 0.0551,
      "grad_norm": 0.6839604168952217,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 5510
    },
    {
      "epoch": 0.05511,
      "grad_norm": 0.6139416847259401,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 5511
    },
    {
      "epoch": 0.05512,
      "grad_norm": 0.5621071546906728,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5512
    },
    {
      "epoch": 0.05513,
      "grad_norm": 0.6096185049093434,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5513
    },
    {
      "epoch": 0.05514,
      "grad_norm": 0.6336178682323431,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5514
    },
    {
      "epoch": 0.05515,
      "grad_norm": 0.7031400606321546,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 5515
    },
    {
      "epoch": 0.05516,
      "grad_norm": 0.7447410189527172,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 5516
    },
    {
      "epoch": 0.05517,
      "grad_norm": 0.7841931699582834,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 5517
    },
    {
      "epoch": 0.05518,
      "grad_norm": 0.8917972840062397,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5518
    },
    {
      "epoch": 0.05519,
      "grad_norm": 1.0113040291838757,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 5519
    },
    {
      "epoch": 0.0552,
      "grad_norm": 1.0623735541880837,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 5520
    },
    {
      "epoch": 0.05521,
      "grad_norm": 0.7945011952243602,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 5521
    },
    {
      "epoch": 0.05522,
      "grad_norm": 0.6236986309747409,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 5522
    },
    {
      "epoch": 0.05523,
      "grad_norm": 0.6915416985072932,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5523
    },
    {
      "epoch": 0.05524,
      "grad_norm": 0.7772328845162221,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5524
    },
    {
      "epoch": 0.05525,
      "grad_norm": 0.8883412809335927,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5525
    },
    {
      "epoch": 0.05526,
      "grad_norm": 0.8086300513135104,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 5526
    },
    {
      "epoch": 0.05527,
      "grad_norm": 0.6933818876338046,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 5527
    },
    {
      "epoch": 0.05528,
      "grad_norm": 0.7255857040307583,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 5528
    },
    {
      "epoch": 0.05529,
      "grad_norm": 0.7262305634108036,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 5529
    },
    {
      "epoch": 0.0553,
      "grad_norm": 0.6799075981942326,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5530
    },
    {
      "epoch": 0.05531,
      "grad_norm": 0.7288072066854816,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5531
    },
    {
      "epoch": 0.05532,
      "grad_norm": 0.7212131851938903,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 5532
    },
    {
      "epoch": 0.05533,
      "grad_norm": 0.7813370825497535,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 5533
    },
    {
      "epoch": 0.05534,
      "grad_norm": 0.7625719940297025,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 5534
    },
    {
      "epoch": 0.05535,
      "grad_norm": 0.8190303715949169,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5535
    },
    {
      "epoch": 0.05536,
      "grad_norm": 0.8250338882829615,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 5536
    },
    {
      "epoch": 0.05537,
      "grad_norm": 0.8788954790980155,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 5537
    },
    {
      "epoch": 0.05538,
      "grad_norm": 0.9012442797992367,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5538
    },
    {
      "epoch": 0.05539,
      "grad_norm": 0.8122983939229457,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 5539
    },
    {
      "epoch": 0.0554,
      "grad_norm": 0.7598578272137851,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 5540
    },
    {
      "epoch": 0.05541,
      "grad_norm": 0.7025045153629659,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 5541
    },
    {
      "epoch": 0.05542,
      "grad_norm": 0.7249716315464921,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 5542
    },
    {
      "epoch": 0.05543,
      "grad_norm": 0.7653470879890454,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 5543
    },
    {
      "epoch": 0.05544,
      "grad_norm": 0.9152516311386858,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 5544
    },
    {
      "epoch": 0.05545,
      "grad_norm": 0.9264085795715772,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 5545
    },
    {
      "epoch": 0.05546,
      "grad_norm": 0.75980006084045,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 5546
    },
    {
      "epoch": 0.05547,
      "grad_norm": 0.7568411552111042,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 5547
    },
    {
      "epoch": 0.05548,
      "grad_norm": 0.8058342997945618,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 5548
    },
    {
      "epoch": 0.05549,
      "grad_norm": 0.7052563470892956,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 5549
    },
    {
      "epoch": 0.0555,
      "grad_norm": 0.6538707744806174,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 5550
    },
    {
      "epoch": 0.05551,
      "grad_norm": 0.6754492446230133,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 5551
    },
    {
      "epoch": 0.05552,
      "grad_norm": 0.7000361531636052,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 5552
    },
    {
      "epoch": 0.05553,
      "grad_norm": 0.7471036467384345,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5553
    },
    {
      "epoch": 0.05554,
      "grad_norm": 0.6519148066664208,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 5554
    },
    {
      "epoch": 0.05555,
      "grad_norm": 0.6843431614210626,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 5555
    },
    {
      "epoch": 0.05556,
      "grad_norm": 0.6534346041116017,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 5556
    },
    {
      "epoch": 0.05557,
      "grad_norm": 0.816868082356806,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5557
    },
    {
      "epoch": 0.05558,
      "grad_norm": 1.0929732507789733,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 5558
    },
    {
      "epoch": 0.05559,
      "grad_norm": 0.8960362084356669,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 5559
    },
    {
      "epoch": 0.0556,
      "grad_norm": 0.6973978644009409,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 5560
    },
    {
      "epoch": 0.05561,
      "grad_norm": 0.7611043817319628,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 5561
    },
    {
      "epoch": 0.05562,
      "grad_norm": 0.8608099588467866,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5562
    },
    {
      "epoch": 0.05563,
      "grad_norm": 0.9082804694980892,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5563
    },
    {
      "epoch": 0.05564,
      "grad_norm": 0.9906908375188082,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5564
    },
    {
      "epoch": 0.05565,
      "grad_norm": 1.0732377297521,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 5565
    },
    {
      "epoch": 0.05566,
      "grad_norm": 1.0637503448630574,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 5566
    },
    {
      "epoch": 0.05567,
      "grad_norm": 1.0859660134329547,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 5567
    },
    {
      "epoch": 0.05568,
      "grad_norm": 1.0189336160886215,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 5568
    },
    {
      "epoch": 0.05569,
      "grad_norm": 1.004138283903561,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5569
    },
    {
      "epoch": 0.0557,
      "grad_norm": 1.190958221299472,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 5570
    },
    {
      "epoch": 0.05571,
      "grad_norm": 1.010721856861485,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 5571
    },
    {
      "epoch": 0.05572,
      "grad_norm": 0.9197968716418343,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 5572
    },
    {
      "epoch": 0.05573,
      "grad_norm": 0.9807472494013146,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 5573
    },
    {
      "epoch": 0.05574,
      "grad_norm": 0.9948127166172148,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 5574
    },
    {
      "epoch": 0.05575,
      "grad_norm": 0.971313458379033,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 5575
    },
    {
      "epoch": 0.05576,
      "grad_norm": 1.0277422588399054,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 5576
    },
    {
      "epoch": 0.05577,
      "grad_norm": 0.9995940449974644,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 5577
    },
    {
      "epoch": 0.05578,
      "grad_norm": 1.0859216843814685,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 5578
    },
    {
      "epoch": 0.05579,
      "grad_norm": 0.9574713713729739,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 5579
    },
    {
      "epoch": 0.0558,
      "grad_norm": 0.8737247794468992,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 5580
    },
    {
      "epoch": 0.05581,
      "grad_norm": 0.7539148254380033,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 5581
    },
    {
      "epoch": 0.05582,
      "grad_norm": 0.7731731504235131,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 5582
    },
    {
      "epoch": 0.05583,
      "grad_norm": 0.786183465585757,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 5583
    },
    {
      "epoch": 0.05584,
      "grad_norm": 0.8712656140654648,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 5584
    },
    {
      "epoch": 0.05585,
      "grad_norm": 0.8461555674443455,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 5585
    },
    {
      "epoch": 0.05586,
      "grad_norm": 0.6370593274231787,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 5586
    },
    {
      "epoch": 0.05587,
      "grad_norm": 0.6466924043657243,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 5587
    },
    {
      "epoch": 0.05588,
      "grad_norm": 0.6622345802465658,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5588
    },
    {
      "epoch": 0.05589,
      "grad_norm": 0.5913876457137929,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5589
    },
    {
      "epoch": 0.0559,
      "grad_norm": 0.5270714516413212,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5590
    },
    {
      "epoch": 0.05591,
      "grad_norm": 0.5304582733694935,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 5591
    },
    {
      "epoch": 0.05592,
      "grad_norm": 0.5313660136101308,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 5592
    },
    {
      "epoch": 0.05593,
      "grad_norm": 0.5134736329805021,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 5593
    },
    {
      "epoch": 0.05594,
      "grad_norm": 0.4831292187313499,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 5594
    },
    {
      "epoch": 0.05595,
      "grad_norm": 0.47641374196121894,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 5595
    },
    {
      "epoch": 0.05596,
      "grad_norm": 0.5262150385564768,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 5596
    },
    {
      "epoch": 0.05597,
      "grad_norm": 0.5970056001301876,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 5597
    },
    {
      "epoch": 0.05598,
      "grad_norm": 0.5741914372250703,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 5598
    },
    {
      "epoch": 0.05599,
      "grad_norm": 0.6464864543941242,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 5599
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.9195309880740932,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 5600
    },
    {
      "epoch": 0.05601,
      "grad_norm": 1.2480023251412764,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 5601
    },
    {
      "epoch": 0.05602,
      "grad_norm": 0.7822871448594372,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 5602
    },
    {
      "epoch": 0.05603,
      "grad_norm": 0.7289334276521395,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 5603
    },
    {
      "epoch": 0.05604,
      "grad_norm": 0.679359088573528,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 5604
    },
    {
      "epoch": 0.05605,
      "grad_norm": 0.7513465211377279,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 5605
    },
    {
      "epoch": 0.05606,
      "grad_norm": 0.7553929167731772,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 5606
    },
    {
      "epoch": 0.05607,
      "grad_norm": 0.7847903163107046,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 5607
    },
    {
      "epoch": 0.05608,
      "grad_norm": 0.7298569984980766,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 5608
    },
    {
      "epoch": 0.05609,
      "grad_norm": 0.8968015359645949,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5609
    },
    {
      "epoch": 0.0561,
      "grad_norm": 0.911039007619847,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 5610
    },
    {
      "epoch": 0.05611,
      "grad_norm": 0.6357740462584217,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 5611
    },
    {
      "epoch": 0.05612,
      "grad_norm": 0.6371928488100147,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 5612
    },
    {
      "epoch": 0.05613,
      "grad_norm": 0.7120154136241356,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 5613
    },
    {
      "epoch": 0.05614,
      "grad_norm": 0.6531842861897988,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5614
    },
    {
      "epoch": 0.05615,
      "grad_norm": 0.723116589410996,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 5615
    },
    {
      "epoch": 0.05616,
      "grad_norm": 0.6683266115411998,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 5616
    },
    {
      "epoch": 0.05617,
      "grad_norm": 0.5821372860770843,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 5617
    },
    {
      "epoch": 0.05618,
      "grad_norm": 0.5360283472097199,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 5618
    },
    {
      "epoch": 0.05619,
      "grad_norm": 0.5939904129474893,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 5619
    },
    {
      "epoch": 0.0562,
      "grad_norm": 0.667235826498158,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 5620
    },
    {
      "epoch": 0.05621,
      "grad_norm": 0.7593944756058572,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 5621
    },
    {
      "epoch": 0.05622,
      "grad_norm": 0.7560267363483694,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 5622
    },
    {
      "epoch": 0.05623,
      "grad_norm": 0.9633744968283545,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 5623
    },
    {
      "epoch": 0.05624,
      "grad_norm": 1.3159764636382347,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 5624
    },
    {
      "epoch": 0.05625,
      "grad_norm": 0.8598542406723987,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5625
    },
    {
      "epoch": 0.05626,
      "grad_norm": 0.8928264883638428,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 5626
    },
    {
      "epoch": 0.05627,
      "grad_norm": 0.9035581162578888,
      "learning_rate": 0.003,
      "loss": 4.1314,
      "step": 5627
    },
    {
      "epoch": 0.05628,
      "grad_norm": 0.9326763323161109,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5628
    },
    {
      "epoch": 0.05629,
      "grad_norm": 0.9858524239497275,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 5629
    },
    {
      "epoch": 0.0563,
      "grad_norm": 1.0278927211295703,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5630
    },
    {
      "epoch": 0.05631,
      "grad_norm": 0.8158010640041221,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 5631
    },
    {
      "epoch": 0.05632,
      "grad_norm": 0.7904116451209343,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 5632
    },
    {
      "epoch": 0.05633,
      "grad_norm": 0.6674336130552363,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5633
    },
    {
      "epoch": 0.05634,
      "grad_norm": 0.7131647955202249,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 5634
    },
    {
      "epoch": 0.05635,
      "grad_norm": 0.7762681609027913,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5635
    },
    {
      "epoch": 0.05636,
      "grad_norm": 0.8021027171917042,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5636
    },
    {
      "epoch": 0.05637,
      "grad_norm": 0.925350555541478,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 5637
    },
    {
      "epoch": 0.05638,
      "grad_norm": 0.9727914008655373,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 5638
    },
    {
      "epoch": 0.05639,
      "grad_norm": 0.7920727200781233,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5639
    },
    {
      "epoch": 0.0564,
      "grad_norm": 0.6724027259224868,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5640
    },
    {
      "epoch": 0.05641,
      "grad_norm": 0.6404204234893655,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 5641
    },
    {
      "epoch": 0.05642,
      "grad_norm": 0.638929472798256,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 5642
    },
    {
      "epoch": 0.05643,
      "grad_norm": 0.6279646959643067,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5643
    },
    {
      "epoch": 0.05644,
      "grad_norm": 0.6050520516714298,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 5644
    },
    {
      "epoch": 0.05645,
      "grad_norm": 0.7210392134789482,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 5645
    },
    {
      "epoch": 0.05646,
      "grad_norm": 0.9082430985310681,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5646
    },
    {
      "epoch": 0.05647,
      "grad_norm": 1.0629749955090315,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 5647
    },
    {
      "epoch": 0.05648,
      "grad_norm": 0.9337407114146831,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 5648
    },
    {
      "epoch": 0.05649,
      "grad_norm": 0.7908235143189736,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5649
    },
    {
      "epoch": 0.0565,
      "grad_norm": 0.6719463167631939,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 5650
    },
    {
      "epoch": 0.05651,
      "grad_norm": 0.6366536023365831,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 5651
    },
    {
      "epoch": 0.05652,
      "grad_norm": 0.6620856715240755,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 5652
    },
    {
      "epoch": 0.05653,
      "grad_norm": 0.6375701410325152,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 5653
    },
    {
      "epoch": 0.05654,
      "grad_norm": 0.7756105771774413,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 5654
    },
    {
      "epoch": 0.05655,
      "grad_norm": 0.8746926551877258,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 5655
    },
    {
      "epoch": 0.05656,
      "grad_norm": 0.824346564649059,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 5656
    },
    {
      "epoch": 0.05657,
      "grad_norm": 0.9029529758933031,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 5657
    },
    {
      "epoch": 0.05658,
      "grad_norm": 0.9351306065482489,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5658
    },
    {
      "epoch": 0.05659,
      "grad_norm": 0.8596436492121928,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5659
    },
    {
      "epoch": 0.0566,
      "grad_norm": 0.9772647763222754,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 5660
    },
    {
      "epoch": 0.05661,
      "grad_norm": 1.009709800322154,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 5661
    },
    {
      "epoch": 0.05662,
      "grad_norm": 0.8700878215220617,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 5662
    },
    {
      "epoch": 0.05663,
      "grad_norm": 0.7314812319835777,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5663
    },
    {
      "epoch": 0.05664,
      "grad_norm": 0.7218667255886607,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 5664
    },
    {
      "epoch": 0.05665,
      "grad_norm": 0.7545885907055334,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5665
    },
    {
      "epoch": 0.05666,
      "grad_norm": 0.750379441945731,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 5666
    },
    {
      "epoch": 0.05667,
      "grad_norm": 0.7649794680001714,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 5667
    },
    {
      "epoch": 0.05668,
      "grad_norm": 0.7398186277471477,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 5668
    },
    {
      "epoch": 0.05669,
      "grad_norm": 0.7190311775637344,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 5669
    },
    {
      "epoch": 0.0567,
      "grad_norm": 0.7385720576479917,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 5670
    },
    {
      "epoch": 0.05671,
      "grad_norm": 0.7190124419045905,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 5671
    },
    {
      "epoch": 0.05672,
      "grad_norm": 0.6906608031924452,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5672
    },
    {
      "epoch": 0.05673,
      "grad_norm": 0.7440346855522176,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 5673
    },
    {
      "epoch": 0.05674,
      "grad_norm": 0.8058800416308596,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5674
    },
    {
      "epoch": 0.05675,
      "grad_norm": 0.9524033317076179,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 5675
    },
    {
      "epoch": 0.05676,
      "grad_norm": 1.0466364934113919,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 5676
    },
    {
      "epoch": 0.05677,
      "grad_norm": 1.1290737878900445,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 5677
    },
    {
      "epoch": 0.05678,
      "grad_norm": 0.8056761897248448,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5678
    },
    {
      "epoch": 0.05679,
      "grad_norm": 0.8416448974640103,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 5679
    },
    {
      "epoch": 0.0568,
      "grad_norm": 0.8654292108286987,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 5680
    },
    {
      "epoch": 0.05681,
      "grad_norm": 0.8669869662711412,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 5681
    },
    {
      "epoch": 0.05682,
      "grad_norm": 0.7948737245049556,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5682
    },
    {
      "epoch": 0.05683,
      "grad_norm": 0.7377380328343832,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 5683
    },
    {
      "epoch": 0.05684,
      "grad_norm": 0.8063644138611359,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 5684
    },
    {
      "epoch": 0.05685,
      "grad_norm": 0.8527467662490903,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 5685
    },
    {
      "epoch": 0.05686,
      "grad_norm": 0.727039984777113,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 5686
    },
    {
      "epoch": 0.05687,
      "grad_norm": 0.6543891605340604,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 5687
    },
    {
      "epoch": 0.05688,
      "grad_norm": 0.6732327830992805,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 5688
    },
    {
      "epoch": 0.05689,
      "grad_norm": 0.6988792787125268,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 5689
    },
    {
      "epoch": 0.0569,
      "grad_norm": 0.7481389272798462,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5690
    },
    {
      "epoch": 0.05691,
      "grad_norm": 0.8277047142206917,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 5691
    },
    {
      "epoch": 0.05692,
      "grad_norm": 0.8689994157613129,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5692
    },
    {
      "epoch": 0.05693,
      "grad_norm": 0.8286352092123737,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 5693
    },
    {
      "epoch": 0.05694,
      "grad_norm": 0.8815887145121191,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 5694
    },
    {
      "epoch": 0.05695,
      "grad_norm": 0.8249748591708475,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 5695
    },
    {
      "epoch": 0.05696,
      "grad_norm": 0.8653183020137561,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 5696
    },
    {
      "epoch": 0.05697,
      "grad_norm": 0.8835391111395854,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 5697
    },
    {
      "epoch": 0.05698,
      "grad_norm": 0.8593431463479471,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 5698
    },
    {
      "epoch": 0.05699,
      "grad_norm": 0.8457692338051964,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5699
    },
    {
      "epoch": 0.057,
      "grad_norm": 0.7057109559383719,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 5700
    },
    {
      "epoch": 0.05701,
      "grad_norm": 0.7143700073271365,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 5701
    },
    {
      "epoch": 0.05702,
      "grad_norm": 0.7926557745297437,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 5702
    },
    {
      "epoch": 0.05703,
      "grad_norm": 0.788953414454626,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 5703
    },
    {
      "epoch": 0.05704,
      "grad_norm": 0.8987044741769469,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5704
    },
    {
      "epoch": 0.05705,
      "grad_norm": 0.9475911671053671,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 5705
    },
    {
      "epoch": 0.05706,
      "grad_norm": 0.9351503494399143,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 5706
    },
    {
      "epoch": 0.05707,
      "grad_norm": 0.908791879595692,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 5707
    },
    {
      "epoch": 0.05708,
      "grad_norm": 0.9702984464047565,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 5708
    },
    {
      "epoch": 0.05709,
      "grad_norm": 1.02607623723065,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5709
    },
    {
      "epoch": 0.0571,
      "grad_norm": 0.8383478585082147,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 5710
    },
    {
      "epoch": 0.05711,
      "grad_norm": 0.8409807556043147,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 5711
    },
    {
      "epoch": 0.05712,
      "grad_norm": 0.8246373489572517,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 5712
    },
    {
      "epoch": 0.05713,
      "grad_norm": 0.750695304399167,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5713
    },
    {
      "epoch": 0.05714,
      "grad_norm": 0.6256785887976535,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 5714
    },
    {
      "epoch": 0.05715,
      "grad_norm": 0.6405112496698065,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 5715
    },
    {
      "epoch": 0.05716,
      "grad_norm": 0.7045480948725663,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 5716
    },
    {
      "epoch": 0.05717,
      "grad_norm": 0.7251180513932567,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 5717
    },
    {
      "epoch": 0.05718,
      "grad_norm": 0.7144875502995738,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5718
    },
    {
      "epoch": 0.05719,
      "grad_norm": 0.7691901405002558,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 5719
    },
    {
      "epoch": 0.0572,
      "grad_norm": 0.733838672256027,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5720
    },
    {
      "epoch": 0.05721,
      "grad_norm": 0.7007983767497351,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 5721
    },
    {
      "epoch": 0.05722,
      "grad_norm": 0.6786326167442739,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 5722
    },
    {
      "epoch": 0.05723,
      "grad_norm": 0.7298075380572556,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 5723
    },
    {
      "epoch": 0.05724,
      "grad_norm": 0.8701417935987675,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 5724
    },
    {
      "epoch": 0.05725,
      "grad_norm": 0.9317116533192469,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5725
    },
    {
      "epoch": 0.05726,
      "grad_norm": 1.2423360551944322,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 5726
    },
    {
      "epoch": 0.05727,
      "grad_norm": 0.9566322040091452,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 5727
    },
    {
      "epoch": 0.05728,
      "grad_norm": 0.8630235197631045,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 5728
    },
    {
      "epoch": 0.05729,
      "grad_norm": 0.6367381629677203,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 5729
    },
    {
      "epoch": 0.0573,
      "grad_norm": 0.6307754247308351,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 5730
    },
    {
      "epoch": 0.05731,
      "grad_norm": 0.6637990852153008,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 5731
    },
    {
      "epoch": 0.05732,
      "grad_norm": 0.7152262181674999,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 5732
    },
    {
      "epoch": 0.05733,
      "grad_norm": 0.7583908888024418,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 5733
    },
    {
      "epoch": 0.05734,
      "grad_norm": 0.7979697396059119,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 5734
    },
    {
      "epoch": 0.05735,
      "grad_norm": 0.6659788683114847,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 5735
    },
    {
      "epoch": 0.05736,
      "grad_norm": 0.6718526951634197,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 5736
    },
    {
      "epoch": 0.05737,
      "grad_norm": 0.5932737131818991,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 5737
    },
    {
      "epoch": 0.05738,
      "grad_norm": 0.5552311466351124,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 5738
    },
    {
      "epoch": 0.05739,
      "grad_norm": 0.5192428824433908,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 5739
    },
    {
      "epoch": 0.0574,
      "grad_norm": 0.5521978223598561,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 5740
    },
    {
      "epoch": 0.05741,
      "grad_norm": 0.6827764477275866,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 5741
    },
    {
      "epoch": 0.05742,
      "grad_norm": 0.9661334292481325,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 5742
    },
    {
      "epoch": 0.05743,
      "grad_norm": 1.3808942777208146,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5743
    },
    {
      "epoch": 0.05744,
      "grad_norm": 0.6634000683140943,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5744
    },
    {
      "epoch": 0.05745,
      "grad_norm": 0.8051574353421863,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 5745
    },
    {
      "epoch": 0.05746,
      "grad_norm": 0.8451000982869576,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5746
    },
    {
      "epoch": 0.05747,
      "grad_norm": 0.7005027290316844,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 5747
    },
    {
      "epoch": 0.05748,
      "grad_norm": 0.7417025932554691,
      "learning_rate": 0.003,
      "loss": 4.1526,
      "step": 5748
    },
    {
      "epoch": 0.05749,
      "grad_norm": 0.7770045474271049,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 5749
    },
    {
      "epoch": 0.0575,
      "grad_norm": 0.8496769014517581,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5750
    },
    {
      "epoch": 0.05751,
      "grad_norm": 0.9049063166808049,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 5751
    },
    {
      "epoch": 0.05752,
      "grad_norm": 1.0297317331778117,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5752
    },
    {
      "epoch": 0.05753,
      "grad_norm": 0.9151964474023491,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 5753
    },
    {
      "epoch": 0.05754,
      "grad_norm": 0.8031670531802124,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 5754
    },
    {
      "epoch": 0.05755,
      "grad_norm": 1.0112539909545384,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5755
    },
    {
      "epoch": 0.05756,
      "grad_norm": 1.1823863497130285,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 5756
    },
    {
      "epoch": 0.05757,
      "grad_norm": 0.82463840960989,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 5757
    },
    {
      "epoch": 0.05758,
      "grad_norm": 0.8370537741924009,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 5758
    },
    {
      "epoch": 0.05759,
      "grad_norm": 0.7690512085008506,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5759
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.8543370684675727,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5760
    },
    {
      "epoch": 0.05761,
      "grad_norm": 0.8953621498467661,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 5761
    },
    {
      "epoch": 0.05762,
      "grad_norm": 0.8489500157996731,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 5762
    },
    {
      "epoch": 0.05763,
      "grad_norm": 0.7805963155950815,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 5763
    },
    {
      "epoch": 0.05764,
      "grad_norm": 0.7082991317232272,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 5764
    },
    {
      "epoch": 0.05765,
      "grad_norm": 0.7124286749136263,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5765
    },
    {
      "epoch": 0.05766,
      "grad_norm": 0.6760677428953438,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5766
    },
    {
      "epoch": 0.05767,
      "grad_norm": 0.8489631258318778,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 5767
    },
    {
      "epoch": 0.05768,
      "grad_norm": 0.9362514658122035,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 5768
    },
    {
      "epoch": 0.05769,
      "grad_norm": 0.8865918581298181,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 5769
    },
    {
      "epoch": 0.0577,
      "grad_norm": 0.7665977076446682,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 5770
    },
    {
      "epoch": 0.05771,
      "grad_norm": 0.692978560597086,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 5771
    },
    {
      "epoch": 0.05772,
      "grad_norm": 0.5670375020911395,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 5772
    },
    {
      "epoch": 0.05773,
      "grad_norm": 0.5994073546273513,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5773
    },
    {
      "epoch": 0.05774,
      "grad_norm": 0.6697442229469948,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 5774
    },
    {
      "epoch": 0.05775,
      "grad_norm": 0.7260764971579977,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5775
    },
    {
      "epoch": 0.05776,
      "grad_norm": 0.827091034088083,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 5776
    },
    {
      "epoch": 0.05777,
      "grad_norm": 0.85821320032967,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5777
    },
    {
      "epoch": 0.05778,
      "grad_norm": 0.75564632841601,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5778
    },
    {
      "epoch": 0.05779,
      "grad_norm": 0.758336533791715,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 5779
    },
    {
      "epoch": 0.0578,
      "grad_norm": 0.8762931646419935,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 5780
    },
    {
      "epoch": 0.05781,
      "grad_norm": 0.8498076917904277,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 5781
    },
    {
      "epoch": 0.05782,
      "grad_norm": 0.9973539387616522,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 5782
    },
    {
      "epoch": 0.05783,
      "grad_norm": 1.1963721721828537,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 5783
    },
    {
      "epoch": 0.05784,
      "grad_norm": 0.8859060956969492,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5784
    },
    {
      "epoch": 0.05785,
      "grad_norm": 0.8870211693131135,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 5785
    },
    {
      "epoch": 0.05786,
      "grad_norm": 0.8028500867550125,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 5786
    },
    {
      "epoch": 0.05787,
      "grad_norm": 0.9175527403883458,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5787
    },
    {
      "epoch": 0.05788,
      "grad_norm": 0.9094044732533373,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 5788
    },
    {
      "epoch": 0.05789,
      "grad_norm": 0.9198435831448731,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 5789
    },
    {
      "epoch": 0.0579,
      "grad_norm": 0.8903452476156254,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 5790
    },
    {
      "epoch": 0.05791,
      "grad_norm": 0.7995965533827011,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 5791
    },
    {
      "epoch": 0.05792,
      "grad_norm": 0.7374601576105686,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 5792
    },
    {
      "epoch": 0.05793,
      "grad_norm": 0.8694986815433677,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 5793
    },
    {
      "epoch": 0.05794,
      "grad_norm": 1.0476632248094724,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 5794
    },
    {
      "epoch": 0.05795,
      "grad_norm": 1.1869938993812021,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 5795
    },
    {
      "epoch": 0.05796,
      "grad_norm": 0.8360088466789314,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5796
    },
    {
      "epoch": 0.05797,
      "grad_norm": 0.8084090857296047,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5797
    },
    {
      "epoch": 0.05798,
      "grad_norm": 0.7637314143049846,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 5798
    },
    {
      "epoch": 0.05799,
      "grad_norm": 0.8732093010633248,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 5799
    },
    {
      "epoch": 0.058,
      "grad_norm": 0.7122155596601744,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 5800
    },
    {
      "epoch": 0.05801,
      "grad_norm": 0.6818953681284188,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 5801
    },
    {
      "epoch": 0.05802,
      "grad_norm": 0.7665388126779251,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5802
    },
    {
      "epoch": 0.05803,
      "grad_norm": 0.8847770862686954,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 5803
    },
    {
      "epoch": 0.05804,
      "grad_norm": 0.8330623324317175,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 5804
    },
    {
      "epoch": 0.05805,
      "grad_norm": 0.667711721351076,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 5805
    },
    {
      "epoch": 0.05806,
      "grad_norm": 0.7138990216020714,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 5806
    },
    {
      "epoch": 0.05807,
      "grad_norm": 0.6844079777026275,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 5807
    },
    {
      "epoch": 0.05808,
      "grad_norm": 0.7329662618338171,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 5808
    },
    {
      "epoch": 0.05809,
      "grad_norm": 0.7127193982748581,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 5809
    },
    {
      "epoch": 0.0581,
      "grad_norm": 0.7849473665098132,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 5810
    },
    {
      "epoch": 0.05811,
      "grad_norm": 0.7772533700540266,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 5811
    },
    {
      "epoch": 0.05812,
      "grad_norm": 0.8535932209668625,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 5812
    },
    {
      "epoch": 0.05813,
      "grad_norm": 1.050675078146715,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 5813
    },
    {
      "epoch": 0.05814,
      "grad_norm": 0.9859964212140094,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 5814
    },
    {
      "epoch": 0.05815,
      "grad_norm": 0.8071168638710576,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 5815
    },
    {
      "epoch": 0.05816,
      "grad_norm": 0.744822186311271,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 5816
    },
    {
      "epoch": 0.05817,
      "grad_norm": 0.859723113139702,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5817
    },
    {
      "epoch": 0.05818,
      "grad_norm": 0.9076124598026192,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 5818
    },
    {
      "epoch": 0.05819,
      "grad_norm": 0.8614969748668542,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 5819
    },
    {
      "epoch": 0.0582,
      "grad_norm": 0.7596720316772002,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5820
    },
    {
      "epoch": 0.05821,
      "grad_norm": 0.7261229816186945,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5821
    },
    {
      "epoch": 0.05822,
      "grad_norm": 0.6389335810166767,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 5822
    },
    {
      "epoch": 0.05823,
      "grad_norm": 0.650431814244154,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 5823
    },
    {
      "epoch": 0.05824,
      "grad_norm": 0.7021908556345984,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5824
    },
    {
      "epoch": 0.05825,
      "grad_norm": 0.7075707130594027,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 5825
    },
    {
      "epoch": 0.05826,
      "grad_norm": 0.6632684340681474,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 5826
    },
    {
      "epoch": 0.05827,
      "grad_norm": 0.6153406505756235,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 5827
    },
    {
      "epoch": 0.05828,
      "grad_norm": 0.5151089751197206,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 5828
    },
    {
      "epoch": 0.05829,
      "grad_norm": 0.5657176931414479,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 5829
    },
    {
      "epoch": 0.0583,
      "grad_norm": 0.7071018057037146,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 5830
    },
    {
      "epoch": 0.05831,
      "grad_norm": 1.0078552329007178,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 5831
    },
    {
      "epoch": 0.05832,
      "grad_norm": 1.461038643232985,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 5832
    },
    {
      "epoch": 0.05833,
      "grad_norm": 0.515512286205366,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 5833
    },
    {
      "epoch": 0.05834,
      "grad_norm": 0.8693462452576697,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 5834
    },
    {
      "epoch": 0.05835,
      "grad_norm": 0.8652446702771586,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 5835
    },
    {
      "epoch": 0.05836,
      "grad_norm": 0.7247395539295343,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 5836
    },
    {
      "epoch": 0.05837,
      "grad_norm": 0.6702651347004529,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 5837
    },
    {
      "epoch": 0.05838,
      "grad_norm": 0.7045650962837332,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5838
    },
    {
      "epoch": 0.05839,
      "grad_norm": 0.785589688267228,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 5839
    },
    {
      "epoch": 0.0584,
      "grad_norm": 0.8133814023137367,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 5840
    },
    {
      "epoch": 0.05841,
      "grad_norm": 0.832422818436443,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 5841
    },
    {
      "epoch": 0.05842,
      "grad_norm": 0.8619710688010511,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 5842
    },
    {
      "epoch": 0.05843,
      "grad_norm": 0.9243101513524902,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 5843
    },
    {
      "epoch": 0.05844,
      "grad_norm": 0.8736875279868362,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5844
    },
    {
      "epoch": 0.05845,
      "grad_norm": 0.8068837365535566,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5845
    },
    {
      "epoch": 0.05846,
      "grad_norm": 0.711931455880162,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 5846
    },
    {
      "epoch": 0.05847,
      "grad_norm": 0.7993811775050275,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 5847
    },
    {
      "epoch": 0.05848,
      "grad_norm": 0.754358214475092,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5848
    },
    {
      "epoch": 0.05849,
      "grad_norm": 0.8702234783293419,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 5849
    },
    {
      "epoch": 0.0585,
      "grad_norm": 0.9876577919554057,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 5850
    },
    {
      "epoch": 0.05851,
      "grad_norm": 1.0940891749043231,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 5851
    },
    {
      "epoch": 0.05852,
      "grad_norm": 0.9378030392921249,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5852
    },
    {
      "epoch": 0.05853,
      "grad_norm": 0.7566439324877372,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 5853
    },
    {
      "epoch": 0.05854,
      "grad_norm": 0.673296794630987,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 5854
    },
    {
      "epoch": 0.05855,
      "grad_norm": 0.578895079904098,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 5855
    },
    {
      "epoch": 0.05856,
      "grad_norm": 0.624693508028921,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 5856
    },
    {
      "epoch": 0.05857,
      "grad_norm": 0.7368589206607341,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 5857
    },
    {
      "epoch": 0.05858,
      "grad_norm": 0.8538648315734454,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 5858
    },
    {
      "epoch": 0.05859,
      "grad_norm": 0.8556665179737083,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 5859
    },
    {
      "epoch": 0.0586,
      "grad_norm": 0.7826151609045512,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 5860
    },
    {
      "epoch": 0.05861,
      "grad_norm": 0.7886112191620424,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 5861
    },
    {
      "epoch": 0.05862,
      "grad_norm": 0.9292901295633138,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 5862
    },
    {
      "epoch": 0.05863,
      "grad_norm": 0.8121851405516025,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 5863
    },
    {
      "epoch": 0.05864,
      "grad_norm": 0.6827602835508371,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 5864
    },
    {
      "epoch": 0.05865,
      "grad_norm": 0.7544438864360793,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5865
    },
    {
      "epoch": 0.05866,
      "grad_norm": 0.8838045357890466,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 5866
    },
    {
      "epoch": 0.05867,
      "grad_norm": 0.8168835101024318,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 5867
    },
    {
      "epoch": 0.05868,
      "grad_norm": 0.6536324933405178,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 5868
    },
    {
      "epoch": 0.05869,
      "grad_norm": 0.683464201492265,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 5869
    },
    {
      "epoch": 0.0587,
      "grad_norm": 0.7963857853576817,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 5870
    },
    {
      "epoch": 0.05871,
      "grad_norm": 0.9792801347682168,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5871
    },
    {
      "epoch": 0.05872,
      "grad_norm": 1.4172796436924118,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 5872
    },
    {
      "epoch": 0.05873,
      "grad_norm": 0.7027081325886863,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 5873
    },
    {
      "epoch": 0.05874,
      "grad_norm": 0.7473953186030906,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5874
    },
    {
      "epoch": 0.05875,
      "grad_norm": 0.809528681305329,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 5875
    },
    {
      "epoch": 0.05876,
      "grad_norm": 0.772298806074322,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5876
    },
    {
      "epoch": 0.05877,
      "grad_norm": 0.8257282875546335,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 5877
    },
    {
      "epoch": 0.05878,
      "grad_norm": 0.9455870472025797,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 5878
    },
    {
      "epoch": 0.05879,
      "grad_norm": 1.1569037258481754,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 5879
    },
    {
      "epoch": 0.0588,
      "grad_norm": 0.8577210204570376,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 5880
    },
    {
      "epoch": 0.05881,
      "grad_norm": 0.8032693830476965,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 5881
    },
    {
      "epoch": 0.05882,
      "grad_norm": 0.8937763672419452,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 5882
    },
    {
      "epoch": 0.05883,
      "grad_norm": 0.9593530484800846,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 5883
    },
    {
      "epoch": 0.05884,
      "grad_norm": 0.8900684740385257,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 5884
    },
    {
      "epoch": 0.05885,
      "grad_norm": 0.8643772869409747,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 5885
    },
    {
      "epoch": 0.05886,
      "grad_norm": 0.9219910015321089,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 5886
    },
    {
      "epoch": 0.05887,
      "grad_norm": 0.9867677093915318,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 5887
    },
    {
      "epoch": 0.05888,
      "grad_norm": 0.9465735583992112,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 5888
    },
    {
      "epoch": 0.05889,
      "grad_norm": 0.8466592224711629,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 5889
    },
    {
      "epoch": 0.0589,
      "grad_norm": 0.886972464545185,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 5890
    },
    {
      "epoch": 0.05891,
      "grad_norm": 0.8652857070984449,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 5891
    },
    {
      "epoch": 0.05892,
      "grad_norm": 0.8664086289943967,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 5892
    },
    {
      "epoch": 0.05893,
      "grad_norm": 0.8708892579951463,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 5893
    },
    {
      "epoch": 0.05894,
      "grad_norm": 0.8078932753206562,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 5894
    },
    {
      "epoch": 0.05895,
      "grad_norm": 0.7575994215774198,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 5895
    },
    {
      "epoch": 0.05896,
      "grad_norm": 0.8549826754505118,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 5896
    },
    {
      "epoch": 0.05897,
      "grad_norm": 1.0144405805900867,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 5897
    },
    {
      "epoch": 0.05898,
      "grad_norm": 0.9395405641055523,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 5898
    },
    {
      "epoch": 0.05899,
      "grad_norm": 0.81787258885681,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 5899
    },
    {
      "epoch": 0.059,
      "grad_norm": 0.7866920164370645,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 5900
    },
    {
      "epoch": 0.05901,
      "grad_norm": 0.910804015247061,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 5901
    },
    {
      "epoch": 0.05902,
      "grad_norm": 0.8747094738532013,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 5902
    },
    {
      "epoch": 0.05903,
      "grad_norm": 0.8330955629581883,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 5903
    },
    {
      "epoch": 0.05904,
      "grad_norm": 0.8149396050827394,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5904
    },
    {
      "epoch": 0.05905,
      "grad_norm": 0.8504004488105502,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 5905
    },
    {
      "epoch": 0.05906,
      "grad_norm": 0.8430930663249769,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 5906
    },
    {
      "epoch": 0.05907,
      "grad_norm": 0.7769082221377612,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 5907
    },
    {
      "epoch": 0.05908,
      "grad_norm": 0.7772454693694203,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 5908
    },
    {
      "epoch": 0.05909,
      "grad_norm": 0.7555830902663292,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 5909
    },
    {
      "epoch": 0.0591,
      "grad_norm": 0.852614526791537,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 5910
    },
    {
      "epoch": 0.05911,
      "grad_norm": 0.9020994358765787,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 5911
    },
    {
      "epoch": 0.05912,
      "grad_norm": 0.8745419850417824,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 5912
    },
    {
      "epoch": 0.05913,
      "grad_norm": 0.8119839119136394,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 5913
    },
    {
      "epoch": 0.05914,
      "grad_norm": 0.8566768807668204,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 5914
    },
    {
      "epoch": 0.05915,
      "grad_norm": 0.8415977723508733,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 5915
    },
    {
      "epoch": 0.05916,
      "grad_norm": 0.8568674252517323,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 5916
    },
    {
      "epoch": 0.05917,
      "grad_norm": 0.9558112684544686,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 5917
    },
    {
      "epoch": 0.05918,
      "grad_norm": 0.841796281005891,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 5918
    },
    {
      "epoch": 0.05919,
      "grad_norm": 0.7555449033676591,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 5919
    },
    {
      "epoch": 0.0592,
      "grad_norm": 0.6802701845197375,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 5920
    },
    {
      "epoch": 0.05921,
      "grad_norm": 0.7044168918621094,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 5921
    },
    {
      "epoch": 0.05922,
      "grad_norm": 0.6836628304390027,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 5922
    },
    {
      "epoch": 0.05923,
      "grad_norm": 0.6799364137142957,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 5923
    },
    {
      "epoch": 0.05924,
      "grad_norm": 0.6507013910949614,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 5924
    },
    {
      "epoch": 0.05925,
      "grad_norm": 0.7486761778135521,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 5925
    },
    {
      "epoch": 0.05926,
      "grad_norm": 1.047983816202794,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 5926
    },
    {
      "epoch": 0.05927,
      "grad_norm": 1.1278171371548802,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 5927
    },
    {
      "epoch": 0.05928,
      "grad_norm": 0.6906724039529817,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 5928
    },
    {
      "epoch": 0.05929,
      "grad_norm": 0.6279774605894174,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 5929
    },
    {
      "epoch": 0.0593,
      "grad_norm": 0.7848995148521671,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 5930
    },
    {
      "epoch": 0.05931,
      "grad_norm": 0.7630809841555396,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 5931
    },
    {
      "epoch": 0.05932,
      "grad_norm": 0.787461470705689,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 5932
    },
    {
      "epoch": 0.05933,
      "grad_norm": 0.9168988991566422,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 5933
    },
    {
      "epoch": 0.05934,
      "grad_norm": 0.9816745251843256,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 5934
    },
    {
      "epoch": 0.05935,
      "grad_norm": 0.9381079865102622,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 5935
    },
    {
      "epoch": 0.05936,
      "grad_norm": 0.7956503064418853,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 5936
    },
    {
      "epoch": 0.05937,
      "grad_norm": 0.7624269514648494,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 5937
    },
    {
      "epoch": 0.05938,
      "grad_norm": 0.7591454779489837,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 5938
    },
    {
      "epoch": 0.05939,
      "grad_norm": 0.8950444564122585,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 5939
    },
    {
      "epoch": 0.0594,
      "grad_norm": 1.0147229565164109,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 5940
    },
    {
      "epoch": 0.05941,
      "grad_norm": 0.8688261982567362,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 5941
    },
    {
      "epoch": 0.05942,
      "grad_norm": 0.8203220118126857,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 5942
    },
    {
      "epoch": 0.05943,
      "grad_norm": 0.6982096168862657,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 5943
    },
    {
      "epoch": 0.05944,
      "grad_norm": 0.7108234697493028,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 5944
    },
    {
      "epoch": 0.05945,
      "grad_norm": 0.7441575726527577,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 5945
    },
    {
      "epoch": 0.05946,
      "grad_norm": 0.693966267768056,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 5946
    },
    {
      "epoch": 0.05947,
      "grad_norm": 0.7492920047746423,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 5947
    },
    {
      "epoch": 0.05948,
      "grad_norm": 0.9531204208790365,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 5948
    },
    {
      "epoch": 0.05949,
      "grad_norm": 1.0780044113765543,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 5949
    },
    {
      "epoch": 0.0595,
      "grad_norm": 1.1155592756719261,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 5950
    },
    {
      "epoch": 0.05951,
      "grad_norm": 0.9010475516605746,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 5951
    },
    {
      "epoch": 0.05952,
      "grad_norm": 0.8136117524297802,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 5952
    },
    {
      "epoch": 0.05953,
      "grad_norm": 0.9497081402575908,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 5953
    },
    {
      "epoch": 0.05954,
      "grad_norm": 1.2015146969192423,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 5954
    },
    {
      "epoch": 0.05955,
      "grad_norm": 1.1019093784349765,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 5955
    },
    {
      "epoch": 0.05956,
      "grad_norm": 0.9208936016687609,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 5956
    },
    {
      "epoch": 0.05957,
      "grad_norm": 0.9862765579790758,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 5957
    },
    {
      "epoch": 0.05958,
      "grad_norm": 0.9376388993356572,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 5958
    },
    {
      "epoch": 0.05959,
      "grad_norm": 0.8845077986280037,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 5959
    },
    {
      "epoch": 0.0596,
      "grad_norm": 0.7693050418942645,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 5960
    },
    {
      "epoch": 0.05961,
      "grad_norm": 0.7082701488122094,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 5961
    },
    {
      "epoch": 0.05962,
      "grad_norm": 0.735178214010553,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 5962
    },
    {
      "epoch": 0.05963,
      "grad_norm": 0.7296949063274745,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 5963
    },
    {
      "epoch": 0.05964,
      "grad_norm": 0.6884040487243323,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 5964
    },
    {
      "epoch": 0.05965,
      "grad_norm": 0.6613609811586886,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 5965
    },
    {
      "epoch": 0.05966,
      "grad_norm": 0.6081326960838325,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 5966
    },
    {
      "epoch": 0.05967,
      "grad_norm": 0.7025319127350109,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 5967
    },
    {
      "epoch": 0.05968,
      "grad_norm": 0.8710659914493641,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 5968
    },
    {
      "epoch": 0.05969,
      "grad_norm": 1.1435903384618258,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 5969
    },
    {
      "epoch": 0.0597,
      "grad_norm": 0.9448264927162972,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 5970
    },
    {
      "epoch": 0.05971,
      "grad_norm": 0.7976448783438047,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 5971
    },
    {
      "epoch": 0.05972,
      "grad_norm": 0.6796283942705301,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 5972
    },
    {
      "epoch": 0.05973,
      "grad_norm": 0.7020776410943722,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 5973
    },
    {
      "epoch": 0.05974,
      "grad_norm": 0.6191112756816605,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 5974
    },
    {
      "epoch": 0.05975,
      "grad_norm": 0.5989871789758782,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 5975
    },
    {
      "epoch": 0.05976,
      "grad_norm": 0.6739238404846538,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 5976
    },
    {
      "epoch": 0.05977,
      "grad_norm": 0.7039494119632328,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 5977
    },
    {
      "epoch": 0.05978,
      "grad_norm": 0.6642661875712308,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 5978
    },
    {
      "epoch": 0.05979,
      "grad_norm": 0.6685425716431447,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 5979
    },
    {
      "epoch": 0.0598,
      "grad_norm": 0.6261981100897624,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 5980
    },
    {
      "epoch": 0.05981,
      "grad_norm": 0.5985779168949105,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 5981
    },
    {
      "epoch": 0.05982,
      "grad_norm": 0.6136781919006569,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 5982
    },
    {
      "epoch": 0.05983,
      "grad_norm": 0.6537718426165939,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5983
    },
    {
      "epoch": 0.05984,
      "grad_norm": 0.714227533885321,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 5984
    },
    {
      "epoch": 0.05985,
      "grad_norm": 0.9506528476418039,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 5985
    },
    {
      "epoch": 0.05986,
      "grad_norm": 1.2746078074067864,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 5986
    },
    {
      "epoch": 0.05987,
      "grad_norm": 0.6791043539335656,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 5987
    },
    {
      "epoch": 0.05988,
      "grad_norm": 0.6346472513213476,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 5988
    },
    {
      "epoch": 0.05989,
      "grad_norm": 0.7500925942098075,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 5989
    },
    {
      "epoch": 0.0599,
      "grad_norm": 0.7561227959508446,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 5990
    },
    {
      "epoch": 0.05991,
      "grad_norm": 0.8673565510034621,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 5991
    },
    {
      "epoch": 0.05992,
      "grad_norm": 0.8299139122744383,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 5992
    },
    {
      "epoch": 0.05993,
      "grad_norm": 0.8298749382371114,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 5993
    },
    {
      "epoch": 0.05994,
      "grad_norm": 0.9205333072442913,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 5994
    },
    {
      "epoch": 0.05995,
      "grad_norm": 1.028139546571208,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 5995
    },
    {
      "epoch": 0.05996,
      "grad_norm": 1.0308864569555287,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 5996
    },
    {
      "epoch": 0.05997,
      "grad_norm": 0.9909951484527934,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 5997
    },
    {
      "epoch": 0.05998,
      "grad_norm": 0.9523678206761688,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 5998
    },
    {
      "epoch": 0.05999,
      "grad_norm": 0.8423597146329871,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 5999
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7574405785566366,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 6000
    },
    {
      "epoch": 0.06001,
      "grad_norm": 0.7753384956998035,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 6001
    },
    {
      "epoch": 0.06002,
      "grad_norm": 0.6748903991582674,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6002
    },
    {
      "epoch": 0.06003,
      "grad_norm": 0.6986365521153964,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6003
    },
    {
      "epoch": 0.06004,
      "grad_norm": 0.7256641113413403,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 6004
    },
    {
      "epoch": 0.06005,
      "grad_norm": 0.9044735035650205,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6005
    },
    {
      "epoch": 0.06006,
      "grad_norm": 0.9755346452792312,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6006
    },
    {
      "epoch": 0.06007,
      "grad_norm": 1.0182340535787422,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 6007
    },
    {
      "epoch": 0.06008,
      "grad_norm": 0.9652283579037206,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6008
    },
    {
      "epoch": 0.06009,
      "grad_norm": 0.9368301156725635,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 6009
    },
    {
      "epoch": 0.0601,
      "grad_norm": 0.8091764219084102,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6010
    },
    {
      "epoch": 0.06011,
      "grad_norm": 0.8468775113048598,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 6011
    },
    {
      "epoch": 0.06012,
      "grad_norm": 0.9029973593955182,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 6012
    },
    {
      "epoch": 0.06013,
      "grad_norm": 0.967030390027878,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6013
    },
    {
      "epoch": 0.06014,
      "grad_norm": 0.8441165732003634,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6014
    },
    {
      "epoch": 0.06015,
      "grad_norm": 0.8268570933571939,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 6015
    },
    {
      "epoch": 0.06016,
      "grad_norm": 0.9865898228103146,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 6016
    },
    {
      "epoch": 0.06017,
      "grad_norm": 1.1052131183203031,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 6017
    },
    {
      "epoch": 0.06018,
      "grad_norm": 0.8257250500559014,
      "learning_rate": 0.003,
      "loss": 4.1614,
      "step": 6018
    },
    {
      "epoch": 0.06019,
      "grad_norm": 0.9972106824107583,
      "learning_rate": 0.003,
      "loss": 4.1339,
      "step": 6019
    },
    {
      "epoch": 0.0602,
      "grad_norm": 0.9191337383387227,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 6020
    },
    {
      "epoch": 0.06021,
      "grad_norm": 0.8616408991185872,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 6021
    },
    {
      "epoch": 0.06022,
      "grad_norm": 0.846750898429168,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 6022
    },
    {
      "epoch": 0.06023,
      "grad_norm": 0.7999915956085327,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 6023
    },
    {
      "epoch": 0.06024,
      "grad_norm": 0.7601013634660952,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 6024
    },
    {
      "epoch": 0.06025,
      "grad_norm": 0.7551843609513086,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 6025
    },
    {
      "epoch": 0.06026,
      "grad_norm": 0.7858596116292706,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6026
    },
    {
      "epoch": 0.06027,
      "grad_norm": 0.9347452644357562,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 6027
    },
    {
      "epoch": 0.06028,
      "grad_norm": 1.235123054449198,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 6028
    },
    {
      "epoch": 0.06029,
      "grad_norm": 0.6475597045882941,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 6029
    },
    {
      "epoch": 0.0603,
      "grad_norm": 0.7127227106122453,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6030
    },
    {
      "epoch": 0.06031,
      "grad_norm": 0.6537901329104299,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 6031
    },
    {
      "epoch": 0.06032,
      "grad_norm": 0.6475267434508163,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6032
    },
    {
      "epoch": 0.06033,
      "grad_norm": 0.612865958283774,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6033
    },
    {
      "epoch": 0.06034,
      "grad_norm": 0.6061910811704923,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 6034
    },
    {
      "epoch": 0.06035,
      "grad_norm": 0.6873968709597013,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 6035
    },
    {
      "epoch": 0.06036,
      "grad_norm": 0.739626150320404,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 6036
    },
    {
      "epoch": 0.06037,
      "grad_norm": 0.8044747493310122,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 6037
    },
    {
      "epoch": 0.06038,
      "grad_norm": 0.7763911045180569,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6038
    },
    {
      "epoch": 0.06039,
      "grad_norm": 0.6590222868385746,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 6039
    },
    {
      "epoch": 0.0604,
      "grad_norm": 0.6350119436647222,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 6040
    },
    {
      "epoch": 0.06041,
      "grad_norm": 0.6853702741345424,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 6041
    },
    {
      "epoch": 0.06042,
      "grad_norm": 0.812479618716859,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6042
    },
    {
      "epoch": 0.06043,
      "grad_norm": 1.0164590695928413,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6043
    },
    {
      "epoch": 0.06044,
      "grad_norm": 1.0986220369036817,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 6044
    },
    {
      "epoch": 0.06045,
      "grad_norm": 0.7560995542851634,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 6045
    },
    {
      "epoch": 0.06046,
      "grad_norm": 0.7425869786438686,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 6046
    },
    {
      "epoch": 0.06047,
      "grad_norm": 0.7895703623032488,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 6047
    },
    {
      "epoch": 0.06048,
      "grad_norm": 0.7640848771237746,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 6048
    },
    {
      "epoch": 0.06049,
      "grad_norm": 0.8191791885730467,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6049
    },
    {
      "epoch": 0.0605,
      "grad_norm": 0.9304880173731273,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6050
    },
    {
      "epoch": 0.06051,
      "grad_norm": 1.0615295174903012,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 6051
    },
    {
      "epoch": 0.06052,
      "grad_norm": 0.9073447027396274,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6052
    },
    {
      "epoch": 0.06053,
      "grad_norm": 0.8100014586142358,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 6053
    },
    {
      "epoch": 0.06054,
      "grad_norm": 0.751311503119369,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6054
    },
    {
      "epoch": 0.06055,
      "grad_norm": 0.7759515169731848,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 6055
    },
    {
      "epoch": 0.06056,
      "grad_norm": 0.8914450153166513,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6056
    },
    {
      "epoch": 0.06057,
      "grad_norm": 1.024211037077155,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 6057
    },
    {
      "epoch": 0.06058,
      "grad_norm": 0.8879874628037983,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6058
    },
    {
      "epoch": 0.06059,
      "grad_norm": 0.6906829909411741,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6059
    },
    {
      "epoch": 0.0606,
      "grad_norm": 0.8104217253555417,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6060
    },
    {
      "epoch": 0.06061,
      "grad_norm": 0.9436958041364509,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 6061
    },
    {
      "epoch": 0.06062,
      "grad_norm": 0.9932174935557901,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 6062
    },
    {
      "epoch": 0.06063,
      "grad_norm": 0.9401163733568618,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 6063
    },
    {
      "epoch": 0.06064,
      "grad_norm": 0.8183782327422785,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 6064
    },
    {
      "epoch": 0.06065,
      "grad_norm": 0.7768072754191233,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 6065
    },
    {
      "epoch": 0.06066,
      "grad_norm": 0.7909830641550517,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 6066
    },
    {
      "epoch": 0.06067,
      "grad_norm": 0.8761823413532669,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 6067
    },
    {
      "epoch": 0.06068,
      "grad_norm": 0.8406889367451033,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6068
    },
    {
      "epoch": 0.06069,
      "grad_norm": 0.8176623171241324,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 6069
    },
    {
      "epoch": 0.0607,
      "grad_norm": 0.8480636571382947,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 6070
    },
    {
      "epoch": 0.06071,
      "grad_norm": 0.9315249540383046,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6071
    },
    {
      "epoch": 0.06072,
      "grad_norm": 0.9083643027747857,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 6072
    },
    {
      "epoch": 0.06073,
      "grad_norm": 0.921573456841131,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 6073
    },
    {
      "epoch": 0.06074,
      "grad_norm": 0.9225097184432787,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6074
    },
    {
      "epoch": 0.06075,
      "grad_norm": 0.7677224007706127,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 6075
    },
    {
      "epoch": 0.06076,
      "grad_norm": 0.6736519570783548,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6076
    },
    {
      "epoch": 0.06077,
      "grad_norm": 0.6250416081043346,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6077
    },
    {
      "epoch": 0.06078,
      "grad_norm": 0.6163479466470738,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6078
    },
    {
      "epoch": 0.06079,
      "grad_norm": 0.5999049345196155,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 6079
    },
    {
      "epoch": 0.0608,
      "grad_norm": 0.601162886689798,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 6080
    },
    {
      "epoch": 0.06081,
      "grad_norm": 0.6240043403002632,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 6081
    },
    {
      "epoch": 0.06082,
      "grad_norm": 0.614892243322943,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 6082
    },
    {
      "epoch": 0.06083,
      "grad_norm": 0.6688855801850269,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 6083
    },
    {
      "epoch": 0.06084,
      "grad_norm": 0.7690977511826896,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 6084
    },
    {
      "epoch": 0.06085,
      "grad_norm": 0.8541611833462315,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 6085
    },
    {
      "epoch": 0.06086,
      "grad_norm": 0.7865155489463524,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 6086
    },
    {
      "epoch": 0.06087,
      "grad_norm": 0.7600736797530553,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 6087
    },
    {
      "epoch": 0.06088,
      "grad_norm": 0.6616973340921765,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 6088
    },
    {
      "epoch": 0.06089,
      "grad_norm": 0.6692763449618849,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 6089
    },
    {
      "epoch": 0.0609,
      "grad_norm": 0.8696023229449554,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 6090
    },
    {
      "epoch": 0.06091,
      "grad_norm": 0.9918899593451429,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6091
    },
    {
      "epoch": 0.06092,
      "grad_norm": 1.0384619852543377,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 6092
    },
    {
      "epoch": 0.06093,
      "grad_norm": 1.2074123219045159,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 6093
    },
    {
      "epoch": 0.06094,
      "grad_norm": 0.9389995680148735,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 6094
    },
    {
      "epoch": 0.06095,
      "grad_norm": 1.0313863077072658,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 6095
    },
    {
      "epoch": 0.06096,
      "grad_norm": 0.9094326514641806,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6096
    },
    {
      "epoch": 0.06097,
      "grad_norm": 0.8515923116062766,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 6097
    },
    {
      "epoch": 0.06098,
      "grad_norm": 0.8412113738456533,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 6098
    },
    {
      "epoch": 0.06099,
      "grad_norm": 0.7500388004828196,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 6099
    },
    {
      "epoch": 0.061,
      "grad_norm": 0.8105097769600541,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 6100
    },
    {
      "epoch": 0.06101,
      "grad_norm": 0.9271622318300181,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 6101
    },
    {
      "epoch": 0.06102,
      "grad_norm": 1.3781789498192056,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 6102
    },
    {
      "epoch": 0.06103,
      "grad_norm": 0.7983625286469737,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 6103
    },
    {
      "epoch": 0.06104,
      "grad_norm": 0.7600704963358303,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 6104
    },
    {
      "epoch": 0.06105,
      "grad_norm": 0.7601314430280259,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 6105
    },
    {
      "epoch": 0.06106,
      "grad_norm": 0.6953532621943936,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 6106
    },
    {
      "epoch": 0.06107,
      "grad_norm": 0.5743377587831264,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 6107
    },
    {
      "epoch": 0.06108,
      "grad_norm": 0.5287888866406477,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6108
    },
    {
      "epoch": 0.06109,
      "grad_norm": 0.4973630164083057,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 6109
    },
    {
      "epoch": 0.0611,
      "grad_norm": 0.5158580251680166,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 6110
    },
    {
      "epoch": 0.06111,
      "grad_norm": 0.5630354022802069,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6111
    },
    {
      "epoch": 0.06112,
      "grad_norm": 0.5923550014514884,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6112
    },
    {
      "epoch": 0.06113,
      "grad_norm": 0.6699929210319788,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6113
    },
    {
      "epoch": 0.06114,
      "grad_norm": 0.7386163212187459,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 6114
    },
    {
      "epoch": 0.06115,
      "grad_norm": 0.8404980906457252,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 6115
    },
    {
      "epoch": 0.06116,
      "grad_norm": 0.9766819519937263,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6116
    },
    {
      "epoch": 0.06117,
      "grad_norm": 1.1615586503825515,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6117
    },
    {
      "epoch": 0.06118,
      "grad_norm": 0.8758457447215398,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 6118
    },
    {
      "epoch": 0.06119,
      "grad_norm": 0.8049538639676183,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 6119
    },
    {
      "epoch": 0.0612,
      "grad_norm": 0.9210216013334553,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 6120
    },
    {
      "epoch": 0.06121,
      "grad_norm": 0.9600155278920769,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6121
    },
    {
      "epoch": 0.06122,
      "grad_norm": 0.9978512170322461,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 6122
    },
    {
      "epoch": 0.06123,
      "grad_norm": 0.9967240552614798,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 6123
    },
    {
      "epoch": 0.06124,
      "grad_norm": 0.974088850622343,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 6124
    },
    {
      "epoch": 0.06125,
      "grad_norm": 0.9600836830620103,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 6125
    },
    {
      "epoch": 0.06126,
      "grad_norm": 0.8700792342844844,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 6126
    },
    {
      "epoch": 0.06127,
      "grad_norm": 0.9271241446223905,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6127
    },
    {
      "epoch": 0.06128,
      "grad_norm": 0.9422494813351958,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6128
    },
    {
      "epoch": 0.06129,
      "grad_norm": 1.059583879974644,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 6129
    },
    {
      "epoch": 0.0613,
      "grad_norm": 1.0947813947090914,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 6130
    },
    {
      "epoch": 0.06131,
      "grad_norm": 1.040398206093915,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 6131
    },
    {
      "epoch": 0.06132,
      "grad_norm": 1.2039883962519413,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 6132
    },
    {
      "epoch": 0.06133,
      "grad_norm": 0.8566788727645377,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 6133
    },
    {
      "epoch": 0.06134,
      "grad_norm": 0.855848356512777,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 6134
    },
    {
      "epoch": 0.06135,
      "grad_norm": 0.7583857270651828,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 6135
    },
    {
      "epoch": 0.06136,
      "grad_norm": 0.7061925305043327,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6136
    },
    {
      "epoch": 0.06137,
      "grad_norm": 0.7565804099972333,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6137
    },
    {
      "epoch": 0.06138,
      "grad_norm": 0.7437903267496562,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6138
    },
    {
      "epoch": 0.06139,
      "grad_norm": 0.7409304538323409,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6139
    },
    {
      "epoch": 0.0614,
      "grad_norm": 0.7536092482113058,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 6140
    },
    {
      "epoch": 0.06141,
      "grad_norm": 0.7483522220373314,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 6141
    },
    {
      "epoch": 0.06142,
      "grad_norm": 0.7859436301216659,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6142
    },
    {
      "epoch": 0.06143,
      "grad_norm": 0.7728075385999797,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6143
    },
    {
      "epoch": 0.06144,
      "grad_norm": 0.7053828491804723,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6144
    },
    {
      "epoch": 0.06145,
      "grad_norm": 0.7192002729715705,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6145
    },
    {
      "epoch": 0.06146,
      "grad_norm": 0.8251909641474612,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6146
    },
    {
      "epoch": 0.06147,
      "grad_norm": 0.9886778253887212,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 6147
    },
    {
      "epoch": 0.06148,
      "grad_norm": 1.155593610839946,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 6148
    },
    {
      "epoch": 0.06149,
      "grad_norm": 0.8070815232350429,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 6149
    },
    {
      "epoch": 0.0615,
      "grad_norm": 0.5204054968874773,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6150
    },
    {
      "epoch": 0.06151,
      "grad_norm": 0.6505364378426297,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6151
    },
    {
      "epoch": 0.06152,
      "grad_norm": 0.7431735128096151,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 6152
    },
    {
      "epoch": 0.06153,
      "grad_norm": 0.7930578549673836,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6153
    },
    {
      "epoch": 0.06154,
      "grad_norm": 0.6504483402063065,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 6154
    },
    {
      "epoch": 0.06155,
      "grad_norm": 0.6346938940236132,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 6155
    },
    {
      "epoch": 0.06156,
      "grad_norm": 0.6953414800898144,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 6156
    },
    {
      "epoch": 0.06157,
      "grad_norm": 0.6331665507038173,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 6157
    },
    {
      "epoch": 0.06158,
      "grad_norm": 0.5534718559856988,
      "learning_rate": 0.003,
      "loss": 4.1388,
      "step": 6158
    },
    {
      "epoch": 0.06159,
      "grad_norm": 0.6648240189265773,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6159
    },
    {
      "epoch": 0.0616,
      "grad_norm": 0.8461177280021167,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 6160
    },
    {
      "epoch": 0.06161,
      "grad_norm": 1.0823146365874887,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 6161
    },
    {
      "epoch": 0.06162,
      "grad_norm": 1.0938191916380575,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 6162
    },
    {
      "epoch": 0.06163,
      "grad_norm": 0.9951722082510337,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 6163
    },
    {
      "epoch": 0.06164,
      "grad_norm": 1.0715869196930037,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 6164
    },
    {
      "epoch": 0.06165,
      "grad_norm": 0.9385585217959526,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 6165
    },
    {
      "epoch": 0.06166,
      "grad_norm": 1.1322976552908572,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 6166
    },
    {
      "epoch": 0.06167,
      "grad_norm": 0.782731511695347,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6167
    },
    {
      "epoch": 0.06168,
      "grad_norm": 0.7067978886921109,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 6168
    },
    {
      "epoch": 0.06169,
      "grad_norm": 0.8159491893240778,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 6169
    },
    {
      "epoch": 0.0617,
      "grad_norm": 0.8210084830581018,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 6170
    },
    {
      "epoch": 0.06171,
      "grad_norm": 0.8561350609180014,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 6171
    },
    {
      "epoch": 0.06172,
      "grad_norm": 0.7783998258196728,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6172
    },
    {
      "epoch": 0.06173,
      "grad_norm": 0.7897242515178252,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 6173
    },
    {
      "epoch": 0.06174,
      "grad_norm": 0.8018311525812623,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 6174
    },
    {
      "epoch": 0.06175,
      "grad_norm": 0.8029275912295462,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6175
    },
    {
      "epoch": 0.06176,
      "grad_norm": 0.8180919970020917,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6176
    },
    {
      "epoch": 0.06177,
      "grad_norm": 0.8315943443066572,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6177
    },
    {
      "epoch": 0.06178,
      "grad_norm": 0.83549072750134,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 6178
    },
    {
      "epoch": 0.06179,
      "grad_norm": 0.7234740322597832,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6179
    },
    {
      "epoch": 0.0618,
      "grad_norm": 0.6996794046268644,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6180
    },
    {
      "epoch": 0.06181,
      "grad_norm": 0.8092448833420763,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6181
    },
    {
      "epoch": 0.06182,
      "grad_norm": 0.967878116319822,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 6182
    },
    {
      "epoch": 0.06183,
      "grad_norm": 1.2411533171982605,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 6183
    },
    {
      "epoch": 0.06184,
      "grad_norm": 0.9159714235499898,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6184
    },
    {
      "epoch": 0.06185,
      "grad_norm": 0.8292810109595354,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 6185
    },
    {
      "epoch": 0.06186,
      "grad_norm": 0.730826666849849,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 6186
    },
    {
      "epoch": 0.06187,
      "grad_norm": 0.7722325434278592,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 6187
    },
    {
      "epoch": 0.06188,
      "grad_norm": 0.7338999457057217,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6188
    },
    {
      "epoch": 0.06189,
      "grad_norm": 0.6370529217825343,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 6189
    },
    {
      "epoch": 0.0619,
      "grad_norm": 0.5776382912555511,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 6190
    },
    {
      "epoch": 0.06191,
      "grad_norm": 0.5174409949409816,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 6191
    },
    {
      "epoch": 0.06192,
      "grad_norm": 0.5449741549919427,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 6192
    },
    {
      "epoch": 0.06193,
      "grad_norm": 0.5306253078122207,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 6193
    },
    {
      "epoch": 0.06194,
      "grad_norm": 0.6667082111771107,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 6194
    },
    {
      "epoch": 0.06195,
      "grad_norm": 1.0558812938178088,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6195
    },
    {
      "epoch": 0.06196,
      "grad_norm": 1.2872949574727126,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 6196
    },
    {
      "epoch": 0.06197,
      "grad_norm": 0.65715222645378,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 6197
    },
    {
      "epoch": 0.06198,
      "grad_norm": 0.8745169993094318,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 6198
    },
    {
      "epoch": 0.06199,
      "grad_norm": 0.9987733008927346,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6199
    },
    {
      "epoch": 0.062,
      "grad_norm": 0.9846095446431814,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 6200
    },
    {
      "epoch": 0.06201,
      "grad_norm": 0.9485154036502214,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6201
    },
    {
      "epoch": 0.06202,
      "grad_norm": 0.8778602800304401,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 6202
    },
    {
      "epoch": 0.06203,
      "grad_norm": 0.8448721905084925,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6203
    },
    {
      "epoch": 0.06204,
      "grad_norm": 0.8474005822451897,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 6204
    },
    {
      "epoch": 0.06205,
      "grad_norm": 0.8173367189925055,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 6205
    },
    {
      "epoch": 0.06206,
      "grad_norm": 0.7111672778214739,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6206
    },
    {
      "epoch": 0.06207,
      "grad_norm": 0.7478977056662977,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 6207
    },
    {
      "epoch": 0.06208,
      "grad_norm": 0.7799493342189914,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 6208
    },
    {
      "epoch": 0.06209,
      "grad_norm": 0.8467347312009388,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 6209
    },
    {
      "epoch": 0.0621,
      "grad_norm": 0.9239306201168563,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 6210
    },
    {
      "epoch": 0.06211,
      "grad_norm": 0.9596942352000194,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 6211
    },
    {
      "epoch": 0.06212,
      "grad_norm": 0.8948519882056875,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 6212
    },
    {
      "epoch": 0.06213,
      "grad_norm": 0.7933444432212635,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 6213
    },
    {
      "epoch": 0.06214,
      "grad_norm": 0.809317950973503,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 6214
    },
    {
      "epoch": 0.06215,
      "grad_norm": 0.876601204166201,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 6215
    },
    {
      "epoch": 0.06216,
      "grad_norm": 0.9644768045541879,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 6216
    },
    {
      "epoch": 0.06217,
      "grad_norm": 1.043411427164528,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6217
    },
    {
      "epoch": 0.06218,
      "grad_norm": 1.072871876469517,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6218
    },
    {
      "epoch": 0.06219,
      "grad_norm": 0.8571177965245572,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 6219
    },
    {
      "epoch": 0.0622,
      "grad_norm": 0.7413732732655652,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6220
    },
    {
      "epoch": 0.06221,
      "grad_norm": 0.68817509861139,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 6221
    },
    {
      "epoch": 0.06222,
      "grad_norm": 0.6601184764512894,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6222
    },
    {
      "epoch": 0.06223,
      "grad_norm": 0.6189770004058355,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6223
    },
    {
      "epoch": 0.06224,
      "grad_norm": 0.5126985323827083,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 6224
    },
    {
      "epoch": 0.06225,
      "grad_norm": 0.549305548358553,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6225
    },
    {
      "epoch": 0.06226,
      "grad_norm": 0.6038854471945458,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 6226
    },
    {
      "epoch": 0.06227,
      "grad_norm": 0.7264264034452508,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6227
    },
    {
      "epoch": 0.06228,
      "grad_norm": 0.8139873862590992,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6228
    },
    {
      "epoch": 0.06229,
      "grad_norm": 0.9132950691176969,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6229
    },
    {
      "epoch": 0.0623,
      "grad_norm": 0.8608762645557783,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6230
    },
    {
      "epoch": 0.06231,
      "grad_norm": 0.7323701212842743,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 6231
    },
    {
      "epoch": 0.06232,
      "grad_norm": 0.7533201360665337,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 6232
    },
    {
      "epoch": 0.06233,
      "grad_norm": 0.7962432286199019,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 6233
    },
    {
      "epoch": 0.06234,
      "grad_norm": 0.8001002251441555,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6234
    },
    {
      "epoch": 0.06235,
      "grad_norm": 0.8282187152840139,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6235
    },
    {
      "epoch": 0.06236,
      "grad_norm": 0.8529888372634628,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6236
    },
    {
      "epoch": 0.06237,
      "grad_norm": 0.8924649306049437,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 6237
    },
    {
      "epoch": 0.06238,
      "grad_norm": 0.9517024598895053,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 6238
    },
    {
      "epoch": 0.06239,
      "grad_norm": 1.090896941369981,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 6239
    },
    {
      "epoch": 0.0624,
      "grad_norm": 0.8988025177459298,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6240
    },
    {
      "epoch": 0.06241,
      "grad_norm": 0.8863798221813798,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 6241
    },
    {
      "epoch": 0.06242,
      "grad_norm": 1.0387484289967834,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 6242
    },
    {
      "epoch": 0.06243,
      "grad_norm": 0.8979931905156554,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6243
    },
    {
      "epoch": 0.06244,
      "grad_norm": 0.8356118354466084,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 6244
    },
    {
      "epoch": 0.06245,
      "grad_norm": 0.8502805150632947,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 6245
    },
    {
      "epoch": 0.06246,
      "grad_norm": 0.9834636405295031,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 6246
    },
    {
      "epoch": 0.06247,
      "grad_norm": 1.1088932528341893,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6247
    },
    {
      "epoch": 0.06248,
      "grad_norm": 1.018053258070267,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6248
    },
    {
      "epoch": 0.06249,
      "grad_norm": 0.9988289602301562,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 6249
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.9321492794691144,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6250
    },
    {
      "epoch": 0.06251,
      "grad_norm": 0.9254351424798546,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 6251
    },
    {
      "epoch": 0.06252,
      "grad_norm": 0.9487439513448277,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 6252
    },
    {
      "epoch": 0.06253,
      "grad_norm": 0.8766138636695966,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6253
    },
    {
      "epoch": 0.06254,
      "grad_norm": 0.8485254787917414,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 6254
    },
    {
      "epoch": 0.06255,
      "grad_norm": 0.7291384807358328,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6255
    },
    {
      "epoch": 0.06256,
      "grad_norm": 0.7872134593442462,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 6256
    },
    {
      "epoch": 0.06257,
      "grad_norm": 0.9363917740480254,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 6257
    },
    {
      "epoch": 0.06258,
      "grad_norm": 1.0738791582349567,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 6258
    },
    {
      "epoch": 0.06259,
      "grad_norm": 1.213752532113539,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 6259
    },
    {
      "epoch": 0.0626,
      "grad_norm": 0.8439975974645768,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 6260
    },
    {
      "epoch": 0.06261,
      "grad_norm": 0.7967008971115906,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6261
    },
    {
      "epoch": 0.06262,
      "grad_norm": 0.8203007205386644,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6262
    },
    {
      "epoch": 0.06263,
      "grad_norm": 0.7563771758779613,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 6263
    },
    {
      "epoch": 0.06264,
      "grad_norm": 0.868641577107568,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6264
    },
    {
      "epoch": 0.06265,
      "grad_norm": 0.8732294846971428,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 6265
    },
    {
      "epoch": 0.06266,
      "grad_norm": 0.7353975928746133,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 6266
    },
    {
      "epoch": 0.06267,
      "grad_norm": 0.7283040841735945,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6267
    },
    {
      "epoch": 0.06268,
      "grad_norm": 0.6976022361269695,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 6268
    },
    {
      "epoch": 0.06269,
      "grad_norm": 0.5944661023232062,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 6269
    },
    {
      "epoch": 0.0627,
      "grad_norm": 0.6219309338060965,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 6270
    },
    {
      "epoch": 0.06271,
      "grad_norm": 0.6423109090656182,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6271
    },
    {
      "epoch": 0.06272,
      "grad_norm": 0.6497521209683546,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 6272
    },
    {
      "epoch": 0.06273,
      "grad_norm": 0.6408710867659153,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6273
    },
    {
      "epoch": 0.06274,
      "grad_norm": 0.6192887789752485,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6274
    },
    {
      "epoch": 0.06275,
      "grad_norm": 0.6261230406806092,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 6275
    },
    {
      "epoch": 0.06276,
      "grad_norm": 0.5976816107591381,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 6276
    },
    {
      "epoch": 0.06277,
      "grad_norm": 0.566538285689989,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6277
    },
    {
      "epoch": 0.06278,
      "grad_norm": 0.6008204917237797,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 6278
    },
    {
      "epoch": 0.06279,
      "grad_norm": 0.7578953746913875,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 6279
    },
    {
      "epoch": 0.0628,
      "grad_norm": 0.9512660338098725,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 6280
    },
    {
      "epoch": 0.06281,
      "grad_norm": 1.441612819184275,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6281
    },
    {
      "epoch": 0.06282,
      "grad_norm": 0.9355215042556002,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 6282
    },
    {
      "epoch": 0.06283,
      "grad_norm": 0.9775330592179509,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6283
    },
    {
      "epoch": 0.06284,
      "grad_norm": 1.152004214068161,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 6284
    },
    {
      "epoch": 0.06285,
      "grad_norm": 0.8632280844541238,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 6285
    },
    {
      "epoch": 0.06286,
      "grad_norm": 0.8688132213283226,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 6286
    },
    {
      "epoch": 0.06287,
      "grad_norm": 0.7521943262262261,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 6287
    },
    {
      "epoch": 0.06288,
      "grad_norm": 0.7590966433142204,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 6288
    },
    {
      "epoch": 0.06289,
      "grad_norm": 0.8547449463965621,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 6289
    },
    {
      "epoch": 0.0629,
      "grad_norm": 0.995426203878218,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6290
    },
    {
      "epoch": 0.06291,
      "grad_norm": 1.0501998061136046,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6291
    },
    {
      "epoch": 0.06292,
      "grad_norm": 0.8258937115933653,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6292
    },
    {
      "epoch": 0.06293,
      "grad_norm": 0.744681874493206,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 6293
    },
    {
      "epoch": 0.06294,
      "grad_norm": 0.6844598828137828,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6294
    },
    {
      "epoch": 0.06295,
      "grad_norm": 0.709427198581793,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 6295
    },
    {
      "epoch": 0.06296,
      "grad_norm": 0.7742480202450152,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 6296
    },
    {
      "epoch": 0.06297,
      "grad_norm": 0.9065810778029665,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 6297
    },
    {
      "epoch": 0.06298,
      "grad_norm": 1.0205749038805025,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 6298
    },
    {
      "epoch": 0.06299,
      "grad_norm": 1.0014978552686065,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6299
    },
    {
      "epoch": 0.063,
      "grad_norm": 0.9906041215988247,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 6300
    },
    {
      "epoch": 0.06301,
      "grad_norm": 1.0237044308753132,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 6301
    },
    {
      "epoch": 0.06302,
      "grad_norm": 0.8716511028680038,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6302
    },
    {
      "epoch": 0.06303,
      "grad_norm": 0.9180601021684964,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6303
    },
    {
      "epoch": 0.06304,
      "grad_norm": 0.8595930720169722,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 6304
    },
    {
      "epoch": 0.06305,
      "grad_norm": 0.8891037067579485,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 6305
    },
    {
      "epoch": 0.06306,
      "grad_norm": 0.835929826094222,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 6306
    },
    {
      "epoch": 0.06307,
      "grad_norm": 0.8222800944063304,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 6307
    },
    {
      "epoch": 0.06308,
      "grad_norm": 0.7978612799259077,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 6308
    },
    {
      "epoch": 0.06309,
      "grad_norm": 0.8169417224723879,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 6309
    },
    {
      "epoch": 0.0631,
      "grad_norm": 0.7885494048599513,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 6310
    },
    {
      "epoch": 0.06311,
      "grad_norm": 0.7141125014011859,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 6311
    },
    {
      "epoch": 0.06312,
      "grad_norm": 0.8834006613536179,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6312
    },
    {
      "epoch": 0.06313,
      "grad_norm": 1.0782426771166267,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6313
    },
    {
      "epoch": 0.06314,
      "grad_norm": 0.9270596290945304,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 6314
    },
    {
      "epoch": 0.06315,
      "grad_norm": 0.866614344975497,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 6315
    },
    {
      "epoch": 0.06316,
      "grad_norm": 0.8025239448074284,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 6316
    },
    {
      "epoch": 0.06317,
      "grad_norm": 0.7483988975313902,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 6317
    },
    {
      "epoch": 0.06318,
      "grad_norm": 0.7026343077048057,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 6318
    },
    {
      "epoch": 0.06319,
      "grad_norm": 0.6258340242384319,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 6319
    },
    {
      "epoch": 0.0632,
      "grad_norm": 0.6022905615388534,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 6320
    },
    {
      "epoch": 0.06321,
      "grad_norm": 0.6537192595187358,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6321
    },
    {
      "epoch": 0.06322,
      "grad_norm": 0.7679534667703762,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 6322
    },
    {
      "epoch": 0.06323,
      "grad_norm": 0.9777865112549871,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6323
    },
    {
      "epoch": 0.06324,
      "grad_norm": 1.1128170132105777,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 6324
    },
    {
      "epoch": 0.06325,
      "grad_norm": 0.9056105963919202,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 6325
    },
    {
      "epoch": 0.06326,
      "grad_norm": 0.9396881242509516,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 6326
    },
    {
      "epoch": 0.06327,
      "grad_norm": 0.9848953169243714,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6327
    },
    {
      "epoch": 0.06328,
      "grad_norm": 1.1839671933476679,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 6328
    },
    {
      "epoch": 0.06329,
      "grad_norm": 0.7896842525783743,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6329
    },
    {
      "epoch": 0.0633,
      "grad_norm": 0.7869900869544297,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 6330
    },
    {
      "epoch": 0.06331,
      "grad_norm": 0.7444958476349693,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6331
    },
    {
      "epoch": 0.06332,
      "grad_norm": 0.7071721517704226,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 6332
    },
    {
      "epoch": 0.06333,
      "grad_norm": 0.6531451777187667,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 6333
    },
    {
      "epoch": 0.06334,
      "grad_norm": 0.5871978280186811,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 6334
    },
    {
      "epoch": 0.06335,
      "grad_norm": 0.5799498876324073,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 6335
    },
    {
      "epoch": 0.06336,
      "grad_norm": 0.5706118531443688,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6336
    },
    {
      "epoch": 0.06337,
      "grad_norm": 0.7146855029967338,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 6337
    },
    {
      "epoch": 0.06338,
      "grad_norm": 0.9100626823469105,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 6338
    },
    {
      "epoch": 0.06339,
      "grad_norm": 1.2461283232196485,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 6339
    },
    {
      "epoch": 0.0634,
      "grad_norm": 0.8399519595087078,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 6340
    },
    {
      "epoch": 0.06341,
      "grad_norm": 0.6934313188966001,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 6341
    },
    {
      "epoch": 0.06342,
      "grad_norm": 0.7330598162651885,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6342
    },
    {
      "epoch": 0.06343,
      "grad_norm": 0.8671585484671817,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 6343
    },
    {
      "epoch": 0.06344,
      "grad_norm": 0.8753238271910776,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 6344
    },
    {
      "epoch": 0.06345,
      "grad_norm": 0.880493691893159,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6345
    },
    {
      "epoch": 0.06346,
      "grad_norm": 0.8605398856134917,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6346
    },
    {
      "epoch": 0.06347,
      "grad_norm": 0.7991693733841341,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 6347
    },
    {
      "epoch": 0.06348,
      "grad_norm": 0.8445072089673568,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6348
    },
    {
      "epoch": 0.06349,
      "grad_norm": 1.013610824242872,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6349
    },
    {
      "epoch": 0.0635,
      "grad_norm": 1.1525443266019426,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 6350
    },
    {
      "epoch": 0.06351,
      "grad_norm": 0.9068123910947427,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 6351
    },
    {
      "epoch": 0.06352,
      "grad_norm": 0.9267793019839832,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 6352
    },
    {
      "epoch": 0.06353,
      "grad_norm": 0.9507357268589157,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6353
    },
    {
      "epoch": 0.06354,
      "grad_norm": 0.8933158322283256,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6354
    },
    {
      "epoch": 0.06355,
      "grad_norm": 0.8410685705766945,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 6355
    },
    {
      "epoch": 0.06356,
      "grad_norm": 0.8005298650181658,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6356
    },
    {
      "epoch": 0.06357,
      "grad_norm": 0.9221272355695364,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 6357
    },
    {
      "epoch": 0.06358,
      "grad_norm": 0.9028874971232729,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 6358
    },
    {
      "epoch": 0.06359,
      "grad_norm": 0.8042560641443137,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6359
    },
    {
      "epoch": 0.0636,
      "grad_norm": 0.9188716749774963,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 6360
    },
    {
      "epoch": 0.06361,
      "grad_norm": 0.95062392680164,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 6361
    },
    {
      "epoch": 0.06362,
      "grad_norm": 1.0546076113357394,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 6362
    },
    {
      "epoch": 0.06363,
      "grad_norm": 0.9203663169616263,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6363
    },
    {
      "epoch": 0.06364,
      "grad_norm": 0.7499435174266673,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6364
    },
    {
      "epoch": 0.06365,
      "grad_norm": 0.6331693994024129,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 6365
    },
    {
      "epoch": 0.06366,
      "grad_norm": 0.5836138837711193,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6366
    },
    {
      "epoch": 0.06367,
      "grad_norm": 0.5905679268129659,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 6367
    },
    {
      "epoch": 0.06368,
      "grad_norm": 0.5546690802747907,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 6368
    },
    {
      "epoch": 0.06369,
      "grad_norm": 0.5409525915376348,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 6369
    },
    {
      "epoch": 0.0637,
      "grad_norm": 0.6041426965862612,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 6370
    },
    {
      "epoch": 0.06371,
      "grad_norm": 0.6134926672713396,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 6371
    },
    {
      "epoch": 0.06372,
      "grad_norm": 0.5585920732749725,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 6372
    },
    {
      "epoch": 0.06373,
      "grad_norm": 0.5617821036929576,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 6373
    },
    {
      "epoch": 0.06374,
      "grad_norm": 0.5884001712501294,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 6374
    },
    {
      "epoch": 0.06375,
      "grad_norm": 0.5836596540025951,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 6375
    },
    {
      "epoch": 0.06376,
      "grad_norm": 0.6350421742417018,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 6376
    },
    {
      "epoch": 0.06377,
      "grad_norm": 0.7167562203574344,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 6377
    },
    {
      "epoch": 0.06378,
      "grad_norm": 0.8951058433252564,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 6378
    },
    {
      "epoch": 0.06379,
      "grad_norm": 1.3635620798766852,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 6379
    },
    {
      "epoch": 0.0638,
      "grad_norm": 1.2248674028316016,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 6380
    },
    {
      "epoch": 0.06381,
      "grad_norm": 0.9100897202260229,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 6381
    },
    {
      "epoch": 0.06382,
      "grad_norm": 0.8036704486012445,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 6382
    },
    {
      "epoch": 0.06383,
      "grad_norm": 0.743174482288171,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6383
    },
    {
      "epoch": 0.06384,
      "grad_norm": 0.8513937818435598,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6384
    },
    {
      "epoch": 0.06385,
      "grad_norm": 1.0786453242708027,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 6385
    },
    {
      "epoch": 0.06386,
      "grad_norm": 1.0238890384448516,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6386
    },
    {
      "epoch": 0.06387,
      "grad_norm": 1.027593845316165,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6387
    },
    {
      "epoch": 0.06388,
      "grad_norm": 0.839413084100022,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 6388
    },
    {
      "epoch": 0.06389,
      "grad_norm": 0.7756185982524284,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 6389
    },
    {
      "epoch": 0.0639,
      "grad_norm": 0.8713850515068945,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 6390
    },
    {
      "epoch": 0.06391,
      "grad_norm": 1.0408467215495305,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 6391
    },
    {
      "epoch": 0.06392,
      "grad_norm": 1.0034083196993608,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 6392
    },
    {
      "epoch": 0.06393,
      "grad_norm": 0.8723434890415933,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6393
    },
    {
      "epoch": 0.06394,
      "grad_norm": 0.8362827621512328,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 6394
    },
    {
      "epoch": 0.06395,
      "grad_norm": 1.0579394140223388,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 6395
    },
    {
      "epoch": 0.06396,
      "grad_norm": 1.0118284529885586,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 6396
    },
    {
      "epoch": 0.06397,
      "grad_norm": 1.0625315745598696,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 6397
    },
    {
      "epoch": 0.06398,
      "grad_norm": 0.9743429452862477,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 6398
    },
    {
      "epoch": 0.06399,
      "grad_norm": 0.9436595857616809,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 6399
    },
    {
      "epoch": 0.064,
      "grad_norm": 1.018780717912782,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 6400
    },
    {
      "epoch": 0.06401,
      "grad_norm": 1.0102187640595444,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 6401
    },
    {
      "epoch": 0.06402,
      "grad_norm": 0.9216577375444452,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 6402
    },
    {
      "epoch": 0.06403,
      "grad_norm": 0.7975560893939748,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 6403
    },
    {
      "epoch": 0.06404,
      "grad_norm": 0.7840951281577243,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 6404
    },
    {
      "epoch": 0.06405,
      "grad_norm": 0.8211264085131749,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6405
    },
    {
      "epoch": 0.06406,
      "grad_norm": 0.8035929209152411,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 6406
    },
    {
      "epoch": 0.06407,
      "grad_norm": 0.8175467906216495,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 6407
    },
    {
      "epoch": 0.06408,
      "grad_norm": 1.001836035975777,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6408
    },
    {
      "epoch": 0.06409,
      "grad_norm": 1.2009899389705134,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 6409
    },
    {
      "epoch": 0.0641,
      "grad_norm": 0.8972232960554846,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 6410
    },
    {
      "epoch": 0.06411,
      "grad_norm": 0.8367366713616154,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 6411
    },
    {
      "epoch": 0.06412,
      "grad_norm": 0.8972350190353556,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 6412
    },
    {
      "epoch": 0.06413,
      "grad_norm": 1.0400706852657329,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 6413
    },
    {
      "epoch": 0.06414,
      "grad_norm": 1.020312511829939,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6414
    },
    {
      "epoch": 0.06415,
      "grad_norm": 1.028531990172739,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6415
    },
    {
      "epoch": 0.06416,
      "grad_norm": 0.8253303199298359,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 6416
    },
    {
      "epoch": 0.06417,
      "grad_norm": 0.6680387166207186,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 6417
    },
    {
      "epoch": 0.06418,
      "grad_norm": 0.6168723571778798,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6418
    },
    {
      "epoch": 0.06419,
      "grad_norm": 0.5368546076983477,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6419
    },
    {
      "epoch": 0.0642,
      "grad_norm": 0.6584077181684485,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 6420
    },
    {
      "epoch": 0.06421,
      "grad_norm": 0.8212576516224894,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 6421
    },
    {
      "epoch": 0.06422,
      "grad_norm": 1.1150116071999399,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6422
    },
    {
      "epoch": 0.06423,
      "grad_norm": 0.9785393710656575,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 6423
    },
    {
      "epoch": 0.06424,
      "grad_norm": 0.84840591881521,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6424
    },
    {
      "epoch": 0.06425,
      "grad_norm": 0.7746501989148333,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6425
    },
    {
      "epoch": 0.06426,
      "grad_norm": 0.7476002784072737,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 6426
    },
    {
      "epoch": 0.06427,
      "grad_norm": 0.7423456692827792,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 6427
    },
    {
      "epoch": 0.06428,
      "grad_norm": 0.7587323475557202,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 6428
    },
    {
      "epoch": 0.06429,
      "grad_norm": 0.746195830101206,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6429
    },
    {
      "epoch": 0.0643,
      "grad_norm": 0.8255820271958185,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 6430
    },
    {
      "epoch": 0.06431,
      "grad_norm": 0.819510641482716,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 6431
    },
    {
      "epoch": 0.06432,
      "grad_norm": 0.6815867999333811,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 6432
    },
    {
      "epoch": 0.06433,
      "grad_norm": 0.6178051256251746,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 6433
    },
    {
      "epoch": 0.06434,
      "grad_norm": 0.6523329798252133,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 6434
    },
    {
      "epoch": 0.06435,
      "grad_norm": 0.6691394409492107,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 6435
    },
    {
      "epoch": 0.06436,
      "grad_norm": 0.6367136162669638,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 6436
    },
    {
      "epoch": 0.06437,
      "grad_norm": 0.7151244539764642,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 6437
    },
    {
      "epoch": 0.06438,
      "grad_norm": 0.9303185507422409,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 6438
    },
    {
      "epoch": 0.06439,
      "grad_norm": 1.155547956758966,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 6439
    },
    {
      "epoch": 0.0644,
      "grad_norm": 0.8106227904496509,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 6440
    },
    {
      "epoch": 0.06441,
      "grad_norm": 0.6604648817786103,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 6441
    },
    {
      "epoch": 0.06442,
      "grad_norm": 0.7373200154591052,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 6442
    },
    {
      "epoch": 0.06443,
      "grad_norm": 0.8450728654117171,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 6443
    },
    {
      "epoch": 0.06444,
      "grad_norm": 1.0151152080080585,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 6444
    },
    {
      "epoch": 0.06445,
      "grad_norm": 1.0956617337515981,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6445
    },
    {
      "epoch": 0.06446,
      "grad_norm": 0.8878041786727323,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 6446
    },
    {
      "epoch": 0.06447,
      "grad_norm": 0.9130724650918877,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 6447
    },
    {
      "epoch": 0.06448,
      "grad_norm": 0.9311141225976478,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 6448
    },
    {
      "epoch": 0.06449,
      "grad_norm": 0.9266467844539056,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 6449
    },
    {
      "epoch": 0.0645,
      "grad_norm": 0.8842097985531295,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 6450
    },
    {
      "epoch": 0.06451,
      "grad_norm": 0.9231514896987845,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 6451
    },
    {
      "epoch": 0.06452,
      "grad_norm": 1.0541860942458612,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 6452
    },
    {
      "epoch": 0.06453,
      "grad_norm": 0.9145490815901766,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 6453
    },
    {
      "epoch": 0.06454,
      "grad_norm": 0.826571008490979,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6454
    },
    {
      "epoch": 0.06455,
      "grad_norm": 0.7020677977773648,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 6455
    },
    {
      "epoch": 0.06456,
      "grad_norm": 0.6693276104354243,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 6456
    },
    {
      "epoch": 0.06457,
      "grad_norm": 0.7996629696131489,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6457
    },
    {
      "epoch": 0.06458,
      "grad_norm": 0.9270705106073829,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6458
    },
    {
      "epoch": 0.06459,
      "grad_norm": 1.074497828894076,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 6459
    },
    {
      "epoch": 0.0646,
      "grad_norm": 0.9551554373739378,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6460
    },
    {
      "epoch": 0.06461,
      "grad_norm": 0.8033640397123838,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 6461
    },
    {
      "epoch": 0.06462,
      "grad_norm": 0.8571886698508542,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 6462
    },
    {
      "epoch": 0.06463,
      "grad_norm": 0.8476034867616584,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 6463
    },
    {
      "epoch": 0.06464,
      "grad_norm": 0.9308796780668124,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 6464
    },
    {
      "epoch": 0.06465,
      "grad_norm": 0.8577677132113817,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 6465
    },
    {
      "epoch": 0.06466,
      "grad_norm": 0.9685221965234145,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6466
    },
    {
      "epoch": 0.06467,
      "grad_norm": 0.947967210756858,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 6467
    },
    {
      "epoch": 0.06468,
      "grad_norm": 1.2313807236375918,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 6468
    },
    {
      "epoch": 0.06469,
      "grad_norm": 1.013089232311945,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6469
    },
    {
      "epoch": 0.0647,
      "grad_norm": 0.7873532128436814,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6470
    },
    {
      "epoch": 0.06471,
      "grad_norm": 0.7728490694203982,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 6471
    },
    {
      "epoch": 0.06472,
      "grad_norm": 0.8569138079643775,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 6472
    },
    {
      "epoch": 0.06473,
      "grad_norm": 0.8903789705045198,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 6473
    },
    {
      "epoch": 0.06474,
      "grad_norm": 0.8414339883577647,
      "learning_rate": 0.003,
      "loss": 4.1485,
      "step": 6474
    },
    {
      "epoch": 0.06475,
      "grad_norm": 0.8185119706104915,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 6475
    },
    {
      "epoch": 0.06476,
      "grad_norm": 0.7735489074806816,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 6476
    },
    {
      "epoch": 0.06477,
      "grad_norm": 0.692541672437415,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 6477
    },
    {
      "epoch": 0.06478,
      "grad_norm": 0.6687914180889757,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 6478
    },
    {
      "epoch": 0.06479,
      "grad_norm": 0.6777500505762841,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6479
    },
    {
      "epoch": 0.0648,
      "grad_norm": 0.6486993051596719,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 6480
    },
    {
      "epoch": 0.06481,
      "grad_norm": 0.6227452111637245,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 6481
    },
    {
      "epoch": 0.06482,
      "grad_norm": 0.6060284390551969,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 6482
    },
    {
      "epoch": 0.06483,
      "grad_norm": 0.7307292793303055,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6483
    },
    {
      "epoch": 0.06484,
      "grad_norm": 0.8150760871337055,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 6484
    },
    {
      "epoch": 0.06485,
      "grad_norm": 0.9208909630123016,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6485
    },
    {
      "epoch": 0.06486,
      "grad_norm": 1.1799194145477874,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 6486
    },
    {
      "epoch": 0.06487,
      "grad_norm": 0.8822420101848714,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6487
    },
    {
      "epoch": 0.06488,
      "grad_norm": 0.8457525546814305,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 6488
    },
    {
      "epoch": 0.06489,
      "grad_norm": 0.8600818951575393,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 6489
    },
    {
      "epoch": 0.0649,
      "grad_norm": 0.8039264767899832,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 6490
    },
    {
      "epoch": 0.06491,
      "grad_norm": 0.8760540483969935,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 6491
    },
    {
      "epoch": 0.06492,
      "grad_norm": 0.82749202495857,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 6492
    },
    {
      "epoch": 0.06493,
      "grad_norm": 0.8943255943541918,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 6493
    },
    {
      "epoch": 0.06494,
      "grad_norm": 0.9825172024300408,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 6494
    },
    {
      "epoch": 0.06495,
      "grad_norm": 0.9527896010651128,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6495
    },
    {
      "epoch": 0.06496,
      "grad_norm": 0.8119745595177673,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 6496
    },
    {
      "epoch": 0.06497,
      "grad_norm": 0.6621216446190086,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 6497
    },
    {
      "epoch": 0.06498,
      "grad_norm": 0.6064887728246845,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 6498
    },
    {
      "epoch": 0.06499,
      "grad_norm": 0.5707453577641534,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6499
    },
    {
      "epoch": 0.065,
      "grad_norm": 0.5073270473104914,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 6500
    },
    {
      "epoch": 0.06501,
      "grad_norm": 0.5208394304660492,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 6501
    },
    {
      "epoch": 0.06502,
      "grad_norm": 0.5859156335669509,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 6502
    },
    {
      "epoch": 0.06503,
      "grad_norm": 0.6319278647879405,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6503
    },
    {
      "epoch": 0.06504,
      "grad_norm": 0.8345350267691076,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 6504
    },
    {
      "epoch": 0.06505,
      "grad_norm": 1.022594666565787,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 6505
    },
    {
      "epoch": 0.06506,
      "grad_norm": 1.1082118647709895,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 6506
    },
    {
      "epoch": 0.06507,
      "grad_norm": 0.9631150466533694,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 6507
    },
    {
      "epoch": 0.06508,
      "grad_norm": 0.9358226724585814,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 6508
    },
    {
      "epoch": 0.06509,
      "grad_norm": 0.903320053092754,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 6509
    },
    {
      "epoch": 0.0651,
      "grad_norm": 0.7988134798037461,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 6510
    },
    {
      "epoch": 0.06511,
      "grad_norm": 1.0115086044546666,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 6511
    },
    {
      "epoch": 0.06512,
      "grad_norm": 1.1339344083529093,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6512
    },
    {
      "epoch": 0.06513,
      "grad_norm": 0.9322613192351319,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6513
    },
    {
      "epoch": 0.06514,
      "grad_norm": 1.0489869510777938,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 6514
    },
    {
      "epoch": 0.06515,
      "grad_norm": 0.8551726865667978,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 6515
    },
    {
      "epoch": 0.06516,
      "grad_norm": 0.8303598702103568,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6516
    },
    {
      "epoch": 0.06517,
      "grad_norm": 0.9044563142027361,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 6517
    },
    {
      "epoch": 0.06518,
      "grad_norm": 0.8703001988372849,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 6518
    },
    {
      "epoch": 0.06519,
      "grad_norm": 0.7847889819210776,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 6519
    },
    {
      "epoch": 0.0652,
      "grad_norm": 0.7604897553062402,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 6520
    },
    {
      "epoch": 0.06521,
      "grad_norm": 0.7045491663860287,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6521
    },
    {
      "epoch": 0.06522,
      "grad_norm": 0.5967359913316697,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6522
    },
    {
      "epoch": 0.06523,
      "grad_norm": 0.5994954551174616,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 6523
    },
    {
      "epoch": 0.06524,
      "grad_norm": 0.5929713355650309,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 6524
    },
    {
      "epoch": 0.06525,
      "grad_norm": 0.5725345126769195,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 6525
    },
    {
      "epoch": 0.06526,
      "grad_norm": 0.6769384736460269,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 6526
    },
    {
      "epoch": 0.06527,
      "grad_norm": 0.9123198261221063,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 6527
    },
    {
      "epoch": 0.06528,
      "grad_norm": 1.297782323536123,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 6528
    },
    {
      "epoch": 0.06529,
      "grad_norm": 0.9911311317864455,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 6529
    },
    {
      "epoch": 0.0653,
      "grad_norm": 1.0305368070734895,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 6530
    },
    {
      "epoch": 0.06531,
      "grad_norm": 0.9401086376942769,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 6531
    },
    {
      "epoch": 0.06532,
      "grad_norm": 0.9614097857861146,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 6532
    },
    {
      "epoch": 0.06533,
      "grad_norm": 1.0868494247089455,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 6533
    },
    {
      "epoch": 0.06534,
      "grad_norm": 1.0009233535813384,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 6534
    },
    {
      "epoch": 0.06535,
      "grad_norm": 1.0081228563907538,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 6535
    },
    {
      "epoch": 0.06536,
      "grad_norm": 0.9645958863688807,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 6536
    },
    {
      "epoch": 0.06537,
      "grad_norm": 0.8769309680301732,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 6537
    },
    {
      "epoch": 0.06538,
      "grad_norm": 0.8137137300201961,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 6538
    },
    {
      "epoch": 0.06539,
      "grad_norm": 0.727085328822687,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 6539
    },
    {
      "epoch": 0.0654,
      "grad_norm": 0.7005089827719883,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 6540
    },
    {
      "epoch": 0.06541,
      "grad_norm": 0.6228763105598434,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6541
    },
    {
      "epoch": 0.06542,
      "grad_norm": 0.7516112721925631,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 6542
    },
    {
      "epoch": 0.06543,
      "grad_norm": 1.0509494597212574,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 6543
    },
    {
      "epoch": 0.06544,
      "grad_norm": 1.1940069365357546,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6544
    },
    {
      "epoch": 0.06545,
      "grad_norm": 0.8150174480762931,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 6545
    },
    {
      "epoch": 0.06546,
      "grad_norm": 0.9408126901018795,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6546
    },
    {
      "epoch": 0.06547,
      "grad_norm": 0.985809825256504,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6547
    },
    {
      "epoch": 0.06548,
      "grad_norm": 1.016850972033843,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 6548
    },
    {
      "epoch": 0.06549,
      "grad_norm": 0.9915312968382956,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 6549
    },
    {
      "epoch": 0.0655,
      "grad_norm": 0.7584853863997532,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 6550
    },
    {
      "epoch": 0.06551,
      "grad_norm": 0.7532414275895373,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 6551
    },
    {
      "epoch": 0.06552,
      "grad_norm": 0.7132878191635362,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 6552
    },
    {
      "epoch": 0.06553,
      "grad_norm": 0.7251259028750098,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 6553
    },
    {
      "epoch": 0.06554,
      "grad_norm": 0.6869756855220046,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6554
    },
    {
      "epoch": 0.06555,
      "grad_norm": 0.6789180373444202,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 6555
    },
    {
      "epoch": 0.06556,
      "grad_norm": 0.7688068758022717,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 6556
    },
    {
      "epoch": 0.06557,
      "grad_norm": 0.820506804409578,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 6557
    },
    {
      "epoch": 0.06558,
      "grad_norm": 1.0451835481907836,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 6558
    },
    {
      "epoch": 0.06559,
      "grad_norm": 1.0506338412291107,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 6559
    },
    {
      "epoch": 0.0656,
      "grad_norm": 0.8227023399945246,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6560
    },
    {
      "epoch": 0.06561,
      "grad_norm": 0.8263825289183352,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 6561
    },
    {
      "epoch": 0.06562,
      "grad_norm": 0.9118099179127191,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 6562
    },
    {
      "epoch": 0.06563,
      "grad_norm": 1.008868981921078,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 6563
    },
    {
      "epoch": 0.06564,
      "grad_norm": 0.8794384654321471,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 6564
    },
    {
      "epoch": 0.06565,
      "grad_norm": 0.7799483866337282,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 6565
    },
    {
      "epoch": 0.06566,
      "grad_norm": 0.7247305841084201,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 6566
    },
    {
      "epoch": 0.06567,
      "grad_norm": 0.7781944614519534,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6567
    },
    {
      "epoch": 0.06568,
      "grad_norm": 0.8470435504646778,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 6568
    },
    {
      "epoch": 0.06569,
      "grad_norm": 0.8975263588278858,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 6569
    },
    {
      "epoch": 0.0657,
      "grad_norm": 0.9290752582286405,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 6570
    },
    {
      "epoch": 0.06571,
      "grad_norm": 0.9170887562858809,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6571
    },
    {
      "epoch": 0.06572,
      "grad_norm": 0.9006929422375534,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 6572
    },
    {
      "epoch": 0.06573,
      "grad_norm": 0.9322962718617643,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6573
    },
    {
      "epoch": 0.06574,
      "grad_norm": 0.7833680952020777,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 6574
    },
    {
      "epoch": 0.06575,
      "grad_norm": 0.7445025969000303,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 6575
    },
    {
      "epoch": 0.06576,
      "grad_norm": 0.7569699043405889,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 6576
    },
    {
      "epoch": 0.06577,
      "grad_norm": 0.9003358954292493,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 6577
    },
    {
      "epoch": 0.06578,
      "grad_norm": 1.0621423432401027,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 6578
    },
    {
      "epoch": 0.06579,
      "grad_norm": 1.1726402516475503,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 6579
    },
    {
      "epoch": 0.0658,
      "grad_norm": 0.8552429990963719,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 6580
    },
    {
      "epoch": 0.06581,
      "grad_norm": 0.8514271483398563,
      "learning_rate": 0.003,
      "loss": 4.1324,
      "step": 6581
    },
    {
      "epoch": 0.06582,
      "grad_norm": 0.9602464463060446,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6582
    },
    {
      "epoch": 0.06583,
      "grad_norm": 0.9829530100461856,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 6583
    },
    {
      "epoch": 0.06584,
      "grad_norm": 0.9171258754858127,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 6584
    },
    {
      "epoch": 0.06585,
      "grad_norm": 0.84935508857,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6585
    },
    {
      "epoch": 0.06586,
      "grad_norm": 0.8705786102009245,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 6586
    },
    {
      "epoch": 0.06587,
      "grad_norm": 0.9714069205255046,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 6587
    },
    {
      "epoch": 0.06588,
      "grad_norm": 0.9127013611939936,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 6588
    },
    {
      "epoch": 0.06589,
      "grad_norm": 0.864633992723417,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 6589
    },
    {
      "epoch": 0.0659,
      "grad_norm": 0.7734108219438173,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 6590
    },
    {
      "epoch": 0.06591,
      "grad_norm": 0.7864629956935438,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 6591
    },
    {
      "epoch": 0.06592,
      "grad_norm": 0.7865367848858412,
      "learning_rate": 0.003,
      "loss": 4.1272,
      "step": 6592
    },
    {
      "epoch": 0.06593,
      "grad_norm": 0.795726928953009,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6593
    },
    {
      "epoch": 0.06594,
      "grad_norm": 0.9228436883463198,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 6594
    },
    {
      "epoch": 0.06595,
      "grad_norm": 0.999263149903344,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 6595
    },
    {
      "epoch": 0.06596,
      "grad_norm": 0.8576980929926036,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6596
    },
    {
      "epoch": 0.06597,
      "grad_norm": 0.7302418341288966,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6597
    },
    {
      "epoch": 0.06598,
      "grad_norm": 0.7765021212690375,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 6598
    },
    {
      "epoch": 0.06599,
      "grad_norm": 0.8935858983881297,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 6599
    },
    {
      "epoch": 0.066,
      "grad_norm": 1.0268301145305159,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 6600
    },
    {
      "epoch": 0.06601,
      "grad_norm": 0.9441826567712125,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 6601
    },
    {
      "epoch": 0.06602,
      "grad_norm": 0.8826431920881573,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 6602
    },
    {
      "epoch": 0.06603,
      "grad_norm": 0.6782729476695759,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 6603
    },
    {
      "epoch": 0.06604,
      "grad_norm": 0.721923977940998,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 6604
    },
    {
      "epoch": 0.06605,
      "grad_norm": 0.8126707635116521,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 6605
    },
    {
      "epoch": 0.06606,
      "grad_norm": 0.9139955529917244,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 6606
    },
    {
      "epoch": 0.06607,
      "grad_norm": 0.9787805099146487,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6607
    },
    {
      "epoch": 0.06608,
      "grad_norm": 1.0120287022530385,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 6608
    },
    {
      "epoch": 0.06609,
      "grad_norm": 0.9553503023983144,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6609
    },
    {
      "epoch": 0.0661,
      "grad_norm": 0.773415990938354,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 6610
    },
    {
      "epoch": 0.06611,
      "grad_norm": 0.7654329601140009,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 6611
    },
    {
      "epoch": 0.06612,
      "grad_norm": 0.7452194084183761,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 6612
    },
    {
      "epoch": 0.06613,
      "grad_norm": 0.8299039489543768,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 6613
    },
    {
      "epoch": 0.06614,
      "grad_norm": 0.8093687816033562,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 6614
    },
    {
      "epoch": 0.06615,
      "grad_norm": 0.8207953880857137,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 6615
    },
    {
      "epoch": 0.06616,
      "grad_norm": 0.8552296704671903,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 6616
    },
    {
      "epoch": 0.06617,
      "grad_norm": 0.9244751774090344,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 6617
    },
    {
      "epoch": 0.06618,
      "grad_norm": 0.971219029341592,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 6618
    },
    {
      "epoch": 0.06619,
      "grad_norm": 0.8506209302501583,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 6619
    },
    {
      "epoch": 0.0662,
      "grad_norm": 0.8713201673630117,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6620
    },
    {
      "epoch": 0.06621,
      "grad_norm": 0.8914133581100084,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 6621
    },
    {
      "epoch": 0.06622,
      "grad_norm": 0.9005033881236821,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 6622
    },
    {
      "epoch": 0.06623,
      "grad_norm": 1.0929062045922742,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 6623
    },
    {
      "epoch": 0.06624,
      "grad_norm": 1.0309406977127722,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 6624
    },
    {
      "epoch": 0.06625,
      "grad_norm": 0.9349272080200425,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 6625
    },
    {
      "epoch": 0.06626,
      "grad_norm": 0.9406574766318107,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 6626
    },
    {
      "epoch": 0.06627,
      "grad_norm": 0.8215318070242014,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 6627
    },
    {
      "epoch": 0.06628,
      "grad_norm": 0.8411166912060155,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 6628
    },
    {
      "epoch": 0.06629,
      "grad_norm": 0.778414099268401,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 6629
    },
    {
      "epoch": 0.0663,
      "grad_norm": 0.7994566410875287,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 6630
    },
    {
      "epoch": 0.06631,
      "grad_norm": 0.9203945265715889,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6631
    },
    {
      "epoch": 0.06632,
      "grad_norm": 0.9593833055548883,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6632
    },
    {
      "epoch": 0.06633,
      "grad_norm": 0.8574352076483307,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 6633
    },
    {
      "epoch": 0.06634,
      "grad_norm": 0.7496261857566855,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 6634
    },
    {
      "epoch": 0.06635,
      "grad_norm": 0.6760158781401536,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 6635
    },
    {
      "epoch": 0.06636,
      "grad_norm": 0.6916307102085072,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 6636
    },
    {
      "epoch": 0.06637,
      "grad_norm": 0.6966845658280828,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 6637
    },
    {
      "epoch": 0.06638,
      "grad_norm": 0.6883147167526921,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 6638
    },
    {
      "epoch": 0.06639,
      "grad_norm": 0.6473901715950106,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 6639
    },
    {
      "epoch": 0.0664,
      "grad_norm": 0.7765981197562362,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6640
    },
    {
      "epoch": 0.06641,
      "grad_norm": 1.115678151117138,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6641
    },
    {
      "epoch": 0.06642,
      "grad_norm": 0.9790163721838566,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 6642
    },
    {
      "epoch": 0.06643,
      "grad_norm": 0.8361408902689975,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 6643
    },
    {
      "epoch": 0.06644,
      "grad_norm": 0.7584196343566021,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6644
    },
    {
      "epoch": 0.06645,
      "grad_norm": 0.816065178623302,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 6645
    },
    {
      "epoch": 0.06646,
      "grad_norm": 0.8060094631825453,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 6646
    },
    {
      "epoch": 0.06647,
      "grad_norm": 0.757636506404951,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 6647
    },
    {
      "epoch": 0.06648,
      "grad_norm": 0.8404369742133107,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 6648
    },
    {
      "epoch": 0.06649,
      "grad_norm": 0.8868264232337089,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 6649
    },
    {
      "epoch": 0.0665,
      "grad_norm": 0.8472513295914429,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 6650
    },
    {
      "epoch": 0.06651,
      "grad_norm": 0.7624647071594521,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 6651
    },
    {
      "epoch": 0.06652,
      "grad_norm": 0.70113702299994,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 6652
    },
    {
      "epoch": 0.06653,
      "grad_norm": 0.7170179450934846,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 6653
    },
    {
      "epoch": 0.06654,
      "grad_norm": 0.6408955553595038,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 6654
    },
    {
      "epoch": 0.06655,
      "grad_norm": 0.7303405365737384,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6655
    },
    {
      "epoch": 0.06656,
      "grad_norm": 0.8170077189908054,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 6656
    },
    {
      "epoch": 0.06657,
      "grad_norm": 1.0373997656423108,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 6657
    },
    {
      "epoch": 0.06658,
      "grad_norm": 1.2877634414846078,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 6658
    },
    {
      "epoch": 0.06659,
      "grad_norm": 0.7003902163594509,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 6659
    },
    {
      "epoch": 0.0666,
      "grad_norm": 0.7777938489458038,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6660
    },
    {
      "epoch": 0.06661,
      "grad_norm": 0.759383450915605,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 6661
    },
    {
      "epoch": 0.06662,
      "grad_norm": 0.6357206487995134,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 6662
    },
    {
      "epoch": 0.06663,
      "grad_norm": 0.5997016664182901,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 6663
    },
    {
      "epoch": 0.06664,
      "grad_norm": 0.6928418683619769,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 6664
    },
    {
      "epoch": 0.06665,
      "grad_norm": 0.6725431198503523,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 6665
    },
    {
      "epoch": 0.06666,
      "grad_norm": 0.7185786516891489,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 6666
    },
    {
      "epoch": 0.06667,
      "grad_norm": 0.7583846091674911,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 6667
    },
    {
      "epoch": 0.06668,
      "grad_norm": 0.7727383750975167,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 6668
    },
    {
      "epoch": 0.06669,
      "grad_norm": 0.7990602062120472,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 6669
    },
    {
      "epoch": 0.0667,
      "grad_norm": 0.8932825213449198,
      "learning_rate": 0.003,
      "loss": 4.1163,
      "step": 6670
    },
    {
      "epoch": 0.06671,
      "grad_norm": 0.9161270016145233,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 6671
    },
    {
      "epoch": 0.06672,
      "grad_norm": 0.8948099688910036,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6672
    },
    {
      "epoch": 0.06673,
      "grad_norm": 0.8131754380634141,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 6673
    },
    {
      "epoch": 0.06674,
      "grad_norm": 0.9822219202437968,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 6674
    },
    {
      "epoch": 0.06675,
      "grad_norm": 1.4384917377302466,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 6675
    },
    {
      "epoch": 0.06676,
      "grad_norm": 1.0715294490450158,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 6676
    },
    {
      "epoch": 0.06677,
      "grad_norm": 0.9943884055205793,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 6677
    },
    {
      "epoch": 0.06678,
      "grad_norm": 1.1472283624566484,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 6678
    },
    {
      "epoch": 0.06679,
      "grad_norm": 0.837081045625505,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6679
    },
    {
      "epoch": 0.0668,
      "grad_norm": 0.8632696497966131,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 6680
    },
    {
      "epoch": 0.06681,
      "grad_norm": 0.881128794076206,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 6681
    },
    {
      "epoch": 0.06682,
      "grad_norm": 0.8577139639731307,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 6682
    },
    {
      "epoch": 0.06683,
      "grad_norm": 0.9681904531791742,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 6683
    },
    {
      "epoch": 0.06684,
      "grad_norm": 1.045816856105673,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6684
    },
    {
      "epoch": 0.06685,
      "grad_norm": 1.1727987324167213,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 6685
    },
    {
      "epoch": 0.06686,
      "grad_norm": 1.027054896660675,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6686
    },
    {
      "epoch": 0.06687,
      "grad_norm": 0.9524644542845865,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 6687
    },
    {
      "epoch": 0.06688,
      "grad_norm": 0.8156341843328404,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 6688
    },
    {
      "epoch": 0.06689,
      "grad_norm": 0.7545435705045717,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 6689
    },
    {
      "epoch": 0.0669,
      "grad_norm": 0.7352426544757619,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6690
    },
    {
      "epoch": 0.06691,
      "grad_norm": 0.6665883072114859,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 6691
    },
    {
      "epoch": 0.06692,
      "grad_norm": 0.6725148756303375,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 6692
    },
    {
      "epoch": 0.06693,
      "grad_norm": 0.7130842452638179,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6693
    },
    {
      "epoch": 0.06694,
      "grad_norm": 0.8096745511420714,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6694
    },
    {
      "epoch": 0.06695,
      "grad_norm": 0.904033151495042,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6695
    },
    {
      "epoch": 0.06696,
      "grad_norm": 0.9485066116960444,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6696
    },
    {
      "epoch": 0.06697,
      "grad_norm": 0.9268917013823743,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 6697
    },
    {
      "epoch": 0.06698,
      "grad_norm": 0.9285500128581322,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 6698
    },
    {
      "epoch": 0.06699,
      "grad_norm": 1.0584540045258635,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6699
    },
    {
      "epoch": 0.067,
      "grad_norm": 1.0373189607989615,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 6700
    },
    {
      "epoch": 0.06701,
      "grad_norm": 0.9623285914421408,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 6701
    },
    {
      "epoch": 0.06702,
      "grad_norm": 0.9701353847551325,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 6702
    },
    {
      "epoch": 0.06703,
      "grad_norm": 0.8858538800012404,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 6703
    },
    {
      "epoch": 0.06704,
      "grad_norm": 0.9016995986468413,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 6704
    },
    {
      "epoch": 0.06705,
      "grad_norm": 0.925225581835428,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 6705
    },
    {
      "epoch": 0.06706,
      "grad_norm": 0.896855586772984,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 6706
    },
    {
      "epoch": 0.06707,
      "grad_norm": 0.7977187311047266,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 6707
    },
    {
      "epoch": 0.06708,
      "grad_norm": 0.8299509664392212,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 6708
    },
    {
      "epoch": 0.06709,
      "grad_norm": 0.9460884720074625,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 6709
    },
    {
      "epoch": 0.0671,
      "grad_norm": 0.8571195032444293,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6710
    },
    {
      "epoch": 0.06711,
      "grad_norm": 0.9023183298505612,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 6711
    },
    {
      "epoch": 0.06712,
      "grad_norm": 1.1030390737820472,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6712
    },
    {
      "epoch": 0.06713,
      "grad_norm": 0.9582329185489545,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 6713
    },
    {
      "epoch": 0.06714,
      "grad_norm": 0.8269350900476661,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 6714
    },
    {
      "epoch": 0.06715,
      "grad_norm": 0.7110607112083103,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 6715
    },
    {
      "epoch": 0.06716,
      "grad_norm": 0.7100774219805601,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 6716
    },
    {
      "epoch": 0.06717,
      "grad_norm": 0.855459300866275,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 6717
    },
    {
      "epoch": 0.06718,
      "grad_norm": 0.9053889950699634,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 6718
    },
    {
      "epoch": 0.06719,
      "grad_norm": 0.8719291055963075,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 6719
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.8359415545682723,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 6720
    },
    {
      "epoch": 0.06721,
      "grad_norm": 0.8906370466300249,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 6721
    },
    {
      "epoch": 0.06722,
      "grad_norm": 1.0176449340197307,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 6722
    },
    {
      "epoch": 0.06723,
      "grad_norm": 0.9805817985293129,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 6723
    },
    {
      "epoch": 0.06724,
      "grad_norm": 0.9703246693491563,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6724
    },
    {
      "epoch": 0.06725,
      "grad_norm": 0.9242585229701251,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 6725
    },
    {
      "epoch": 0.06726,
      "grad_norm": 0.8600589534698727,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 6726
    },
    {
      "epoch": 0.06727,
      "grad_norm": 0.950790968925123,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 6727
    },
    {
      "epoch": 0.06728,
      "grad_norm": 1.1631164741777629,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 6728
    },
    {
      "epoch": 0.06729,
      "grad_norm": 0.7915866355485529,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 6729
    },
    {
      "epoch": 0.0673,
      "grad_norm": 0.8274594227494144,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 6730
    },
    {
      "epoch": 0.06731,
      "grad_norm": 0.6766878958185836,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 6731
    },
    {
      "epoch": 0.06732,
      "grad_norm": 0.6976909564398467,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 6732
    },
    {
      "epoch": 0.06733,
      "grad_norm": 0.7614581431023822,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 6733
    },
    {
      "epoch": 0.06734,
      "grad_norm": 1.057194128915908,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 6734
    },
    {
      "epoch": 0.06735,
      "grad_norm": 1.213622998381686,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 6735
    },
    {
      "epoch": 0.06736,
      "grad_norm": 0.734105473964094,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 6736
    },
    {
      "epoch": 0.06737,
      "grad_norm": 0.6158143832281706,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 6737
    },
    {
      "epoch": 0.06738,
      "grad_norm": 0.5821932035460526,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 6738
    },
    {
      "epoch": 0.06739,
      "grad_norm": 0.6445881524656328,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 6739
    },
    {
      "epoch": 0.0674,
      "grad_norm": 0.7083800475248121,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 6740
    },
    {
      "epoch": 0.06741,
      "grad_norm": 0.8121921490974171,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 6741
    },
    {
      "epoch": 0.06742,
      "grad_norm": 0.9424065146650864,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 6742
    },
    {
      "epoch": 0.06743,
      "grad_norm": 1.0699042142898014,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 6743
    },
    {
      "epoch": 0.06744,
      "grad_norm": 1.047616453763929,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6744
    },
    {
      "epoch": 0.06745,
      "grad_norm": 0.897040372045614,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 6745
    },
    {
      "epoch": 0.06746,
      "grad_norm": 0.8107749141615237,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6746
    },
    {
      "epoch": 0.06747,
      "grad_norm": 0.8070857428863515,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 6747
    },
    {
      "epoch": 0.06748,
      "grad_norm": 0.8359697636031163,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 6748
    },
    {
      "epoch": 0.06749,
      "grad_norm": 0.827772533552076,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 6749
    },
    {
      "epoch": 0.0675,
      "grad_norm": 0.859730671541346,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 6750
    },
    {
      "epoch": 0.06751,
      "grad_norm": 0.7946849688098641,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6751
    },
    {
      "epoch": 0.06752,
      "grad_norm": 0.7787584913556823,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6752
    },
    {
      "epoch": 0.06753,
      "grad_norm": 0.7233106355478739,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 6753
    },
    {
      "epoch": 0.06754,
      "grad_norm": 0.5923969384902699,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 6754
    },
    {
      "epoch": 0.06755,
      "grad_norm": 0.620061843467779,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 6755
    },
    {
      "epoch": 0.06756,
      "grad_norm": 0.6491047345358799,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 6756
    },
    {
      "epoch": 0.06757,
      "grad_norm": 0.647987484033838,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 6757
    },
    {
      "epoch": 0.06758,
      "grad_norm": 0.6452806428391407,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 6758
    },
    {
      "epoch": 0.06759,
      "grad_norm": 0.7117568970383578,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 6759
    },
    {
      "epoch": 0.0676,
      "grad_norm": 0.9554260901760981,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 6760
    },
    {
      "epoch": 0.06761,
      "grad_norm": 1.4624912759082085,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6761
    },
    {
      "epoch": 0.06762,
      "grad_norm": 0.5931283354167579,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 6762
    },
    {
      "epoch": 0.06763,
      "grad_norm": 0.8396616773751427,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 6763
    },
    {
      "epoch": 0.06764,
      "grad_norm": 0.9612365334414132,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 6764
    },
    {
      "epoch": 0.06765,
      "grad_norm": 0.895825369939231,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 6765
    },
    {
      "epoch": 0.06766,
      "grad_norm": 0.9141273493748527,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 6766
    },
    {
      "epoch": 0.06767,
      "grad_norm": 1.0179298922982885,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 6767
    },
    {
      "epoch": 0.06768,
      "grad_norm": 1.0094086260801791,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 6768
    },
    {
      "epoch": 0.06769,
      "grad_norm": 1.0112198539456203,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 6769
    },
    {
      "epoch": 0.0677,
      "grad_norm": 0.9926517409036734,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 6770
    },
    {
      "epoch": 0.06771,
      "grad_norm": 0.8812766249718238,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 6771
    },
    {
      "epoch": 0.06772,
      "grad_norm": 0.8389743876880504,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 6772
    },
    {
      "epoch": 0.06773,
      "grad_norm": 0.7951452996932417,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6773
    },
    {
      "epoch": 0.06774,
      "grad_norm": 0.8837761133765221,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 6774
    },
    {
      "epoch": 0.06775,
      "grad_norm": 0.957136507400834,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 6775
    },
    {
      "epoch": 0.06776,
      "grad_norm": 1.0481666958281863,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 6776
    },
    {
      "epoch": 0.06777,
      "grad_norm": 0.8631871107785212,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 6777
    },
    {
      "epoch": 0.06778,
      "grad_norm": 0.7576677776928321,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 6778
    },
    {
      "epoch": 0.06779,
      "grad_norm": 0.7621818727362122,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 6779
    },
    {
      "epoch": 0.0678,
      "grad_norm": 0.7354090321987549,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 6780
    },
    {
      "epoch": 0.06781,
      "grad_norm": 0.796508496988399,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 6781
    },
    {
      "epoch": 0.06782,
      "grad_norm": 0.8416596876225658,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 6782
    },
    {
      "epoch": 0.06783,
      "grad_norm": 0.9169424759930719,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 6783
    },
    {
      "epoch": 0.06784,
      "grad_norm": 0.8178011962748643,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 6784
    },
    {
      "epoch": 0.06785,
      "grad_norm": 0.7835817405720389,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 6785
    },
    {
      "epoch": 0.06786,
      "grad_norm": 0.8242013907791396,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 6786
    },
    {
      "epoch": 0.06787,
      "grad_norm": 0.7961787365055251,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 6787
    },
    {
      "epoch": 0.06788,
      "grad_norm": 0.8721165254426617,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 6788
    },
    {
      "epoch": 0.06789,
      "grad_norm": 1.1396955949704233,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 6789
    },
    {
      "epoch": 0.0679,
      "grad_norm": 1.0568810715247237,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6790
    },
    {
      "epoch": 0.06791,
      "grad_norm": 1.0143944196991193,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 6791
    },
    {
      "epoch": 0.06792,
      "grad_norm": 0.884818963074791,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 6792
    },
    {
      "epoch": 0.06793,
      "grad_norm": 0.8478088580149906,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 6793
    },
    {
      "epoch": 0.06794,
      "grad_norm": 0.8056286827298742,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 6794
    },
    {
      "epoch": 0.06795,
      "grad_norm": 0.7411740341729665,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 6795
    },
    {
      "epoch": 0.06796,
      "grad_norm": 0.7889975862725216,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 6796
    },
    {
      "epoch": 0.06797,
      "grad_norm": 0.9643036551361494,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 6797
    },
    {
      "epoch": 0.06798,
      "grad_norm": 0.9344988567161295,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6798
    },
    {
      "epoch": 0.06799,
      "grad_norm": 0.9452355458669397,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 6799
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.8857614730909316,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 6800
    },
    {
      "epoch": 0.06801,
      "grad_norm": 0.809752329455322,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 6801
    },
    {
      "epoch": 0.06802,
      "grad_norm": 0.8227719693453557,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 6802
    },
    {
      "epoch": 0.06803,
      "grad_norm": 0.7412508487470881,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6803
    },
    {
      "epoch": 0.06804,
      "grad_norm": 0.7054837765584971,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 6804
    },
    {
      "epoch": 0.06805,
      "grad_norm": 0.7568713263104572,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6805
    },
    {
      "epoch": 0.06806,
      "grad_norm": 0.9817473684527529,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 6806
    },
    {
      "epoch": 0.06807,
      "grad_norm": 1.1600981786410642,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 6807
    },
    {
      "epoch": 0.06808,
      "grad_norm": 1.159887129703928,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6808
    },
    {
      "epoch": 0.06809,
      "grad_norm": 0.8456896116982721,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 6809
    },
    {
      "epoch": 0.0681,
      "grad_norm": 0.858282322637232,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 6810
    },
    {
      "epoch": 0.06811,
      "grad_norm": 0.9611788401234072,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 6811
    },
    {
      "epoch": 0.06812,
      "grad_norm": 0.8997598486326439,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 6812
    },
    {
      "epoch": 0.06813,
      "grad_norm": 0.7739640351300833,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 6813
    },
    {
      "epoch": 0.06814,
      "grad_norm": 0.811869754027927,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6814
    },
    {
      "epoch": 0.06815,
      "grad_norm": 0.7998801552358381,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 6815
    },
    {
      "epoch": 0.06816,
      "grad_norm": 0.7960758830426774,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 6816
    },
    {
      "epoch": 0.06817,
      "grad_norm": 0.863740767658441,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6817
    },
    {
      "epoch": 0.06818,
      "grad_norm": 0.8459457179335972,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 6818
    },
    {
      "epoch": 0.06819,
      "grad_norm": 0.8884032934270286,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6819
    },
    {
      "epoch": 0.0682,
      "grad_norm": 0.9225686576604342,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 6820
    },
    {
      "epoch": 0.06821,
      "grad_norm": 0.844036272472802,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 6821
    },
    {
      "epoch": 0.06822,
      "grad_norm": 0.8174293547589027,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 6822
    },
    {
      "epoch": 0.06823,
      "grad_norm": 0.836645021687918,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 6823
    },
    {
      "epoch": 0.06824,
      "grad_norm": 0.9752883654153214,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 6824
    },
    {
      "epoch": 0.06825,
      "grad_norm": 1.079600387760812,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 6825
    },
    {
      "epoch": 0.06826,
      "grad_norm": 0.8197898525120547,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 6826
    },
    {
      "epoch": 0.06827,
      "grad_norm": 0.9956479850125667,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 6827
    },
    {
      "epoch": 0.06828,
      "grad_norm": 1.152216850473556,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6828
    },
    {
      "epoch": 0.06829,
      "grad_norm": 0.8571944350147577,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 6829
    },
    {
      "epoch": 0.0683,
      "grad_norm": 0.757993990915538,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 6830
    },
    {
      "epoch": 0.06831,
      "grad_norm": 0.864695692673034,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 6831
    },
    {
      "epoch": 0.06832,
      "grad_norm": 0.9705812470204103,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 6832
    },
    {
      "epoch": 0.06833,
      "grad_norm": 0.9213907790898336,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 6833
    },
    {
      "epoch": 0.06834,
      "grad_norm": 0.792235057764182,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 6834
    },
    {
      "epoch": 0.06835,
      "grad_norm": 0.7066677455983338,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6835
    },
    {
      "epoch": 0.06836,
      "grad_norm": 0.6670788363565453,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 6836
    },
    {
      "epoch": 0.06837,
      "grad_norm": 0.6464856752678825,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 6837
    },
    {
      "epoch": 0.06838,
      "grad_norm": 0.6108028457392457,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 6838
    },
    {
      "epoch": 0.06839,
      "grad_norm": 0.6813565301110693,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 6839
    },
    {
      "epoch": 0.0684,
      "grad_norm": 0.794881760436682,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 6840
    },
    {
      "epoch": 0.06841,
      "grad_norm": 0.961174574044574,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 6841
    },
    {
      "epoch": 0.06842,
      "grad_norm": 1.0476574337633355,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 6842
    },
    {
      "epoch": 0.06843,
      "grad_norm": 0.8907102890021011,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 6843
    },
    {
      "epoch": 0.06844,
      "grad_norm": 0.8811872643928172,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 6844
    },
    {
      "epoch": 0.06845,
      "grad_norm": 0.9952993463626575,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 6845
    },
    {
      "epoch": 0.06846,
      "grad_norm": 1.2628089603731774,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 6846
    },
    {
      "epoch": 0.06847,
      "grad_norm": 0.901670711894547,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 6847
    },
    {
      "epoch": 0.06848,
      "grad_norm": 0.956123404918695,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 6848
    },
    {
      "epoch": 0.06849,
      "grad_norm": 1.0968210993157013,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 6849
    },
    {
      "epoch": 0.0685,
      "grad_norm": 0.936347183022355,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 6850
    },
    {
      "epoch": 0.06851,
      "grad_norm": 0.8968792133031543,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 6851
    },
    {
      "epoch": 0.06852,
      "grad_norm": 0.8507324682860469,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 6852
    },
    {
      "epoch": 0.06853,
      "grad_norm": 0.7108631167713592,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 6853
    },
    {
      "epoch": 0.06854,
      "grad_norm": 0.5899558073031355,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 6854
    },
    {
      "epoch": 0.06855,
      "grad_norm": 0.6714377951019126,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 6855
    },
    {
      "epoch": 0.06856,
      "grad_norm": 0.7264783579127705,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 6856
    },
    {
      "epoch": 0.06857,
      "grad_norm": 0.8461334990738569,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 6857
    },
    {
      "epoch": 0.06858,
      "grad_norm": 0.9692288911903976,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 6858
    },
    {
      "epoch": 0.06859,
      "grad_norm": 0.9198005654114306,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 6859
    },
    {
      "epoch": 0.0686,
      "grad_norm": 0.6854066787324575,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 6860
    },
    {
      "epoch": 0.06861,
      "grad_norm": 0.6574071277975999,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 6861
    },
    {
      "epoch": 0.06862,
      "grad_norm": 0.930478597926848,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 6862
    },
    {
      "epoch": 0.06863,
      "grad_norm": 1.0592073313127546,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 6863
    },
    {
      "epoch": 0.06864,
      "grad_norm": 0.9388318477503523,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 6864
    },
    {
      "epoch": 0.06865,
      "grad_norm": 0.746039107432962,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 6865
    },
    {
      "epoch": 0.06866,
      "grad_norm": 0.7626047298181141,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 6866
    },
    {
      "epoch": 0.06867,
      "grad_norm": 0.8192485910979884,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 6867
    },
    {
      "epoch": 0.06868,
      "grad_norm": 0.8943577260151027,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 6868
    },
    {
      "epoch": 0.06869,
      "grad_norm": 1.0994368832659123,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 6869
    },
    {
      "epoch": 0.0687,
      "grad_norm": 0.8861962197230898,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 6870
    },
    {
      "epoch": 0.06871,
      "grad_norm": 0.8815941784538627,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 6871
    },
    {
      "epoch": 0.06872,
      "grad_norm": 0.9924460757229603,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 6872
    },
    {
      "epoch": 0.06873,
      "grad_norm": 1.0509114751207171,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 6873
    },
    {
      "epoch": 0.06874,
      "grad_norm": 0.9532056718212061,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6874
    },
    {
      "epoch": 0.06875,
      "grad_norm": 1.023680307403794,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 6875
    },
    {
      "epoch": 0.06876,
      "grad_norm": 0.9902008565775648,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 6876
    },
    {
      "epoch": 0.06877,
      "grad_norm": 0.9531958381228303,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 6877
    },
    {
      "epoch": 0.06878,
      "grad_norm": 0.9710478342391432,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 6878
    },
    {
      "epoch": 0.06879,
      "grad_norm": 0.909639330602345,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 6879
    },
    {
      "epoch": 0.0688,
      "grad_norm": 0.8095671998298988,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 6880
    },
    {
      "epoch": 0.06881,
      "grad_norm": 0.9113186941401037,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 6881
    },
    {
      "epoch": 0.06882,
      "grad_norm": 0.9258946157268206,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 6882
    },
    {
      "epoch": 0.06883,
      "grad_norm": 1.0573211690290847,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 6883
    },
    {
      "epoch": 0.06884,
      "grad_norm": 1.0560959741409428,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 6884
    },
    {
      "epoch": 0.06885,
      "grad_norm": 0.9393005675185666,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 6885
    },
    {
      "epoch": 0.06886,
      "grad_norm": 0.8873501258981987,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 6886
    },
    {
      "epoch": 0.06887,
      "grad_norm": 0.9371568421249765,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 6887
    },
    {
      "epoch": 0.06888,
      "grad_norm": 0.960033318581805,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 6888
    },
    {
      "epoch": 0.06889,
      "grad_norm": 1.0030573810363002,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 6889
    },
    {
      "epoch": 0.0689,
      "grad_norm": 1.0144314201908178,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 6890
    },
    {
      "epoch": 0.06891,
      "grad_norm": 1.1068906368874076,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 6891
    },
    {
      "epoch": 0.06892,
      "grad_norm": 0.9054892903304268,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 6892
    },
    {
      "epoch": 0.06893,
      "grad_norm": 0.7540684414365276,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6893
    },
    {
      "epoch": 0.06894,
      "grad_norm": 0.6663108621237684,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 6894
    },
    {
      "epoch": 0.06895,
      "grad_norm": 0.5979163007296868,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 6895
    },
    {
      "epoch": 0.06896,
      "grad_norm": 0.5872771645065226,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 6896
    },
    {
      "epoch": 0.06897,
      "grad_norm": 0.6173404544816022,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 6897
    },
    {
      "epoch": 0.06898,
      "grad_norm": 0.762140883361191,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 6898
    },
    {
      "epoch": 0.06899,
      "grad_norm": 1.0242084853160343,
      "learning_rate": 0.003,
      "loss": 4.1402,
      "step": 6899
    },
    {
      "epoch": 0.069,
      "grad_norm": 1.2323653252300188,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 6900
    },
    {
      "epoch": 0.06901,
      "grad_norm": 0.5820472612244612,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 6901
    },
    {
      "epoch": 0.06902,
      "grad_norm": 0.7237387908075824,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 6902
    },
    {
      "epoch": 0.06903,
      "grad_norm": 0.8664255236861678,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 6903
    },
    {
      "epoch": 0.06904,
      "grad_norm": 0.8559541713855746,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 6904
    },
    {
      "epoch": 0.06905,
      "grad_norm": 0.8471025018164605,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 6905
    },
    {
      "epoch": 0.06906,
      "grad_norm": 0.8690540035946629,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 6906
    },
    {
      "epoch": 0.06907,
      "grad_norm": 0.8178132983249496,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 6907
    },
    {
      "epoch": 0.06908,
      "grad_norm": 0.7564578666671172,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 6908
    },
    {
      "epoch": 0.06909,
      "grad_norm": 0.7790308250713,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 6909
    },
    {
      "epoch": 0.0691,
      "grad_norm": 0.7351729319289587,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 6910
    },
    {
      "epoch": 0.06911,
      "grad_norm": 0.7267083412234101,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 6911
    },
    {
      "epoch": 0.06912,
      "grad_norm": 0.6661025595948273,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 6912
    },
    {
      "epoch": 0.06913,
      "grad_norm": 0.674553980244978,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 6913
    },
    {
      "epoch": 0.06914,
      "grad_norm": 0.7348105212280468,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 6914
    },
    {
      "epoch": 0.06915,
      "grad_norm": 0.7493265450674638,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 6915
    },
    {
      "epoch": 0.06916,
      "grad_norm": 0.808956213639763,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 6916
    },
    {
      "epoch": 0.06917,
      "grad_norm": 0.8113376674561755,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 6917
    },
    {
      "epoch": 0.06918,
      "grad_norm": 0.8283810471683984,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 6918
    },
    {
      "epoch": 0.06919,
      "grad_norm": 0.8686223282408305,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 6919
    },
    {
      "epoch": 0.0692,
      "grad_norm": 0.7951171895581218,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 6920
    },
    {
      "epoch": 0.06921,
      "grad_norm": 0.9445545003116608,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 6921
    },
    {
      "epoch": 0.06922,
      "grad_norm": 0.9810026720945391,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 6922
    },
    {
      "epoch": 0.06923,
      "grad_norm": 0.9535589717974995,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 6923
    },
    {
      "epoch": 0.06924,
      "grad_norm": 0.9826106732792211,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 6924
    },
    {
      "epoch": 0.06925,
      "grad_norm": 0.958775832947445,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 6925
    },
    {
      "epoch": 0.06926,
      "grad_norm": 0.9026835208656796,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6926
    },
    {
      "epoch": 0.06927,
      "grad_norm": 0.8065987081198023,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 6927
    },
    {
      "epoch": 0.06928,
      "grad_norm": 0.7325070632512863,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 6928
    },
    {
      "epoch": 0.06929,
      "grad_norm": 0.7844696991539358,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 6929
    },
    {
      "epoch": 0.0693,
      "grad_norm": 0.7321103434207817,
      "learning_rate": 0.003,
      "loss": 4.1448,
      "step": 6930
    },
    {
      "epoch": 0.06931,
      "grad_norm": 0.6348052074469944,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 6931
    },
    {
      "epoch": 0.06932,
      "grad_norm": 0.6156078545430647,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 6932
    },
    {
      "epoch": 0.06933,
      "grad_norm": 0.7281337302342974,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 6933
    },
    {
      "epoch": 0.06934,
      "grad_norm": 0.9326406238175676,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 6934
    },
    {
      "epoch": 0.06935,
      "grad_norm": 1.348898577949277,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 6935
    },
    {
      "epoch": 0.06936,
      "grad_norm": 0.8658725709964428,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 6936
    },
    {
      "epoch": 0.06937,
      "grad_norm": 0.8735566248577522,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 6937
    },
    {
      "epoch": 0.06938,
      "grad_norm": 0.9320220136220663,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 6938
    },
    {
      "epoch": 0.06939,
      "grad_norm": 0.9133167311611287,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 6939
    },
    {
      "epoch": 0.0694,
      "grad_norm": 0.9890705299908502,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 6940
    },
    {
      "epoch": 0.06941,
      "grad_norm": 1.3060885860531548,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6941
    },
    {
      "epoch": 0.06942,
      "grad_norm": 0.7699335742222013,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 6942
    },
    {
      "epoch": 0.06943,
      "grad_norm": 0.7643000254330296,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 6943
    },
    {
      "epoch": 0.06944,
      "grad_norm": 0.7529705113455923,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 6944
    },
    {
      "epoch": 0.06945,
      "grad_norm": 0.882823727029274,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 6945
    },
    {
      "epoch": 0.06946,
      "grad_norm": 1.0250594327340659,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 6946
    },
    {
      "epoch": 0.06947,
      "grad_norm": 1.1502658138369475,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 6947
    },
    {
      "epoch": 0.06948,
      "grad_norm": 0.8444220868745179,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6948
    },
    {
      "epoch": 0.06949,
      "grad_norm": 0.7711423294691977,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 6949
    },
    {
      "epoch": 0.0695,
      "grad_norm": 0.8610365574220729,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 6950
    },
    {
      "epoch": 0.06951,
      "grad_norm": 0.9731685606617599,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 6951
    },
    {
      "epoch": 0.06952,
      "grad_norm": 1.1267620956159707,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 6952
    },
    {
      "epoch": 0.06953,
      "grad_norm": 0.9862000527924848,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 6953
    },
    {
      "epoch": 0.06954,
      "grad_norm": 0.9666325788788848,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 6954
    },
    {
      "epoch": 0.06955,
      "grad_norm": 0.9599709329069231,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 6955
    },
    {
      "epoch": 0.06956,
      "grad_norm": 0.9866026018546916,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6956
    },
    {
      "epoch": 0.06957,
      "grad_norm": 0.9845733907627405,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 6957
    },
    {
      "epoch": 0.06958,
      "grad_norm": 0.9943021661775551,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 6958
    },
    {
      "epoch": 0.06959,
      "grad_norm": 1.0279554280201155,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 6959
    },
    {
      "epoch": 0.0696,
      "grad_norm": 1.1047587660616394,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 6960
    },
    {
      "epoch": 0.06961,
      "grad_norm": 0.9950639197668989,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 6961
    },
    {
      "epoch": 0.06962,
      "grad_norm": 1.027748422776989,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 6962
    },
    {
      "epoch": 0.06963,
      "grad_norm": 0.9781377480333107,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 6963
    },
    {
      "epoch": 0.06964,
      "grad_norm": 0.9482629019532105,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 6964
    },
    {
      "epoch": 0.06965,
      "grad_norm": 0.9356422484428207,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 6965
    },
    {
      "epoch": 0.06966,
      "grad_norm": 0.9056612242786197,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 6966
    },
    {
      "epoch": 0.06967,
      "grad_norm": 0.7203106230136754,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 6967
    },
    {
      "epoch": 0.06968,
      "grad_norm": 0.6599503868887251,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 6968
    },
    {
      "epoch": 0.06969,
      "grad_norm": 0.6987003782313111,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 6969
    },
    {
      "epoch": 0.0697,
      "grad_norm": 0.8342591105193845,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 6970
    },
    {
      "epoch": 0.06971,
      "grad_norm": 1.018817534245797,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 6971
    },
    {
      "epoch": 0.06972,
      "grad_norm": 1.2395063765633032,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 6972
    },
    {
      "epoch": 0.06973,
      "grad_norm": 0.8545094616038666,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 6973
    },
    {
      "epoch": 0.06974,
      "grad_norm": 0.8121109776419817,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 6974
    },
    {
      "epoch": 0.06975,
      "grad_norm": 0.7584233055394562,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 6975
    },
    {
      "epoch": 0.06976,
      "grad_norm": 0.715581943047214,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 6976
    },
    {
      "epoch": 0.06977,
      "grad_norm": 0.8238558659092502,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 6977
    },
    {
      "epoch": 0.06978,
      "grad_norm": 0.9171894622076492,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 6978
    },
    {
      "epoch": 0.06979,
      "grad_norm": 0.8537393028092058,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 6979
    },
    {
      "epoch": 0.0698,
      "grad_norm": 0.6494522431850996,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 6980
    },
    {
      "epoch": 0.06981,
      "grad_norm": 0.6394410355891653,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 6981
    },
    {
      "epoch": 0.06982,
      "grad_norm": 0.655752618433206,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 6982
    },
    {
      "epoch": 0.06983,
      "grad_norm": 0.6522649714130653,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 6983
    },
    {
      "epoch": 0.06984,
      "grad_norm": 0.6745056056178718,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 6984
    },
    {
      "epoch": 0.06985,
      "grad_norm": 0.7167454235222351,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 6985
    },
    {
      "epoch": 0.06986,
      "grad_norm": 0.7745691656939917,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 6986
    },
    {
      "epoch": 0.06987,
      "grad_norm": 0.7824556430219963,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 6987
    },
    {
      "epoch": 0.06988,
      "grad_norm": 0.713073188685409,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 6988
    },
    {
      "epoch": 0.06989,
      "grad_norm": 0.6918866773648449,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 6989
    },
    {
      "epoch": 0.0699,
      "grad_norm": 0.6887212376040721,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 6990
    },
    {
      "epoch": 0.06991,
      "grad_norm": 0.728960146591993,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 6991
    },
    {
      "epoch": 0.06992,
      "grad_norm": 0.7773350928913978,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 6992
    },
    {
      "epoch": 0.06993,
      "grad_norm": 1.0282095814384735,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 6993
    },
    {
      "epoch": 0.06994,
      "grad_norm": 1.2480624334513428,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 6994
    },
    {
      "epoch": 0.06995,
      "grad_norm": 0.8763541675945252,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 6995
    },
    {
      "epoch": 0.06996,
      "grad_norm": 0.825844759458018,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 6996
    },
    {
      "epoch": 0.06997,
      "grad_norm": 0.9173767828845192,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 6997
    },
    {
      "epoch": 0.06998,
      "grad_norm": 0.9510745593290524,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 6998
    },
    {
      "epoch": 0.06999,
      "grad_norm": 0.9779021385996265,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 6999
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1827894871717832,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7000
    },
    {
      "epoch": 0.07001,
      "grad_norm": 0.7823428891121994,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 7001
    },
    {
      "epoch": 0.07002,
      "grad_norm": 0.7502844132003458,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 7002
    },
    {
      "epoch": 0.07003,
      "grad_norm": 0.6431900619854782,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7003
    },
    {
      "epoch": 0.07004,
      "grad_norm": 0.6545830386408341,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 7004
    },
    {
      "epoch": 0.07005,
      "grad_norm": 0.683875895653165,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 7005
    },
    {
      "epoch": 0.07006,
      "grad_norm": 0.7166463834547604,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 7006
    },
    {
      "epoch": 0.07007,
      "grad_norm": 0.7142802784000205,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7007
    },
    {
      "epoch": 0.07008,
      "grad_norm": 0.8137440162068434,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7008
    },
    {
      "epoch": 0.07009,
      "grad_norm": 0.9813188383797056,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 7009
    },
    {
      "epoch": 0.0701,
      "grad_norm": 1.1360396784480946,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 7010
    },
    {
      "epoch": 0.07011,
      "grad_norm": 1.122021954079669,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7011
    },
    {
      "epoch": 0.07012,
      "grad_norm": 0.9025536928180181,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7012
    },
    {
      "epoch": 0.07013,
      "grad_norm": 0.8888035296110233,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 7013
    },
    {
      "epoch": 0.07014,
      "grad_norm": 0.9010860607271816,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 7014
    },
    {
      "epoch": 0.07015,
      "grad_norm": 0.9906713805217798,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7015
    },
    {
      "epoch": 0.07016,
      "grad_norm": 1.1465687996483969,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7016
    },
    {
      "epoch": 0.07017,
      "grad_norm": 1.0247558863315256,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 7017
    },
    {
      "epoch": 0.07018,
      "grad_norm": 1.0613439192576388,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 7018
    },
    {
      "epoch": 0.07019,
      "grad_norm": 0.9990910934627305,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7019
    },
    {
      "epoch": 0.0702,
      "grad_norm": 0.9540598745983695,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7020
    },
    {
      "epoch": 0.07021,
      "grad_norm": 0.9831178649652754,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 7021
    },
    {
      "epoch": 0.07022,
      "grad_norm": 0.9737401392446607,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 7022
    },
    {
      "epoch": 0.07023,
      "grad_norm": 1.1732936942454621,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 7023
    },
    {
      "epoch": 0.07024,
      "grad_norm": 0.8945111721502813,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7024
    },
    {
      "epoch": 0.07025,
      "grad_norm": 0.867087064716649,
      "learning_rate": 0.003,
      "loss": 4.1268,
      "step": 7025
    },
    {
      "epoch": 0.07026,
      "grad_norm": 0.8901204165200368,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7026
    },
    {
      "epoch": 0.07027,
      "grad_norm": 0.7820722242424083,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7027
    },
    {
      "epoch": 0.07028,
      "grad_norm": 0.8644909122379497,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7028
    },
    {
      "epoch": 0.07029,
      "grad_norm": 0.8953906928512193,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 7029
    },
    {
      "epoch": 0.0703,
      "grad_norm": 0.8991569997104605,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 7030
    },
    {
      "epoch": 0.07031,
      "grad_norm": 1.1088920936393387,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 7031
    },
    {
      "epoch": 0.07032,
      "grad_norm": 1.0567421197796187,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 7032
    },
    {
      "epoch": 0.07033,
      "grad_norm": 1.1300939504610534,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 7033
    },
    {
      "epoch": 0.07034,
      "grad_norm": 0.8874141777977914,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 7034
    },
    {
      "epoch": 0.07035,
      "grad_norm": 0.752043314504715,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7035
    },
    {
      "epoch": 0.07036,
      "grad_norm": 0.7572808680785389,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 7036
    },
    {
      "epoch": 0.07037,
      "grad_norm": 0.7616378265590469,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 7037
    },
    {
      "epoch": 0.07038,
      "grad_norm": 0.7319105489860215,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 7038
    },
    {
      "epoch": 0.07039,
      "grad_norm": 0.6986810220580801,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 7039
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.6919082493656014,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 7040
    },
    {
      "epoch": 0.07041,
      "grad_norm": 0.6752147196309027,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7041
    },
    {
      "epoch": 0.07042,
      "grad_norm": 0.7597403656610957,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 7042
    },
    {
      "epoch": 0.07043,
      "grad_norm": 0.8702208068936226,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 7043
    },
    {
      "epoch": 0.07044,
      "grad_norm": 0.997721683934861,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 7044
    },
    {
      "epoch": 0.07045,
      "grad_norm": 1.0590434408909246,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 7045
    },
    {
      "epoch": 0.07046,
      "grad_norm": 0.7262778214035177,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7046
    },
    {
      "epoch": 0.07047,
      "grad_norm": 0.6105175041151503,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7047
    },
    {
      "epoch": 0.07048,
      "grad_norm": 0.6919653137461556,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 7048
    },
    {
      "epoch": 0.07049,
      "grad_norm": 0.7158702927734397,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7049
    },
    {
      "epoch": 0.0705,
      "grad_norm": 0.7091325935961303,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 7050
    },
    {
      "epoch": 0.07051,
      "grad_norm": 0.8551664598496885,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7051
    },
    {
      "epoch": 0.07052,
      "grad_norm": 0.9183031935565956,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 7052
    },
    {
      "epoch": 0.07053,
      "grad_norm": 0.9122596997693425,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7053
    },
    {
      "epoch": 0.07054,
      "grad_norm": 1.0481518040269315,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7054
    },
    {
      "epoch": 0.07055,
      "grad_norm": 0.8836917687242147,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7055
    },
    {
      "epoch": 0.07056,
      "grad_norm": 0.8241724236620276,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 7056
    },
    {
      "epoch": 0.07057,
      "grad_norm": 0.8198719505381878,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 7057
    },
    {
      "epoch": 0.07058,
      "grad_norm": 1.0373640533731205,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7058
    },
    {
      "epoch": 0.07059,
      "grad_norm": 1.1298985200213496,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 7059
    },
    {
      "epoch": 0.0706,
      "grad_norm": 0.917680527288264,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7060
    },
    {
      "epoch": 0.07061,
      "grad_norm": 1.1487741663485906,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7061
    },
    {
      "epoch": 0.07062,
      "grad_norm": 0.9765073918800828,
      "learning_rate": 0.003,
      "loss": 4.1421,
      "step": 7062
    },
    {
      "epoch": 0.07063,
      "grad_norm": 1.0320830994762968,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 7063
    },
    {
      "epoch": 0.07064,
      "grad_norm": 0.8433226609779286,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 7064
    },
    {
      "epoch": 0.07065,
      "grad_norm": 0.7826897468256804,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 7065
    },
    {
      "epoch": 0.07066,
      "grad_norm": 0.6869110843573996,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7066
    },
    {
      "epoch": 0.07067,
      "grad_norm": 0.6492482664561744,
      "learning_rate": 0.003,
      "loss": 4.1481,
      "step": 7067
    },
    {
      "epoch": 0.07068,
      "grad_norm": 0.6736220554775696,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 7068
    },
    {
      "epoch": 0.07069,
      "grad_norm": 0.803916767312363,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 7069
    },
    {
      "epoch": 0.0707,
      "grad_norm": 1.1648389249713418,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 7070
    },
    {
      "epoch": 0.07071,
      "grad_norm": 1.05402571939476,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7071
    },
    {
      "epoch": 0.07072,
      "grad_norm": 0.8354323461428795,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 7072
    },
    {
      "epoch": 0.07073,
      "grad_norm": 0.799265495194416,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 7073
    },
    {
      "epoch": 0.07074,
      "grad_norm": 0.8870813126594803,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 7074
    },
    {
      "epoch": 0.07075,
      "grad_norm": 0.9983580451793913,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 7075
    },
    {
      "epoch": 0.07076,
      "grad_norm": 0.9665581584289197,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7076
    },
    {
      "epoch": 0.07077,
      "grad_norm": 0.8830123185073006,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7077
    },
    {
      "epoch": 0.07078,
      "grad_norm": 0.8590966319766036,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 7078
    },
    {
      "epoch": 0.07079,
      "grad_norm": 0.8754624480453634,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 7079
    },
    {
      "epoch": 0.0708,
      "grad_norm": 0.8357253588784314,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 7080
    },
    {
      "epoch": 0.07081,
      "grad_norm": 0.6853886151525752,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7081
    },
    {
      "epoch": 0.07082,
      "grad_norm": 0.746488253274503,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 7082
    },
    {
      "epoch": 0.07083,
      "grad_norm": 0.7589362058517924,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7083
    },
    {
      "epoch": 0.07084,
      "grad_norm": 0.9173953249459876,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7084
    },
    {
      "epoch": 0.07085,
      "grad_norm": 1.1273331172700023,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 7085
    },
    {
      "epoch": 0.07086,
      "grad_norm": 0.9129945566559451,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7086
    },
    {
      "epoch": 0.07087,
      "grad_norm": 0.782084497555971,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7087
    },
    {
      "epoch": 0.07088,
      "grad_norm": 0.7191976397005901,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 7088
    },
    {
      "epoch": 0.07089,
      "grad_norm": 0.7254306742811141,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 7089
    },
    {
      "epoch": 0.0709,
      "grad_norm": 0.936953014758606,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 7090
    },
    {
      "epoch": 0.07091,
      "grad_norm": 1.1774180392198823,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 7091
    },
    {
      "epoch": 0.07092,
      "grad_norm": 0.8878181062918875,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 7092
    },
    {
      "epoch": 0.07093,
      "grad_norm": 0.9002651333364067,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 7093
    },
    {
      "epoch": 0.07094,
      "grad_norm": 0.9329648399723189,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 7094
    },
    {
      "epoch": 0.07095,
      "grad_norm": 0.9632962134461859,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 7095
    },
    {
      "epoch": 0.07096,
      "grad_norm": 0.9420914940041324,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 7096
    },
    {
      "epoch": 0.07097,
      "grad_norm": 1.141780989450593,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 7097
    },
    {
      "epoch": 0.07098,
      "grad_norm": 1.200591969334309,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 7098
    },
    {
      "epoch": 0.07099,
      "grad_norm": 0.8009834169786036,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 7099
    },
    {
      "epoch": 0.071,
      "grad_norm": 0.7941069430643708,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 7100
    },
    {
      "epoch": 0.07101,
      "grad_norm": 0.8289765206226791,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7101
    },
    {
      "epoch": 0.07102,
      "grad_norm": 0.7502407355058645,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7102
    },
    {
      "epoch": 0.07103,
      "grad_norm": 0.7118775207721677,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7103
    },
    {
      "epoch": 0.07104,
      "grad_norm": 0.7238556977567164,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7104
    },
    {
      "epoch": 0.07105,
      "grad_norm": 0.6978584610652161,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 7105
    },
    {
      "epoch": 0.07106,
      "grad_norm": 0.7371502133162132,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 7106
    },
    {
      "epoch": 0.07107,
      "grad_norm": 0.8360787228838016,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7107
    },
    {
      "epoch": 0.07108,
      "grad_norm": 0.9175088055451589,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 7108
    },
    {
      "epoch": 0.07109,
      "grad_norm": 1.162644092744131,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7109
    },
    {
      "epoch": 0.0711,
      "grad_norm": 0.9424749205513693,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 7110
    },
    {
      "epoch": 0.07111,
      "grad_norm": 1.017081150522671,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7111
    },
    {
      "epoch": 0.07112,
      "grad_norm": 1.0662436592570368,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7112
    },
    {
      "epoch": 0.07113,
      "grad_norm": 1.054393130448686,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 7113
    },
    {
      "epoch": 0.07114,
      "grad_norm": 0.9257779285013658,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7114
    },
    {
      "epoch": 0.07115,
      "grad_norm": 0.9034958469747977,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7115
    },
    {
      "epoch": 0.07116,
      "grad_norm": 0.8243521449434865,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7116
    },
    {
      "epoch": 0.07117,
      "grad_norm": 0.7541154662606151,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 7117
    },
    {
      "epoch": 0.07118,
      "grad_norm": 0.8201537536206699,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7118
    },
    {
      "epoch": 0.07119,
      "grad_norm": 0.8555300267392018,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 7119
    },
    {
      "epoch": 0.0712,
      "grad_norm": 0.9725671974482196,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7120
    },
    {
      "epoch": 0.07121,
      "grad_norm": 0.9287551042992966,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 7121
    },
    {
      "epoch": 0.07122,
      "grad_norm": 0.8730257071456973,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 7122
    },
    {
      "epoch": 0.07123,
      "grad_norm": 0.9240063776082804,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7123
    },
    {
      "epoch": 0.07124,
      "grad_norm": 0.8897100879971401,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7124
    },
    {
      "epoch": 0.07125,
      "grad_norm": 0.7255092544625131,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7125
    },
    {
      "epoch": 0.07126,
      "grad_norm": 0.64410022138994,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 7126
    },
    {
      "epoch": 0.07127,
      "grad_norm": 0.6757358271574939,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 7127
    },
    {
      "epoch": 0.07128,
      "grad_norm": 0.8691171001871797,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 7128
    },
    {
      "epoch": 0.07129,
      "grad_norm": 1.1128442451767677,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 7129
    },
    {
      "epoch": 0.0713,
      "grad_norm": 1.0139741345763373,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7130
    },
    {
      "epoch": 0.07131,
      "grad_norm": 0.9159528876799511,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 7131
    },
    {
      "epoch": 0.07132,
      "grad_norm": 0.7898416755233202,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 7132
    },
    {
      "epoch": 0.07133,
      "grad_norm": 0.7848369474841521,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 7133
    },
    {
      "epoch": 0.07134,
      "grad_norm": 0.7751588573196545,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 7134
    },
    {
      "epoch": 0.07135,
      "grad_norm": 0.8777005315917537,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7135
    },
    {
      "epoch": 0.07136,
      "grad_norm": 1.039823448097075,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 7136
    },
    {
      "epoch": 0.07137,
      "grad_norm": 0.9365493853127754,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 7137
    },
    {
      "epoch": 0.07138,
      "grad_norm": 1.0284030256383379,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7138
    },
    {
      "epoch": 0.07139,
      "grad_norm": 1.1288999030524856,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7139
    },
    {
      "epoch": 0.0714,
      "grad_norm": 1.057387996892944,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 7140
    },
    {
      "epoch": 0.07141,
      "grad_norm": 1.068538739459718,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7141
    },
    {
      "epoch": 0.07142,
      "grad_norm": 0.9626388123201881,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 7142
    },
    {
      "epoch": 0.07143,
      "grad_norm": 0.9093794766008101,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 7143
    },
    {
      "epoch": 0.07144,
      "grad_norm": 0.7746289446671863,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 7144
    },
    {
      "epoch": 0.07145,
      "grad_norm": 0.7339182870303272,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7145
    },
    {
      "epoch": 0.07146,
      "grad_norm": 0.733959174720137,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7146
    },
    {
      "epoch": 0.07147,
      "grad_norm": 0.6952850136145936,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 7147
    },
    {
      "epoch": 0.07148,
      "grad_norm": 0.7615226403621356,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 7148
    },
    {
      "epoch": 0.07149,
      "grad_norm": 0.8936711792505057,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 7149
    },
    {
      "epoch": 0.0715,
      "grad_norm": 1.273915364248639,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 7150
    },
    {
      "epoch": 0.07151,
      "grad_norm": 0.7956720748067053,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7151
    },
    {
      "epoch": 0.07152,
      "grad_norm": 0.6838464347817911,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7152
    },
    {
      "epoch": 0.07153,
      "grad_norm": 0.704544600749531,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7153
    },
    {
      "epoch": 0.07154,
      "grad_norm": 0.8027032921273827,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 7154
    },
    {
      "epoch": 0.07155,
      "grad_norm": 0.8232429663470329,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 7155
    },
    {
      "epoch": 0.07156,
      "grad_norm": 0.7266009407623805,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 7156
    },
    {
      "epoch": 0.07157,
      "grad_norm": 0.74357634876916,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7157
    },
    {
      "epoch": 0.07158,
      "grad_norm": 0.6961339201769408,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 7158
    },
    {
      "epoch": 0.07159,
      "grad_norm": 0.7163813835599939,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 7159
    },
    {
      "epoch": 0.0716,
      "grad_norm": 0.6021787870861519,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 7160
    },
    {
      "epoch": 0.07161,
      "grad_norm": 0.5996751260865746,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7161
    },
    {
      "epoch": 0.07162,
      "grad_norm": 0.7383165393295259,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7162
    },
    {
      "epoch": 0.07163,
      "grad_norm": 0.8624369790330983,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 7163
    },
    {
      "epoch": 0.07164,
      "grad_norm": 0.9983212047149563,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7164
    },
    {
      "epoch": 0.07165,
      "grad_norm": 1.1984220097038951,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7165
    },
    {
      "epoch": 0.07166,
      "grad_norm": 1.0158195249204152,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 7166
    },
    {
      "epoch": 0.07167,
      "grad_norm": 1.2477237244225603,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 7167
    },
    {
      "epoch": 0.07168,
      "grad_norm": 0.8759979215569342,
      "learning_rate": 0.003,
      "loss": 4.1327,
      "step": 7168
    },
    {
      "epoch": 0.07169,
      "grad_norm": 0.9293872636545353,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 7169
    },
    {
      "epoch": 0.0717,
      "grad_norm": 0.9602891801389307,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 7170
    },
    {
      "epoch": 0.07171,
      "grad_norm": 1.1372428448834564,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 7171
    },
    {
      "epoch": 0.07172,
      "grad_norm": 1.1238803169197797,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 7172
    },
    {
      "epoch": 0.07173,
      "grad_norm": 0.9931562441621579,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 7173
    },
    {
      "epoch": 0.07174,
      "grad_norm": 0.9403706846070115,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 7174
    },
    {
      "epoch": 0.07175,
      "grad_norm": 0.9295449986782115,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 7175
    },
    {
      "epoch": 0.07176,
      "grad_norm": 0.911114790596918,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 7176
    },
    {
      "epoch": 0.07177,
      "grad_norm": 0.7839495659287221,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7177
    },
    {
      "epoch": 0.07178,
      "grad_norm": 0.7947159198866947,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 7178
    },
    {
      "epoch": 0.07179,
      "grad_norm": 0.8603443469449197,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 7179
    },
    {
      "epoch": 0.0718,
      "grad_norm": 0.9414623094280078,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 7180
    },
    {
      "epoch": 0.07181,
      "grad_norm": 0.9660522402559355,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7181
    },
    {
      "epoch": 0.07182,
      "grad_norm": 0.9278959841877216,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 7182
    },
    {
      "epoch": 0.07183,
      "grad_norm": 0.9339118033397995,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 7183
    },
    {
      "epoch": 0.07184,
      "grad_norm": 1.1778253262437426,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 7184
    },
    {
      "epoch": 0.07185,
      "grad_norm": 0.894326222770007,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 7185
    },
    {
      "epoch": 0.07186,
      "grad_norm": 1.0074070633488665,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7186
    },
    {
      "epoch": 0.07187,
      "grad_norm": 0.990447843885946,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7187
    },
    {
      "epoch": 0.07188,
      "grad_norm": 0.964041753045374,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 7188
    },
    {
      "epoch": 0.07189,
      "grad_norm": 0.8917382273902423,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 7189
    },
    {
      "epoch": 0.0719,
      "grad_norm": 0.8732047816874139,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 7190
    },
    {
      "epoch": 0.07191,
      "grad_norm": 0.972536173342455,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 7191
    },
    {
      "epoch": 0.07192,
      "grad_norm": 1.0780834708500115,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7192
    },
    {
      "epoch": 0.07193,
      "grad_norm": 0.8395279271863723,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 7193
    },
    {
      "epoch": 0.07194,
      "grad_norm": 0.7648413650848259,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7194
    },
    {
      "epoch": 0.07195,
      "grad_norm": 0.8198768874993303,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 7195
    },
    {
      "epoch": 0.07196,
      "grad_norm": 0.9730451749490292,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7196
    },
    {
      "epoch": 0.07197,
      "grad_norm": 1.078359113639344,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 7197
    },
    {
      "epoch": 0.07198,
      "grad_norm": 0.872979660810934,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7198
    },
    {
      "epoch": 0.07199,
      "grad_norm": 0.7594430118398644,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 7199
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.8418417578675667,
      "learning_rate": 0.003,
      "loss": 4.1276,
      "step": 7200
    },
    {
      "epoch": 0.07201,
      "grad_norm": 1.0033287381546427,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 7201
    },
    {
      "epoch": 0.07202,
      "grad_norm": 1.091821079155893,
      "learning_rate": 0.003,
      "loss": 4.137,
      "step": 7202
    },
    {
      "epoch": 0.07203,
      "grad_norm": 0.8990512736019646,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7203
    },
    {
      "epoch": 0.07204,
      "grad_norm": 0.8939835184490502,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 7204
    },
    {
      "epoch": 0.07205,
      "grad_norm": 1.0106951711454872,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 7205
    },
    {
      "epoch": 0.07206,
      "grad_norm": 1.0501076234127416,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7206
    },
    {
      "epoch": 0.07207,
      "grad_norm": 0.8683748356773303,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7207
    },
    {
      "epoch": 0.07208,
      "grad_norm": 0.9807707638242467,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7208
    },
    {
      "epoch": 0.07209,
      "grad_norm": 1.313720322166701,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 7209
    },
    {
      "epoch": 0.0721,
      "grad_norm": 0.9036061216874292,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7210
    },
    {
      "epoch": 0.07211,
      "grad_norm": 0.8532801365525488,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7211
    },
    {
      "epoch": 0.07212,
      "grad_norm": 0.8177440544033299,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 7212
    },
    {
      "epoch": 0.07213,
      "grad_norm": 0.7877402885555771,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 7213
    },
    {
      "epoch": 0.07214,
      "grad_norm": 0.7962574349869463,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 7214
    },
    {
      "epoch": 0.07215,
      "grad_norm": 0.840821954584076,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7215
    },
    {
      "epoch": 0.07216,
      "grad_norm": 1.0202453719622064,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 7216
    },
    {
      "epoch": 0.07217,
      "grad_norm": 1.0210177210663058,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 7217
    },
    {
      "epoch": 0.07218,
      "grad_norm": 0.9291393542941405,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7218
    },
    {
      "epoch": 0.07219,
      "grad_norm": 0.9125365057037023,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7219
    },
    {
      "epoch": 0.0722,
      "grad_norm": 0.8684648065476432,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 7220
    },
    {
      "epoch": 0.07221,
      "grad_norm": 0.8203475478863702,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7221
    },
    {
      "epoch": 0.07222,
      "grad_norm": 0.7337104645338505,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7222
    },
    {
      "epoch": 0.07223,
      "grad_norm": 0.7540489703643396,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 7223
    },
    {
      "epoch": 0.07224,
      "grad_norm": 0.791141223365698,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 7224
    },
    {
      "epoch": 0.07225,
      "grad_norm": 0.8289724417410262,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7225
    },
    {
      "epoch": 0.07226,
      "grad_norm": 0.7971187373872961,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 7226
    },
    {
      "epoch": 0.07227,
      "grad_norm": 0.8010700923864582,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 7227
    },
    {
      "epoch": 0.07228,
      "grad_norm": 0.867491024834857,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 7228
    },
    {
      "epoch": 0.07229,
      "grad_norm": 0.9726581300415823,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7229
    },
    {
      "epoch": 0.0723,
      "grad_norm": 1.0772680849810623,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 7230
    },
    {
      "epoch": 0.07231,
      "grad_norm": 1.0210940982142689,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 7231
    },
    {
      "epoch": 0.07232,
      "grad_norm": 0.8347979960369515,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 7232
    },
    {
      "epoch": 0.07233,
      "grad_norm": 0.7798104407644068,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 7233
    },
    {
      "epoch": 0.07234,
      "grad_norm": 0.7699473287189199,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 7234
    },
    {
      "epoch": 0.07235,
      "grad_norm": 0.8633138422643347,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 7235
    },
    {
      "epoch": 0.07236,
      "grad_norm": 1.0997904141948405,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 7236
    },
    {
      "epoch": 0.07237,
      "grad_norm": 0.9961327053280845,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 7237
    },
    {
      "epoch": 0.07238,
      "grad_norm": 0.8483548613066104,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 7238
    },
    {
      "epoch": 0.07239,
      "grad_norm": 0.7701019036531008,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 7239
    },
    {
      "epoch": 0.0724,
      "grad_norm": 0.7417934914192521,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 7240
    },
    {
      "epoch": 0.07241,
      "grad_norm": 0.7191524525608475,
      "learning_rate": 0.003,
      "loss": 4.1154,
      "step": 7241
    },
    {
      "epoch": 0.07242,
      "grad_norm": 0.8072871682611079,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 7242
    },
    {
      "epoch": 0.07243,
      "grad_norm": 0.8184708621816302,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 7243
    },
    {
      "epoch": 0.07244,
      "grad_norm": 0.7093054794981116,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 7244
    },
    {
      "epoch": 0.07245,
      "grad_norm": 0.675330894388966,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 7245
    },
    {
      "epoch": 0.07246,
      "grad_norm": 0.7323266139017566,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 7246
    },
    {
      "epoch": 0.07247,
      "grad_norm": 0.780943196371678,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 7247
    },
    {
      "epoch": 0.07248,
      "grad_norm": 0.899671652675055,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 7248
    },
    {
      "epoch": 0.07249,
      "grad_norm": 1.1022793278108771,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 7249
    },
    {
      "epoch": 0.0725,
      "grad_norm": 1.1608600206621666,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 7250
    },
    {
      "epoch": 0.07251,
      "grad_norm": 0.9173391017419072,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7251
    },
    {
      "epoch": 0.07252,
      "grad_norm": 0.8543519564138532,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 7252
    },
    {
      "epoch": 0.07253,
      "grad_norm": 0.7929389671267115,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 7253
    },
    {
      "epoch": 0.07254,
      "grad_norm": 0.8566224768735904,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 7254
    },
    {
      "epoch": 0.07255,
      "grad_norm": 0.9509981257694082,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 7255
    },
    {
      "epoch": 0.07256,
      "grad_norm": 1.0323475737017942,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 7256
    },
    {
      "epoch": 0.07257,
      "grad_norm": 1.125862787361634,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 7257
    },
    {
      "epoch": 0.07258,
      "grad_norm": 0.9874067143143448,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7258
    },
    {
      "epoch": 0.07259,
      "grad_norm": 1.1005675012845613,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7259
    },
    {
      "epoch": 0.0726,
      "grad_norm": 0.8454037889791309,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 7260
    },
    {
      "epoch": 0.07261,
      "grad_norm": 0.856806639173543,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7261
    },
    {
      "epoch": 0.07262,
      "grad_norm": 1.1774126393139672,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 7262
    },
    {
      "epoch": 0.07263,
      "grad_norm": 0.98856001091982,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 7263
    },
    {
      "epoch": 0.07264,
      "grad_norm": 1.052918037838861,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 7264
    },
    {
      "epoch": 0.07265,
      "grad_norm": 0.9481848938716576,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7265
    },
    {
      "epoch": 0.07266,
      "grad_norm": 0.8686413426498576,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 7266
    },
    {
      "epoch": 0.07267,
      "grad_norm": 0.8525273425159318,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 7267
    },
    {
      "epoch": 0.07268,
      "grad_norm": 0.9533691584951004,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 7268
    },
    {
      "epoch": 0.07269,
      "grad_norm": 0.9840691218693606,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7269
    },
    {
      "epoch": 0.0727,
      "grad_norm": 1.022133557552684,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7270
    },
    {
      "epoch": 0.07271,
      "grad_norm": 0.9427872988130471,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 7271
    },
    {
      "epoch": 0.07272,
      "grad_norm": 0.8835705515323187,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7272
    },
    {
      "epoch": 0.07273,
      "grad_norm": 0.7662262254321305,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 7273
    },
    {
      "epoch": 0.07274,
      "grad_norm": 0.6861881669104918,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 7274
    },
    {
      "epoch": 0.07275,
      "grad_norm": 0.6940748807512555,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 7275
    },
    {
      "epoch": 0.07276,
      "grad_norm": 0.7255703327321986,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 7276
    },
    {
      "epoch": 0.07277,
      "grad_norm": 0.826898534170412,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 7277
    },
    {
      "epoch": 0.07278,
      "grad_norm": 0.7205286139763815,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 7278
    },
    {
      "epoch": 0.07279,
      "grad_norm": 0.7870100781810526,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 7279
    },
    {
      "epoch": 0.0728,
      "grad_norm": 1.1224906772039385,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 7280
    },
    {
      "epoch": 0.07281,
      "grad_norm": 1.064756942044224,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 7281
    },
    {
      "epoch": 0.07282,
      "grad_norm": 0.755328815035219,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 7282
    },
    {
      "epoch": 0.07283,
      "grad_norm": 0.7480564984223254,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 7283
    },
    {
      "epoch": 0.07284,
      "grad_norm": 0.7969334161992938,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 7284
    },
    {
      "epoch": 0.07285,
      "grad_norm": 0.8037999842824531,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 7285
    },
    {
      "epoch": 0.07286,
      "grad_norm": 0.8432356062305908,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 7286
    },
    {
      "epoch": 0.07287,
      "grad_norm": 0.7074733000617406,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7287
    },
    {
      "epoch": 0.07288,
      "grad_norm": 0.6862151638972047,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 7288
    },
    {
      "epoch": 0.07289,
      "grad_norm": 0.7256205161629711,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 7289
    },
    {
      "epoch": 0.0729,
      "grad_norm": 0.7506778983987582,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7290
    },
    {
      "epoch": 0.07291,
      "grad_norm": 0.9175171243997577,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 7291
    },
    {
      "epoch": 0.07292,
      "grad_norm": 1.1748119483133153,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 7292
    },
    {
      "epoch": 0.07293,
      "grad_norm": 0.946454415717222,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 7293
    },
    {
      "epoch": 0.07294,
      "grad_norm": 0.8588138723902505,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 7294
    },
    {
      "epoch": 0.07295,
      "grad_norm": 0.9200655141576249,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 7295
    },
    {
      "epoch": 0.07296,
      "grad_norm": 0.945584786641691,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 7296
    },
    {
      "epoch": 0.07297,
      "grad_norm": 0.9076831566613616,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7297
    },
    {
      "epoch": 0.07298,
      "grad_norm": 0.9132887839709575,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 7298
    },
    {
      "epoch": 0.07299,
      "grad_norm": 1.0600320642602021,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 7299
    },
    {
      "epoch": 0.073,
      "grad_norm": 1.1555254508950479,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 7300
    },
    {
      "epoch": 0.07301,
      "grad_norm": 0.8091012187372432,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 7301
    },
    {
      "epoch": 0.07302,
      "grad_norm": 0.7969703558310967,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 7302
    },
    {
      "epoch": 0.07303,
      "grad_norm": 0.8742412667236584,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7303
    },
    {
      "epoch": 0.07304,
      "grad_norm": 0.9295281273271904,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7304
    },
    {
      "epoch": 0.07305,
      "grad_norm": 0.9710795112852725,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 7305
    },
    {
      "epoch": 0.07306,
      "grad_norm": 0.969596840488141,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 7306
    },
    {
      "epoch": 0.07307,
      "grad_norm": 1.1571657345709758,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 7307
    },
    {
      "epoch": 0.07308,
      "grad_norm": 0.8780928378824396,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7308
    },
    {
      "epoch": 0.07309,
      "grad_norm": 0.7381472139260883,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 7309
    },
    {
      "epoch": 0.0731,
      "grad_norm": 0.7936367882676433,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 7310
    },
    {
      "epoch": 0.07311,
      "grad_norm": 0.8771254185491402,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7311
    },
    {
      "epoch": 0.07312,
      "grad_norm": 0.9049549301553881,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 7312
    },
    {
      "epoch": 0.07313,
      "grad_norm": 0.8671243317374602,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 7313
    },
    {
      "epoch": 0.07314,
      "grad_norm": 0.8712423511074834,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 7314
    },
    {
      "epoch": 0.07315,
      "grad_norm": 0.9012186242781552,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7315
    },
    {
      "epoch": 0.07316,
      "grad_norm": 1.00783380326016,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 7316
    },
    {
      "epoch": 0.07317,
      "grad_norm": 0.9312278344910015,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 7317
    },
    {
      "epoch": 0.07318,
      "grad_norm": 0.9073283549997471,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 7318
    },
    {
      "epoch": 0.07319,
      "grad_norm": 0.9182845586703946,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 7319
    },
    {
      "epoch": 0.0732,
      "grad_norm": 1.0130838674437885,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7320
    },
    {
      "epoch": 0.07321,
      "grad_norm": 1.1175547994945165,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 7321
    },
    {
      "epoch": 0.07322,
      "grad_norm": 0.9871197560308737,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 7322
    },
    {
      "epoch": 0.07323,
      "grad_norm": 0.9778441606954984,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7323
    },
    {
      "epoch": 0.07324,
      "grad_norm": 0.8298462372409252,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 7324
    },
    {
      "epoch": 0.07325,
      "grad_norm": 0.7996270399874803,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 7325
    },
    {
      "epoch": 0.07326,
      "grad_norm": 0.8480564833291753,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 7326
    },
    {
      "epoch": 0.07327,
      "grad_norm": 0.9655678078823727,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 7327
    },
    {
      "epoch": 0.07328,
      "grad_norm": 0.9383954669497225,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 7328
    },
    {
      "epoch": 0.07329,
      "grad_norm": 0.9699028773325669,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7329
    },
    {
      "epoch": 0.0733,
      "grad_norm": 0.9768618937856877,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 7330
    },
    {
      "epoch": 0.07331,
      "grad_norm": 0.9467415617801669,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 7331
    },
    {
      "epoch": 0.07332,
      "grad_norm": 0.8567564686017997,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7332
    },
    {
      "epoch": 0.07333,
      "grad_norm": 0.84392576775453,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 7333
    },
    {
      "epoch": 0.07334,
      "grad_norm": 0.8137901445810288,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7334
    },
    {
      "epoch": 0.07335,
      "grad_norm": 0.9003483859560396,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 7335
    },
    {
      "epoch": 0.07336,
      "grad_norm": 1.0011860597622482,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 7336
    },
    {
      "epoch": 0.07337,
      "grad_norm": 1.0810806000621536,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7337
    },
    {
      "epoch": 0.07338,
      "grad_norm": 0.987420151810005,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 7338
    },
    {
      "epoch": 0.07339,
      "grad_norm": 0.8647937623676709,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 7339
    },
    {
      "epoch": 0.0734,
      "grad_norm": 0.823939663089722,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 7340
    },
    {
      "epoch": 0.07341,
      "grad_norm": 0.8020510247193195,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7341
    },
    {
      "epoch": 0.07342,
      "grad_norm": 0.8621826681855989,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 7342
    },
    {
      "epoch": 0.07343,
      "grad_norm": 0.96982132901119,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 7343
    },
    {
      "epoch": 0.07344,
      "grad_norm": 1.0089151394572116,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7344
    },
    {
      "epoch": 0.07345,
      "grad_norm": 1.0012381625820561,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 7345
    },
    {
      "epoch": 0.07346,
      "grad_norm": 1.0578613234075809,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 7346
    },
    {
      "epoch": 0.07347,
      "grad_norm": 0.9368355879723315,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 7347
    },
    {
      "epoch": 0.07348,
      "grad_norm": 0.98538363953966,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7348
    },
    {
      "epoch": 0.07349,
      "grad_norm": 1.0421211993700539,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7349
    },
    {
      "epoch": 0.0735,
      "grad_norm": 0.8354136210538131,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 7350
    },
    {
      "epoch": 0.07351,
      "grad_norm": 0.7274907728499738,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 7351
    },
    {
      "epoch": 0.07352,
      "grad_norm": 0.7679883952167378,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 7352
    },
    {
      "epoch": 0.07353,
      "grad_norm": 0.8616796265202713,
      "learning_rate": 0.003,
      "loss": 4.126,
      "step": 7353
    },
    {
      "epoch": 0.07354,
      "grad_norm": 0.9798851246937682,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 7354
    },
    {
      "epoch": 0.07355,
      "grad_norm": 1.0511225960032606,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 7355
    },
    {
      "epoch": 0.07356,
      "grad_norm": 0.9587450442062034,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 7356
    },
    {
      "epoch": 0.07357,
      "grad_norm": 0.8805872619152303,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7357
    },
    {
      "epoch": 0.07358,
      "grad_norm": 0.849654200097335,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 7358
    },
    {
      "epoch": 0.07359,
      "grad_norm": 0.8872628692069536,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7359
    },
    {
      "epoch": 0.0736,
      "grad_norm": 1.0989657984839474,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 7360
    },
    {
      "epoch": 0.07361,
      "grad_norm": 0.8695135396560348,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 7361
    },
    {
      "epoch": 0.07362,
      "grad_norm": 0.8528139392382852,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 7362
    },
    {
      "epoch": 0.07363,
      "grad_norm": 0.8772352719466499,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 7363
    },
    {
      "epoch": 0.07364,
      "grad_norm": 0.9572152378472161,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 7364
    },
    {
      "epoch": 0.07365,
      "grad_norm": 1.2418185340432195,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 7365
    },
    {
      "epoch": 0.07366,
      "grad_norm": 0.7569511177947986,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 7366
    },
    {
      "epoch": 0.07367,
      "grad_norm": 0.7408105950286488,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7367
    },
    {
      "epoch": 0.07368,
      "grad_norm": 0.9113307753760953,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 7368
    },
    {
      "epoch": 0.07369,
      "grad_norm": 1.1054144696822306,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 7369
    },
    {
      "epoch": 0.0737,
      "grad_norm": 0.9173693391865863,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 7370
    },
    {
      "epoch": 0.07371,
      "grad_norm": 0.7126602835933623,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 7371
    },
    {
      "epoch": 0.07372,
      "grad_norm": 0.7070143475341606,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7372
    },
    {
      "epoch": 0.07373,
      "grad_norm": 0.7912761016620928,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7373
    },
    {
      "epoch": 0.07374,
      "grad_norm": 0.7382445142368581,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 7374
    },
    {
      "epoch": 0.07375,
      "grad_norm": 0.792351842888862,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 7375
    },
    {
      "epoch": 0.07376,
      "grad_norm": 0.8906583846931676,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 7376
    },
    {
      "epoch": 0.07377,
      "grad_norm": 0.9447731004050475,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 7377
    },
    {
      "epoch": 0.07378,
      "grad_norm": 0.8173782820222226,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 7378
    },
    {
      "epoch": 0.07379,
      "grad_norm": 0.8061166590769808,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 7379
    },
    {
      "epoch": 0.0738,
      "grad_norm": 0.9117507479699275,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 7380
    },
    {
      "epoch": 0.07381,
      "grad_norm": 0.8833902157946827,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 7381
    },
    {
      "epoch": 0.07382,
      "grad_norm": 0.859726728050116,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 7382
    },
    {
      "epoch": 0.07383,
      "grad_norm": 0.8848213921508847,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7383
    },
    {
      "epoch": 0.07384,
      "grad_norm": 1.0298301316190803,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7384
    },
    {
      "epoch": 0.07385,
      "grad_norm": 1.172557569337942,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7385
    },
    {
      "epoch": 0.07386,
      "grad_norm": 0.8669312655544931,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7386
    },
    {
      "epoch": 0.07387,
      "grad_norm": 0.8392913033776324,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 7387
    },
    {
      "epoch": 0.07388,
      "grad_norm": 0.7447523316906665,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 7388
    },
    {
      "epoch": 0.07389,
      "grad_norm": 0.6495275179258706,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 7389
    },
    {
      "epoch": 0.0739,
      "grad_norm": 0.6025258391763327,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 7390
    },
    {
      "epoch": 0.07391,
      "grad_norm": 0.6318049445146262,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 7391
    },
    {
      "epoch": 0.07392,
      "grad_norm": 0.7343060747804614,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 7392
    },
    {
      "epoch": 0.07393,
      "grad_norm": 1.0000247571371808,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7393
    },
    {
      "epoch": 0.07394,
      "grad_norm": 1.5060068670788715,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 7394
    },
    {
      "epoch": 0.07395,
      "grad_norm": 0.6499300765174943,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 7395
    },
    {
      "epoch": 0.07396,
      "grad_norm": 0.8568555410525963,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 7396
    },
    {
      "epoch": 0.07397,
      "grad_norm": 0.9063071745959117,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7397
    },
    {
      "epoch": 0.07398,
      "grad_norm": 0.9019215085358515,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 7398
    },
    {
      "epoch": 0.07399,
      "grad_norm": 0.9546957281863762,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 7399
    },
    {
      "epoch": 0.074,
      "grad_norm": 0.8849692017723262,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7400
    },
    {
      "epoch": 0.07401,
      "grad_norm": 0.8546205777376308,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 7401
    },
    {
      "epoch": 0.07402,
      "grad_norm": 0.7032841794462815,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7402
    },
    {
      "epoch": 0.07403,
      "grad_norm": 0.7639352112565744,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 7403
    },
    {
      "epoch": 0.07404,
      "grad_norm": 0.8684415164326015,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 7404
    },
    {
      "epoch": 0.07405,
      "grad_norm": 1.0166394680109212,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7405
    },
    {
      "epoch": 0.07406,
      "grad_norm": 1.1062291356675835,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7406
    },
    {
      "epoch": 0.07407,
      "grad_norm": 0.9316736131089458,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 7407
    },
    {
      "epoch": 0.07408,
      "grad_norm": 0.8517821079613881,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7408
    },
    {
      "epoch": 0.07409,
      "grad_norm": 0.8096451130433067,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7409
    },
    {
      "epoch": 0.0741,
      "grad_norm": 0.7521940521106885,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7410
    },
    {
      "epoch": 0.07411,
      "grad_norm": 0.7925702752865725,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 7411
    },
    {
      "epoch": 0.07412,
      "grad_norm": 0.8215884643463309,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7412
    },
    {
      "epoch": 0.07413,
      "grad_norm": 0.9250271828845864,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7413
    },
    {
      "epoch": 0.07414,
      "grad_norm": 1.1770944944492254,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7414
    },
    {
      "epoch": 0.07415,
      "grad_norm": 1.0576816988803364,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 7415
    },
    {
      "epoch": 0.07416,
      "grad_norm": 0.9147150961080485,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7416
    },
    {
      "epoch": 0.07417,
      "grad_norm": 0.8841312717180329,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 7417
    },
    {
      "epoch": 0.07418,
      "grad_norm": 1.0311827398926998,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 7418
    },
    {
      "epoch": 0.07419,
      "grad_norm": 1.0768166118437081,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 7419
    },
    {
      "epoch": 0.0742,
      "grad_norm": 0.988151970685207,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 7420
    },
    {
      "epoch": 0.07421,
      "grad_norm": 0.9184883764372472,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 7421
    },
    {
      "epoch": 0.07422,
      "grad_norm": 0.8685292845726943,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7422
    },
    {
      "epoch": 0.07423,
      "grad_norm": 0.8906473884300916,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 7423
    },
    {
      "epoch": 0.07424,
      "grad_norm": 1.0992726932466628,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 7424
    },
    {
      "epoch": 0.07425,
      "grad_norm": 1.169164042468968,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7425
    },
    {
      "epoch": 0.07426,
      "grad_norm": 0.8731095700449631,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 7426
    },
    {
      "epoch": 0.07427,
      "grad_norm": 0.8974625019426375,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 7427
    },
    {
      "epoch": 0.07428,
      "grad_norm": 0.8619370872170999,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 7428
    },
    {
      "epoch": 0.07429,
      "grad_norm": 0.8606884636965143,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7429
    },
    {
      "epoch": 0.0743,
      "grad_norm": 0.9104655756270731,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 7430
    },
    {
      "epoch": 0.07431,
      "grad_norm": 1.0461049195727867,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 7431
    },
    {
      "epoch": 0.07432,
      "grad_norm": 1.030513769267123,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 7432
    },
    {
      "epoch": 0.07433,
      "grad_norm": 0.7826646595179413,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 7433
    },
    {
      "epoch": 0.07434,
      "grad_norm": 0.8449733596632347,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 7434
    },
    {
      "epoch": 0.07435,
      "grad_norm": 0.829034109699188,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7435
    },
    {
      "epoch": 0.07436,
      "grad_norm": 0.7276311889102272,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7436
    },
    {
      "epoch": 0.07437,
      "grad_norm": 0.6962567087169638,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 7437
    },
    {
      "epoch": 0.07438,
      "grad_norm": 0.754150337723482,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7438
    },
    {
      "epoch": 0.07439,
      "grad_norm": 0.8394913294621994,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7439
    },
    {
      "epoch": 0.0744,
      "grad_norm": 0.9210027114951105,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7440
    },
    {
      "epoch": 0.07441,
      "grad_norm": 0.9677662290482478,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 7441
    },
    {
      "epoch": 0.07442,
      "grad_norm": 0.9984463151295965,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 7442
    },
    {
      "epoch": 0.07443,
      "grad_norm": 0.8595131868015653,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 7443
    },
    {
      "epoch": 0.07444,
      "grad_norm": 0.7683813726273543,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 7444
    },
    {
      "epoch": 0.07445,
      "grad_norm": 0.753682686098993,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7445
    },
    {
      "epoch": 0.07446,
      "grad_norm": 0.8268762150034119,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 7446
    },
    {
      "epoch": 0.07447,
      "grad_norm": 1.094602286197476,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7447
    },
    {
      "epoch": 0.07448,
      "grad_norm": 0.9771739644265591,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 7448
    },
    {
      "epoch": 0.07449,
      "grad_norm": 0.8384542889699795,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 7449
    },
    {
      "epoch": 0.0745,
      "grad_norm": 0.7590190855947146,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7450
    },
    {
      "epoch": 0.07451,
      "grad_norm": 0.8937331488483308,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 7451
    },
    {
      "epoch": 0.07452,
      "grad_norm": 0.9693727157247541,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 7452
    },
    {
      "epoch": 0.07453,
      "grad_norm": 1.079219331582194,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 7453
    },
    {
      "epoch": 0.07454,
      "grad_norm": 0.9744818238494153,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 7454
    },
    {
      "epoch": 0.07455,
      "grad_norm": 0.9536322310393928,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 7455
    },
    {
      "epoch": 0.07456,
      "grad_norm": 1.0769846542491077,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7456
    },
    {
      "epoch": 0.07457,
      "grad_norm": 1.0409702107301997,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 7457
    },
    {
      "epoch": 0.07458,
      "grad_norm": 1.001295432389869,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 7458
    },
    {
      "epoch": 0.07459,
      "grad_norm": 0.8889128511582081,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 7459
    },
    {
      "epoch": 0.0746,
      "grad_norm": 0.9411931447305963,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 7460
    },
    {
      "epoch": 0.07461,
      "grad_norm": 0.8519572681750707,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 7461
    },
    {
      "epoch": 0.07462,
      "grad_norm": 0.7905764728430406,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 7462
    },
    {
      "epoch": 0.07463,
      "grad_norm": 0.8052221816614337,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 7463
    },
    {
      "epoch": 0.07464,
      "grad_norm": 0.9184878977339718,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 7464
    },
    {
      "epoch": 0.07465,
      "grad_norm": 0.9096124012536108,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 7465
    },
    {
      "epoch": 0.07466,
      "grad_norm": 0.8868750508524639,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 7466
    },
    {
      "epoch": 0.07467,
      "grad_norm": 1.1739238328463417,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7467
    },
    {
      "epoch": 0.07468,
      "grad_norm": 1.2731034343951069,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 7468
    },
    {
      "epoch": 0.07469,
      "grad_norm": 0.9599056408935575,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 7469
    },
    {
      "epoch": 0.0747,
      "grad_norm": 0.8545661960822021,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7470
    },
    {
      "epoch": 0.07471,
      "grad_norm": 0.8017399886069515,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 7471
    },
    {
      "epoch": 0.07472,
      "grad_norm": 0.8220636823261487,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7472
    },
    {
      "epoch": 0.07473,
      "grad_norm": 0.8333913762759374,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 7473
    },
    {
      "epoch": 0.07474,
      "grad_norm": 0.9949440880989607,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 7474
    },
    {
      "epoch": 0.07475,
      "grad_norm": 1.0047709217104455,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 7475
    },
    {
      "epoch": 0.07476,
      "grad_norm": 0.9783980895093549,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 7476
    },
    {
      "epoch": 0.07477,
      "grad_norm": 1.089186008059605,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 7477
    },
    {
      "epoch": 0.07478,
      "grad_norm": 1.1072575591847578,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 7478
    },
    {
      "epoch": 0.07479,
      "grad_norm": 1.1298579787016074,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 7479
    },
    {
      "epoch": 0.0748,
      "grad_norm": 0.9998908890872064,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 7480
    },
    {
      "epoch": 0.07481,
      "grad_norm": 0.8169704618263066,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 7481
    },
    {
      "epoch": 0.07482,
      "grad_norm": 0.6487475847193562,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 7482
    },
    {
      "epoch": 0.07483,
      "grad_norm": 0.6863949014856395,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 7483
    },
    {
      "epoch": 0.07484,
      "grad_norm": 0.6587103027574833,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 7484
    },
    {
      "epoch": 0.07485,
      "grad_norm": 0.7160032679100043,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 7485
    },
    {
      "epoch": 0.07486,
      "grad_norm": 0.9672056856968704,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 7486
    },
    {
      "epoch": 0.07487,
      "grad_norm": 1.1696254951341385,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 7487
    },
    {
      "epoch": 0.07488,
      "grad_norm": 0.7863461489638554,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7488
    },
    {
      "epoch": 0.07489,
      "grad_norm": 0.7384563910599533,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7489
    },
    {
      "epoch": 0.0749,
      "grad_norm": 0.7572591385660193,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 7490
    },
    {
      "epoch": 0.07491,
      "grad_norm": 0.9074264182707067,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 7491
    },
    {
      "epoch": 0.07492,
      "grad_norm": 1.1199081661678163,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 7492
    },
    {
      "epoch": 0.07493,
      "grad_norm": 1.1575662283283366,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 7493
    },
    {
      "epoch": 0.07494,
      "grad_norm": 0.9434610130517279,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 7494
    },
    {
      "epoch": 0.07495,
      "grad_norm": 0.8748804539205894,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7495
    },
    {
      "epoch": 0.07496,
      "grad_norm": 0.8175617506999896,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7496
    },
    {
      "epoch": 0.07497,
      "grad_norm": 0.9144161517532557,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7497
    },
    {
      "epoch": 0.07498,
      "grad_norm": 0.9185423406505983,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 7498
    },
    {
      "epoch": 0.07499,
      "grad_norm": 0.8280717177948074,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 7499
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.7960540221263401,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7500
    },
    {
      "epoch": 0.07501,
      "grad_norm": 0.6399790221522081,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 7501
    },
    {
      "epoch": 0.07502,
      "grad_norm": 0.6669586492655686,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 7502
    },
    {
      "epoch": 0.07503,
      "grad_norm": 0.8085394369052066,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 7503
    },
    {
      "epoch": 0.07504,
      "grad_norm": 1.1164579548991722,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7504
    },
    {
      "epoch": 0.07505,
      "grad_norm": 1.1764299489396552,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 7505
    },
    {
      "epoch": 0.07506,
      "grad_norm": 0.6778128482933068,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7506
    },
    {
      "epoch": 0.07507,
      "grad_norm": 0.6887355085828547,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 7507
    },
    {
      "epoch": 0.07508,
      "grad_norm": 0.8305506452463216,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 7508
    },
    {
      "epoch": 0.07509,
      "grad_norm": 0.9381005410193313,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7509
    },
    {
      "epoch": 0.0751,
      "grad_norm": 0.9286555796672374,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 7510
    },
    {
      "epoch": 0.07511,
      "grad_norm": 0.8493832587890462,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 7511
    },
    {
      "epoch": 0.07512,
      "grad_norm": 0.8203675397517558,
      "learning_rate": 0.003,
      "loss": 4.1186,
      "step": 7512
    },
    {
      "epoch": 0.07513,
      "grad_norm": 0.9987084064424658,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7513
    },
    {
      "epoch": 0.07514,
      "grad_norm": 1.0869538532848078,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 7514
    },
    {
      "epoch": 0.07515,
      "grad_norm": 1.0403169728871233,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 7515
    },
    {
      "epoch": 0.07516,
      "grad_norm": 1.011790174275216,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 7516
    },
    {
      "epoch": 0.07517,
      "grad_norm": 0.8942557366391805,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7517
    },
    {
      "epoch": 0.07518,
      "grad_norm": 0.7974760716257956,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 7518
    },
    {
      "epoch": 0.07519,
      "grad_norm": 0.7849755190610048,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 7519
    },
    {
      "epoch": 0.0752,
      "grad_norm": 0.7855656056140153,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 7520
    },
    {
      "epoch": 0.07521,
      "grad_norm": 0.8813805509682716,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 7521
    },
    {
      "epoch": 0.07522,
      "grad_norm": 0.9965079513901988,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 7522
    },
    {
      "epoch": 0.07523,
      "grad_norm": 1.2142599219211911,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 7523
    },
    {
      "epoch": 0.07524,
      "grad_norm": 0.7326420889303906,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 7524
    },
    {
      "epoch": 0.07525,
      "grad_norm": 0.7133952113276287,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 7525
    },
    {
      "epoch": 0.07526,
      "grad_norm": 0.7827807395620044,
      "learning_rate": 0.003,
      "loss": 4.1181,
      "step": 7526
    },
    {
      "epoch": 0.07527,
      "grad_norm": 0.8767676778583422,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 7527
    },
    {
      "epoch": 0.07528,
      "grad_norm": 0.9065608532077762,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 7528
    },
    {
      "epoch": 0.07529,
      "grad_norm": 0.959630193106546,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 7529
    },
    {
      "epoch": 0.0753,
      "grad_norm": 0.9481228165926106,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7530
    },
    {
      "epoch": 0.07531,
      "grad_norm": 1.0127377126343031,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 7531
    },
    {
      "epoch": 0.07532,
      "grad_norm": 1.1381689608302172,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 7532
    },
    {
      "epoch": 0.07533,
      "grad_norm": 0.9088536862602051,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 7533
    },
    {
      "epoch": 0.07534,
      "grad_norm": 0.9695169216091517,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7534
    },
    {
      "epoch": 0.07535,
      "grad_norm": 0.9754230578467733,
      "learning_rate": 0.003,
      "loss": 4.1159,
      "step": 7535
    },
    {
      "epoch": 0.07536,
      "grad_norm": 0.9562163044893859,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 7536
    },
    {
      "epoch": 0.07537,
      "grad_norm": 1.1305394732850846,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7537
    },
    {
      "epoch": 0.07538,
      "grad_norm": 0.9158848504564159,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 7538
    },
    {
      "epoch": 0.07539,
      "grad_norm": 0.9652651208133781,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 7539
    },
    {
      "epoch": 0.0754,
      "grad_norm": 0.9022392107906086,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7540
    },
    {
      "epoch": 0.07541,
      "grad_norm": 0.8225042354592814,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 7541
    },
    {
      "epoch": 0.07542,
      "grad_norm": 1.0114606631383205,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 7542
    },
    {
      "epoch": 0.07543,
      "grad_norm": 1.2258446672383758,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 7543
    },
    {
      "epoch": 0.07544,
      "grad_norm": 0.9065708338931346,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 7544
    },
    {
      "epoch": 0.07545,
      "grad_norm": 0.8413566763887645,
      "learning_rate": 0.003,
      "loss": 4.1358,
      "step": 7545
    },
    {
      "epoch": 0.07546,
      "grad_norm": 0.931446753858521,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7546
    },
    {
      "epoch": 0.07547,
      "grad_norm": 0.8924179590547977,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7547
    },
    {
      "epoch": 0.07548,
      "grad_norm": 0.7981884285863752,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 7548
    },
    {
      "epoch": 0.07549,
      "grad_norm": 0.847248132212302,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 7549
    },
    {
      "epoch": 0.0755,
      "grad_norm": 0.929686503004784,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 7550
    },
    {
      "epoch": 0.07551,
      "grad_norm": 1.0899432730160927,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 7551
    },
    {
      "epoch": 0.07552,
      "grad_norm": 0.9023134743903442,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 7552
    },
    {
      "epoch": 0.07553,
      "grad_norm": 0.9320609493389578,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 7553
    },
    {
      "epoch": 0.07554,
      "grad_norm": 0.9366611277818084,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 7554
    },
    {
      "epoch": 0.07555,
      "grad_norm": 1.1249548849939486,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 7555
    },
    {
      "epoch": 0.07556,
      "grad_norm": 0.8809337436260816,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 7556
    },
    {
      "epoch": 0.07557,
      "grad_norm": 0.8977538234717878,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 7557
    },
    {
      "epoch": 0.07558,
      "grad_norm": 0.9323094114843662,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7558
    },
    {
      "epoch": 0.07559,
      "grad_norm": 1.0365418541296352,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 7559
    },
    {
      "epoch": 0.0756,
      "grad_norm": 0.9700985533938646,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 7560
    },
    {
      "epoch": 0.07561,
      "grad_norm": 0.9945502187488869,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7561
    },
    {
      "epoch": 0.07562,
      "grad_norm": 1.0168375224426294,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 7562
    },
    {
      "epoch": 0.07563,
      "grad_norm": 0.9790876715890733,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7563
    },
    {
      "epoch": 0.07564,
      "grad_norm": 0.9910540055685099,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 7564
    },
    {
      "epoch": 0.07565,
      "grad_norm": 1.0341921379761243,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 7565
    },
    {
      "epoch": 0.07566,
      "grad_norm": 0.9443317462260504,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7566
    },
    {
      "epoch": 0.07567,
      "grad_norm": 1.0159322899398833,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 7567
    },
    {
      "epoch": 0.07568,
      "grad_norm": 1.1086490020888524,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 7568
    },
    {
      "epoch": 0.07569,
      "grad_norm": 0.861288591479629,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 7569
    },
    {
      "epoch": 0.0757,
      "grad_norm": 1.0504830493537256,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 7570
    },
    {
      "epoch": 0.07571,
      "grad_norm": 1.041967555005186,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7571
    },
    {
      "epoch": 0.07572,
      "grad_norm": 1.0344004502845179,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 7572
    },
    {
      "epoch": 0.07573,
      "grad_norm": 0.763379777310832,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 7573
    },
    {
      "epoch": 0.07574,
      "grad_norm": 0.7213694523357767,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 7574
    },
    {
      "epoch": 0.07575,
      "grad_norm": 0.708796097196343,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 7575
    },
    {
      "epoch": 0.07576,
      "grad_norm": 0.7248673136274486,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 7576
    },
    {
      "epoch": 0.07577,
      "grad_norm": 0.8612577774285249,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 7577
    },
    {
      "epoch": 0.07578,
      "grad_norm": 1.005619129828009,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 7578
    },
    {
      "epoch": 0.07579,
      "grad_norm": 1.0450194172356522,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7579
    },
    {
      "epoch": 0.0758,
      "grad_norm": 1.036801784903056,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7580
    },
    {
      "epoch": 0.07581,
      "grad_norm": 0.8846896117250778,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7581
    },
    {
      "epoch": 0.07582,
      "grad_norm": 0.8553042305382026,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 7582
    },
    {
      "epoch": 0.07583,
      "grad_norm": 0.746157598396527,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 7583
    },
    {
      "epoch": 0.07584,
      "grad_norm": 0.703281013230497,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 7584
    },
    {
      "epoch": 0.07585,
      "grad_norm": 0.717776305908099,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 7585
    },
    {
      "epoch": 0.07586,
      "grad_norm": 0.7361220487355621,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 7586
    },
    {
      "epoch": 0.07587,
      "grad_norm": 0.7746201501489567,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 7587
    },
    {
      "epoch": 0.07588,
      "grad_norm": 0.883273441061299,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 7588
    },
    {
      "epoch": 0.07589,
      "grad_norm": 1.0296051718436425,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 7589
    },
    {
      "epoch": 0.0759,
      "grad_norm": 1.0366218527846427,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7590
    },
    {
      "epoch": 0.07591,
      "grad_norm": 0.9785797112728007,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 7591
    },
    {
      "epoch": 0.07592,
      "grad_norm": 1.0644976856872927,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 7592
    },
    {
      "epoch": 0.07593,
      "grad_norm": 0.8193582499715998,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 7593
    },
    {
      "epoch": 0.07594,
      "grad_norm": 0.7786543529015693,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 7594
    },
    {
      "epoch": 0.07595,
      "grad_norm": 0.846428038730107,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 7595
    },
    {
      "epoch": 0.07596,
      "grad_norm": 0.8584163070813771,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 7596
    },
    {
      "epoch": 0.07597,
      "grad_norm": 0.9415235991639561,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 7597
    },
    {
      "epoch": 0.07598,
      "grad_norm": 1.0765153427852032,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 7598
    },
    {
      "epoch": 0.07599,
      "grad_norm": 0.804425870287582,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 7599
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.8318708876658205,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 7600
    },
    {
      "epoch": 0.07601,
      "grad_norm": 1.1185425010615238,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 7601
    },
    {
      "epoch": 0.07602,
      "grad_norm": 1.047634075986158,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 7602
    },
    {
      "epoch": 0.07603,
      "grad_norm": 0.9009409871203874,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7603
    },
    {
      "epoch": 0.07604,
      "grad_norm": 0.8568906673551611,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 7604
    },
    {
      "epoch": 0.07605,
      "grad_norm": 0.7784225426074195,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 7605
    },
    {
      "epoch": 0.07606,
      "grad_norm": 0.839250801955735,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 7606
    },
    {
      "epoch": 0.07607,
      "grad_norm": 0.8782045362404554,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 7607
    },
    {
      "epoch": 0.07608,
      "grad_norm": 0.9200848229872217,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 7608
    },
    {
      "epoch": 0.07609,
      "grad_norm": 0.8082524390097369,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 7609
    },
    {
      "epoch": 0.0761,
      "grad_norm": 0.872081723964123,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 7610
    },
    {
      "epoch": 0.07611,
      "grad_norm": 0.9341942398949311,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 7611
    },
    {
      "epoch": 0.07612,
      "grad_norm": 0.9227398398243862,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 7612
    },
    {
      "epoch": 0.07613,
      "grad_norm": 0.8683378115106041,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 7613
    },
    {
      "epoch": 0.07614,
      "grad_norm": 1.0063037001529425,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 7614
    },
    {
      "epoch": 0.07615,
      "grad_norm": 1.1435774924345972,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7615
    },
    {
      "epoch": 0.07616,
      "grad_norm": 0.8848928857504734,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7616
    },
    {
      "epoch": 0.07617,
      "grad_norm": 0.8941746199995625,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 7617
    },
    {
      "epoch": 0.07618,
      "grad_norm": 0.9902163474565946,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 7618
    },
    {
      "epoch": 0.07619,
      "grad_norm": 1.0531865923482646,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 7619
    },
    {
      "epoch": 0.0762,
      "grad_norm": 0.9255658320995923,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 7620
    },
    {
      "epoch": 0.07621,
      "grad_norm": 0.8649930523728451,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 7621
    },
    {
      "epoch": 0.07622,
      "grad_norm": 0.7989732615276255,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 7622
    },
    {
      "epoch": 0.07623,
      "grad_norm": 0.9110231085001621,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 7623
    },
    {
      "epoch": 0.07624,
      "grad_norm": 1.0834218353454426,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7624
    },
    {
      "epoch": 0.07625,
      "grad_norm": 1.0325495122882569,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 7625
    },
    {
      "epoch": 0.07626,
      "grad_norm": 0.8475068109381595,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7626
    },
    {
      "epoch": 0.07627,
      "grad_norm": 0.727600425780518,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7627
    },
    {
      "epoch": 0.07628,
      "grad_norm": 0.7632302827090037,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 7628
    },
    {
      "epoch": 0.07629,
      "grad_norm": 0.8788639933925416,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7629
    },
    {
      "epoch": 0.0763,
      "grad_norm": 1.0709031796647681,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 7630
    },
    {
      "epoch": 0.07631,
      "grad_norm": 1.074935960825108,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 7631
    },
    {
      "epoch": 0.07632,
      "grad_norm": 1.061277776789126,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7632
    },
    {
      "epoch": 0.07633,
      "grad_norm": 1.077834867567973,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 7633
    },
    {
      "epoch": 0.07634,
      "grad_norm": 0.8857955127757137,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 7634
    },
    {
      "epoch": 0.07635,
      "grad_norm": 0.7954993888470026,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 7635
    },
    {
      "epoch": 0.07636,
      "grad_norm": 0.7860087645669086,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 7636
    },
    {
      "epoch": 0.07637,
      "grad_norm": 0.8285890551300037,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 7637
    },
    {
      "epoch": 0.07638,
      "grad_norm": 0.8451951844334696,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 7638
    },
    {
      "epoch": 0.07639,
      "grad_norm": 0.8191635031988865,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7639
    },
    {
      "epoch": 0.0764,
      "grad_norm": 0.8579832231757687,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 7640
    },
    {
      "epoch": 0.07641,
      "grad_norm": 0.8068346229932589,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 7641
    },
    {
      "epoch": 0.07642,
      "grad_norm": 0.8465566767852095,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 7642
    },
    {
      "epoch": 0.07643,
      "grad_norm": 0.8119061434375141,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7643
    },
    {
      "epoch": 0.07644,
      "grad_norm": 0.793320262503154,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 7644
    },
    {
      "epoch": 0.07645,
      "grad_norm": 0.9691323458548825,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 7645
    },
    {
      "epoch": 0.07646,
      "grad_norm": 1.3768015813021042,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 7646
    },
    {
      "epoch": 0.07647,
      "grad_norm": 0.9773228446390375,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 7647
    },
    {
      "epoch": 0.07648,
      "grad_norm": 1.051227994942795,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 7648
    },
    {
      "epoch": 0.07649,
      "grad_norm": 0.8613190674836249,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7649
    },
    {
      "epoch": 0.0765,
      "grad_norm": 0.8644471785420074,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7650
    },
    {
      "epoch": 0.07651,
      "grad_norm": 0.893229363014541,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 7651
    },
    {
      "epoch": 0.07652,
      "grad_norm": 0.8239815130553648,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 7652
    },
    {
      "epoch": 0.07653,
      "grad_norm": 0.8454893514249822,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 7653
    },
    {
      "epoch": 0.07654,
      "grad_norm": 0.7622344925499536,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7654
    },
    {
      "epoch": 0.07655,
      "grad_norm": 0.7645190916741668,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 7655
    },
    {
      "epoch": 0.07656,
      "grad_norm": 0.813005250974179,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 7656
    },
    {
      "epoch": 0.07657,
      "grad_norm": 0.8896111934604986,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 7657
    },
    {
      "epoch": 0.07658,
      "grad_norm": 0.8706430779245553,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 7658
    },
    {
      "epoch": 0.07659,
      "grad_norm": 0.8081972670264891,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 7659
    },
    {
      "epoch": 0.0766,
      "grad_norm": 0.7729443681983716,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 7660
    },
    {
      "epoch": 0.07661,
      "grad_norm": 0.862803077545191,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 7661
    },
    {
      "epoch": 0.07662,
      "grad_norm": 1.0006964442318478,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 7662
    },
    {
      "epoch": 0.07663,
      "grad_norm": 0.9936477236936112,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7663
    },
    {
      "epoch": 0.07664,
      "grad_norm": 1.0569422427730744,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7664
    },
    {
      "epoch": 0.07665,
      "grad_norm": 0.900621822760911,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 7665
    },
    {
      "epoch": 0.07666,
      "grad_norm": 0.8581843257219371,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 7666
    },
    {
      "epoch": 0.07667,
      "grad_norm": 0.983726396544071,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 7667
    },
    {
      "epoch": 0.07668,
      "grad_norm": 1.1069406233194603,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 7668
    },
    {
      "epoch": 0.07669,
      "grad_norm": 1.0015927610296076,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7669
    },
    {
      "epoch": 0.0767,
      "grad_norm": 1.0573154199589114,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 7670
    },
    {
      "epoch": 0.07671,
      "grad_norm": 0.9825877248296772,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 7671
    },
    {
      "epoch": 0.07672,
      "grad_norm": 0.9631425499922999,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 7672
    },
    {
      "epoch": 0.07673,
      "grad_norm": 1.0326441105372977,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 7673
    },
    {
      "epoch": 0.07674,
      "grad_norm": 1.0631847500181333,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7674
    },
    {
      "epoch": 0.07675,
      "grad_norm": 1.101230295023731,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 7675
    },
    {
      "epoch": 0.07676,
      "grad_norm": 0.9273268109359967,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 7676
    },
    {
      "epoch": 0.07677,
      "grad_norm": 0.8619308640763275,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 7677
    },
    {
      "epoch": 0.07678,
      "grad_norm": 0.9052969792740448,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 7678
    },
    {
      "epoch": 0.07679,
      "grad_norm": 0.9378634519819189,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 7679
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.8521164764973965,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 7680
    },
    {
      "epoch": 0.07681,
      "grad_norm": 0.8595477564055118,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7681
    },
    {
      "epoch": 0.07682,
      "grad_norm": 0.7754677865078394,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 7682
    },
    {
      "epoch": 0.07683,
      "grad_norm": 0.6840059827859887,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 7683
    },
    {
      "epoch": 0.07684,
      "grad_norm": 0.6575974583729506,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 7684
    },
    {
      "epoch": 0.07685,
      "grad_norm": 0.7235799579926786,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 7685
    },
    {
      "epoch": 0.07686,
      "grad_norm": 0.8377398222220133,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 7686
    },
    {
      "epoch": 0.07687,
      "grad_norm": 0.9082625532369022,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 7687
    },
    {
      "epoch": 0.07688,
      "grad_norm": 1.2331010320496771,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 7688
    },
    {
      "epoch": 0.07689,
      "grad_norm": 0.8212537354275941,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 7689
    },
    {
      "epoch": 0.0769,
      "grad_norm": 0.6665825438931752,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7690
    },
    {
      "epoch": 0.07691,
      "grad_norm": 0.7566566405069987,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 7691
    },
    {
      "epoch": 0.07692,
      "grad_norm": 0.8271311664477755,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 7692
    },
    {
      "epoch": 0.07693,
      "grad_norm": 0.8707736692236765,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 7693
    },
    {
      "epoch": 0.07694,
      "grad_norm": 0.8808612418292081,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 7694
    },
    {
      "epoch": 0.07695,
      "grad_norm": 1.1008954469557017,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7695
    },
    {
      "epoch": 0.07696,
      "grad_norm": 1.0606187645237288,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 7696
    },
    {
      "epoch": 0.07697,
      "grad_norm": 1.0672701176658974,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 7697
    },
    {
      "epoch": 0.07698,
      "grad_norm": 0.881431735642276,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 7698
    },
    {
      "epoch": 0.07699,
      "grad_norm": 0.9075150605445227,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 7699
    },
    {
      "epoch": 0.077,
      "grad_norm": 0.9026931164648633,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 7700
    },
    {
      "epoch": 0.07701,
      "grad_norm": 1.0261275216945884,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 7701
    },
    {
      "epoch": 0.07702,
      "grad_norm": 1.0674364416061641,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 7702
    },
    {
      "epoch": 0.07703,
      "grad_norm": 0.8464641737302251,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 7703
    },
    {
      "epoch": 0.07704,
      "grad_norm": 0.7763702415342664,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 7704
    },
    {
      "epoch": 0.07705,
      "grad_norm": 0.7689660168469808,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 7705
    },
    {
      "epoch": 0.07706,
      "grad_norm": 0.8664018808766383,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7706
    },
    {
      "epoch": 0.07707,
      "grad_norm": 1.0667476859474658,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7707
    },
    {
      "epoch": 0.07708,
      "grad_norm": 1.2058062363442252,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 7708
    },
    {
      "epoch": 0.07709,
      "grad_norm": 1.0492356527458138,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 7709
    },
    {
      "epoch": 0.0771,
      "grad_norm": 0.9750451640390347,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 7710
    },
    {
      "epoch": 0.07711,
      "grad_norm": 0.8336813405535649,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 7711
    },
    {
      "epoch": 0.07712,
      "grad_norm": 0.8155486398893903,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 7712
    },
    {
      "epoch": 0.07713,
      "grad_norm": 0.8688818823181422,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 7713
    },
    {
      "epoch": 0.07714,
      "grad_norm": 0.9332384720820431,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 7714
    },
    {
      "epoch": 0.07715,
      "grad_norm": 1.1579449218264937,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7715
    },
    {
      "epoch": 0.07716,
      "grad_norm": 1.0775937449235249,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 7716
    },
    {
      "epoch": 0.07717,
      "grad_norm": 0.8872808956615507,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 7717
    },
    {
      "epoch": 0.07718,
      "grad_norm": 0.8502430408961995,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 7718
    },
    {
      "epoch": 0.07719,
      "grad_norm": 0.8840941630249397,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7719
    },
    {
      "epoch": 0.0772,
      "grad_norm": 0.9698792840195539,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 7720
    },
    {
      "epoch": 0.07721,
      "grad_norm": 0.9769844568177094,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7721
    },
    {
      "epoch": 0.07722,
      "grad_norm": 1.0502547313312145,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 7722
    },
    {
      "epoch": 0.07723,
      "grad_norm": 1.0515641978538086,
      "learning_rate": 0.003,
      "loss": 4.1316,
      "step": 7723
    },
    {
      "epoch": 0.07724,
      "grad_norm": 0.8196988574303467,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 7724
    },
    {
      "epoch": 0.07725,
      "grad_norm": 0.7914284658585109,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7725
    },
    {
      "epoch": 0.07726,
      "grad_norm": 0.8789170064349751,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 7726
    },
    {
      "epoch": 0.07727,
      "grad_norm": 1.010851883887785,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 7727
    },
    {
      "epoch": 0.07728,
      "grad_norm": 1.0707742849762627,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7728
    },
    {
      "epoch": 0.07729,
      "grad_norm": 1.105072577657322,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 7729
    },
    {
      "epoch": 0.0773,
      "grad_norm": 0.8048148723787254,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 7730
    },
    {
      "epoch": 0.07731,
      "grad_norm": 0.804004108549886,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 7731
    },
    {
      "epoch": 0.07732,
      "grad_norm": 0.9865333094479842,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 7732
    },
    {
      "epoch": 0.07733,
      "grad_norm": 1.187532994008932,
      "learning_rate": 0.003,
      "loss": 4.1274,
      "step": 7733
    },
    {
      "epoch": 0.07734,
      "grad_norm": 0.9116123033967836,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 7734
    },
    {
      "epoch": 0.07735,
      "grad_norm": 0.9647065687240539,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 7735
    },
    {
      "epoch": 0.07736,
      "grad_norm": 0.8163482733718864,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 7736
    },
    {
      "epoch": 0.07737,
      "grad_norm": 0.879564638557422,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 7737
    },
    {
      "epoch": 0.07738,
      "grad_norm": 0.922084879137366,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 7738
    },
    {
      "epoch": 0.07739,
      "grad_norm": 1.0445880792762618,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 7739
    },
    {
      "epoch": 0.0774,
      "grad_norm": 1.0375087691471465,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7740
    },
    {
      "epoch": 0.07741,
      "grad_norm": 0.9145323840861375,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 7741
    },
    {
      "epoch": 0.07742,
      "grad_norm": 0.8266492697370303,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 7742
    },
    {
      "epoch": 0.07743,
      "grad_norm": 0.9110104456277404,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 7743
    },
    {
      "epoch": 0.07744,
      "grad_norm": 1.0371705555718609,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7744
    },
    {
      "epoch": 0.07745,
      "grad_norm": 0.969698982306619,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 7745
    },
    {
      "epoch": 0.07746,
      "grad_norm": 1.0255113318693947,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7746
    },
    {
      "epoch": 0.07747,
      "grad_norm": 0.8882628795779053,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 7747
    },
    {
      "epoch": 0.07748,
      "grad_norm": 0.7973386036195474,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 7748
    },
    {
      "epoch": 0.07749,
      "grad_norm": 0.8652226027466258,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 7749
    },
    {
      "epoch": 0.0775,
      "grad_norm": 0.870376274432496,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 7750
    },
    {
      "epoch": 0.07751,
      "grad_norm": 0.9893903819503679,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 7751
    },
    {
      "epoch": 0.07752,
      "grad_norm": 1.3482906255390648,
      "learning_rate": 0.003,
      "loss": 4.1296,
      "step": 7752
    },
    {
      "epoch": 0.07753,
      "grad_norm": 0.8068929176849469,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 7753
    },
    {
      "epoch": 0.07754,
      "grad_norm": 0.8014680097244021,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 7754
    },
    {
      "epoch": 0.07755,
      "grad_norm": 0.8754909218158006,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 7755
    },
    {
      "epoch": 0.07756,
      "grad_norm": 0.9676792258616127,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 7756
    },
    {
      "epoch": 0.07757,
      "grad_norm": 1.0045119040336394,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 7757
    },
    {
      "epoch": 0.07758,
      "grad_norm": 1.111589755893894,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 7758
    },
    {
      "epoch": 0.07759,
      "grad_norm": 0.8968827266122236,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7759
    },
    {
      "epoch": 0.0776,
      "grad_norm": 0.9286993706118074,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 7760
    },
    {
      "epoch": 0.07761,
      "grad_norm": 1.048945024253054,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 7761
    },
    {
      "epoch": 0.07762,
      "grad_norm": 0.9319693389175623,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 7762
    },
    {
      "epoch": 0.07763,
      "grad_norm": 0.813291989320193,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 7763
    },
    {
      "epoch": 0.07764,
      "grad_norm": 0.784853226242816,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 7764
    },
    {
      "epoch": 0.07765,
      "grad_norm": 0.7032656157830414,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 7765
    },
    {
      "epoch": 0.07766,
      "grad_norm": 0.6646150699887404,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 7766
    },
    {
      "epoch": 0.07767,
      "grad_norm": 0.7353992656309892,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 7767
    },
    {
      "epoch": 0.07768,
      "grad_norm": 0.880810970396917,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 7768
    },
    {
      "epoch": 0.07769,
      "grad_norm": 0.9578602477971352,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 7769
    },
    {
      "epoch": 0.0777,
      "grad_norm": 1.1156377007546823,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 7770
    },
    {
      "epoch": 0.07771,
      "grad_norm": 0.8301209539531904,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 7771
    },
    {
      "epoch": 0.07772,
      "grad_norm": 0.7560818213800261,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 7772
    },
    {
      "epoch": 0.07773,
      "grad_norm": 0.8404921927294836,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 7773
    },
    {
      "epoch": 0.07774,
      "grad_norm": 0.8782760933360856,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 7774
    },
    {
      "epoch": 0.07775,
      "grad_norm": 0.7808631321910917,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 7775
    },
    {
      "epoch": 0.07776,
      "grad_norm": 0.7541917305536269,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 7776
    },
    {
      "epoch": 0.07777,
      "grad_norm": 0.8093912075646889,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 7777
    },
    {
      "epoch": 0.07778,
      "grad_norm": 0.9732383115498489,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 7778
    },
    {
      "epoch": 0.07779,
      "grad_norm": 0.9921828686435009,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 7779
    },
    {
      "epoch": 0.0778,
      "grad_norm": 0.9497774890620421,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 7780
    },
    {
      "epoch": 0.07781,
      "grad_norm": 0.9033005759467783,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 7781
    },
    {
      "epoch": 0.07782,
      "grad_norm": 1.0370862427957597,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 7782
    },
    {
      "epoch": 0.07783,
      "grad_norm": 1.30997993322532,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7783
    },
    {
      "epoch": 0.07784,
      "grad_norm": 0.9418736206229247,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 7784
    },
    {
      "epoch": 0.07785,
      "grad_norm": 1.0195051094835668,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 7785
    },
    {
      "epoch": 0.07786,
      "grad_norm": 1.0159016886924566,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 7786
    },
    {
      "epoch": 0.07787,
      "grad_norm": 1.0675515986885633,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 7787
    },
    {
      "epoch": 0.07788,
      "grad_norm": 1.2557775330748935,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 7788
    },
    {
      "epoch": 0.07789,
      "grad_norm": 1.081778503928991,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7789
    },
    {
      "epoch": 0.0779,
      "grad_norm": 0.8537282011122178,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 7790
    },
    {
      "epoch": 0.07791,
      "grad_norm": 0.7874541429569311,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 7791
    },
    {
      "epoch": 0.07792,
      "grad_norm": 0.7999390691767071,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 7792
    },
    {
      "epoch": 0.07793,
      "grad_norm": 0.7277192795700171,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 7793
    },
    {
      "epoch": 0.07794,
      "grad_norm": 0.9514831979826355,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 7794
    },
    {
      "epoch": 0.07795,
      "grad_norm": 1.012601597640212,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 7795
    },
    {
      "epoch": 0.07796,
      "grad_norm": 0.9201926222912106,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 7796
    },
    {
      "epoch": 0.07797,
      "grad_norm": 0.7756213744843218,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 7797
    },
    {
      "epoch": 0.07798,
      "grad_norm": 0.8500195316443565,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 7798
    },
    {
      "epoch": 0.07799,
      "grad_norm": 0.7663934706772912,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 7799
    },
    {
      "epoch": 0.078,
      "grad_norm": 0.6950472452800454,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 7800
    },
    {
      "epoch": 0.07801,
      "grad_norm": 0.7781267723317771,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 7801
    },
    {
      "epoch": 0.07802,
      "grad_norm": 1.0163302306750792,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 7802
    },
    {
      "epoch": 0.07803,
      "grad_norm": 1.280595782619173,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 7803
    },
    {
      "epoch": 0.07804,
      "grad_norm": 0.8778051384904699,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 7804
    },
    {
      "epoch": 0.07805,
      "grad_norm": 0.899483151510845,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 7805
    },
    {
      "epoch": 0.07806,
      "grad_norm": 0.9396336994589124,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 7806
    },
    {
      "epoch": 0.07807,
      "grad_norm": 0.8665558808389224,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 7807
    },
    {
      "epoch": 0.07808,
      "grad_norm": 0.7981525957939095,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 7808
    },
    {
      "epoch": 0.07809,
      "grad_norm": 0.8751564912698292,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 7809
    },
    {
      "epoch": 0.0781,
      "grad_norm": 0.8166553655034294,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 7810
    },
    {
      "epoch": 0.07811,
      "grad_norm": 0.767979955899306,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 7811
    },
    {
      "epoch": 0.07812,
      "grad_norm": 0.7617431224556912,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 7812
    },
    {
      "epoch": 0.07813,
      "grad_norm": 0.7967942604337378,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 7813
    },
    {
      "epoch": 0.07814,
      "grad_norm": 0.9942681164347447,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 7814
    },
    {
      "epoch": 0.07815,
      "grad_norm": 1.1892095427285347,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 7815
    },
    {
      "epoch": 0.07816,
      "grad_norm": 0.876659865594819,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 7816
    },
    {
      "epoch": 0.07817,
      "grad_norm": 0.9631730515288163,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 7817
    },
    {
      "epoch": 0.07818,
      "grad_norm": 1.0715009376405462,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 7818
    },
    {
      "epoch": 0.07819,
      "grad_norm": 1.0520680251193615,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 7819
    },
    {
      "epoch": 0.0782,
      "grad_norm": 0.9439418798912824,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 7820
    },
    {
      "epoch": 0.07821,
      "grad_norm": 0.9069113626074072,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 7821
    },
    {
      "epoch": 0.07822,
      "grad_norm": 1.0012938180643256,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 7822
    },
    {
      "epoch": 0.07823,
      "grad_norm": 1.0048698890522239,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7823
    },
    {
      "epoch": 0.07824,
      "grad_norm": 0.9561751321839956,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 7824
    },
    {
      "epoch": 0.07825,
      "grad_norm": 0.9957350057887719,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 7825
    },
    {
      "epoch": 0.07826,
      "grad_norm": 1.079540426834375,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 7826
    },
    {
      "epoch": 0.07827,
      "grad_norm": 0.9385499777483688,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 7827
    },
    {
      "epoch": 0.07828,
      "grad_norm": 1.0935244049947113,
      "learning_rate": 0.003,
      "loss": 4.1207,
      "step": 7828
    },
    {
      "epoch": 0.07829,
      "grad_norm": 0.9621888212658791,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7829
    },
    {
      "epoch": 0.0783,
      "grad_norm": 1.0488809493534366,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 7830
    },
    {
      "epoch": 0.07831,
      "grad_norm": 1.1546610786818092,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 7831
    },
    {
      "epoch": 0.07832,
      "grad_norm": 0.9707818700073304,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7832
    },
    {
      "epoch": 0.07833,
      "grad_norm": 1.0395346936823906,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 7833
    },
    {
      "epoch": 0.07834,
      "grad_norm": 1.0441939813531944,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 7834
    },
    {
      "epoch": 0.07835,
      "grad_norm": 0.9142633227917905,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 7835
    },
    {
      "epoch": 0.07836,
      "grad_norm": 0.7722831477807496,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 7836
    },
    {
      "epoch": 0.07837,
      "grad_norm": 0.7700267869758111,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 7837
    },
    {
      "epoch": 0.07838,
      "grad_norm": 0.8265142746844435,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 7838
    },
    {
      "epoch": 0.07839,
      "grad_norm": 0.8648999060225745,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 7839
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.8130848977918211,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 7840
    },
    {
      "epoch": 0.07841,
      "grad_norm": 0.834862433110529,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 7841
    },
    {
      "epoch": 0.07842,
      "grad_norm": 0.721640102949708,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 7842
    },
    {
      "epoch": 0.07843,
      "grad_norm": 0.6116456054704642,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 7843
    },
    {
      "epoch": 0.07844,
      "grad_norm": 0.6234499687641849,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 7844
    },
    {
      "epoch": 0.07845,
      "grad_norm": 0.638928055891956,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 7845
    },
    {
      "epoch": 0.07846,
      "grad_norm": 0.7063602100511459,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 7846
    },
    {
      "epoch": 0.07847,
      "grad_norm": 0.7799338866735226,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 7847
    },
    {
      "epoch": 0.07848,
      "grad_norm": 0.898204534805398,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 7848
    },
    {
      "epoch": 0.07849,
      "grad_norm": 1.012005052883131,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 7849
    },
    {
      "epoch": 0.0785,
      "grad_norm": 1.0529431619025262,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 7850
    },
    {
      "epoch": 0.07851,
      "grad_norm": 0.9650822583370998,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 7851
    },
    {
      "epoch": 0.07852,
      "grad_norm": 1.0400516186857025,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 7852
    },
    {
      "epoch": 0.07853,
      "grad_norm": 0.9710595472244659,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 7853
    },
    {
      "epoch": 0.07854,
      "grad_norm": 1.0238516024921551,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 7854
    },
    {
      "epoch": 0.07855,
      "grad_norm": 1.09812820718059,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 7855
    },
    {
      "epoch": 0.07856,
      "grad_norm": 0.9164501284041578,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 7856
    },
    {
      "epoch": 0.07857,
      "grad_norm": 1.0913391162498283,
      "learning_rate": 0.003,
      "loss": 4.1262,
      "step": 7857
    },
    {
      "epoch": 0.07858,
      "grad_norm": 1.0601655274183024,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 7858
    },
    {
      "epoch": 0.07859,
      "grad_norm": 0.9500195926062109,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 7859
    },
    {
      "epoch": 0.0786,
      "grad_norm": 0.9248820458548533,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 7860
    },
    {
      "epoch": 0.07861,
      "grad_norm": 1.0564365177503385,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 7861
    },
    {
      "epoch": 0.07862,
      "grad_norm": 1.0399061246279695,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 7862
    },
    {
      "epoch": 0.07863,
      "grad_norm": 0.9697750588673806,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 7863
    },
    {
      "epoch": 0.07864,
      "grad_norm": 1.0453089078095656,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 7864
    },
    {
      "epoch": 0.07865,
      "grad_norm": 1.009583329358567,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 7865
    },
    {
      "epoch": 0.07866,
      "grad_norm": 1.2318259015249517,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 7866
    },
    {
      "epoch": 0.07867,
      "grad_norm": 0.8590130093094929,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7867
    },
    {
      "epoch": 0.07868,
      "grad_norm": 0.9954328336568087,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 7868
    },
    {
      "epoch": 0.07869,
      "grad_norm": 1.1347719980582816,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 7869
    },
    {
      "epoch": 0.0787,
      "grad_norm": 0.9909944155141596,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 7870
    },
    {
      "epoch": 0.07871,
      "grad_norm": 1.027533472785972,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 7871
    },
    {
      "epoch": 0.07872,
      "grad_norm": 0.9509538791389344,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 7872
    },
    {
      "epoch": 0.07873,
      "grad_norm": 0.9743948500738193,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 7873
    },
    {
      "epoch": 0.07874,
      "grad_norm": 1.076018237392653,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 7874
    },
    {
      "epoch": 0.07875,
      "grad_norm": 0.991291863842112,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 7875
    },
    {
      "epoch": 0.07876,
      "grad_norm": 1.111246410563993,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 7876
    },
    {
      "epoch": 0.07877,
      "grad_norm": 0.9900378915913184,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 7877
    },
    {
      "epoch": 0.07878,
      "grad_norm": 0.8784205316420091,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 7878
    },
    {
      "epoch": 0.07879,
      "grad_norm": 0.9417046644892657,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 7879
    },
    {
      "epoch": 0.0788,
      "grad_norm": 1.037927767240949,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7880
    },
    {
      "epoch": 0.07881,
      "grad_norm": 1.2229688137388648,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7881
    },
    {
      "epoch": 0.07882,
      "grad_norm": 0.8555608429927865,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 7882
    },
    {
      "epoch": 0.07883,
      "grad_norm": 0.8611709341175292,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 7883
    },
    {
      "epoch": 0.07884,
      "grad_norm": 0.77157174436383,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 7884
    },
    {
      "epoch": 0.07885,
      "grad_norm": 0.8455320691314467,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 7885
    },
    {
      "epoch": 0.07886,
      "grad_norm": 0.9014648447792224,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 7886
    },
    {
      "epoch": 0.07887,
      "grad_norm": 0.9366329641994356,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 7887
    },
    {
      "epoch": 0.07888,
      "grad_norm": 0.8053760771673537,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7888
    },
    {
      "epoch": 0.07889,
      "grad_norm": 0.7828342950771663,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 7889
    },
    {
      "epoch": 0.0789,
      "grad_norm": 0.6756189697361212,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 7890
    },
    {
      "epoch": 0.07891,
      "grad_norm": 0.7330290536805933,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 7891
    },
    {
      "epoch": 0.07892,
      "grad_norm": 0.7188111647939824,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 7892
    },
    {
      "epoch": 0.07893,
      "grad_norm": 0.7586433320705798,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 7893
    },
    {
      "epoch": 0.07894,
      "grad_norm": 0.9966525848375563,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 7894
    },
    {
      "epoch": 0.07895,
      "grad_norm": 1.3442975582134642,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 7895
    },
    {
      "epoch": 0.07896,
      "grad_norm": 0.729365353256303,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 7896
    },
    {
      "epoch": 0.07897,
      "grad_norm": 0.6629638171890595,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 7897
    },
    {
      "epoch": 0.07898,
      "grad_norm": 0.781346999366163,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 7898
    },
    {
      "epoch": 0.07899,
      "grad_norm": 0.793772205969321,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 7899
    },
    {
      "epoch": 0.079,
      "grad_norm": 0.9389063494976632,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 7900
    },
    {
      "epoch": 0.07901,
      "grad_norm": 1.1343911954848018,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 7901
    },
    {
      "epoch": 0.07902,
      "grad_norm": 0.872291018671974,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 7902
    },
    {
      "epoch": 0.07903,
      "grad_norm": 0.8012280475594034,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 7903
    },
    {
      "epoch": 0.07904,
      "grad_norm": 0.8332934673059036,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 7904
    },
    {
      "epoch": 0.07905,
      "grad_norm": 0.9334487509044569,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 7905
    },
    {
      "epoch": 0.07906,
      "grad_norm": 0.9873155951012756,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 7906
    },
    {
      "epoch": 0.07907,
      "grad_norm": 0.9584124647739157,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7907
    },
    {
      "epoch": 0.07908,
      "grad_norm": 0.883721552574731,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 7908
    },
    {
      "epoch": 0.07909,
      "grad_norm": 0.870325820513853,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 7909
    },
    {
      "epoch": 0.0791,
      "grad_norm": 0.850686094030066,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 7910
    },
    {
      "epoch": 0.07911,
      "grad_norm": 0.9894998327802038,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 7911
    },
    {
      "epoch": 0.07912,
      "grad_norm": 1.0699204166192853,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 7912
    },
    {
      "epoch": 0.07913,
      "grad_norm": 1.1014338916357802,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 7913
    },
    {
      "epoch": 0.07914,
      "grad_norm": 0.8829338047891474,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 7914
    },
    {
      "epoch": 0.07915,
      "grad_norm": 0.8054943769351637,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 7915
    },
    {
      "epoch": 0.07916,
      "grad_norm": 0.8730309155139634,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 7916
    },
    {
      "epoch": 0.07917,
      "grad_norm": 0.8263944569948347,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 7917
    },
    {
      "epoch": 0.07918,
      "grad_norm": 0.8908263616181593,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7918
    },
    {
      "epoch": 0.07919,
      "grad_norm": 1.199853472611651,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 7919
    },
    {
      "epoch": 0.0792,
      "grad_norm": 1.0475806506501837,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 7920
    },
    {
      "epoch": 0.07921,
      "grad_norm": 1.0110599460498464,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 7921
    },
    {
      "epoch": 0.07922,
      "grad_norm": 0.9433898207574585,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 7922
    },
    {
      "epoch": 0.07923,
      "grad_norm": 0.8348723789778444,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 7923
    },
    {
      "epoch": 0.07924,
      "grad_norm": 0.833140296619527,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 7924
    },
    {
      "epoch": 0.07925,
      "grad_norm": 0.9794878682061134,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 7925
    },
    {
      "epoch": 0.07926,
      "grad_norm": 1.2612535875768331,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 7926
    },
    {
      "epoch": 0.07927,
      "grad_norm": 0.7405583869267955,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 7927
    },
    {
      "epoch": 0.07928,
      "grad_norm": 0.651164411585338,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 7928
    },
    {
      "epoch": 0.07929,
      "grad_norm": 0.7197387026951962,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 7929
    },
    {
      "epoch": 0.0793,
      "grad_norm": 0.8263085239318726,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 7930
    },
    {
      "epoch": 0.07931,
      "grad_norm": 0.834126319233174,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 7931
    },
    {
      "epoch": 0.07932,
      "grad_norm": 0.7483212923735366,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 7932
    },
    {
      "epoch": 0.07933,
      "grad_norm": 0.8006850146197332,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 7933
    },
    {
      "epoch": 0.07934,
      "grad_norm": 0.8874417895778808,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 7934
    },
    {
      "epoch": 0.07935,
      "grad_norm": 0.9335481424769582,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 7935
    },
    {
      "epoch": 0.07936,
      "grad_norm": 0.9488902483540785,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 7936
    },
    {
      "epoch": 0.07937,
      "grad_norm": 0.9210487807966519,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 7937
    },
    {
      "epoch": 0.07938,
      "grad_norm": 0.8731093404086347,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 7938
    },
    {
      "epoch": 0.07939,
      "grad_norm": 0.8216594478774197,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 7939
    },
    {
      "epoch": 0.0794,
      "grad_norm": 0.7735027055172842,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 7940
    },
    {
      "epoch": 0.07941,
      "grad_norm": 0.9167989593985493,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 7941
    },
    {
      "epoch": 0.07942,
      "grad_norm": 1.0793942883220813,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 7942
    },
    {
      "epoch": 0.07943,
      "grad_norm": 1.2258293772455378,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 7943
    },
    {
      "epoch": 0.07944,
      "grad_norm": 0.8299843295889483,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 7944
    },
    {
      "epoch": 0.07945,
      "grad_norm": 0.8715797526894794,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 7945
    },
    {
      "epoch": 0.07946,
      "grad_norm": 1.0070541220348046,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 7946
    },
    {
      "epoch": 0.07947,
      "grad_norm": 1.0559591654746912,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 7947
    },
    {
      "epoch": 0.07948,
      "grad_norm": 1.1246304232141864,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 7948
    },
    {
      "epoch": 0.07949,
      "grad_norm": 1.035646997049254,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 7949
    },
    {
      "epoch": 0.0795,
      "grad_norm": 1.1462152538006722,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 7950
    },
    {
      "epoch": 0.07951,
      "grad_norm": 1.045473251959421,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 7951
    },
    {
      "epoch": 0.07952,
      "grad_norm": 0.9603579292671465,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 7952
    },
    {
      "epoch": 0.07953,
      "grad_norm": 0.9846533672672768,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 7953
    },
    {
      "epoch": 0.07954,
      "grad_norm": 1.0311061580889707,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 7954
    },
    {
      "epoch": 0.07955,
      "grad_norm": 1.1129291561662322,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 7955
    },
    {
      "epoch": 0.07956,
      "grad_norm": 1.3157210126198073,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 7956
    },
    {
      "epoch": 0.07957,
      "grad_norm": 0.9656061575557505,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 7957
    },
    {
      "epoch": 0.07958,
      "grad_norm": 1.0363457502099291,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 7958
    },
    {
      "epoch": 0.07959,
      "grad_norm": 0.9519287205545679,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 7959
    },
    {
      "epoch": 0.0796,
      "grad_norm": 0.8874892938612973,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 7960
    },
    {
      "epoch": 0.07961,
      "grad_norm": 0.8499204692701227,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 7961
    },
    {
      "epoch": 0.07962,
      "grad_norm": 0.7782187703657507,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 7962
    },
    {
      "epoch": 0.07963,
      "grad_norm": 0.8258839990795058,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 7963
    },
    {
      "epoch": 0.07964,
      "grad_norm": 0.8106755098862167,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 7964
    },
    {
      "epoch": 0.07965,
      "grad_norm": 0.8950386164564889,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7965
    },
    {
      "epoch": 0.07966,
      "grad_norm": 0.9664103829230193,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 7966
    },
    {
      "epoch": 0.07967,
      "grad_norm": 1.0360852176776754,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 7967
    },
    {
      "epoch": 0.07968,
      "grad_norm": 0.9475801661247285,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 7968
    },
    {
      "epoch": 0.07969,
      "grad_norm": 0.9150409614473582,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 7969
    },
    {
      "epoch": 0.0797,
      "grad_norm": 0.9289409320596667,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 7970
    },
    {
      "epoch": 0.07971,
      "grad_norm": 0.910519318449528,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 7971
    },
    {
      "epoch": 0.07972,
      "grad_norm": 1.1071345490604145,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 7972
    },
    {
      "epoch": 0.07973,
      "grad_norm": 0.9397068669706325,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 7973
    },
    {
      "epoch": 0.07974,
      "grad_norm": 0.9921359001502011,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 7974
    },
    {
      "epoch": 0.07975,
      "grad_norm": 1.001495620220728,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 7975
    },
    {
      "epoch": 0.07976,
      "grad_norm": 0.803218925866123,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 7976
    },
    {
      "epoch": 0.07977,
      "grad_norm": 0.7357551979506947,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 7977
    },
    {
      "epoch": 0.07978,
      "grad_norm": 0.7441367320460851,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 7978
    },
    {
      "epoch": 0.07979,
      "grad_norm": 0.7100854694405272,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 7979
    },
    {
      "epoch": 0.0798,
      "grad_norm": 0.6890360258648266,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 7980
    },
    {
      "epoch": 0.07981,
      "grad_norm": 0.7537645173886749,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 7981
    },
    {
      "epoch": 0.07982,
      "grad_norm": 0.8812941996862754,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 7982
    },
    {
      "epoch": 0.07983,
      "grad_norm": 0.973402696610273,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 7983
    },
    {
      "epoch": 0.07984,
      "grad_norm": 1.1594045747733872,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 7984
    },
    {
      "epoch": 0.07985,
      "grad_norm": 0.9192740280114771,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 7985
    },
    {
      "epoch": 0.07986,
      "grad_norm": 0.9228179720848998,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 7986
    },
    {
      "epoch": 0.07987,
      "grad_norm": 1.003241676572412,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 7987
    },
    {
      "epoch": 0.07988,
      "grad_norm": 1.103347656691755,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 7988
    },
    {
      "epoch": 0.07989,
      "grad_norm": 0.9947380880010859,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 7989
    },
    {
      "epoch": 0.0799,
      "grad_norm": 1.0074317800416395,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 7990
    },
    {
      "epoch": 0.07991,
      "grad_norm": 1.0652610782678547,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 7991
    },
    {
      "epoch": 0.07992,
      "grad_norm": 0.9264232521401602,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 7992
    },
    {
      "epoch": 0.07993,
      "grad_norm": 0.9708215893637503,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 7993
    },
    {
      "epoch": 0.07994,
      "grad_norm": 0.9185573371093531,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 7994
    },
    {
      "epoch": 0.07995,
      "grad_norm": 1.058085876800838,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 7995
    },
    {
      "epoch": 0.07996,
      "grad_norm": 1.085515813637621,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 7996
    },
    {
      "epoch": 0.07997,
      "grad_norm": 1.0465886940521179,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 7997
    },
    {
      "epoch": 0.07998,
      "grad_norm": 0.8881454041255957,
      "learning_rate": 0.003,
      "loss": 4.1295,
      "step": 7998
    },
    {
      "epoch": 0.07999,
      "grad_norm": 0.8240517740099984,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 7999
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.945864278006696,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 8000
    },
    {
      "epoch": 0.08001,
      "grad_norm": 1.0212774414123797,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8001
    },
    {
      "epoch": 0.08002,
      "grad_norm": 1.1204835979046452,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8002
    },
    {
      "epoch": 0.08003,
      "grad_norm": 0.9520420391468847,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 8003
    },
    {
      "epoch": 0.08004,
      "grad_norm": 0.9448651646661593,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8004
    },
    {
      "epoch": 0.08005,
      "grad_norm": 0.9170614863205321,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 8005
    },
    {
      "epoch": 0.08006,
      "grad_norm": 0.8779333843730515,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8006
    },
    {
      "epoch": 0.08007,
      "grad_norm": 0.9266707111712874,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 8007
    },
    {
      "epoch": 0.08008,
      "grad_norm": 0.9901624722228317,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 8008
    },
    {
      "epoch": 0.08009,
      "grad_norm": 0.9651216411589141,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8009
    },
    {
      "epoch": 0.0801,
      "grad_norm": 1.0042760352203992,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8010
    },
    {
      "epoch": 0.08011,
      "grad_norm": 1.1768489281812973,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 8011
    },
    {
      "epoch": 0.08012,
      "grad_norm": 0.969355311788046,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8012
    },
    {
      "epoch": 0.08013,
      "grad_norm": 1.1120606338609458,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8013
    },
    {
      "epoch": 0.08014,
      "grad_norm": 0.9782866986068071,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8014
    },
    {
      "epoch": 0.08015,
      "grad_norm": 1.0758211992496314,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 8015
    },
    {
      "epoch": 0.08016,
      "grad_norm": 0.880271638602275,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 8016
    },
    {
      "epoch": 0.08017,
      "grad_norm": 0.7502263255876486,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 8017
    },
    {
      "epoch": 0.08018,
      "grad_norm": 0.7734348211878546,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8018
    },
    {
      "epoch": 0.08019,
      "grad_norm": 0.7684078035607653,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 8019
    },
    {
      "epoch": 0.0802,
      "grad_norm": 0.7708407552479869,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8020
    },
    {
      "epoch": 0.08021,
      "grad_norm": 0.848673360124088,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 8021
    },
    {
      "epoch": 0.08022,
      "grad_norm": 0.8845796435472464,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8022
    },
    {
      "epoch": 0.08023,
      "grad_norm": 1.016380065852941,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8023
    },
    {
      "epoch": 0.08024,
      "grad_norm": 1.0460153574644655,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 8024
    },
    {
      "epoch": 0.08025,
      "grad_norm": 1.0453001878006873,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 8025
    },
    {
      "epoch": 0.08026,
      "grad_norm": 0.9965656351708417,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8026
    },
    {
      "epoch": 0.08027,
      "grad_norm": 1.0457492474459331,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 8027
    },
    {
      "epoch": 0.08028,
      "grad_norm": 0.8875637028878758,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 8028
    },
    {
      "epoch": 0.08029,
      "grad_norm": 0.8862575547070733,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 8029
    },
    {
      "epoch": 0.0803,
      "grad_norm": 0.8727888553607356,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8030
    },
    {
      "epoch": 0.08031,
      "grad_norm": 0.91179863708708,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 8031
    },
    {
      "epoch": 0.08032,
      "grad_norm": 0.8917368648696833,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8032
    },
    {
      "epoch": 0.08033,
      "grad_norm": 0.9249513133662625,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 8033
    },
    {
      "epoch": 0.08034,
      "grad_norm": 1.056594929957501,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8034
    },
    {
      "epoch": 0.08035,
      "grad_norm": 1.2140600766666199,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 8035
    },
    {
      "epoch": 0.08036,
      "grad_norm": 0.9604814329204158,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 8036
    },
    {
      "epoch": 0.08037,
      "grad_norm": 0.9718783518855493,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8037
    },
    {
      "epoch": 0.08038,
      "grad_norm": 1.0788267825486346,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8038
    },
    {
      "epoch": 0.08039,
      "grad_norm": 1.0426960236669471,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8039
    },
    {
      "epoch": 0.0804,
      "grad_norm": 0.8990767990638859,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 8040
    },
    {
      "epoch": 0.08041,
      "grad_norm": 0.8206146604776955,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 8041
    },
    {
      "epoch": 0.08042,
      "grad_norm": 0.9315698270340985,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8042
    },
    {
      "epoch": 0.08043,
      "grad_norm": 0.9892501547865945,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 8043
    },
    {
      "epoch": 0.08044,
      "grad_norm": 1.0447417437517303,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8044
    },
    {
      "epoch": 0.08045,
      "grad_norm": 1.0149509159686925,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8045
    },
    {
      "epoch": 0.08046,
      "grad_norm": 0.9358159055498079,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8046
    },
    {
      "epoch": 0.08047,
      "grad_norm": 0.8933533196863428,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8047
    },
    {
      "epoch": 0.08048,
      "grad_norm": 0.7664757213485384,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 8048
    },
    {
      "epoch": 0.08049,
      "grad_norm": 0.750556226478064,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8049
    },
    {
      "epoch": 0.0805,
      "grad_norm": 0.8246113468903432,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8050
    },
    {
      "epoch": 0.08051,
      "grad_norm": 0.7466306423459229,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 8051
    },
    {
      "epoch": 0.08052,
      "grad_norm": 0.7461882411581569,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 8052
    },
    {
      "epoch": 0.08053,
      "grad_norm": 0.7594324683486439,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8053
    },
    {
      "epoch": 0.08054,
      "grad_norm": 0.7783815553299155,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 8054
    },
    {
      "epoch": 0.08055,
      "grad_norm": 0.8662143832724243,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 8055
    },
    {
      "epoch": 0.08056,
      "grad_norm": 1.0448541115220396,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8056
    },
    {
      "epoch": 0.08057,
      "grad_norm": 1.1089887487369232,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 8057
    },
    {
      "epoch": 0.08058,
      "grad_norm": 0.7539792902391563,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8058
    },
    {
      "epoch": 0.08059,
      "grad_norm": 0.7389100005311129,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 8059
    },
    {
      "epoch": 0.0806,
      "grad_norm": 0.9262456642006766,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8060
    },
    {
      "epoch": 0.08061,
      "grad_norm": 0.9795001713625814,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 8061
    },
    {
      "epoch": 0.08062,
      "grad_norm": 0.971892968973883,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 8062
    },
    {
      "epoch": 0.08063,
      "grad_norm": 0.9179217197155219,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 8063
    },
    {
      "epoch": 0.08064,
      "grad_norm": 0.8057836726183596,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 8064
    },
    {
      "epoch": 0.08065,
      "grad_norm": 0.7998960995576136,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 8065
    },
    {
      "epoch": 0.08066,
      "grad_norm": 0.8277143695899347,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 8066
    },
    {
      "epoch": 0.08067,
      "grad_norm": 0.8590956136100433,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8067
    },
    {
      "epoch": 0.08068,
      "grad_norm": 0.9394200313521635,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8068
    },
    {
      "epoch": 0.08069,
      "grad_norm": 1.090000328128135,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 8069
    },
    {
      "epoch": 0.0807,
      "grad_norm": 0.995341685424494,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8070
    },
    {
      "epoch": 0.08071,
      "grad_norm": 1.1683102498178706,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 8071
    },
    {
      "epoch": 0.08072,
      "grad_norm": 1.144707212294291,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8072
    },
    {
      "epoch": 0.08073,
      "grad_norm": 1.0813210978226437,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8073
    },
    {
      "epoch": 0.08074,
      "grad_norm": 0.9566230271673927,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 8074
    },
    {
      "epoch": 0.08075,
      "grad_norm": 0.95779239752232,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 8075
    },
    {
      "epoch": 0.08076,
      "grad_norm": 0.9663973763723217,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 8076
    },
    {
      "epoch": 0.08077,
      "grad_norm": 0.887449762689371,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 8077
    },
    {
      "epoch": 0.08078,
      "grad_norm": 0.8084711156241483,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8078
    },
    {
      "epoch": 0.08079,
      "grad_norm": 0.8723111575946362,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 8079
    },
    {
      "epoch": 0.0808,
      "grad_norm": 0.9387061640341692,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8080
    },
    {
      "epoch": 0.08081,
      "grad_norm": 0.8355180753509858,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8081
    },
    {
      "epoch": 0.08082,
      "grad_norm": 0.7557460372416088,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 8082
    },
    {
      "epoch": 0.08083,
      "grad_norm": 0.7660773749158244,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 8083
    },
    {
      "epoch": 0.08084,
      "grad_norm": 0.8306004005127731,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 8084
    },
    {
      "epoch": 0.08085,
      "grad_norm": 0.8877741787140743,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 8085
    },
    {
      "epoch": 0.08086,
      "grad_norm": 0.8576233891044043,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 8086
    },
    {
      "epoch": 0.08087,
      "grad_norm": 0.9238061867576346,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8087
    },
    {
      "epoch": 0.08088,
      "grad_norm": 0.8883115357991481,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 8088
    },
    {
      "epoch": 0.08089,
      "grad_norm": 0.799907814286212,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 8089
    },
    {
      "epoch": 0.0809,
      "grad_norm": 0.8787224459097344,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 8090
    },
    {
      "epoch": 0.08091,
      "grad_norm": 1.1325378931178327,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 8091
    },
    {
      "epoch": 0.08092,
      "grad_norm": 1.1720491273192712,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8092
    },
    {
      "epoch": 0.08093,
      "grad_norm": 1.21086412881328,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 8093
    },
    {
      "epoch": 0.08094,
      "grad_norm": 1.2858697519995719,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 8094
    },
    {
      "epoch": 0.08095,
      "grad_norm": 0.984302984232329,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8095
    },
    {
      "epoch": 0.08096,
      "grad_norm": 0.996396135629695,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 8096
    },
    {
      "epoch": 0.08097,
      "grad_norm": 1.0159214458021701,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8097
    },
    {
      "epoch": 0.08098,
      "grad_norm": 0.8969056697661771,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 8098
    },
    {
      "epoch": 0.08099,
      "grad_norm": 0.8810506298085342,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 8099
    },
    {
      "epoch": 0.081,
      "grad_norm": 0.9078212526479597,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 8100
    },
    {
      "epoch": 0.08101,
      "grad_norm": 0.949131998327021,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 8101
    },
    {
      "epoch": 0.08102,
      "grad_norm": 0.9893750375976563,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8102
    },
    {
      "epoch": 0.08103,
      "grad_norm": 0.9257798083015237,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8103
    },
    {
      "epoch": 0.08104,
      "grad_norm": 0.8487015013060492,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 8104
    },
    {
      "epoch": 0.08105,
      "grad_norm": 0.8526181950103013,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 8105
    },
    {
      "epoch": 0.08106,
      "grad_norm": 1.0786969740224188,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 8106
    },
    {
      "epoch": 0.08107,
      "grad_norm": 0.962314671540165,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8107
    },
    {
      "epoch": 0.08108,
      "grad_norm": 0.9135863883972013,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 8108
    },
    {
      "epoch": 0.08109,
      "grad_norm": 0.9401520709060528,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8109
    },
    {
      "epoch": 0.0811,
      "grad_norm": 0.9325748612282647,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 8110
    },
    {
      "epoch": 0.08111,
      "grad_norm": 0.743009273471666,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 8111
    },
    {
      "epoch": 0.08112,
      "grad_norm": 0.7374812555980911,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 8112
    },
    {
      "epoch": 0.08113,
      "grad_norm": 0.791116362217403,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 8113
    },
    {
      "epoch": 0.08114,
      "grad_norm": 0.8262660445589118,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8114
    },
    {
      "epoch": 0.08115,
      "grad_norm": 0.885927726638974,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8115
    },
    {
      "epoch": 0.08116,
      "grad_norm": 1.0776387999266541,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 8116
    },
    {
      "epoch": 0.08117,
      "grad_norm": 1.0225223453354517,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 8117
    },
    {
      "epoch": 0.08118,
      "grad_norm": 0.9085632221138596,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8118
    },
    {
      "epoch": 0.08119,
      "grad_norm": 0.9197333700250085,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8119
    },
    {
      "epoch": 0.0812,
      "grad_norm": 1.0084531260316627,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 8120
    },
    {
      "epoch": 0.08121,
      "grad_norm": 1.1783087596297812,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 8121
    },
    {
      "epoch": 0.08122,
      "grad_norm": 0.9639565880131747,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8122
    },
    {
      "epoch": 0.08123,
      "grad_norm": 0.9390472599240589,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 8123
    },
    {
      "epoch": 0.08124,
      "grad_norm": 0.8142870814265253,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 8124
    },
    {
      "epoch": 0.08125,
      "grad_norm": 0.788177789413209,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 8125
    },
    {
      "epoch": 0.08126,
      "grad_norm": 0.6912524998631848,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 8126
    },
    {
      "epoch": 0.08127,
      "grad_norm": 0.7226577365051999,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 8127
    },
    {
      "epoch": 0.08128,
      "grad_norm": 0.8021905080797842,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 8128
    },
    {
      "epoch": 0.08129,
      "grad_norm": 0.8332217552376096,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 8129
    },
    {
      "epoch": 0.0813,
      "grad_norm": 0.81623281600983,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 8130
    },
    {
      "epoch": 0.08131,
      "grad_norm": 0.8995463311569188,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8131
    },
    {
      "epoch": 0.08132,
      "grad_norm": 1.2007744161280947,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8132
    },
    {
      "epoch": 0.08133,
      "grad_norm": 0.954576606375035,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8133
    },
    {
      "epoch": 0.08134,
      "grad_norm": 0.9799625763539351,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 8134
    },
    {
      "epoch": 0.08135,
      "grad_norm": 0.9363301593817257,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 8135
    },
    {
      "epoch": 0.08136,
      "grad_norm": 1.0237255724901253,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 8136
    },
    {
      "epoch": 0.08137,
      "grad_norm": 1.2813349124004845,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 8137
    },
    {
      "epoch": 0.08138,
      "grad_norm": 0.7579009503454592,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8138
    },
    {
      "epoch": 0.08139,
      "grad_norm": 0.7301283473485153,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8139
    },
    {
      "epoch": 0.0814,
      "grad_norm": 0.8224340230135018,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8140
    },
    {
      "epoch": 0.08141,
      "grad_norm": 1.0125830693831113,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 8141
    },
    {
      "epoch": 0.08142,
      "grad_norm": 1.0662048042861398,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 8142
    },
    {
      "epoch": 0.08143,
      "grad_norm": 1.0043431337989337,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8143
    },
    {
      "epoch": 0.08144,
      "grad_norm": 1.0395035240004515,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8144
    },
    {
      "epoch": 0.08145,
      "grad_norm": 0.9759886158509355,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8145
    },
    {
      "epoch": 0.08146,
      "grad_norm": 0.9920073789649562,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 8146
    },
    {
      "epoch": 0.08147,
      "grad_norm": 1.180853967281421,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 8147
    },
    {
      "epoch": 0.08148,
      "grad_norm": 0.9563016175288455,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 8148
    },
    {
      "epoch": 0.08149,
      "grad_norm": 0.9375460459019592,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 8149
    },
    {
      "epoch": 0.0815,
      "grad_norm": 1.115020042058792,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 8150
    },
    {
      "epoch": 0.08151,
      "grad_norm": 0.9762709072033995,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 8151
    },
    {
      "epoch": 0.08152,
      "grad_norm": 0.8951996280257692,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 8152
    },
    {
      "epoch": 0.08153,
      "grad_norm": 0.8142401236996037,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8153
    },
    {
      "epoch": 0.08154,
      "grad_norm": 0.9155275103932833,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 8154
    },
    {
      "epoch": 0.08155,
      "grad_norm": 0.9719599421713861,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 8155
    },
    {
      "epoch": 0.08156,
      "grad_norm": 1.0665316378206227,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8156
    },
    {
      "epoch": 0.08157,
      "grad_norm": 1.135966914316917,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 8157
    },
    {
      "epoch": 0.08158,
      "grad_norm": 1.066902527807956,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 8158
    },
    {
      "epoch": 0.08159,
      "grad_norm": 0.9717089917395534,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 8159
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.869490845983841,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8160
    },
    {
      "epoch": 0.08161,
      "grad_norm": 0.6959074303014611,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8161
    },
    {
      "epoch": 0.08162,
      "grad_norm": 0.9986967763832338,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8162
    },
    {
      "epoch": 0.08163,
      "grad_norm": 1.3187472317613815,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 8163
    },
    {
      "epoch": 0.08164,
      "grad_norm": 0.7687805545417059,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8164
    },
    {
      "epoch": 0.08165,
      "grad_norm": 0.7627252335014355,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 8165
    },
    {
      "epoch": 0.08166,
      "grad_norm": 0.9603073324718235,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8166
    },
    {
      "epoch": 0.08167,
      "grad_norm": 1.1172406788641254,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 8167
    },
    {
      "epoch": 0.08168,
      "grad_norm": 0.8451325272847822,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 8168
    },
    {
      "epoch": 0.08169,
      "grad_norm": 0.9763570797625406,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 8169
    },
    {
      "epoch": 0.0817,
      "grad_norm": 1.0675923212188487,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8170
    },
    {
      "epoch": 0.08171,
      "grad_norm": 1.0080405192206545,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8171
    },
    {
      "epoch": 0.08172,
      "grad_norm": 1.0205645004410517,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8172
    },
    {
      "epoch": 0.08173,
      "grad_norm": 1.2927512926089266,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 8173
    },
    {
      "epoch": 0.08174,
      "grad_norm": 0.8968040898130804,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 8174
    },
    {
      "epoch": 0.08175,
      "grad_norm": 0.8905007870405737,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8175
    },
    {
      "epoch": 0.08176,
      "grad_norm": 0.9541562811374488,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 8176
    },
    {
      "epoch": 0.08177,
      "grad_norm": 1.2033461912639642,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 8177
    },
    {
      "epoch": 0.08178,
      "grad_norm": 0.8207723835191075,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8178
    },
    {
      "epoch": 0.08179,
      "grad_norm": 0.7950044143539097,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 8179
    },
    {
      "epoch": 0.0818,
      "grad_norm": 0.8278043595515765,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8180
    },
    {
      "epoch": 0.08181,
      "grad_norm": 0.7898952196220245,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8181
    },
    {
      "epoch": 0.08182,
      "grad_norm": 0.8195849298752939,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 8182
    },
    {
      "epoch": 0.08183,
      "grad_norm": 0.9060159380953525,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8183
    },
    {
      "epoch": 0.08184,
      "grad_norm": 0.9695006051669894,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 8184
    },
    {
      "epoch": 0.08185,
      "grad_norm": 0.920077323996504,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 8185
    },
    {
      "epoch": 0.08186,
      "grad_norm": 0.8684848648180729,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8186
    },
    {
      "epoch": 0.08187,
      "grad_norm": 0.8950480855490158,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 8187
    },
    {
      "epoch": 0.08188,
      "grad_norm": 0.9082754445038497,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 8188
    },
    {
      "epoch": 0.08189,
      "grad_norm": 0.9345206345561519,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 8189
    },
    {
      "epoch": 0.0819,
      "grad_norm": 0.9538918294206803,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8190
    },
    {
      "epoch": 0.08191,
      "grad_norm": 1.1712240807589007,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 8191
    },
    {
      "epoch": 0.08192,
      "grad_norm": 0.8889111819447577,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 8192
    },
    {
      "epoch": 0.08193,
      "grad_norm": 0.9174624290372992,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8193
    },
    {
      "epoch": 0.08194,
      "grad_norm": 0.8802928153982901,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 8194
    },
    {
      "epoch": 0.08195,
      "grad_norm": 0.8173704619648245,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 8195
    },
    {
      "epoch": 0.08196,
      "grad_norm": 0.8797287670114331,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 8196
    },
    {
      "epoch": 0.08197,
      "grad_norm": 0.937319930269984,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8197
    },
    {
      "epoch": 0.08198,
      "grad_norm": 1.1242588728185605,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8198
    },
    {
      "epoch": 0.08199,
      "grad_norm": 0.8700428118547184,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8199
    },
    {
      "epoch": 0.082,
      "grad_norm": 0.8767979055978028,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 8200
    },
    {
      "epoch": 0.08201,
      "grad_norm": 0.8900605291980226,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 8201
    },
    {
      "epoch": 0.08202,
      "grad_norm": 0.9096496752774041,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 8202
    },
    {
      "epoch": 0.08203,
      "grad_norm": 0.9792634921450283,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 8203
    },
    {
      "epoch": 0.08204,
      "grad_norm": 1.0490129381297,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8204
    },
    {
      "epoch": 0.08205,
      "grad_norm": 1.0814614534865807,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8205
    },
    {
      "epoch": 0.08206,
      "grad_norm": 0.9134513944524054,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 8206
    },
    {
      "epoch": 0.08207,
      "grad_norm": 1.00344956911268,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8207
    },
    {
      "epoch": 0.08208,
      "grad_norm": 0.9993535843590942,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 8208
    },
    {
      "epoch": 0.08209,
      "grad_norm": 1.0850834633621271,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 8209
    },
    {
      "epoch": 0.0821,
      "grad_norm": 1.113381435981985,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8210
    },
    {
      "epoch": 0.08211,
      "grad_norm": 1.1005650428474547,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8211
    },
    {
      "epoch": 0.08212,
      "grad_norm": 0.9040714007129831,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 8212
    },
    {
      "epoch": 0.08213,
      "grad_norm": 0.9492708093617471,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 8213
    },
    {
      "epoch": 0.08214,
      "grad_norm": 1.117679266472159,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 8214
    },
    {
      "epoch": 0.08215,
      "grad_norm": 1.0190111526945924,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 8215
    },
    {
      "epoch": 0.08216,
      "grad_norm": 1.0878093466039667,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8216
    },
    {
      "epoch": 0.08217,
      "grad_norm": 0.9383248153244881,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 8217
    },
    {
      "epoch": 0.08218,
      "grad_norm": 0.8917500362995279,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8218
    },
    {
      "epoch": 0.08219,
      "grad_norm": 0.9147124289798197,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 8219
    },
    {
      "epoch": 0.0822,
      "grad_norm": 1.0017630355328986,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8220
    },
    {
      "epoch": 0.08221,
      "grad_norm": 1.1284879557676335,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 8221
    },
    {
      "epoch": 0.08222,
      "grad_norm": 1.003441827304382,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 8222
    },
    {
      "epoch": 0.08223,
      "grad_norm": 0.9619782927365945,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 8223
    },
    {
      "epoch": 0.08224,
      "grad_norm": 0.9756162129814597,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8224
    },
    {
      "epoch": 0.08225,
      "grad_norm": 0.9888357301690992,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 8225
    },
    {
      "epoch": 0.08226,
      "grad_norm": 0.9663381395639372,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 8226
    },
    {
      "epoch": 0.08227,
      "grad_norm": 1.0183404083190035,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 8227
    },
    {
      "epoch": 0.08228,
      "grad_norm": 1.140292961870533,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 8228
    },
    {
      "epoch": 0.08229,
      "grad_norm": 1.0203708219405985,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 8229
    },
    {
      "epoch": 0.0823,
      "grad_norm": 1.0163195425352944,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 8230
    },
    {
      "epoch": 0.08231,
      "grad_norm": 1.1712795476862792,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8231
    },
    {
      "epoch": 0.08232,
      "grad_norm": 1.0333865533848603,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8232
    },
    {
      "epoch": 0.08233,
      "grad_norm": 1.0259646369301396,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 8233
    },
    {
      "epoch": 0.08234,
      "grad_norm": 0.9259284266883602,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 8234
    },
    {
      "epoch": 0.08235,
      "grad_norm": 0.8058762523422451,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 8235
    },
    {
      "epoch": 0.08236,
      "grad_norm": 0.858811923745795,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 8236
    },
    {
      "epoch": 0.08237,
      "grad_norm": 0.9339055212007007,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8237
    },
    {
      "epoch": 0.08238,
      "grad_norm": 0.9168589072673792,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 8238
    },
    {
      "epoch": 0.08239,
      "grad_norm": 0.9459787857366488,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 8239
    },
    {
      "epoch": 0.0824,
      "grad_norm": 1.0167156904516566,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 8240
    },
    {
      "epoch": 0.08241,
      "grad_norm": 1.074340977515434,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 8241
    },
    {
      "epoch": 0.08242,
      "grad_norm": 0.9810666497367793,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 8242
    },
    {
      "epoch": 0.08243,
      "grad_norm": 0.9995000094053831,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 8243
    },
    {
      "epoch": 0.08244,
      "grad_norm": 1.0082704251177956,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8244
    },
    {
      "epoch": 0.08245,
      "grad_norm": 1.245333937722013,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8245
    },
    {
      "epoch": 0.08246,
      "grad_norm": 0.9836272618729974,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8246
    },
    {
      "epoch": 0.08247,
      "grad_norm": 0.9425039973065011,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8247
    },
    {
      "epoch": 0.08248,
      "grad_norm": 0.9823287792469457,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 8248
    },
    {
      "epoch": 0.08249,
      "grad_norm": 0.9588202707053501,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8249
    },
    {
      "epoch": 0.0825,
      "grad_norm": 0.9720158461900901,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8250
    },
    {
      "epoch": 0.08251,
      "grad_norm": 1.0653919600638315,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 8251
    },
    {
      "epoch": 0.08252,
      "grad_norm": 0.9189080871525903,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 8252
    },
    {
      "epoch": 0.08253,
      "grad_norm": 0.9618627197874235,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8253
    },
    {
      "epoch": 0.08254,
      "grad_norm": 0.9241157927248576,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 8254
    },
    {
      "epoch": 0.08255,
      "grad_norm": 1.05514630292859,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 8255
    },
    {
      "epoch": 0.08256,
      "grad_norm": 1.0556785701037765,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 8256
    },
    {
      "epoch": 0.08257,
      "grad_norm": 1.0625228873286099,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8257
    },
    {
      "epoch": 0.08258,
      "grad_norm": 0.8915720985402055,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 8258
    },
    {
      "epoch": 0.08259,
      "grad_norm": 0.7484394189351964,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 8259
    },
    {
      "epoch": 0.0826,
      "grad_norm": 0.7755690062742758,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 8260
    },
    {
      "epoch": 0.08261,
      "grad_norm": 0.7913615993039677,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 8261
    },
    {
      "epoch": 0.08262,
      "grad_norm": 0.7726032835071411,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 8262
    },
    {
      "epoch": 0.08263,
      "grad_norm": 0.8015224842829364,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8263
    },
    {
      "epoch": 0.08264,
      "grad_norm": 0.9833954108334119,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 8264
    },
    {
      "epoch": 0.08265,
      "grad_norm": 1.3245892086877982,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8265
    },
    {
      "epoch": 0.08266,
      "grad_norm": 0.6411106525224666,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 8266
    },
    {
      "epoch": 0.08267,
      "grad_norm": 0.6656579659129879,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8267
    },
    {
      "epoch": 0.08268,
      "grad_norm": 0.7261153634127828,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 8268
    },
    {
      "epoch": 0.08269,
      "grad_norm": 0.8544934625491182,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 8269
    },
    {
      "epoch": 0.0827,
      "grad_norm": 1.0458857129455876,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8270
    },
    {
      "epoch": 0.08271,
      "grad_norm": 0.9911064176984495,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 8271
    },
    {
      "epoch": 0.08272,
      "grad_norm": 0.9393573176422051,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 8272
    },
    {
      "epoch": 0.08273,
      "grad_norm": 0.8646719994209546,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8273
    },
    {
      "epoch": 0.08274,
      "grad_norm": 0.833681481435964,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 8274
    },
    {
      "epoch": 0.08275,
      "grad_norm": 0.8499700263111327,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 8275
    },
    {
      "epoch": 0.08276,
      "grad_norm": 0.8761764628765608,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 8276
    },
    {
      "epoch": 0.08277,
      "grad_norm": 1.0009606164430283,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 8277
    },
    {
      "epoch": 0.08278,
      "grad_norm": 1.1219748051066134,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8278
    },
    {
      "epoch": 0.08279,
      "grad_norm": 1.0353428874496953,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8279
    },
    {
      "epoch": 0.0828,
      "grad_norm": 1.032366749690802,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 8280
    },
    {
      "epoch": 0.08281,
      "grad_norm": 0.88361996817911,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 8281
    },
    {
      "epoch": 0.08282,
      "grad_norm": 0.7805065196095217,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 8282
    },
    {
      "epoch": 0.08283,
      "grad_norm": 0.7078532197092947,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 8283
    },
    {
      "epoch": 0.08284,
      "grad_norm": 0.8334930096391483,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8284
    },
    {
      "epoch": 0.08285,
      "grad_norm": 0.9418443321092326,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8285
    },
    {
      "epoch": 0.08286,
      "grad_norm": 0.9196371932960127,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 8286
    },
    {
      "epoch": 0.08287,
      "grad_norm": 0.7266922490894868,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8287
    },
    {
      "epoch": 0.08288,
      "grad_norm": 0.7577686283368227,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 8288
    },
    {
      "epoch": 0.08289,
      "grad_norm": 0.9147385160263024,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8289
    },
    {
      "epoch": 0.0829,
      "grad_norm": 0.9970454387546689,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 8290
    },
    {
      "epoch": 0.08291,
      "grad_norm": 1.0525426776153293,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 8291
    },
    {
      "epoch": 0.08292,
      "grad_norm": 1.0216758548976381,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 8292
    },
    {
      "epoch": 0.08293,
      "grad_norm": 0.9754705735289286,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 8293
    },
    {
      "epoch": 0.08294,
      "grad_norm": 1.210783738859885,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8294
    },
    {
      "epoch": 0.08295,
      "grad_norm": 1.0394514713837457,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 8295
    },
    {
      "epoch": 0.08296,
      "grad_norm": 1.0556248494716691,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8296
    },
    {
      "epoch": 0.08297,
      "grad_norm": 1.0074376752131393,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8297
    },
    {
      "epoch": 0.08298,
      "grad_norm": 1.0587475783665163,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8298
    },
    {
      "epoch": 0.08299,
      "grad_norm": 1.0609037924375446,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8299
    },
    {
      "epoch": 0.083,
      "grad_norm": 0.9013525969374929,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 8300
    },
    {
      "epoch": 0.08301,
      "grad_norm": 0.8461074908827423,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8301
    },
    {
      "epoch": 0.08302,
      "grad_norm": 0.7808808902520736,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 8302
    },
    {
      "epoch": 0.08303,
      "grad_norm": 0.9240730304280933,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 8303
    },
    {
      "epoch": 0.08304,
      "grad_norm": 0.9410222419739239,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 8304
    },
    {
      "epoch": 0.08305,
      "grad_norm": 0.861091610388214,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8305
    },
    {
      "epoch": 0.08306,
      "grad_norm": 0.8150239293199747,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8306
    },
    {
      "epoch": 0.08307,
      "grad_norm": 0.9976234336488442,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 8307
    },
    {
      "epoch": 0.08308,
      "grad_norm": 1.1560035728708618,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 8308
    },
    {
      "epoch": 0.08309,
      "grad_norm": 1.118929485449606,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 8309
    },
    {
      "epoch": 0.0831,
      "grad_norm": 0.8558868180407445,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8310
    },
    {
      "epoch": 0.08311,
      "grad_norm": 0.7250276766729903,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8311
    },
    {
      "epoch": 0.08312,
      "grad_norm": 0.7096377864215281,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 8312
    },
    {
      "epoch": 0.08313,
      "grad_norm": 0.8317014820764579,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 8313
    },
    {
      "epoch": 0.08314,
      "grad_norm": 0.9061323428480831,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 8314
    },
    {
      "epoch": 0.08315,
      "grad_norm": 0.9783087646693294,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 8315
    },
    {
      "epoch": 0.08316,
      "grad_norm": 1.1250180316240839,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8316
    },
    {
      "epoch": 0.08317,
      "grad_norm": 1.1731269535904751,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 8317
    },
    {
      "epoch": 0.08318,
      "grad_norm": 1.0696733664284113,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8318
    },
    {
      "epoch": 0.08319,
      "grad_norm": 0.9282818531386554,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 8319
    },
    {
      "epoch": 0.0832,
      "grad_norm": 1.108581382048844,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 8320
    },
    {
      "epoch": 0.08321,
      "grad_norm": 0.8810037509971127,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 8321
    },
    {
      "epoch": 0.08322,
      "grad_norm": 0.8426847706169508,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 8322
    },
    {
      "epoch": 0.08323,
      "grad_norm": 0.8922307194607451,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 8323
    },
    {
      "epoch": 0.08324,
      "grad_norm": 1.0048199819448347,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 8324
    },
    {
      "epoch": 0.08325,
      "grad_norm": 1.1134236440962169,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 8325
    },
    {
      "epoch": 0.08326,
      "grad_norm": 0.9733022633639508,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 8326
    },
    {
      "epoch": 0.08327,
      "grad_norm": 1.0054631036308757,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 8327
    },
    {
      "epoch": 0.08328,
      "grad_norm": 1.038948492795079,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8328
    },
    {
      "epoch": 0.08329,
      "grad_norm": 1.0033030320157774,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 8329
    },
    {
      "epoch": 0.0833,
      "grad_norm": 0.9755743730426134,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8330
    },
    {
      "epoch": 0.08331,
      "grad_norm": 1.0383272686924012,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8331
    },
    {
      "epoch": 0.08332,
      "grad_norm": 1.0593791914963704,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 8332
    },
    {
      "epoch": 0.08333,
      "grad_norm": 0.9791667631234542,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 8333
    },
    {
      "epoch": 0.08334,
      "grad_norm": 1.2529702981986344,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 8334
    },
    {
      "epoch": 0.08335,
      "grad_norm": 0.7364862616928067,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8335
    },
    {
      "epoch": 0.08336,
      "grad_norm": 0.7413619580049867,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 8336
    },
    {
      "epoch": 0.08337,
      "grad_norm": 0.7371250991151914,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8337
    },
    {
      "epoch": 0.08338,
      "grad_norm": 0.7373545100133702,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8338
    },
    {
      "epoch": 0.08339,
      "grad_norm": 0.8437067988637226,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 8339
    },
    {
      "epoch": 0.0834,
      "grad_norm": 0.9548820887564872,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 8340
    },
    {
      "epoch": 0.08341,
      "grad_norm": 0.9986280957539666,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 8341
    },
    {
      "epoch": 0.08342,
      "grad_norm": 1.0692833169544227,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 8342
    },
    {
      "epoch": 0.08343,
      "grad_norm": 0.8537320866239007,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 8343
    },
    {
      "epoch": 0.08344,
      "grad_norm": 0.8521570469099049,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 8344
    },
    {
      "epoch": 0.08345,
      "grad_norm": 0.9526963753585949,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 8345
    },
    {
      "epoch": 0.08346,
      "grad_norm": 1.0064844318689625,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 8346
    },
    {
      "epoch": 0.08347,
      "grad_norm": 1.1864005892522749,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 8347
    },
    {
      "epoch": 0.08348,
      "grad_norm": 0.9763480219332881,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 8348
    },
    {
      "epoch": 0.08349,
      "grad_norm": 1.1231533499463706,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8349
    },
    {
      "epoch": 0.0835,
      "grad_norm": 0.992323166083528,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 8350
    },
    {
      "epoch": 0.08351,
      "grad_norm": 1.0428277918008841,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 8351
    },
    {
      "epoch": 0.08352,
      "grad_norm": 1.020903771905812,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 8352
    },
    {
      "epoch": 0.08353,
      "grad_norm": 0.8986777589795104,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 8353
    },
    {
      "epoch": 0.08354,
      "grad_norm": 0.7969639569922113,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 8354
    },
    {
      "epoch": 0.08355,
      "grad_norm": 0.8429266001908198,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8355
    },
    {
      "epoch": 0.08356,
      "grad_norm": 0.8533752836936277,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 8356
    },
    {
      "epoch": 0.08357,
      "grad_norm": 0.8779051945830572,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 8357
    },
    {
      "epoch": 0.08358,
      "grad_norm": 1.1304931543239005,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 8358
    },
    {
      "epoch": 0.08359,
      "grad_norm": 1.1755980604584186,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 8359
    },
    {
      "epoch": 0.0836,
      "grad_norm": 0.8820015012892566,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 8360
    },
    {
      "epoch": 0.08361,
      "grad_norm": 0.7968932521237145,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8361
    },
    {
      "epoch": 0.08362,
      "grad_norm": 0.7705615939438236,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 8362
    },
    {
      "epoch": 0.08363,
      "grad_norm": 0.8336591440869687,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 8363
    },
    {
      "epoch": 0.08364,
      "grad_norm": 0.8681252935444068,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 8364
    },
    {
      "epoch": 0.08365,
      "grad_norm": 0.7925044346326383,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 8365
    },
    {
      "epoch": 0.08366,
      "grad_norm": 0.9118473686485629,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 8366
    },
    {
      "epoch": 0.08367,
      "grad_norm": 0.9863807106347243,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 8367
    },
    {
      "epoch": 0.08368,
      "grad_norm": 1.074863795592151,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 8368
    },
    {
      "epoch": 0.08369,
      "grad_norm": 1.0787611844668858,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 8369
    },
    {
      "epoch": 0.0837,
      "grad_norm": 1.085781743841614,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 8370
    },
    {
      "epoch": 0.08371,
      "grad_norm": 0.9121831664388997,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 8371
    },
    {
      "epoch": 0.08372,
      "grad_norm": 0.9425613813014645,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 8372
    },
    {
      "epoch": 0.08373,
      "grad_norm": 0.8475164269153014,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 8373
    },
    {
      "epoch": 0.08374,
      "grad_norm": 0.8365839586448178,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 8374
    },
    {
      "epoch": 0.08375,
      "grad_norm": 0.9788780251329858,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8375
    },
    {
      "epoch": 0.08376,
      "grad_norm": 1.1289608704104126,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 8376
    },
    {
      "epoch": 0.08377,
      "grad_norm": 0.8654509843602431,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 8377
    },
    {
      "epoch": 0.08378,
      "grad_norm": 0.8972839474588501,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8378
    },
    {
      "epoch": 0.08379,
      "grad_norm": 0.8716796230605194,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8379
    },
    {
      "epoch": 0.0838,
      "grad_norm": 0.8997359385164568,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8380
    },
    {
      "epoch": 0.08381,
      "grad_norm": 1.1955867056419545,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8381
    },
    {
      "epoch": 0.08382,
      "grad_norm": 1.100200909479212,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8382
    },
    {
      "epoch": 0.08383,
      "grad_norm": 1.0062419897932509,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 8383
    },
    {
      "epoch": 0.08384,
      "grad_norm": 1.04090096158985,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 8384
    },
    {
      "epoch": 0.08385,
      "grad_norm": 1.0428807613061897,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 8385
    },
    {
      "epoch": 0.08386,
      "grad_norm": 1.092704813209603,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 8386
    },
    {
      "epoch": 0.08387,
      "grad_norm": 1.0018898539564087,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 8387
    },
    {
      "epoch": 0.08388,
      "grad_norm": 1.0467770477939047,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 8388
    },
    {
      "epoch": 0.08389,
      "grad_norm": 0.9696307153718503,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8389
    },
    {
      "epoch": 0.0839,
      "grad_norm": 0.9834999621136172,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8390
    },
    {
      "epoch": 0.08391,
      "grad_norm": 0.9555752100171525,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 8391
    },
    {
      "epoch": 0.08392,
      "grad_norm": 0.9505299882219872,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 8392
    },
    {
      "epoch": 0.08393,
      "grad_norm": 1.0183911107738324,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 8393
    },
    {
      "epoch": 0.08394,
      "grad_norm": 0.9640779833177097,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 8394
    },
    {
      "epoch": 0.08395,
      "grad_norm": 0.8142550950148915,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 8395
    },
    {
      "epoch": 0.08396,
      "grad_norm": 0.8371291056267166,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8396
    },
    {
      "epoch": 0.08397,
      "grad_norm": 0.9596925071423696,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 8397
    },
    {
      "epoch": 0.08398,
      "grad_norm": 1.1271033499051497,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 8398
    },
    {
      "epoch": 0.08399,
      "grad_norm": 0.9775306799027432,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8399
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.9959146702560995,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 8400
    },
    {
      "epoch": 0.08401,
      "grad_norm": 1.0833441388968759,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8401
    },
    {
      "epoch": 0.08402,
      "grad_norm": 0.8392391540233762,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 8402
    },
    {
      "epoch": 0.08403,
      "grad_norm": 0.7315853340798093,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 8403
    },
    {
      "epoch": 0.08404,
      "grad_norm": 0.8000480577303863,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8404
    },
    {
      "epoch": 0.08405,
      "grad_norm": 0.8795325707481986,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 8405
    },
    {
      "epoch": 0.08406,
      "grad_norm": 0.9618122878448181,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 8406
    },
    {
      "epoch": 0.08407,
      "grad_norm": 1.046916378488563,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 8407
    },
    {
      "epoch": 0.08408,
      "grad_norm": 1.1743216286266962,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 8408
    },
    {
      "epoch": 0.08409,
      "grad_norm": 0.8200986022886277,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 8409
    },
    {
      "epoch": 0.0841,
      "grad_norm": 0.7601736533748236,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 8410
    },
    {
      "epoch": 0.08411,
      "grad_norm": 0.7930919482404809,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 8411
    },
    {
      "epoch": 0.08412,
      "grad_norm": 0.8236918876291852,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 8412
    },
    {
      "epoch": 0.08413,
      "grad_norm": 0.7255830504638703,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 8413
    },
    {
      "epoch": 0.08414,
      "grad_norm": 0.6364226525778399,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 8414
    },
    {
      "epoch": 0.08415,
      "grad_norm": 0.6283569979695853,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 8415
    },
    {
      "epoch": 0.08416,
      "grad_norm": 0.5584551617182978,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8416
    },
    {
      "epoch": 0.08417,
      "grad_norm": 0.6047478150553158,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8417
    },
    {
      "epoch": 0.08418,
      "grad_norm": 0.8178534299204772,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 8418
    },
    {
      "epoch": 0.08419,
      "grad_norm": 1.2445876162198983,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8419
    },
    {
      "epoch": 0.0842,
      "grad_norm": 1.2032672789944496,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 8420
    },
    {
      "epoch": 0.08421,
      "grad_norm": 0.9301781996834642,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 8421
    },
    {
      "epoch": 0.08422,
      "grad_norm": 1.043640415248368,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 8422
    },
    {
      "epoch": 0.08423,
      "grad_norm": 1.177651483822221,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 8423
    },
    {
      "epoch": 0.08424,
      "grad_norm": 0.8367571494130656,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 8424
    },
    {
      "epoch": 0.08425,
      "grad_norm": 0.7649028313402926,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 8425
    },
    {
      "epoch": 0.08426,
      "grad_norm": 0.7796505596514975,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8426
    },
    {
      "epoch": 0.08427,
      "grad_norm": 0.8939607809315887,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 8427
    },
    {
      "epoch": 0.08428,
      "grad_norm": 1.1205858181962776,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8428
    },
    {
      "epoch": 0.08429,
      "grad_norm": 0.9351377901801635,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 8429
    },
    {
      "epoch": 0.0843,
      "grad_norm": 1.1917523797683225,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 8430
    },
    {
      "epoch": 0.08431,
      "grad_norm": 1.0769729281874565,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 8431
    },
    {
      "epoch": 0.08432,
      "grad_norm": 1.0675313880133899,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 8432
    },
    {
      "epoch": 0.08433,
      "grad_norm": 1.0814649771944809,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 8433
    },
    {
      "epoch": 0.08434,
      "grad_norm": 1.1374803833691784,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 8434
    },
    {
      "epoch": 0.08435,
      "grad_norm": 1.042128215570702,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 8435
    },
    {
      "epoch": 0.08436,
      "grad_norm": 1.121784999344552,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 8436
    },
    {
      "epoch": 0.08437,
      "grad_norm": 1.0213150253295122,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 8437
    },
    {
      "epoch": 0.08438,
      "grad_norm": 0.9921329826792672,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8438
    },
    {
      "epoch": 0.08439,
      "grad_norm": 0.9095075680838063,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 8439
    },
    {
      "epoch": 0.0844,
      "grad_norm": 0.9142515498264308,
      "learning_rate": 0.003,
      "loss": 4.1189,
      "step": 8440
    },
    {
      "epoch": 0.08441,
      "grad_norm": 0.9369650500084913,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 8441
    },
    {
      "epoch": 0.08442,
      "grad_norm": 1.0283524745080719,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 8442
    },
    {
      "epoch": 0.08443,
      "grad_norm": 1.3918574878910746,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 8443
    },
    {
      "epoch": 0.08444,
      "grad_norm": 1.0797096461735152,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 8444
    },
    {
      "epoch": 0.08445,
      "grad_norm": 1.0810997939018532,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 8445
    },
    {
      "epoch": 0.08446,
      "grad_norm": 1.0629108011545507,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 8446
    },
    {
      "epoch": 0.08447,
      "grad_norm": 1.1155516321536225,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 8447
    },
    {
      "epoch": 0.08448,
      "grad_norm": 0.8902814958665589,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8448
    },
    {
      "epoch": 0.08449,
      "grad_norm": 0.8921467481709783,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 8449
    },
    {
      "epoch": 0.0845,
      "grad_norm": 0.9566675588218161,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8450
    },
    {
      "epoch": 0.08451,
      "grad_norm": 0.8487439459049356,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8451
    },
    {
      "epoch": 0.08452,
      "grad_norm": 0.8216051671856134,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8452
    },
    {
      "epoch": 0.08453,
      "grad_norm": 0.71083164292099,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 8453
    },
    {
      "epoch": 0.08454,
      "grad_norm": 0.6932399778796303,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8454
    },
    {
      "epoch": 0.08455,
      "grad_norm": 0.6245955308624751,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 8455
    },
    {
      "epoch": 0.08456,
      "grad_norm": 0.710232969209134,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8456
    },
    {
      "epoch": 0.08457,
      "grad_norm": 0.8382923593356233,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 8457
    },
    {
      "epoch": 0.08458,
      "grad_norm": 1.1260446393029737,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 8458
    },
    {
      "epoch": 0.08459,
      "grad_norm": 1.1147365995229561,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 8459
    },
    {
      "epoch": 0.0846,
      "grad_norm": 0.952014546149753,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 8460
    },
    {
      "epoch": 0.08461,
      "grad_norm": 1.0093825873691928,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 8461
    },
    {
      "epoch": 0.08462,
      "grad_norm": 1.082164079498455,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 8462
    },
    {
      "epoch": 0.08463,
      "grad_norm": 0.9339020056620597,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 8463
    },
    {
      "epoch": 0.08464,
      "grad_norm": 1.1038150626697216,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 8464
    },
    {
      "epoch": 0.08465,
      "grad_norm": 0.8668252080144988,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8465
    },
    {
      "epoch": 0.08466,
      "grad_norm": 1.0004324146635026,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 8466
    },
    {
      "epoch": 0.08467,
      "grad_norm": 1.1348520322515445,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 8467
    },
    {
      "epoch": 0.08468,
      "grad_norm": 0.839905944482792,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 8468
    },
    {
      "epoch": 0.08469,
      "grad_norm": 0.8611295736656586,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8469
    },
    {
      "epoch": 0.0847,
      "grad_norm": 1.0042989946600784,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 8470
    },
    {
      "epoch": 0.08471,
      "grad_norm": 1.2058096444243045,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 8471
    },
    {
      "epoch": 0.08472,
      "grad_norm": 0.9217538372683004,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8472
    },
    {
      "epoch": 0.08473,
      "grad_norm": 0.9762750638792925,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 8473
    },
    {
      "epoch": 0.08474,
      "grad_norm": 0.9744705402329853,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 8474
    },
    {
      "epoch": 0.08475,
      "grad_norm": 0.9522311246673508,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 8475
    },
    {
      "epoch": 0.08476,
      "grad_norm": 0.8380797124582146,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 8476
    },
    {
      "epoch": 0.08477,
      "grad_norm": 0.9117904343682559,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 8477
    },
    {
      "epoch": 0.08478,
      "grad_norm": 0.9942329046979015,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8478
    },
    {
      "epoch": 0.08479,
      "grad_norm": 1.1633063589143064,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 8479
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.8464171119669712,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 8480
    },
    {
      "epoch": 0.08481,
      "grad_norm": 0.7888781982408283,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 8481
    },
    {
      "epoch": 0.08482,
      "grad_norm": 0.6832850025599434,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8482
    },
    {
      "epoch": 0.08483,
      "grad_norm": 0.647607987664764,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 8483
    },
    {
      "epoch": 0.08484,
      "grad_norm": 0.8013913578651249,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 8484
    },
    {
      "epoch": 0.08485,
      "grad_norm": 0.9673099824178485,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8485
    },
    {
      "epoch": 0.08486,
      "grad_norm": 1.3943328728885203,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 8486
    },
    {
      "epoch": 0.08487,
      "grad_norm": 0.9092491054040347,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 8487
    },
    {
      "epoch": 0.08488,
      "grad_norm": 0.9362284721268309,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 8488
    },
    {
      "epoch": 0.08489,
      "grad_norm": 1.0548388718733714,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8489
    },
    {
      "epoch": 0.0849,
      "grad_norm": 1.007922010373895,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 8490
    },
    {
      "epoch": 0.08491,
      "grad_norm": 0.950728262769965,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 8491
    },
    {
      "epoch": 0.08492,
      "grad_norm": 0.9646047582212579,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8492
    },
    {
      "epoch": 0.08493,
      "grad_norm": 1.1639240847412275,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 8493
    },
    {
      "epoch": 0.08494,
      "grad_norm": 0.8714062193959657,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8494
    },
    {
      "epoch": 0.08495,
      "grad_norm": 0.8652047802203818,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 8495
    },
    {
      "epoch": 0.08496,
      "grad_norm": 0.76430691704875,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 8496
    },
    {
      "epoch": 0.08497,
      "grad_norm": 0.8622147074710873,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 8497
    },
    {
      "epoch": 0.08498,
      "grad_norm": 1.0688759380024537,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 8498
    },
    {
      "epoch": 0.08499,
      "grad_norm": 1.3345686099220913,
      "learning_rate": 0.003,
      "loss": 4.1308,
      "step": 8499
    },
    {
      "epoch": 0.085,
      "grad_norm": 1.0074196125905852,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 8500
    },
    {
      "epoch": 0.08501,
      "grad_norm": 1.0205831966113184,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8501
    },
    {
      "epoch": 0.08502,
      "grad_norm": 0.8448826348859688,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 8502
    },
    {
      "epoch": 0.08503,
      "grad_norm": 0.8429552443487274,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8503
    },
    {
      "epoch": 0.08504,
      "grad_norm": 0.8110637286450764,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 8504
    },
    {
      "epoch": 0.08505,
      "grad_norm": 0.7803465412418662,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 8505
    },
    {
      "epoch": 0.08506,
      "grad_norm": 0.7741778086269289,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 8506
    },
    {
      "epoch": 0.08507,
      "grad_norm": 0.9284973996173289,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 8507
    },
    {
      "epoch": 0.08508,
      "grad_norm": 1.219393337434796,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8508
    },
    {
      "epoch": 0.08509,
      "grad_norm": 0.9541456980504919,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8509
    },
    {
      "epoch": 0.0851,
      "grad_norm": 1.0650148464854814,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 8510
    },
    {
      "epoch": 0.08511,
      "grad_norm": 1.0213709394078312,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 8511
    },
    {
      "epoch": 0.08512,
      "grad_norm": 1.02423392398546,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 8512
    },
    {
      "epoch": 0.08513,
      "grad_norm": 0.9925200002172757,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8513
    },
    {
      "epoch": 0.08514,
      "grad_norm": 1.0517186955602054,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8514
    },
    {
      "epoch": 0.08515,
      "grad_norm": 1.051372765446143,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8515
    },
    {
      "epoch": 0.08516,
      "grad_norm": 0.961139345826629,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 8516
    },
    {
      "epoch": 0.08517,
      "grad_norm": 1.1383847613698732,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8517
    },
    {
      "epoch": 0.08518,
      "grad_norm": 1.0614310515013552,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8518
    },
    {
      "epoch": 0.08519,
      "grad_norm": 0.9909535453090349,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8519
    },
    {
      "epoch": 0.0852,
      "grad_norm": 0.9606661139384731,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 8520
    },
    {
      "epoch": 0.08521,
      "grad_norm": 1.050436970571876,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 8521
    },
    {
      "epoch": 0.08522,
      "grad_norm": 0.9610076466521358,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 8522
    },
    {
      "epoch": 0.08523,
      "grad_norm": 0.8824080000960122,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8523
    },
    {
      "epoch": 0.08524,
      "grad_norm": 0.829155013045862,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8524
    },
    {
      "epoch": 0.08525,
      "grad_norm": 0.9207597550538581,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 8525
    },
    {
      "epoch": 0.08526,
      "grad_norm": 1.2774950411838009,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 8526
    },
    {
      "epoch": 0.08527,
      "grad_norm": 1.0681627254256,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 8527
    },
    {
      "epoch": 0.08528,
      "grad_norm": 1.1873985636491318,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8528
    },
    {
      "epoch": 0.08529,
      "grad_norm": 0.9005438289637654,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 8529
    },
    {
      "epoch": 0.0853,
      "grad_norm": 0.943871451895167,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8530
    },
    {
      "epoch": 0.08531,
      "grad_norm": 0.9930340272788102,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 8531
    },
    {
      "epoch": 0.08532,
      "grad_norm": 1.0464636343499278,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 8532
    },
    {
      "epoch": 0.08533,
      "grad_norm": 1.1742467639306242,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 8533
    },
    {
      "epoch": 0.08534,
      "grad_norm": 0.9125996800869809,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 8534
    },
    {
      "epoch": 0.08535,
      "grad_norm": 0.9589681901028572,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 8535
    },
    {
      "epoch": 0.08536,
      "grad_norm": 1.003312873568851,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 8536
    },
    {
      "epoch": 0.08537,
      "grad_norm": 1.0579497731198382,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 8537
    },
    {
      "epoch": 0.08538,
      "grad_norm": 1.029713587209702,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 8538
    },
    {
      "epoch": 0.08539,
      "grad_norm": 1.0616355170105047,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 8539
    },
    {
      "epoch": 0.0854,
      "grad_norm": 0.8632828924317381,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 8540
    },
    {
      "epoch": 0.08541,
      "grad_norm": 0.8004540770861811,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 8541
    },
    {
      "epoch": 0.08542,
      "grad_norm": 0.8044230686905137,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 8542
    },
    {
      "epoch": 0.08543,
      "grad_norm": 0.892873539934706,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 8543
    },
    {
      "epoch": 0.08544,
      "grad_norm": 0.961046548554565,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 8544
    },
    {
      "epoch": 0.08545,
      "grad_norm": 1.0248813672026638,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8545
    },
    {
      "epoch": 0.08546,
      "grad_norm": 0.9312362207459531,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 8546
    },
    {
      "epoch": 0.08547,
      "grad_norm": 1.044567535355616,
      "learning_rate": 0.003,
      "loss": 4.113,
      "step": 8547
    },
    {
      "epoch": 0.08548,
      "grad_norm": 1.3226993174635777,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 8548
    },
    {
      "epoch": 0.08549,
      "grad_norm": 1.1046724941315957,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 8549
    },
    {
      "epoch": 0.0855,
      "grad_norm": 1.0576792234018175,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 8550
    },
    {
      "epoch": 0.08551,
      "grad_norm": 0.9612912995427461,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 8551
    },
    {
      "epoch": 0.08552,
      "grad_norm": 1.054325780580547,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 8552
    },
    {
      "epoch": 0.08553,
      "grad_norm": 0.9892609567397835,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 8553
    },
    {
      "epoch": 0.08554,
      "grad_norm": 1.1262764733880528,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 8554
    },
    {
      "epoch": 0.08555,
      "grad_norm": 1.0051709334035899,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 8555
    },
    {
      "epoch": 0.08556,
      "grad_norm": 0.9899672021307735,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 8556
    },
    {
      "epoch": 0.08557,
      "grad_norm": 0.8190451863778596,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 8557
    },
    {
      "epoch": 0.08558,
      "grad_norm": 0.7138030450904894,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 8558
    },
    {
      "epoch": 0.08559,
      "grad_norm": 0.7010492899381877,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 8559
    },
    {
      "epoch": 0.0856,
      "grad_norm": 0.6839596608467561,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 8560
    },
    {
      "epoch": 0.08561,
      "grad_norm": 0.8009807285545935,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 8561
    },
    {
      "epoch": 0.08562,
      "grad_norm": 0.9527168047081275,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 8562
    },
    {
      "epoch": 0.08563,
      "grad_norm": 1.203093505131277,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8563
    },
    {
      "epoch": 0.08564,
      "grad_norm": 0.8506805712070058,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 8564
    },
    {
      "epoch": 0.08565,
      "grad_norm": 0.8867525740638995,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 8565
    },
    {
      "epoch": 0.08566,
      "grad_norm": 0.8312063974139594,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8566
    },
    {
      "epoch": 0.08567,
      "grad_norm": 0.9998824441733825,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 8567
    },
    {
      "epoch": 0.08568,
      "grad_norm": 0.932656507572589,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 8568
    },
    {
      "epoch": 0.08569,
      "grad_norm": 0.9868388758897252,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 8569
    },
    {
      "epoch": 0.0857,
      "grad_norm": 1.2304091841301328,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 8570
    },
    {
      "epoch": 0.08571,
      "grad_norm": 1.0204084145493275,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 8571
    },
    {
      "epoch": 0.08572,
      "grad_norm": 1.034976370606038,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 8572
    },
    {
      "epoch": 0.08573,
      "grad_norm": 1.0524807086959482,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8573
    },
    {
      "epoch": 0.08574,
      "grad_norm": 0.877955924468013,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 8574
    },
    {
      "epoch": 0.08575,
      "grad_norm": 0.8300598690466507,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 8575
    },
    {
      "epoch": 0.08576,
      "grad_norm": 0.9308275145300718,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 8576
    },
    {
      "epoch": 0.08577,
      "grad_norm": 1.0891867015620795,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 8577
    },
    {
      "epoch": 0.08578,
      "grad_norm": 1.1639439297733545,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 8578
    },
    {
      "epoch": 0.08579,
      "grad_norm": 0.8114637979777802,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 8579
    },
    {
      "epoch": 0.0858,
      "grad_norm": 0.8219563908248404,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8580
    },
    {
      "epoch": 0.08581,
      "grad_norm": 0.9514354441568278,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 8581
    },
    {
      "epoch": 0.08582,
      "grad_norm": 1.14862016991355,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 8582
    },
    {
      "epoch": 0.08583,
      "grad_norm": 0.9554904081552709,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 8583
    },
    {
      "epoch": 0.08584,
      "grad_norm": 1.1919961547318443,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 8584
    },
    {
      "epoch": 0.08585,
      "grad_norm": 0.9220063547173044,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 8585
    },
    {
      "epoch": 0.08586,
      "grad_norm": 0.7917210109825877,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 8586
    },
    {
      "epoch": 0.08587,
      "grad_norm": 0.6974880795367134,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 8587
    },
    {
      "epoch": 0.08588,
      "grad_norm": 0.7500850079717578,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 8588
    },
    {
      "epoch": 0.08589,
      "grad_norm": 0.9188442664538239,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 8589
    },
    {
      "epoch": 0.0859,
      "grad_norm": 0.9381614785643198,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8590
    },
    {
      "epoch": 0.08591,
      "grad_norm": 0.8381703502207993,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8591
    },
    {
      "epoch": 0.08592,
      "grad_norm": 0.9957238942437086,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 8592
    },
    {
      "epoch": 0.08593,
      "grad_norm": 1.092680779855448,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 8593
    },
    {
      "epoch": 0.08594,
      "grad_norm": 1.0701608564717937,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 8594
    },
    {
      "epoch": 0.08595,
      "grad_norm": 1.3280866738943609,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8595
    },
    {
      "epoch": 0.08596,
      "grad_norm": 0.8332740836636172,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 8596
    },
    {
      "epoch": 0.08597,
      "grad_norm": 0.8133158050233575,
      "learning_rate": 0.003,
      "loss": 4.108,
      "step": 8597
    },
    {
      "epoch": 0.08598,
      "grad_norm": 0.8677960828642273,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 8598
    },
    {
      "epoch": 0.08599,
      "grad_norm": 0.9453091797543772,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 8599
    },
    {
      "epoch": 0.086,
      "grad_norm": 1.101573205386506,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 8600
    },
    {
      "epoch": 0.08601,
      "grad_norm": 0.917760372367751,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8601
    },
    {
      "epoch": 0.08602,
      "grad_norm": 0.9154001124148273,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 8602
    },
    {
      "epoch": 0.08603,
      "grad_norm": 0.9398969707343389,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 8603
    },
    {
      "epoch": 0.08604,
      "grad_norm": 0.8825392522698661,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8604
    },
    {
      "epoch": 0.08605,
      "grad_norm": 0.8845588681011858,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 8605
    },
    {
      "epoch": 0.08606,
      "grad_norm": 0.8767686307766559,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 8606
    },
    {
      "epoch": 0.08607,
      "grad_norm": 0.9979908090017543,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 8607
    },
    {
      "epoch": 0.08608,
      "grad_norm": 1.167260745143609,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 8608
    },
    {
      "epoch": 0.08609,
      "grad_norm": 1.0047803494363299,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 8609
    },
    {
      "epoch": 0.0861,
      "grad_norm": 1.1790950381285876,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 8610
    },
    {
      "epoch": 0.08611,
      "grad_norm": 0.9155454006163548,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 8611
    },
    {
      "epoch": 0.08612,
      "grad_norm": 1.1200626341917086,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 8612
    },
    {
      "epoch": 0.08613,
      "grad_norm": 1.0417155737120185,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 8613
    },
    {
      "epoch": 0.08614,
      "grad_norm": 0.99622616457739,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8614
    },
    {
      "epoch": 0.08615,
      "grad_norm": 1.1192206628076375,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 8615
    },
    {
      "epoch": 0.08616,
      "grad_norm": 1.2384683745781875,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 8616
    },
    {
      "epoch": 0.08617,
      "grad_norm": 1.0970291121489821,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8617
    },
    {
      "epoch": 0.08618,
      "grad_norm": 1.1560410206984835,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 8618
    },
    {
      "epoch": 0.08619,
      "grad_norm": 0.8941331840757659,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 8619
    },
    {
      "epoch": 0.0862,
      "grad_norm": 0.8951808980836901,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8620
    },
    {
      "epoch": 0.08621,
      "grad_norm": 0.9144905457902658,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 8621
    },
    {
      "epoch": 0.08622,
      "grad_norm": 0.846593765826877,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 8622
    },
    {
      "epoch": 0.08623,
      "grad_norm": 0.8296474727794095,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 8623
    },
    {
      "epoch": 0.08624,
      "grad_norm": 0.857115826300225,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 8624
    },
    {
      "epoch": 0.08625,
      "grad_norm": 0.9770816255055488,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 8625
    },
    {
      "epoch": 0.08626,
      "grad_norm": 1.054544039763614,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 8626
    },
    {
      "epoch": 0.08627,
      "grad_norm": 0.976150638507909,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 8627
    },
    {
      "epoch": 0.08628,
      "grad_norm": 0.9236639095745208,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 8628
    },
    {
      "epoch": 0.08629,
      "grad_norm": 0.8825959695592736,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 8629
    },
    {
      "epoch": 0.0863,
      "grad_norm": 0.8902473144031249,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 8630
    },
    {
      "epoch": 0.08631,
      "grad_norm": 0.9149383790153159,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8631
    },
    {
      "epoch": 0.08632,
      "grad_norm": 0.9182837062894001,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 8632
    },
    {
      "epoch": 0.08633,
      "grad_norm": 1.0899428012063783,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 8633
    },
    {
      "epoch": 0.08634,
      "grad_norm": 1.0629044913328984,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 8634
    },
    {
      "epoch": 0.08635,
      "grad_norm": 0.9717726584129625,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8635
    },
    {
      "epoch": 0.08636,
      "grad_norm": 1.1164187292125398,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8636
    },
    {
      "epoch": 0.08637,
      "grad_norm": 1.3025486362018337,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 8637
    },
    {
      "epoch": 0.08638,
      "grad_norm": 0.8541936401852307,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 8638
    },
    {
      "epoch": 0.08639,
      "grad_norm": 0.771724595769981,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8639
    },
    {
      "epoch": 0.0864,
      "grad_norm": 0.866492675972138,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8640
    },
    {
      "epoch": 0.08641,
      "grad_norm": 1.3350454450812863,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 8641
    },
    {
      "epoch": 0.08642,
      "grad_norm": 0.8473052034583577,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 8642
    },
    {
      "epoch": 0.08643,
      "grad_norm": 0.9423298490396024,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8643
    },
    {
      "epoch": 0.08644,
      "grad_norm": 0.9804863052777438,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 8644
    },
    {
      "epoch": 0.08645,
      "grad_norm": 0.9176586395090423,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8645
    },
    {
      "epoch": 0.08646,
      "grad_norm": 1.051248679692915,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 8646
    },
    {
      "epoch": 0.08647,
      "grad_norm": 1.2067666871909615,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 8647
    },
    {
      "epoch": 0.08648,
      "grad_norm": 0.8541324948901896,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 8648
    },
    {
      "epoch": 0.08649,
      "grad_norm": 0.952863199924152,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 8649
    },
    {
      "epoch": 0.0865,
      "grad_norm": 1.0020719863767489,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 8650
    },
    {
      "epoch": 0.08651,
      "grad_norm": 1.0541904737375236,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8651
    },
    {
      "epoch": 0.08652,
      "grad_norm": 0.965588617649747,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 8652
    },
    {
      "epoch": 0.08653,
      "grad_norm": 1.008009301735876,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 8653
    },
    {
      "epoch": 0.08654,
      "grad_norm": 1.0616740428111644,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 8654
    },
    {
      "epoch": 0.08655,
      "grad_norm": 0.9641551606343974,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 8655
    },
    {
      "epoch": 0.08656,
      "grad_norm": 0.92701904040581,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8656
    },
    {
      "epoch": 0.08657,
      "grad_norm": 0.9477211259009536,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 8657
    },
    {
      "epoch": 0.08658,
      "grad_norm": 0.8613906735799333,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 8658
    },
    {
      "epoch": 0.08659,
      "grad_norm": 0.8582341328171613,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 8659
    },
    {
      "epoch": 0.0866,
      "grad_norm": 0.7554921765534043,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 8660
    },
    {
      "epoch": 0.08661,
      "grad_norm": 0.710138619445291,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 8661
    },
    {
      "epoch": 0.08662,
      "grad_norm": 0.6948787363147834,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 8662
    },
    {
      "epoch": 0.08663,
      "grad_norm": 0.8130674241859551,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 8663
    },
    {
      "epoch": 0.08664,
      "grad_norm": 0.8427386250145507,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8664
    },
    {
      "epoch": 0.08665,
      "grad_norm": 0.8151506821681499,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 8665
    },
    {
      "epoch": 0.08666,
      "grad_norm": 0.9564462744936523,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8666
    },
    {
      "epoch": 0.08667,
      "grad_norm": 1.1118303234126337,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 8667
    },
    {
      "epoch": 0.08668,
      "grad_norm": 0.9710629851216804,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 8668
    },
    {
      "epoch": 0.08669,
      "grad_norm": 1.5737744642483429,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8669
    },
    {
      "epoch": 0.0867,
      "grad_norm": 0.6740528009072599,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 8670
    },
    {
      "epoch": 0.08671,
      "grad_norm": 0.8368283275216355,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 8671
    },
    {
      "epoch": 0.08672,
      "grad_norm": 1.0436902189256585,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8672
    },
    {
      "epoch": 0.08673,
      "grad_norm": 1.0127357970798005,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8673
    },
    {
      "epoch": 0.08674,
      "grad_norm": 0.9615865521203145,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8674
    },
    {
      "epoch": 0.08675,
      "grad_norm": 1.0774654469343408,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 8675
    },
    {
      "epoch": 0.08676,
      "grad_norm": 1.0537181165753355,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 8676
    },
    {
      "epoch": 0.08677,
      "grad_norm": 1.0110231179636788,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 8677
    },
    {
      "epoch": 0.08678,
      "grad_norm": 0.9615912561390582,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 8678
    },
    {
      "epoch": 0.08679,
      "grad_norm": 0.9402198527708937,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8679
    },
    {
      "epoch": 0.0868,
      "grad_norm": 0.905661734228063,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 8680
    },
    {
      "epoch": 0.08681,
      "grad_norm": 1.1569378589826407,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 8681
    },
    {
      "epoch": 0.08682,
      "grad_norm": 0.9978572454734634,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 8682
    },
    {
      "epoch": 0.08683,
      "grad_norm": 0.9820424038569685,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 8683
    },
    {
      "epoch": 0.08684,
      "grad_norm": 0.9217316868751142,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 8684
    },
    {
      "epoch": 0.08685,
      "grad_norm": 1.13020305011687,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 8685
    },
    {
      "epoch": 0.08686,
      "grad_norm": 1.135596127784365,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 8686
    },
    {
      "epoch": 0.08687,
      "grad_norm": 0.8991895009637995,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8687
    },
    {
      "epoch": 0.08688,
      "grad_norm": 0.7950591470094579,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8688
    },
    {
      "epoch": 0.08689,
      "grad_norm": 0.9175036427923986,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 8689
    },
    {
      "epoch": 0.0869,
      "grad_norm": 1.1867185106176557,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 8690
    },
    {
      "epoch": 0.08691,
      "grad_norm": 1.150743996358858,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8691
    },
    {
      "epoch": 0.08692,
      "grad_norm": 1.072456612250304,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8692
    },
    {
      "epoch": 0.08693,
      "grad_norm": 0.9820410634402862,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 8693
    },
    {
      "epoch": 0.08694,
      "grad_norm": 0.996739398321439,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 8694
    },
    {
      "epoch": 0.08695,
      "grad_norm": 1.050712916361131,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8695
    },
    {
      "epoch": 0.08696,
      "grad_norm": 0.9400645560048955,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 8696
    },
    {
      "epoch": 0.08697,
      "grad_norm": 0.9974884973769946,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8697
    },
    {
      "epoch": 0.08698,
      "grad_norm": 1.0051997817192861,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8698
    },
    {
      "epoch": 0.08699,
      "grad_norm": 0.9163255483083583,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 8699
    },
    {
      "epoch": 0.087,
      "grad_norm": 1.006810901255363,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 8700
    },
    {
      "epoch": 0.08701,
      "grad_norm": 1.115045098588956,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 8701
    },
    {
      "epoch": 0.08702,
      "grad_norm": 0.9880723518291648,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 8702
    },
    {
      "epoch": 0.08703,
      "grad_norm": 0.8366949205399988,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 8703
    },
    {
      "epoch": 0.08704,
      "grad_norm": 0.7286032282861155,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 8704
    },
    {
      "epoch": 0.08705,
      "grad_norm": 0.818286090640339,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 8705
    },
    {
      "epoch": 0.08706,
      "grad_norm": 1.0020414264801158,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 8706
    },
    {
      "epoch": 0.08707,
      "grad_norm": 1.1058535805008598,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 8707
    },
    {
      "epoch": 0.08708,
      "grad_norm": 1.0637305918614082,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 8708
    },
    {
      "epoch": 0.08709,
      "grad_norm": 1.1622544435094628,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 8709
    },
    {
      "epoch": 0.0871,
      "grad_norm": 1.1107128106540358,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 8710
    },
    {
      "epoch": 0.08711,
      "grad_norm": 0.9646068373714444,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 8711
    },
    {
      "epoch": 0.08712,
      "grad_norm": 0.9931572732752244,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8712
    },
    {
      "epoch": 0.08713,
      "grad_norm": 0.9908778373110699,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 8713
    },
    {
      "epoch": 0.08714,
      "grad_norm": 1.0197596826627089,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8714
    },
    {
      "epoch": 0.08715,
      "grad_norm": 1.014702738749466,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 8715
    },
    {
      "epoch": 0.08716,
      "grad_norm": 1.0045099245365388,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 8716
    },
    {
      "epoch": 0.08717,
      "grad_norm": 1.0710132715620833,
      "learning_rate": 0.003,
      "loss": 4.1203,
      "step": 8717
    },
    {
      "epoch": 0.08718,
      "grad_norm": 0.9699281274500614,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 8718
    },
    {
      "epoch": 0.08719,
      "grad_norm": 1.012536675784157,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 8719
    },
    {
      "epoch": 0.0872,
      "grad_norm": 0.9846464927862566,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 8720
    },
    {
      "epoch": 0.08721,
      "grad_norm": 1.051748323833279,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 8721
    },
    {
      "epoch": 0.08722,
      "grad_norm": 1.1390897577030044,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 8722
    },
    {
      "epoch": 0.08723,
      "grad_norm": 1.014378976687448,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 8723
    },
    {
      "epoch": 0.08724,
      "grad_norm": 1.0096210285347254,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8724
    },
    {
      "epoch": 0.08725,
      "grad_norm": 0.9399244619167805,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 8725
    },
    {
      "epoch": 0.08726,
      "grad_norm": 0.9568485117772358,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 8726
    },
    {
      "epoch": 0.08727,
      "grad_norm": 0.9385590320694874,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 8727
    },
    {
      "epoch": 0.08728,
      "grad_norm": 0.9049360545768217,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 8728
    },
    {
      "epoch": 0.08729,
      "grad_norm": 0.9019422289302671,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 8729
    },
    {
      "epoch": 0.0873,
      "grad_norm": 0.9330390334308929,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 8730
    },
    {
      "epoch": 0.08731,
      "grad_norm": 0.9809468283330093,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 8731
    },
    {
      "epoch": 0.08732,
      "grad_norm": 0.9010326157614988,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8732
    },
    {
      "epoch": 0.08733,
      "grad_norm": 0.7938742404264579,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 8733
    },
    {
      "epoch": 0.08734,
      "grad_norm": 0.8111233684875911,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 8734
    },
    {
      "epoch": 0.08735,
      "grad_norm": 0.7272221907480916,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 8735
    },
    {
      "epoch": 0.08736,
      "grad_norm": 0.679447738423607,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 8736
    },
    {
      "epoch": 0.08737,
      "grad_norm": 0.8396545033444168,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 8737
    },
    {
      "epoch": 0.08738,
      "grad_norm": 1.1738244903309252,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 8738
    },
    {
      "epoch": 0.08739,
      "grad_norm": 1.0681589571108603,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8739
    },
    {
      "epoch": 0.0874,
      "grad_norm": 1.05002028020519,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8740
    },
    {
      "epoch": 0.08741,
      "grad_norm": 0.9219236197609371,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 8741
    },
    {
      "epoch": 0.08742,
      "grad_norm": 1.000316386094372,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 8742
    },
    {
      "epoch": 0.08743,
      "grad_norm": 0.964186838515666,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 8743
    },
    {
      "epoch": 0.08744,
      "grad_norm": 0.8953102034049782,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 8744
    },
    {
      "epoch": 0.08745,
      "grad_norm": 1.0423780413564045,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8745
    },
    {
      "epoch": 0.08746,
      "grad_norm": 1.1156519078431824,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8746
    },
    {
      "epoch": 0.08747,
      "grad_norm": 0.9289740494359398,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8747
    },
    {
      "epoch": 0.08748,
      "grad_norm": 0.9221501347652119,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8748
    },
    {
      "epoch": 0.08749,
      "grad_norm": 1.137448507112466,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 8749
    },
    {
      "epoch": 0.0875,
      "grad_norm": 1.1606946555207784,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 8750
    },
    {
      "epoch": 0.08751,
      "grad_norm": 1.1418896933212783,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 8751
    },
    {
      "epoch": 0.08752,
      "grad_norm": 0.8892781422608295,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 8752
    },
    {
      "epoch": 0.08753,
      "grad_norm": 0.8235048326716268,
      "learning_rate": 0.003,
      "loss": 4.1165,
      "step": 8753
    },
    {
      "epoch": 0.08754,
      "grad_norm": 0.9178459509732138,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8754
    },
    {
      "epoch": 0.08755,
      "grad_norm": 1.043656892193055,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 8755
    },
    {
      "epoch": 0.08756,
      "grad_norm": 0.9634587521325285,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8756
    },
    {
      "epoch": 0.08757,
      "grad_norm": 1.0095694340504555,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 8757
    },
    {
      "epoch": 0.08758,
      "grad_norm": 1.0702638193359773,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 8758
    },
    {
      "epoch": 0.08759,
      "grad_norm": 1.1871597807200291,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8759
    },
    {
      "epoch": 0.0876,
      "grad_norm": 0.8734990186560198,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8760
    },
    {
      "epoch": 0.08761,
      "grad_norm": 0.9459790552922084,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 8761
    },
    {
      "epoch": 0.08762,
      "grad_norm": 1.170255371175725,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 8762
    },
    {
      "epoch": 0.08763,
      "grad_norm": 0.9807952883818362,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 8763
    },
    {
      "epoch": 0.08764,
      "grad_norm": 1.3716972394122473,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 8764
    },
    {
      "epoch": 0.08765,
      "grad_norm": 0.8981483803103426,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 8765
    },
    {
      "epoch": 0.08766,
      "grad_norm": 0.9178267580818741,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8766
    },
    {
      "epoch": 0.08767,
      "grad_norm": 0.8387751062720594,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 8767
    },
    {
      "epoch": 0.08768,
      "grad_norm": 0.765362163097001,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8768
    },
    {
      "epoch": 0.08769,
      "grad_norm": 0.8128170310066327,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 8769
    },
    {
      "epoch": 0.0877,
      "grad_norm": 0.9858951252865733,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 8770
    },
    {
      "epoch": 0.08771,
      "grad_norm": 1.1445554138270546,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 8771
    },
    {
      "epoch": 0.08772,
      "grad_norm": 0.8621127850353041,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 8772
    },
    {
      "epoch": 0.08773,
      "grad_norm": 0.8525885089903674,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8773
    },
    {
      "epoch": 0.08774,
      "grad_norm": 0.8871272119975309,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8774
    },
    {
      "epoch": 0.08775,
      "grad_norm": 0.991806274254639,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 8775
    },
    {
      "epoch": 0.08776,
      "grad_norm": 1.1533994303031723,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 8776
    },
    {
      "epoch": 0.08777,
      "grad_norm": 1.055322012206207,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 8777
    },
    {
      "epoch": 0.08778,
      "grad_norm": 1.071611255711352,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8778
    },
    {
      "epoch": 0.08779,
      "grad_norm": 0.9203284832312962,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 8779
    },
    {
      "epoch": 0.0878,
      "grad_norm": 0.8661451623284229,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 8780
    },
    {
      "epoch": 0.08781,
      "grad_norm": 0.9695475727064722,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 8781
    },
    {
      "epoch": 0.08782,
      "grad_norm": 0.9855488794840954,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 8782
    },
    {
      "epoch": 0.08783,
      "grad_norm": 1.0854676896438462,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 8783
    },
    {
      "epoch": 0.08784,
      "grad_norm": 0.92019280232576,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 8784
    },
    {
      "epoch": 0.08785,
      "grad_norm": 0.8916386749670293,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 8785
    },
    {
      "epoch": 0.08786,
      "grad_norm": 0.9384291792786592,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 8786
    },
    {
      "epoch": 0.08787,
      "grad_norm": 1.1242759626646945,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 8787
    },
    {
      "epoch": 0.08788,
      "grad_norm": 0.9566646203576128,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 8788
    },
    {
      "epoch": 0.08789,
      "grad_norm": 1.1892424954334855,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8789
    },
    {
      "epoch": 0.0879,
      "grad_norm": 0.9588075240707433,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 8790
    },
    {
      "epoch": 0.08791,
      "grad_norm": 1.06193895017149,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 8791
    },
    {
      "epoch": 0.08792,
      "grad_norm": 1.0293430845691305,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 8792
    },
    {
      "epoch": 0.08793,
      "grad_norm": 0.9914591638234613,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8793
    },
    {
      "epoch": 0.08794,
      "grad_norm": 1.1989613224884543,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 8794
    },
    {
      "epoch": 0.08795,
      "grad_norm": 0.9861795106142962,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 8795
    },
    {
      "epoch": 0.08796,
      "grad_norm": 1.0041571125270279,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8796
    },
    {
      "epoch": 0.08797,
      "grad_norm": 1.0698702519124776,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 8797
    },
    {
      "epoch": 0.08798,
      "grad_norm": 1.1705625141993568,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 8798
    },
    {
      "epoch": 0.08799,
      "grad_norm": 0.7915192269652498,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 8799
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.6604270774270279,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 8800
    },
    {
      "epoch": 0.08801,
      "grad_norm": 0.7719368938463326,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 8801
    },
    {
      "epoch": 0.08802,
      "grad_norm": 0.8416958932367354,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 8802
    },
    {
      "epoch": 0.08803,
      "grad_norm": 1.0164740952605043,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 8803
    },
    {
      "epoch": 0.08804,
      "grad_norm": 1.1090228946685283,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 8804
    },
    {
      "epoch": 0.08805,
      "grad_norm": 0.9168056996303934,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 8805
    },
    {
      "epoch": 0.08806,
      "grad_norm": 0.9155273228148718,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 8806
    },
    {
      "epoch": 0.08807,
      "grad_norm": 0.8379245763047375,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 8807
    },
    {
      "epoch": 0.08808,
      "grad_norm": 0.8670250361401486,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 8808
    },
    {
      "epoch": 0.08809,
      "grad_norm": 1.0376022891758734,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 8809
    },
    {
      "epoch": 0.0881,
      "grad_norm": 1.065934829807962,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 8810
    },
    {
      "epoch": 0.08811,
      "grad_norm": 1.0298973690098348,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 8811
    },
    {
      "epoch": 0.08812,
      "grad_norm": 0.8686967291317386,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 8812
    },
    {
      "epoch": 0.08813,
      "grad_norm": 0.9680376583982786,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 8813
    },
    {
      "epoch": 0.08814,
      "grad_norm": 0.8851029994047319,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 8814
    },
    {
      "epoch": 0.08815,
      "grad_norm": 1.016504733957394,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8815
    },
    {
      "epoch": 0.08816,
      "grad_norm": 1.039564849638083,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 8816
    },
    {
      "epoch": 0.08817,
      "grad_norm": 0.9769732491378709,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 8817
    },
    {
      "epoch": 0.08818,
      "grad_norm": 0.7419152032454192,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 8818
    },
    {
      "epoch": 0.08819,
      "grad_norm": 0.7449265064447682,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 8819
    },
    {
      "epoch": 0.0882,
      "grad_norm": 0.8134643932223908,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 8820
    },
    {
      "epoch": 0.08821,
      "grad_norm": 0.8600355770310976,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 8821
    },
    {
      "epoch": 0.08822,
      "grad_norm": 1.0780006611352508,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 8822
    },
    {
      "epoch": 0.08823,
      "grad_norm": 0.9907258089098394,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 8823
    },
    {
      "epoch": 0.08824,
      "grad_norm": 0.9262970676154467,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 8824
    },
    {
      "epoch": 0.08825,
      "grad_norm": 1.0361846461107758,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 8825
    },
    {
      "epoch": 0.08826,
      "grad_norm": 1.1352830686502071,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 8826
    },
    {
      "epoch": 0.08827,
      "grad_norm": 0.9811751503381044,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 8827
    },
    {
      "epoch": 0.08828,
      "grad_norm": 1.1677797594356207,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 8828
    },
    {
      "epoch": 0.08829,
      "grad_norm": 0.8620817289548668,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 8829
    },
    {
      "epoch": 0.0883,
      "grad_norm": 0.9109610741986333,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8830
    },
    {
      "epoch": 0.08831,
      "grad_norm": 0.9792155464654494,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 8831
    },
    {
      "epoch": 0.08832,
      "grad_norm": 1.1813879043570434,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 8832
    },
    {
      "epoch": 0.08833,
      "grad_norm": 1.0032648409430165,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 8833
    },
    {
      "epoch": 0.08834,
      "grad_norm": 0.9769964414966447,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 8834
    },
    {
      "epoch": 0.08835,
      "grad_norm": 1.082737428043474,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 8835
    },
    {
      "epoch": 0.08836,
      "grad_norm": 0.984170203316877,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 8836
    },
    {
      "epoch": 0.08837,
      "grad_norm": 1.056134980376748,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 8837
    },
    {
      "epoch": 0.08838,
      "grad_norm": 1.211754779715747,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 8838
    },
    {
      "epoch": 0.08839,
      "grad_norm": 0.9712974264137377,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 8839
    },
    {
      "epoch": 0.0884,
      "grad_norm": 1.1142216163765637,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 8840
    },
    {
      "epoch": 0.08841,
      "grad_norm": 1.0046432864292048,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 8841
    },
    {
      "epoch": 0.08842,
      "grad_norm": 1.056661438069542,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8842
    },
    {
      "epoch": 0.08843,
      "grad_norm": 1.2236595574756504,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 8843
    },
    {
      "epoch": 0.08844,
      "grad_norm": 1.03881071415002,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 8844
    },
    {
      "epoch": 0.08845,
      "grad_norm": 1.1459947152275722,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 8845
    },
    {
      "epoch": 0.08846,
      "grad_norm": 1.0840904607136326,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 8846
    },
    {
      "epoch": 0.08847,
      "grad_norm": 0.9355054937743638,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 8847
    },
    {
      "epoch": 0.08848,
      "grad_norm": 0.9640890114169899,
      "learning_rate": 0.003,
      "loss": 4.1185,
      "step": 8848
    },
    {
      "epoch": 0.08849,
      "grad_norm": 1.0483305858052685,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8849
    },
    {
      "epoch": 0.0885,
      "grad_norm": 1.0553477746801,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 8850
    },
    {
      "epoch": 0.08851,
      "grad_norm": 0.880408001509907,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 8851
    },
    {
      "epoch": 0.08852,
      "grad_norm": 0.8729293404205195,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8852
    },
    {
      "epoch": 0.08853,
      "grad_norm": 0.8295344564262668,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 8853
    },
    {
      "epoch": 0.08854,
      "grad_norm": 0.8793240956806683,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 8854
    },
    {
      "epoch": 0.08855,
      "grad_norm": 0.9216847733111198,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 8855
    },
    {
      "epoch": 0.08856,
      "grad_norm": 1.0905136756034943,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 8856
    },
    {
      "epoch": 0.08857,
      "grad_norm": 1.0871331725112339,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 8857
    },
    {
      "epoch": 0.08858,
      "grad_norm": 0.93110512066651,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 8858
    },
    {
      "epoch": 0.08859,
      "grad_norm": 0.9090747425713922,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 8859
    },
    {
      "epoch": 0.0886,
      "grad_norm": 0.9685128136141401,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 8860
    },
    {
      "epoch": 0.08861,
      "grad_norm": 0.9015137576718465,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 8861
    },
    {
      "epoch": 0.08862,
      "grad_norm": 1.0226210784829224,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 8862
    },
    {
      "epoch": 0.08863,
      "grad_norm": 1.2542964356410375,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 8863
    },
    {
      "epoch": 0.08864,
      "grad_norm": 0.763921705860707,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 8864
    },
    {
      "epoch": 0.08865,
      "grad_norm": 0.6665256002350811,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 8865
    },
    {
      "epoch": 0.08866,
      "grad_norm": 0.7496478357626905,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 8866
    },
    {
      "epoch": 0.08867,
      "grad_norm": 0.8460797471569417,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 8867
    },
    {
      "epoch": 0.08868,
      "grad_norm": 1.0223040210373233,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8868
    },
    {
      "epoch": 0.08869,
      "grad_norm": 1.1610482768655175,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 8869
    },
    {
      "epoch": 0.0887,
      "grad_norm": 0.9490101679373257,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 8870
    },
    {
      "epoch": 0.08871,
      "grad_norm": 1.0936328384500282,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 8871
    },
    {
      "epoch": 0.08872,
      "grad_norm": 1.0161161364379168,
      "learning_rate": 0.003,
      "loss": 4.1275,
      "step": 8872
    },
    {
      "epoch": 0.08873,
      "grad_norm": 1.092903916933006,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 8873
    },
    {
      "epoch": 0.08874,
      "grad_norm": 0.9262834148165788,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 8874
    },
    {
      "epoch": 0.08875,
      "grad_norm": 0.9453266208102923,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 8875
    },
    {
      "epoch": 0.08876,
      "grad_norm": 1.0239710759961396,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 8876
    },
    {
      "epoch": 0.08877,
      "grad_norm": 0.997162213443831,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 8877
    },
    {
      "epoch": 0.08878,
      "grad_norm": 1.078780262399322,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8878
    },
    {
      "epoch": 0.08879,
      "grad_norm": 1.1714017894712712,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 8879
    },
    {
      "epoch": 0.0888,
      "grad_norm": 1.0249330378361858,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 8880
    },
    {
      "epoch": 0.08881,
      "grad_norm": 1.0028152550391942,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 8881
    },
    {
      "epoch": 0.08882,
      "grad_norm": 1.1161280626520085,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 8882
    },
    {
      "epoch": 0.08883,
      "grad_norm": 0.9449109550267524,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 8883
    },
    {
      "epoch": 0.08884,
      "grad_norm": 0.9509758982206539,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 8884
    },
    {
      "epoch": 0.08885,
      "grad_norm": 1.0088259868099347,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 8885
    },
    {
      "epoch": 0.08886,
      "grad_norm": 1.0034750507067094,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 8886
    },
    {
      "epoch": 0.08887,
      "grad_norm": 1.1564868184220833,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8887
    },
    {
      "epoch": 0.08888,
      "grad_norm": 0.920047239707721,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 8888
    },
    {
      "epoch": 0.08889,
      "grad_norm": 0.8744706857704945,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8889
    },
    {
      "epoch": 0.0889,
      "grad_norm": 0.9285165665395756,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 8890
    },
    {
      "epoch": 0.08891,
      "grad_norm": 1.0099497011549061,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 8891
    },
    {
      "epoch": 0.08892,
      "grad_norm": 1.1488095492811832,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 8892
    },
    {
      "epoch": 0.08893,
      "grad_norm": 0.9091122428129211,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 8893
    },
    {
      "epoch": 0.08894,
      "grad_norm": 0.9429348565271667,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8894
    },
    {
      "epoch": 0.08895,
      "grad_norm": 1.008440730412949,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 8895
    },
    {
      "epoch": 0.08896,
      "grad_norm": 0.9039965851388357,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 8896
    },
    {
      "epoch": 0.08897,
      "grad_norm": 0.7170118026175338,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 8897
    },
    {
      "epoch": 0.08898,
      "grad_norm": 0.7903539062454961,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 8898
    },
    {
      "epoch": 0.08899,
      "grad_norm": 0.7245799759674336,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 8899
    },
    {
      "epoch": 0.089,
      "grad_norm": 0.6887895124684095,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 8900
    },
    {
      "epoch": 0.08901,
      "grad_norm": 0.6703863748825403,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 8901
    },
    {
      "epoch": 0.08902,
      "grad_norm": 0.6971104906184047,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 8902
    },
    {
      "epoch": 0.08903,
      "grad_norm": 0.8088806783931428,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 8903
    },
    {
      "epoch": 0.08904,
      "grad_norm": 1.0528694251570367,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 8904
    },
    {
      "epoch": 0.08905,
      "grad_norm": 1.333318141346917,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 8905
    },
    {
      "epoch": 0.08906,
      "grad_norm": 0.8796950703932015,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 8906
    },
    {
      "epoch": 0.08907,
      "grad_norm": 1.017711138999783,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 8907
    },
    {
      "epoch": 0.08908,
      "grad_norm": 1.2131433849659086,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 8908
    },
    {
      "epoch": 0.08909,
      "grad_norm": 0.910493543162808,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 8909
    },
    {
      "epoch": 0.0891,
      "grad_norm": 0.9813094815777755,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 8910
    },
    {
      "epoch": 0.08911,
      "grad_norm": 1.0796651179058687,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 8911
    },
    {
      "epoch": 0.08912,
      "grad_norm": 0.7841503223559056,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8912
    },
    {
      "epoch": 0.08913,
      "grad_norm": 0.8040837526844624,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 8913
    },
    {
      "epoch": 0.08914,
      "grad_norm": 0.9201111247555188,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 8914
    },
    {
      "epoch": 0.08915,
      "grad_norm": 1.2443487673269766,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 8915
    },
    {
      "epoch": 0.08916,
      "grad_norm": 0.9520842934826077,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 8916
    },
    {
      "epoch": 0.08917,
      "grad_norm": 0.8122356743140442,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 8917
    },
    {
      "epoch": 0.08918,
      "grad_norm": 0.849179194303752,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 8918
    },
    {
      "epoch": 0.08919,
      "grad_norm": 1.0835039742387556,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 8919
    },
    {
      "epoch": 0.0892,
      "grad_norm": 1.0060700478336813,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 8920
    },
    {
      "epoch": 0.08921,
      "grad_norm": 1.0073037824850515,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 8921
    },
    {
      "epoch": 0.08922,
      "grad_norm": 1.1074535105569765,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 8922
    },
    {
      "epoch": 0.08923,
      "grad_norm": 1.3376095505917398,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 8923
    },
    {
      "epoch": 0.08924,
      "grad_norm": 0.8791813131366455,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 8924
    },
    {
      "epoch": 0.08925,
      "grad_norm": 0.8009079909307454,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 8925
    },
    {
      "epoch": 0.08926,
      "grad_norm": 0.8350852915495754,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 8926
    },
    {
      "epoch": 0.08927,
      "grad_norm": 0.8752090800535209,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8927
    },
    {
      "epoch": 0.08928,
      "grad_norm": 0.8879314625475655,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 8928
    },
    {
      "epoch": 0.08929,
      "grad_norm": 1.142327573748278,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 8929
    },
    {
      "epoch": 0.0893,
      "grad_norm": 1.090048369857334,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 8930
    },
    {
      "epoch": 0.08931,
      "grad_norm": 0.9435660692327841,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 8931
    },
    {
      "epoch": 0.08932,
      "grad_norm": 0.9782783037908521,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8932
    },
    {
      "epoch": 0.08933,
      "grad_norm": 1.0634692540653257,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 8933
    },
    {
      "epoch": 0.08934,
      "grad_norm": 1.0125558743673009,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 8934
    },
    {
      "epoch": 0.08935,
      "grad_norm": 0.9385240579578458,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 8935
    },
    {
      "epoch": 0.08936,
      "grad_norm": 0.9949684717325338,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 8936
    },
    {
      "epoch": 0.08937,
      "grad_norm": 1.0884204651036244,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8937
    },
    {
      "epoch": 0.08938,
      "grad_norm": 1.2595959013613358,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 8938
    },
    {
      "epoch": 0.08939,
      "grad_norm": 0.8718352236246899,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 8939
    },
    {
      "epoch": 0.0894,
      "grad_norm": 0.7377382096343973,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 8940
    },
    {
      "epoch": 0.08941,
      "grad_norm": 0.7815043419777372,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 8941
    },
    {
      "epoch": 0.08942,
      "grad_norm": 0.85637053050209,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 8942
    },
    {
      "epoch": 0.08943,
      "grad_norm": 0.809694079364775,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8943
    },
    {
      "epoch": 0.08944,
      "grad_norm": 0.8311218205671262,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 8944
    },
    {
      "epoch": 0.08945,
      "grad_norm": 0.9235692408794608,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 8945
    },
    {
      "epoch": 0.08946,
      "grad_norm": 0.8988805187494495,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 8946
    },
    {
      "epoch": 0.08947,
      "grad_norm": 0.9750913166472722,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 8947
    },
    {
      "epoch": 0.08948,
      "grad_norm": 1.2358006562850736,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 8948
    },
    {
      "epoch": 0.08949,
      "grad_norm": 1.0201308388850814,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 8949
    },
    {
      "epoch": 0.0895,
      "grad_norm": 1.3147224146265035,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 8950
    },
    {
      "epoch": 0.08951,
      "grad_norm": 0.850210561765072,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 8951
    },
    {
      "epoch": 0.08952,
      "grad_norm": 0.7409549525635194,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 8952
    },
    {
      "epoch": 0.08953,
      "grad_norm": 0.7712667656518064,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 8953
    },
    {
      "epoch": 0.08954,
      "grad_norm": 0.8389010542455372,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 8954
    },
    {
      "epoch": 0.08955,
      "grad_norm": 1.0106297979397758,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 8955
    },
    {
      "epoch": 0.08956,
      "grad_norm": 1.1616116590632415,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 8956
    },
    {
      "epoch": 0.08957,
      "grad_norm": 0.9682455671909211,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 8957
    },
    {
      "epoch": 0.08958,
      "grad_norm": 0.9906512714609359,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 8958
    },
    {
      "epoch": 0.08959,
      "grad_norm": 1.0546522223319332,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 8959
    },
    {
      "epoch": 0.0896,
      "grad_norm": 1.1586648477198143,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 8960
    },
    {
      "epoch": 0.08961,
      "grad_norm": 1.0432105881764966,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 8961
    },
    {
      "epoch": 0.08962,
      "grad_norm": 1.0965840342756925,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 8962
    },
    {
      "epoch": 0.08963,
      "grad_norm": 1.0376163837487167,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 8963
    },
    {
      "epoch": 0.08964,
      "grad_norm": 1.0514679891633183,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 8964
    },
    {
      "epoch": 0.08965,
      "grad_norm": 1.0445039391438893,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 8965
    },
    {
      "epoch": 0.08966,
      "grad_norm": 1.0636480223637435,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 8966
    },
    {
      "epoch": 0.08967,
      "grad_norm": 0.8306886055128744,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 8967
    },
    {
      "epoch": 0.08968,
      "grad_norm": 1.0942130565636998,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 8968
    },
    {
      "epoch": 0.08969,
      "grad_norm": 1.0851056474023233,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 8969
    },
    {
      "epoch": 0.0897,
      "grad_norm": 1.0442994580535447,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 8970
    },
    {
      "epoch": 0.08971,
      "grad_norm": 1.0641712074518217,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 8971
    },
    {
      "epoch": 0.08972,
      "grad_norm": 0.8212488342541941,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 8972
    },
    {
      "epoch": 0.08973,
      "grad_norm": 0.9085039288779564,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 8973
    },
    {
      "epoch": 0.08974,
      "grad_norm": 1.05439509788842,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 8974
    },
    {
      "epoch": 0.08975,
      "grad_norm": 1.0378367702298532,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 8975
    },
    {
      "epoch": 0.08976,
      "grad_norm": 1.0587552419566737,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 8976
    },
    {
      "epoch": 0.08977,
      "grad_norm": 0.9882078968290161,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 8977
    },
    {
      "epoch": 0.08978,
      "grad_norm": 1.2634295327865903,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 8978
    },
    {
      "epoch": 0.08979,
      "grad_norm": 1.0868605949551717,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 8979
    },
    {
      "epoch": 0.0898,
      "grad_norm": 0.9024666538735595,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 8980
    },
    {
      "epoch": 0.08981,
      "grad_norm": 0.940643098285581,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 8981
    },
    {
      "epoch": 0.08982,
      "grad_norm": 1.0209299533352179,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 8982
    },
    {
      "epoch": 0.08983,
      "grad_norm": 1.1663864520745926,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 8983
    },
    {
      "epoch": 0.08984,
      "grad_norm": 1.0421843695021389,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 8984
    },
    {
      "epoch": 0.08985,
      "grad_norm": 0.9455839548731014,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 8985
    },
    {
      "epoch": 0.08986,
      "grad_norm": 1.0175793254211578,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 8986
    },
    {
      "epoch": 0.08987,
      "grad_norm": 1.0855539709461204,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 8987
    },
    {
      "epoch": 0.08988,
      "grad_norm": 1.2089023840888569,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 8988
    },
    {
      "epoch": 0.08989,
      "grad_norm": 0.9145464238654417,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 8989
    },
    {
      "epoch": 0.0899,
      "grad_norm": 0.9988559782011223,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 8990
    },
    {
      "epoch": 0.08991,
      "grad_norm": 0.9616640855485232,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 8991
    },
    {
      "epoch": 0.08992,
      "grad_norm": 1.0543039692078464,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 8992
    },
    {
      "epoch": 0.08993,
      "grad_norm": 0.952039320193524,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 8993
    },
    {
      "epoch": 0.08994,
      "grad_norm": 1.24776754345023,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 8994
    },
    {
      "epoch": 0.08995,
      "grad_norm": 0.8960768153711746,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 8995
    },
    {
      "epoch": 0.08996,
      "grad_norm": 0.8299433831762516,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 8996
    },
    {
      "epoch": 0.08997,
      "grad_norm": 0.8434475562979065,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 8997
    },
    {
      "epoch": 0.08998,
      "grad_norm": 0.9434883081551227,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 8998
    },
    {
      "epoch": 0.08999,
      "grad_norm": 1.028623116722872,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 8999
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1483053060085568,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9000
    },
    {
      "epoch": 0.09001,
      "grad_norm": 1.0462603971030422,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 9001
    },
    {
      "epoch": 0.09002,
      "grad_norm": 0.8866854703633072,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 9002
    },
    {
      "epoch": 0.09003,
      "grad_norm": 0.8897374118868974,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 9003
    },
    {
      "epoch": 0.09004,
      "grad_norm": 1.0318758238899706,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9004
    },
    {
      "epoch": 0.09005,
      "grad_norm": 1.140885873667334,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 9005
    },
    {
      "epoch": 0.09006,
      "grad_norm": 0.891376466816432,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9006
    },
    {
      "epoch": 0.09007,
      "grad_norm": 1.1603687488320678,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 9007
    },
    {
      "epoch": 0.09008,
      "grad_norm": 0.8775142494741145,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9008
    },
    {
      "epoch": 0.09009,
      "grad_norm": 0.9654621607739925,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 9009
    },
    {
      "epoch": 0.0901,
      "grad_norm": 1.0294455899118256,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 9010
    },
    {
      "epoch": 0.09011,
      "grad_norm": 1.1050991894320772,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9011
    },
    {
      "epoch": 0.09012,
      "grad_norm": 1.1543143215597398,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9012
    },
    {
      "epoch": 0.09013,
      "grad_norm": 0.8514403902861597,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 9013
    },
    {
      "epoch": 0.09014,
      "grad_norm": 0.869700266149385,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 9014
    },
    {
      "epoch": 0.09015,
      "grad_norm": 0.8302072302834347,
      "learning_rate": 0.003,
      "loss": 4.1265,
      "step": 9015
    },
    {
      "epoch": 0.09016,
      "grad_norm": 0.7449962289937087,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 9016
    },
    {
      "epoch": 0.09017,
      "grad_norm": 0.8441019310672158,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 9017
    },
    {
      "epoch": 0.09018,
      "grad_norm": 0.9224246439163316,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9018
    },
    {
      "epoch": 0.09019,
      "grad_norm": 1.2506310431634982,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 9019
    },
    {
      "epoch": 0.0902,
      "grad_norm": 1.0319725156855537,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 9020
    },
    {
      "epoch": 0.09021,
      "grad_norm": 1.0244238730812105,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 9021
    },
    {
      "epoch": 0.09022,
      "grad_norm": 1.0747193510062496,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9022
    },
    {
      "epoch": 0.09023,
      "grad_norm": 1.0829551491615659,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9023
    },
    {
      "epoch": 0.09024,
      "grad_norm": 0.8854265172004326,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 9024
    },
    {
      "epoch": 0.09025,
      "grad_norm": 0.9624141531169683,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9025
    },
    {
      "epoch": 0.09026,
      "grad_norm": 0.9124575556477138,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 9026
    },
    {
      "epoch": 0.09027,
      "grad_norm": 0.7994369764537259,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9027
    },
    {
      "epoch": 0.09028,
      "grad_norm": 0.9507899093974508,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 9028
    },
    {
      "epoch": 0.09029,
      "grad_norm": 1.0373894988141181,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 9029
    },
    {
      "epoch": 0.0903,
      "grad_norm": 1.0795383394678428,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 9030
    },
    {
      "epoch": 0.09031,
      "grad_norm": 0.9371901971647943,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 9031
    },
    {
      "epoch": 0.09032,
      "grad_norm": 1.3118743983631005,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 9032
    },
    {
      "epoch": 0.09033,
      "grad_norm": 0.9602498029783673,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 9033
    },
    {
      "epoch": 0.09034,
      "grad_norm": 0.9950626958271721,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 9034
    },
    {
      "epoch": 0.09035,
      "grad_norm": 1.0561055356455764,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 9035
    },
    {
      "epoch": 0.09036,
      "grad_norm": 0.9351410832930079,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9036
    },
    {
      "epoch": 0.09037,
      "grad_norm": 0.9375448148491637,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9037
    },
    {
      "epoch": 0.09038,
      "grad_norm": 0.8772536980898831,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 9038
    },
    {
      "epoch": 0.09039,
      "grad_norm": 0.8994311716413951,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9039
    },
    {
      "epoch": 0.0904,
      "grad_norm": 1.010507879277424,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9040
    },
    {
      "epoch": 0.09041,
      "grad_norm": 1.2199604338083574,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 9041
    },
    {
      "epoch": 0.09042,
      "grad_norm": 0.9306883379123116,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9042
    },
    {
      "epoch": 0.09043,
      "grad_norm": 1.0192706712388095,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9043
    },
    {
      "epoch": 0.09044,
      "grad_norm": 1.0720290832366242,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 9044
    },
    {
      "epoch": 0.09045,
      "grad_norm": 0.9851527301184416,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 9045
    },
    {
      "epoch": 0.09046,
      "grad_norm": 1.0610574818669418,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 9046
    },
    {
      "epoch": 0.09047,
      "grad_norm": 0.9500452554503603,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9047
    },
    {
      "epoch": 0.09048,
      "grad_norm": 0.9313318824748757,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9048
    },
    {
      "epoch": 0.09049,
      "grad_norm": 1.2054793227671492,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 9049
    },
    {
      "epoch": 0.0905,
      "grad_norm": 1.00585427743775,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9050
    },
    {
      "epoch": 0.09051,
      "grad_norm": 1.1034780299931397,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9051
    },
    {
      "epoch": 0.09052,
      "grad_norm": 1.0339569510475137,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 9052
    },
    {
      "epoch": 0.09053,
      "grad_norm": 0.9291730063765455,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9053
    },
    {
      "epoch": 0.09054,
      "grad_norm": 0.7598831863125971,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9054
    },
    {
      "epoch": 0.09055,
      "grad_norm": 0.8785025929149661,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 9055
    },
    {
      "epoch": 0.09056,
      "grad_norm": 1.0578467685619801,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 9056
    },
    {
      "epoch": 0.09057,
      "grad_norm": 1.1700566880339585,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 9057
    },
    {
      "epoch": 0.09058,
      "grad_norm": 0.8486453469937698,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 9058
    },
    {
      "epoch": 0.09059,
      "grad_norm": 1.0260084269697587,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9059
    },
    {
      "epoch": 0.0906,
      "grad_norm": 1.0815529657850544,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 9060
    },
    {
      "epoch": 0.09061,
      "grad_norm": 0.9702489090387596,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 9061
    },
    {
      "epoch": 0.09062,
      "grad_norm": 1.0828394503861276,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 9062
    },
    {
      "epoch": 0.09063,
      "grad_norm": 0.9914403527753315,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9063
    },
    {
      "epoch": 0.09064,
      "grad_norm": 1.0178247711714878,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 9064
    },
    {
      "epoch": 0.09065,
      "grad_norm": 1.0639916710421684,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9065
    },
    {
      "epoch": 0.09066,
      "grad_norm": 0.9666784724080716,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9066
    },
    {
      "epoch": 0.09067,
      "grad_norm": 1.1127616550466115,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9067
    },
    {
      "epoch": 0.09068,
      "grad_norm": 0.9649238134507271,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 9068
    },
    {
      "epoch": 0.09069,
      "grad_norm": 0.9887864245443914,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 9069
    },
    {
      "epoch": 0.0907,
      "grad_norm": 1.138533659439355,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 9070
    },
    {
      "epoch": 0.09071,
      "grad_norm": 0.9538235049365951,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9071
    },
    {
      "epoch": 0.09072,
      "grad_norm": 0.9399245548306119,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 9072
    },
    {
      "epoch": 0.09073,
      "grad_norm": 1.0410231512210018,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9073
    },
    {
      "epoch": 0.09074,
      "grad_norm": 0.9825288806775234,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 9074
    },
    {
      "epoch": 0.09075,
      "grad_norm": 1.2536880541064686,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 9075
    },
    {
      "epoch": 0.09076,
      "grad_norm": 1.0802580187439756,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 9076
    },
    {
      "epoch": 0.09077,
      "grad_norm": 1.0000547727846416,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 9077
    },
    {
      "epoch": 0.09078,
      "grad_norm": 1.1810776381316752,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 9078
    },
    {
      "epoch": 0.09079,
      "grad_norm": 0.8675373011852131,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9079
    },
    {
      "epoch": 0.0908,
      "grad_norm": 0.8189772083275174,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 9080
    },
    {
      "epoch": 0.09081,
      "grad_norm": 1.053936858891442,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 9081
    },
    {
      "epoch": 0.09082,
      "grad_norm": 1.1280890556667789,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 9082
    },
    {
      "epoch": 0.09083,
      "grad_norm": 0.8002462051831681,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9083
    },
    {
      "epoch": 0.09084,
      "grad_norm": 0.7580618745930298,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 9084
    },
    {
      "epoch": 0.09085,
      "grad_norm": 0.8404228487181117,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 9085
    },
    {
      "epoch": 0.09086,
      "grad_norm": 0.9092656319720175,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 9086
    },
    {
      "epoch": 0.09087,
      "grad_norm": 1.1575206293331635,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 9087
    },
    {
      "epoch": 0.09088,
      "grad_norm": 1.1392497988218204,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 9088
    },
    {
      "epoch": 0.09089,
      "grad_norm": 1.0999951974096032,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9089
    },
    {
      "epoch": 0.0909,
      "grad_norm": 1.2058860154470423,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 9090
    },
    {
      "epoch": 0.09091,
      "grad_norm": 0.8869769785320447,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 9091
    },
    {
      "epoch": 0.09092,
      "grad_norm": 0.968833257593624,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 9092
    },
    {
      "epoch": 0.09093,
      "grad_norm": 1.030403691084007,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 9093
    },
    {
      "epoch": 0.09094,
      "grad_norm": 1.1575120263478036,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 9094
    },
    {
      "epoch": 0.09095,
      "grad_norm": 1.0658493375061597,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9095
    },
    {
      "epoch": 0.09096,
      "grad_norm": 1.039522944385023,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 9096
    },
    {
      "epoch": 0.09097,
      "grad_norm": 0.9449746348334745,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 9097
    },
    {
      "epoch": 0.09098,
      "grad_norm": 1.112543228949063,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 9098
    },
    {
      "epoch": 0.09099,
      "grad_norm": 1.0356029258565156,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 9099
    },
    {
      "epoch": 0.091,
      "grad_norm": 0.9597566649353041,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 9100
    },
    {
      "epoch": 0.09101,
      "grad_norm": 0.909685008398296,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 9101
    },
    {
      "epoch": 0.09102,
      "grad_norm": 0.9939609422178486,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 9102
    },
    {
      "epoch": 0.09103,
      "grad_norm": 0.9981967075843049,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 9103
    },
    {
      "epoch": 0.09104,
      "grad_norm": 0.901139384150308,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9104
    },
    {
      "epoch": 0.09105,
      "grad_norm": 1.1303987033121574,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 9105
    },
    {
      "epoch": 0.09106,
      "grad_norm": 0.9605403802740986,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 9106
    },
    {
      "epoch": 0.09107,
      "grad_norm": 1.0896925352972906,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9107
    },
    {
      "epoch": 0.09108,
      "grad_norm": 0.9390438282386121,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 9108
    },
    {
      "epoch": 0.09109,
      "grad_norm": 0.9831731988115366,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 9109
    },
    {
      "epoch": 0.0911,
      "grad_norm": 1.0442414761632293,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 9110
    },
    {
      "epoch": 0.09111,
      "grad_norm": 0.9165743816131849,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9111
    },
    {
      "epoch": 0.09112,
      "grad_norm": 0.8395737037341123,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 9112
    },
    {
      "epoch": 0.09113,
      "grad_norm": 0.9188507446416299,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 9113
    },
    {
      "epoch": 0.09114,
      "grad_norm": 1.192161128289549,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 9114
    },
    {
      "epoch": 0.09115,
      "grad_norm": 0.7878424986532916,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9115
    },
    {
      "epoch": 0.09116,
      "grad_norm": 0.756411216963468,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 9116
    },
    {
      "epoch": 0.09117,
      "grad_norm": 0.8675016418146388,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9117
    },
    {
      "epoch": 0.09118,
      "grad_norm": 1.0711225834740048,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9118
    },
    {
      "epoch": 0.09119,
      "grad_norm": 1.3493838045641182,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9119
    },
    {
      "epoch": 0.0912,
      "grad_norm": 0.9734951792402872,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 9120
    },
    {
      "epoch": 0.09121,
      "grad_norm": 0.9595452425717873,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 9121
    },
    {
      "epoch": 0.09122,
      "grad_norm": 0.8608362956129894,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 9122
    },
    {
      "epoch": 0.09123,
      "grad_norm": 0.9267155554468813,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 9123
    },
    {
      "epoch": 0.09124,
      "grad_norm": 1.1438342201379448,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9124
    },
    {
      "epoch": 0.09125,
      "grad_norm": 0.912337828151199,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 9125
    },
    {
      "epoch": 0.09126,
      "grad_norm": 1.0294428453272435,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 9126
    },
    {
      "epoch": 0.09127,
      "grad_norm": 0.9634503152628049,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9127
    },
    {
      "epoch": 0.09128,
      "grad_norm": 0.9285683958788655,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 9128
    },
    {
      "epoch": 0.09129,
      "grad_norm": 1.1437773291811142,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9129
    },
    {
      "epoch": 0.0913,
      "grad_norm": 1.1492290125815292,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 9130
    },
    {
      "epoch": 0.09131,
      "grad_norm": 0.8288575691135607,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 9131
    },
    {
      "epoch": 0.09132,
      "grad_norm": 0.6963154193725737,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 9132
    },
    {
      "epoch": 0.09133,
      "grad_norm": 0.8615993381306877,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 9133
    },
    {
      "epoch": 0.09134,
      "grad_norm": 0.9872135875308844,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 9134
    },
    {
      "epoch": 0.09135,
      "grad_norm": 1.2143403531062356,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 9135
    },
    {
      "epoch": 0.09136,
      "grad_norm": 0.8677902987991879,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 9136
    },
    {
      "epoch": 0.09137,
      "grad_norm": 0.9650359637584537,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 9137
    },
    {
      "epoch": 0.09138,
      "grad_norm": 0.9888560898200366,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9138
    },
    {
      "epoch": 0.09139,
      "grad_norm": 0.9445593852088465,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9139
    },
    {
      "epoch": 0.0914,
      "grad_norm": 1.1631390018706205,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9140
    },
    {
      "epoch": 0.09141,
      "grad_norm": 0.9733863748246883,
      "learning_rate": 0.003,
      "loss": 4.1235,
      "step": 9141
    },
    {
      "epoch": 0.09142,
      "grad_norm": 1.144068894332549,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 9142
    },
    {
      "epoch": 0.09143,
      "grad_norm": 1.1162307413315125,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 9143
    },
    {
      "epoch": 0.09144,
      "grad_norm": 0.9344229915073872,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9144
    },
    {
      "epoch": 0.09145,
      "grad_norm": 0.934174583034849,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 9145
    },
    {
      "epoch": 0.09146,
      "grad_norm": 1.1330945945510411,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9146
    },
    {
      "epoch": 0.09147,
      "grad_norm": 1.0873009118860704,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 9147
    },
    {
      "epoch": 0.09148,
      "grad_norm": 1.063907115512111,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 9148
    },
    {
      "epoch": 0.09149,
      "grad_norm": 0.9492654111198591,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 9149
    },
    {
      "epoch": 0.0915,
      "grad_norm": 0.8348158073927575,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 9150
    },
    {
      "epoch": 0.09151,
      "grad_norm": 0.7907779988545944,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9151
    },
    {
      "epoch": 0.09152,
      "grad_norm": 0.820521579820349,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9152
    },
    {
      "epoch": 0.09153,
      "grad_norm": 1.1762557932817632,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 9153
    },
    {
      "epoch": 0.09154,
      "grad_norm": 1.172758121597814,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9154
    },
    {
      "epoch": 0.09155,
      "grad_norm": 0.7806493977021605,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 9155
    },
    {
      "epoch": 0.09156,
      "grad_norm": 0.7334709809736035,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9156
    },
    {
      "epoch": 0.09157,
      "grad_norm": 0.7096998833904402,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 9157
    },
    {
      "epoch": 0.09158,
      "grad_norm": 0.6552858991619609,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 9158
    },
    {
      "epoch": 0.09159,
      "grad_norm": 0.6654141207947328,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9159
    },
    {
      "epoch": 0.0916,
      "grad_norm": 0.7659679455492391,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 9160
    },
    {
      "epoch": 0.09161,
      "grad_norm": 0.8871298111496974,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 9161
    },
    {
      "epoch": 0.09162,
      "grad_norm": 1.0636906000745443,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 9162
    },
    {
      "epoch": 0.09163,
      "grad_norm": 0.9990259008300131,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 9163
    },
    {
      "epoch": 0.09164,
      "grad_norm": 0.9987224012321043,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 9164
    },
    {
      "epoch": 0.09165,
      "grad_norm": 1.0688733081179889,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 9165
    },
    {
      "epoch": 0.09166,
      "grad_norm": 0.9930601337362358,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9166
    },
    {
      "epoch": 0.09167,
      "grad_norm": 1.0601290119925093,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9167
    },
    {
      "epoch": 0.09168,
      "grad_norm": 1.1036579498016206,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 9168
    },
    {
      "epoch": 0.09169,
      "grad_norm": 1.101183482537871,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 9169
    },
    {
      "epoch": 0.0917,
      "grad_norm": 1.0752491199658576,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 9170
    },
    {
      "epoch": 0.09171,
      "grad_norm": 0.8939745974652651,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 9171
    },
    {
      "epoch": 0.09172,
      "grad_norm": 1.0352574625773354,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 9172
    },
    {
      "epoch": 0.09173,
      "grad_norm": 1.1944014003973107,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 9173
    },
    {
      "epoch": 0.09174,
      "grad_norm": 0.8767007669481375,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 9174
    },
    {
      "epoch": 0.09175,
      "grad_norm": 0.8817169634099795,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 9175
    },
    {
      "epoch": 0.09176,
      "grad_norm": 1.0813839509578795,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 9176
    },
    {
      "epoch": 0.09177,
      "grad_norm": 1.1959325088512973,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 9177
    },
    {
      "epoch": 0.09178,
      "grad_norm": 0.9670692426409625,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 9178
    },
    {
      "epoch": 0.09179,
      "grad_norm": 1.0878516037257497,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 9179
    },
    {
      "epoch": 0.0918,
      "grad_norm": 1.032818751415936,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 9180
    },
    {
      "epoch": 0.09181,
      "grad_norm": 1.1841768292302939,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 9181
    },
    {
      "epoch": 0.09182,
      "grad_norm": 0.8774737598295932,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 9182
    },
    {
      "epoch": 0.09183,
      "grad_norm": 0.9051881615495581,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9183
    },
    {
      "epoch": 0.09184,
      "grad_norm": 0.968743592106802,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 9184
    },
    {
      "epoch": 0.09185,
      "grad_norm": 1.0561355980332867,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 9185
    },
    {
      "epoch": 0.09186,
      "grad_norm": 1.007362715995797,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 9186
    },
    {
      "epoch": 0.09187,
      "grad_norm": 1.0062815389325381,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 9187
    },
    {
      "epoch": 0.09188,
      "grad_norm": 1.2814825677322759,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 9188
    },
    {
      "epoch": 0.09189,
      "grad_norm": 0.9675032030868871,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 9189
    },
    {
      "epoch": 0.0919,
      "grad_norm": 1.1111999953716745,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 9190
    },
    {
      "epoch": 0.09191,
      "grad_norm": 1.1548515391681171,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 9191
    },
    {
      "epoch": 0.09192,
      "grad_norm": 0.8490837287940421,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 9192
    },
    {
      "epoch": 0.09193,
      "grad_norm": 0.9222479120925116,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 9193
    },
    {
      "epoch": 0.09194,
      "grad_norm": 0.9818940470607805,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 9194
    },
    {
      "epoch": 0.09195,
      "grad_norm": 1.0614643603489413,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9195
    },
    {
      "epoch": 0.09196,
      "grad_norm": 1.1294918164841934,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 9196
    },
    {
      "epoch": 0.09197,
      "grad_norm": 1.0156458829127029,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9197
    },
    {
      "epoch": 0.09198,
      "grad_norm": 1.0305620102728879,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9198
    },
    {
      "epoch": 0.09199,
      "grad_norm": 0.9258303899781921,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9199
    },
    {
      "epoch": 0.092,
      "grad_norm": 1.052722229987556,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 9200
    },
    {
      "epoch": 0.09201,
      "grad_norm": 1.1423176862789393,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9201
    },
    {
      "epoch": 0.09202,
      "grad_norm": 0.90405167273345,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9202
    },
    {
      "epoch": 0.09203,
      "grad_norm": 0.8390172177360099,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 9203
    },
    {
      "epoch": 0.09204,
      "grad_norm": 1.0212772428161263,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 9204
    },
    {
      "epoch": 0.09205,
      "grad_norm": 1.3677660056338998,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9205
    },
    {
      "epoch": 0.09206,
      "grad_norm": 0.8123508684473838,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 9206
    },
    {
      "epoch": 0.09207,
      "grad_norm": 0.7671163096769845,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 9207
    },
    {
      "epoch": 0.09208,
      "grad_norm": 0.7996251518237344,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 9208
    },
    {
      "epoch": 0.09209,
      "grad_norm": 0.8386172650640057,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 9209
    },
    {
      "epoch": 0.0921,
      "grad_norm": 0.874525705380657,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9210
    },
    {
      "epoch": 0.09211,
      "grad_norm": 1.0250102800179441,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9211
    },
    {
      "epoch": 0.09212,
      "grad_norm": 1.252928643973771,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 9212
    },
    {
      "epoch": 0.09213,
      "grad_norm": 0.8470741103659993,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 9213
    },
    {
      "epoch": 0.09214,
      "grad_norm": 0.9021459877837885,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 9214
    },
    {
      "epoch": 0.09215,
      "grad_norm": 1.089124576248861,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 9215
    },
    {
      "epoch": 0.09216,
      "grad_norm": 1.1489211291991535,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 9216
    },
    {
      "epoch": 0.09217,
      "grad_norm": 1.0292749044879108,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 9217
    },
    {
      "epoch": 0.09218,
      "grad_norm": 1.1023668228911894,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 9218
    },
    {
      "epoch": 0.09219,
      "grad_norm": 0.9991820142121032,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 9219
    },
    {
      "epoch": 0.0922,
      "grad_norm": 1.0771443271323655,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 9220
    },
    {
      "epoch": 0.09221,
      "grad_norm": 0.8927077451076355,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9221
    },
    {
      "epoch": 0.09222,
      "grad_norm": 0.921168419778457,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 9222
    },
    {
      "epoch": 0.09223,
      "grad_norm": 1.004730682614683,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 9223
    },
    {
      "epoch": 0.09224,
      "grad_norm": 1.1490662225762478,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 9224
    },
    {
      "epoch": 0.09225,
      "grad_norm": 0.8467914181281023,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9225
    },
    {
      "epoch": 0.09226,
      "grad_norm": 0.7997557193738468,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 9226
    },
    {
      "epoch": 0.09227,
      "grad_norm": 0.8425166880932224,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9227
    },
    {
      "epoch": 0.09228,
      "grad_norm": 1.0077895082604096,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 9228
    },
    {
      "epoch": 0.09229,
      "grad_norm": 1.032289932362043,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9229
    },
    {
      "epoch": 0.0923,
      "grad_norm": 1.2099726093934402,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 9230
    },
    {
      "epoch": 0.09231,
      "grad_norm": 0.9700410140732301,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9231
    },
    {
      "epoch": 0.09232,
      "grad_norm": 1.145125305623255,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9232
    },
    {
      "epoch": 0.09233,
      "grad_norm": 0.9857080652159649,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 9233
    },
    {
      "epoch": 0.09234,
      "grad_norm": 0.853541587899831,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 9234
    },
    {
      "epoch": 0.09235,
      "grad_norm": 0.7853583653755962,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9235
    },
    {
      "epoch": 0.09236,
      "grad_norm": 0.8600641615155024,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 9236
    },
    {
      "epoch": 0.09237,
      "grad_norm": 0.8802995694466931,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9237
    },
    {
      "epoch": 0.09238,
      "grad_norm": 0.889473257279724,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 9238
    },
    {
      "epoch": 0.09239,
      "grad_norm": 0.965025798833757,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 9239
    },
    {
      "epoch": 0.0924,
      "grad_norm": 1.045798773841278,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9240
    },
    {
      "epoch": 0.09241,
      "grad_norm": 0.9585578674271291,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9241
    },
    {
      "epoch": 0.09242,
      "grad_norm": 0.9326373066675198,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9242
    },
    {
      "epoch": 0.09243,
      "grad_norm": 1.1674477646861567,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 9243
    },
    {
      "epoch": 0.09244,
      "grad_norm": 1.111314854607193,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9244
    },
    {
      "epoch": 0.09245,
      "grad_norm": 0.8498701271156277,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 9245
    },
    {
      "epoch": 0.09246,
      "grad_norm": 0.8355516969287872,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9246
    },
    {
      "epoch": 0.09247,
      "grad_norm": 0.94621586277171,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 9247
    },
    {
      "epoch": 0.09248,
      "grad_norm": 1.2170065866247142,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9248
    },
    {
      "epoch": 0.09249,
      "grad_norm": 0.9279621491551032,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 9249
    },
    {
      "epoch": 0.0925,
      "grad_norm": 1.1809787158464518,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 9250
    },
    {
      "epoch": 0.09251,
      "grad_norm": 0.9888992006660906,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9251
    },
    {
      "epoch": 0.09252,
      "grad_norm": 0.9847921065738907,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9252
    },
    {
      "epoch": 0.09253,
      "grad_norm": 1.124411687709022,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9253
    },
    {
      "epoch": 0.09254,
      "grad_norm": 0.9386679683697067,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 9254
    },
    {
      "epoch": 0.09255,
      "grad_norm": 0.8865776906497654,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 9255
    },
    {
      "epoch": 0.09256,
      "grad_norm": 1.04809466760138,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9256
    },
    {
      "epoch": 0.09257,
      "grad_norm": 1.244274217288142,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 9257
    },
    {
      "epoch": 0.09258,
      "grad_norm": 0.8510022944180222,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9258
    },
    {
      "epoch": 0.09259,
      "grad_norm": 0.9598837461861952,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 9259
    },
    {
      "epoch": 0.0926,
      "grad_norm": 0.9427395434917412,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9260
    },
    {
      "epoch": 0.09261,
      "grad_norm": 0.8869685161322585,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 9261
    },
    {
      "epoch": 0.09262,
      "grad_norm": 0.8758797089936894,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9262
    },
    {
      "epoch": 0.09263,
      "grad_norm": 1.1037878707078868,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 9263
    },
    {
      "epoch": 0.09264,
      "grad_norm": 1.347143441300766,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 9264
    },
    {
      "epoch": 0.09265,
      "grad_norm": 0.8240409123116929,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 9265
    },
    {
      "epoch": 0.09266,
      "grad_norm": 0.9131281595783721,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 9266
    },
    {
      "epoch": 0.09267,
      "grad_norm": 1.0156568390080056,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 9267
    },
    {
      "epoch": 0.09268,
      "grad_norm": 1.2484963700197804,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9268
    },
    {
      "epoch": 0.09269,
      "grad_norm": 1.0260778636305667,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 9269
    },
    {
      "epoch": 0.0927,
      "grad_norm": 1.125264451825767,
      "learning_rate": 0.003,
      "loss": 4.1217,
      "step": 9270
    },
    {
      "epoch": 0.09271,
      "grad_norm": 0.9337596088484723,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 9271
    },
    {
      "epoch": 0.09272,
      "grad_norm": 0.9755755501712275,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 9272
    },
    {
      "epoch": 0.09273,
      "grad_norm": 0.9365772539075061,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 9273
    },
    {
      "epoch": 0.09274,
      "grad_norm": 1.056284411019321,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9274
    },
    {
      "epoch": 0.09275,
      "grad_norm": 1.0489238634825415,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9275
    },
    {
      "epoch": 0.09276,
      "grad_norm": 0.9904935266729148,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 9276
    },
    {
      "epoch": 0.09277,
      "grad_norm": 0.98765321752019,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 9277
    },
    {
      "epoch": 0.09278,
      "grad_norm": 1.0672587725016252,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 9278
    },
    {
      "epoch": 0.09279,
      "grad_norm": 1.0022153644343905,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 9279
    },
    {
      "epoch": 0.0928,
      "grad_norm": 1.114357821960331,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 9280
    },
    {
      "epoch": 0.09281,
      "grad_norm": 0.9933886219214135,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 9281
    },
    {
      "epoch": 0.09282,
      "grad_norm": 1.2306301044893613,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 9282
    },
    {
      "epoch": 0.09283,
      "grad_norm": 0.894007821463445,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 9283
    },
    {
      "epoch": 0.09284,
      "grad_norm": 0.995785402749101,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 9284
    },
    {
      "epoch": 0.09285,
      "grad_norm": 1.36921097660269,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9285
    },
    {
      "epoch": 0.09286,
      "grad_norm": 0.9689152429230696,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 9286
    },
    {
      "epoch": 0.09287,
      "grad_norm": 0.9373563728794481,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 9287
    },
    {
      "epoch": 0.09288,
      "grad_norm": 1.0159290553086127,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 9288
    },
    {
      "epoch": 0.09289,
      "grad_norm": 0.9130575266656131,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 9289
    },
    {
      "epoch": 0.0929,
      "grad_norm": 0.8501835536575353,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9290
    },
    {
      "epoch": 0.09291,
      "grad_norm": 0.8717656610305028,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 9291
    },
    {
      "epoch": 0.09292,
      "grad_norm": 1.1198959017363963,
      "learning_rate": 0.003,
      "loss": 4.1126,
      "step": 9292
    },
    {
      "epoch": 0.09293,
      "grad_norm": 0.9126173242568293,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 9293
    },
    {
      "epoch": 0.09294,
      "grad_norm": 1.0289742761655396,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 9294
    },
    {
      "epoch": 0.09295,
      "grad_norm": 1.3069141395458677,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9295
    },
    {
      "epoch": 0.09296,
      "grad_norm": 1.032110304900683,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 9296
    },
    {
      "epoch": 0.09297,
      "grad_norm": 1.0647117479317476,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 9297
    },
    {
      "epoch": 0.09298,
      "grad_norm": 1.1600003420191138,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9298
    },
    {
      "epoch": 0.09299,
      "grad_norm": 0.8067138238009587,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9299
    },
    {
      "epoch": 0.093,
      "grad_norm": 0.8762900677643826,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9300
    },
    {
      "epoch": 0.09301,
      "grad_norm": 0.9503109211302515,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9301
    },
    {
      "epoch": 0.09302,
      "grad_norm": 1.1335888930514397,
      "learning_rate": 0.003,
      "loss": 4.1225,
      "step": 9302
    },
    {
      "epoch": 0.09303,
      "grad_norm": 0.8457264426047458,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 9303
    },
    {
      "epoch": 0.09304,
      "grad_norm": 0.9912599414995296,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 9304
    },
    {
      "epoch": 0.09305,
      "grad_norm": 1.0180631509700129,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 9305
    },
    {
      "epoch": 0.09306,
      "grad_norm": 1.1194293041880785,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9306
    },
    {
      "epoch": 0.09307,
      "grad_norm": 1.1133493890272341,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9307
    },
    {
      "epoch": 0.09308,
      "grad_norm": 0.9634161791067534,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 9308
    },
    {
      "epoch": 0.09309,
      "grad_norm": 0.9473486800894216,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 9309
    },
    {
      "epoch": 0.0931,
      "grad_norm": 1.0246728440886619,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 9310
    },
    {
      "epoch": 0.09311,
      "grad_norm": 0.9925465656864717,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9311
    },
    {
      "epoch": 0.09312,
      "grad_norm": 1.2423874194896622,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9312
    },
    {
      "epoch": 0.09313,
      "grad_norm": 0.9586209397059013,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 9313
    },
    {
      "epoch": 0.09314,
      "grad_norm": 0.9862059787685488,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9314
    },
    {
      "epoch": 0.09315,
      "grad_norm": 0.9997247291164021,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 9315
    },
    {
      "epoch": 0.09316,
      "grad_norm": 0.9978795234093828,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9316
    },
    {
      "epoch": 0.09317,
      "grad_norm": 1.0202358866446537,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9317
    },
    {
      "epoch": 0.09318,
      "grad_norm": 1.0358808879366825,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 9318
    },
    {
      "epoch": 0.09319,
      "grad_norm": 0.9359347662036268,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 9319
    },
    {
      "epoch": 0.0932,
      "grad_norm": 1.0516397687574617,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9320
    },
    {
      "epoch": 0.09321,
      "grad_norm": 1.027948417607539,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 9321
    },
    {
      "epoch": 0.09322,
      "grad_norm": 1.2238846396237006,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 9322
    },
    {
      "epoch": 0.09323,
      "grad_norm": 1.0710084791274084,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 9323
    },
    {
      "epoch": 0.09324,
      "grad_norm": 0.8664748503030784,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 9324
    },
    {
      "epoch": 0.09325,
      "grad_norm": 0.8244384523394639,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 9325
    },
    {
      "epoch": 0.09326,
      "grad_norm": 0.8366377301647347,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 9326
    },
    {
      "epoch": 0.09327,
      "grad_norm": 1.002471764215464,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9327
    },
    {
      "epoch": 0.09328,
      "grad_norm": 1.2210088285550147,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9328
    },
    {
      "epoch": 0.09329,
      "grad_norm": 1.0013379117391867,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 9329
    },
    {
      "epoch": 0.0933,
      "grad_norm": 1.1944499976346676,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 9330
    },
    {
      "epoch": 0.09331,
      "grad_norm": 0.7632243261233325,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9331
    },
    {
      "epoch": 0.09332,
      "grad_norm": 0.9200958475962796,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9332
    },
    {
      "epoch": 0.09333,
      "grad_norm": 1.403618569774701,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 9333
    },
    {
      "epoch": 0.09334,
      "grad_norm": 0.9530756028848038,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 9334
    },
    {
      "epoch": 0.09335,
      "grad_norm": 0.8373722239247557,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 9335
    },
    {
      "epoch": 0.09336,
      "grad_norm": 1.07671774620286,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 9336
    },
    {
      "epoch": 0.09337,
      "grad_norm": 1.1688556059409754,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 9337
    },
    {
      "epoch": 0.09338,
      "grad_norm": 0.9004553842817038,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 9338
    },
    {
      "epoch": 0.09339,
      "grad_norm": 0.8997829693537273,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 9339
    },
    {
      "epoch": 0.0934,
      "grad_norm": 1.0364306321074155,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 9340
    },
    {
      "epoch": 0.09341,
      "grad_norm": 1.1496436689471086,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 9341
    },
    {
      "epoch": 0.09342,
      "grad_norm": 0.8404709893004499,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 9342
    },
    {
      "epoch": 0.09343,
      "grad_norm": 0.8638397010633795,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 9343
    },
    {
      "epoch": 0.09344,
      "grad_norm": 1.009502025708084,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 9344
    },
    {
      "epoch": 0.09345,
      "grad_norm": 1.1049442915711438,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 9345
    },
    {
      "epoch": 0.09346,
      "grad_norm": 0.9264275595290813,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 9346
    },
    {
      "epoch": 0.09347,
      "grad_norm": 1.04657561770358,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 9347
    },
    {
      "epoch": 0.09348,
      "grad_norm": 1.1020959605289053,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 9348
    },
    {
      "epoch": 0.09349,
      "grad_norm": 0.8481624664881736,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 9349
    },
    {
      "epoch": 0.0935,
      "grad_norm": 0.8853005377836479,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 9350
    },
    {
      "epoch": 0.09351,
      "grad_norm": 0.8610406536309755,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 9351
    },
    {
      "epoch": 0.09352,
      "grad_norm": 0.8557730123548091,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 9352
    },
    {
      "epoch": 0.09353,
      "grad_norm": 0.8862296744798557,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 9353
    },
    {
      "epoch": 0.09354,
      "grad_norm": 0.9143136262869864,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 9354
    },
    {
      "epoch": 0.09355,
      "grad_norm": 1.0652270291919952,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9355
    },
    {
      "epoch": 0.09356,
      "grad_norm": 1.2739415869779755,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 9356
    },
    {
      "epoch": 0.09357,
      "grad_norm": 0.8507290447629781,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 9357
    },
    {
      "epoch": 0.09358,
      "grad_norm": 0.8382429448084058,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 9358
    },
    {
      "epoch": 0.09359,
      "grad_norm": 1.1026964823449827,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 9359
    },
    {
      "epoch": 0.0936,
      "grad_norm": 0.953788284156056,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9360
    },
    {
      "epoch": 0.09361,
      "grad_norm": 0.9725724420048601,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 9361
    },
    {
      "epoch": 0.09362,
      "grad_norm": 1.0587389958001008,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 9362
    },
    {
      "epoch": 0.09363,
      "grad_norm": 0.9453661891620305,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 9363
    },
    {
      "epoch": 0.09364,
      "grad_norm": 1.166103814168175,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9364
    },
    {
      "epoch": 0.09365,
      "grad_norm": 0.981304472296457,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 9365
    },
    {
      "epoch": 0.09366,
      "grad_norm": 1.3421162985123651,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 9366
    },
    {
      "epoch": 0.09367,
      "grad_norm": 1.112844598418381,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 9367
    },
    {
      "epoch": 0.09368,
      "grad_norm": 0.9409251616261554,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 9368
    },
    {
      "epoch": 0.09369,
      "grad_norm": 0.8769814653061887,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 9369
    },
    {
      "epoch": 0.0937,
      "grad_norm": 0.8344076839182143,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 9370
    },
    {
      "epoch": 0.09371,
      "grad_norm": 0.8198638715880406,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 9371
    },
    {
      "epoch": 0.09372,
      "grad_norm": 0.8780692813529192,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9372
    },
    {
      "epoch": 0.09373,
      "grad_norm": 0.919173903864044,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 9373
    },
    {
      "epoch": 0.09374,
      "grad_norm": 1.1774099587785354,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 9374
    },
    {
      "epoch": 0.09375,
      "grad_norm": 0.9234795661462851,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 9375
    },
    {
      "epoch": 0.09376,
      "grad_norm": 0.8716847861495836,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9376
    },
    {
      "epoch": 0.09377,
      "grad_norm": 0.8578647905954336,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9377
    },
    {
      "epoch": 0.09378,
      "grad_norm": 0.7674758289318644,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 9378
    },
    {
      "epoch": 0.09379,
      "grad_norm": 0.9036010042215633,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 9379
    },
    {
      "epoch": 0.0938,
      "grad_norm": 1.2257343544089907,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9380
    },
    {
      "epoch": 0.09381,
      "grad_norm": 1.1510725609081562,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9381
    },
    {
      "epoch": 0.09382,
      "grad_norm": 1.0854982176578536,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 9382
    },
    {
      "epoch": 0.09383,
      "grad_norm": 1.0776699328606887,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9383
    },
    {
      "epoch": 0.09384,
      "grad_norm": 1.0496928305379307,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 9384
    },
    {
      "epoch": 0.09385,
      "grad_norm": 1.0158952106727617,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9385
    },
    {
      "epoch": 0.09386,
      "grad_norm": 1.1009188691449934,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 9386
    },
    {
      "epoch": 0.09387,
      "grad_norm": 0.9331858264293696,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 9387
    },
    {
      "epoch": 0.09388,
      "grad_norm": 0.9977640976374138,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 9388
    },
    {
      "epoch": 0.09389,
      "grad_norm": 1.09021391133396,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9389
    },
    {
      "epoch": 0.0939,
      "grad_norm": 1.1012507495521715,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 9390
    },
    {
      "epoch": 0.09391,
      "grad_norm": 1.1340299983662043,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 9391
    },
    {
      "epoch": 0.09392,
      "grad_norm": 1.0755966661059326,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 9392
    },
    {
      "epoch": 0.09393,
      "grad_norm": 1.0564351543775305,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 9393
    },
    {
      "epoch": 0.09394,
      "grad_norm": 0.9221702675478415,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 9394
    },
    {
      "epoch": 0.09395,
      "grad_norm": 1.11867912752045,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9395
    },
    {
      "epoch": 0.09396,
      "grad_norm": 1.0650662730096374,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 9396
    },
    {
      "epoch": 0.09397,
      "grad_norm": 0.9989739888141916,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9397
    },
    {
      "epoch": 0.09398,
      "grad_norm": 1.04745389684216,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 9398
    },
    {
      "epoch": 0.09399,
      "grad_norm": 0.9622692280566821,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 9399
    },
    {
      "epoch": 0.094,
      "grad_norm": 1.0129391627458666,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 9400
    },
    {
      "epoch": 0.09401,
      "grad_norm": 1.0179038955491484,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 9401
    },
    {
      "epoch": 0.09402,
      "grad_norm": 1.0037084366352333,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 9402
    },
    {
      "epoch": 0.09403,
      "grad_norm": 1.116193331964759,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 9403
    },
    {
      "epoch": 0.09404,
      "grad_norm": 1.114482641077499,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 9404
    },
    {
      "epoch": 0.09405,
      "grad_norm": 0.9656290178764857,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 9405
    },
    {
      "epoch": 0.09406,
      "grad_norm": 1.0418496678579732,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 9406
    },
    {
      "epoch": 0.09407,
      "grad_norm": 0.8701407322223268,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 9407
    },
    {
      "epoch": 0.09408,
      "grad_norm": 0.8498940051083433,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9408
    },
    {
      "epoch": 0.09409,
      "grad_norm": 0.9515337881634939,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 9409
    },
    {
      "epoch": 0.0941,
      "grad_norm": 0.8952625371005679,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 9410
    },
    {
      "epoch": 0.09411,
      "grad_norm": 0.9904717550526758,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 9411
    },
    {
      "epoch": 0.09412,
      "grad_norm": 1.0747473702120094,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 9412
    },
    {
      "epoch": 0.09413,
      "grad_norm": 1.0544939442524188,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9413
    },
    {
      "epoch": 0.09414,
      "grad_norm": 0.9446059373055282,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 9414
    },
    {
      "epoch": 0.09415,
      "grad_norm": 1.1814575295816008,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 9415
    },
    {
      "epoch": 0.09416,
      "grad_norm": 1.2784405713018843,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 9416
    },
    {
      "epoch": 0.09417,
      "grad_norm": 1.0572081822689008,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 9417
    },
    {
      "epoch": 0.09418,
      "grad_norm": 1.2041574706110378,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9418
    },
    {
      "epoch": 0.09419,
      "grad_norm": 0.9168210345813573,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 9419
    },
    {
      "epoch": 0.0942,
      "grad_norm": 0.9165745133346884,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 9420
    },
    {
      "epoch": 0.09421,
      "grad_norm": 0.8507495772840832,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9421
    },
    {
      "epoch": 0.09422,
      "grad_norm": 0.9415383624526804,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9422
    },
    {
      "epoch": 0.09423,
      "grad_norm": 1.1234380596400146,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 9423
    },
    {
      "epoch": 0.09424,
      "grad_norm": 0.9442798953705628,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9424
    },
    {
      "epoch": 0.09425,
      "grad_norm": 0.9135668408687693,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 9425
    },
    {
      "epoch": 0.09426,
      "grad_norm": 0.9577560361319429,
      "learning_rate": 0.003,
      "loss": 4.1208,
      "step": 9426
    },
    {
      "epoch": 0.09427,
      "grad_norm": 1.0865627308701054,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 9427
    },
    {
      "epoch": 0.09428,
      "grad_norm": 1.0884339478261547,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 9428
    },
    {
      "epoch": 0.09429,
      "grad_norm": 1.059814754063221,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 9429
    },
    {
      "epoch": 0.0943,
      "grad_norm": 1.0809288733936093,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 9430
    },
    {
      "epoch": 0.09431,
      "grad_norm": 0.9441751012722414,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 9431
    },
    {
      "epoch": 0.09432,
      "grad_norm": 1.0069652330680032,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9432
    },
    {
      "epoch": 0.09433,
      "grad_norm": 1.2770781548897892,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9433
    },
    {
      "epoch": 0.09434,
      "grad_norm": 0.9107472747872962,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 9434
    },
    {
      "epoch": 0.09435,
      "grad_norm": 0.9371330665814549,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9435
    },
    {
      "epoch": 0.09436,
      "grad_norm": 0.948797698466554,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 9436
    },
    {
      "epoch": 0.09437,
      "grad_norm": 1.0965285210238682,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 9437
    },
    {
      "epoch": 0.09438,
      "grad_norm": 0.9539276888088277,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9438
    },
    {
      "epoch": 0.09439,
      "grad_norm": 0.9297934842516888,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9439
    },
    {
      "epoch": 0.0944,
      "grad_norm": 0.8833885403510608,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 9440
    },
    {
      "epoch": 0.09441,
      "grad_norm": 0.7745310460762501,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 9441
    },
    {
      "epoch": 0.09442,
      "grad_norm": 0.7444800565907276,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 9442
    },
    {
      "epoch": 0.09443,
      "grad_norm": 0.8414080585325165,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 9443
    },
    {
      "epoch": 0.09444,
      "grad_norm": 1.0036008655746649,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 9444
    },
    {
      "epoch": 0.09445,
      "grad_norm": 1.2505778804758252,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 9445
    },
    {
      "epoch": 0.09446,
      "grad_norm": 1.0393979651836986,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 9446
    },
    {
      "epoch": 0.09447,
      "grad_norm": 0.964114883135747,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 9447
    },
    {
      "epoch": 0.09448,
      "grad_norm": 1.16494849936199,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 9448
    },
    {
      "epoch": 0.09449,
      "grad_norm": 0.9548194627263663,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 9449
    },
    {
      "epoch": 0.0945,
      "grad_norm": 1.0233790768032425,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 9450
    },
    {
      "epoch": 0.09451,
      "grad_norm": 0.9840248139547673,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 9451
    },
    {
      "epoch": 0.09452,
      "grad_norm": 1.10386291686478,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 9452
    },
    {
      "epoch": 0.09453,
      "grad_norm": 1.0942515861458446,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 9453
    },
    {
      "epoch": 0.09454,
      "grad_norm": 1.0520618341084693,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 9454
    },
    {
      "epoch": 0.09455,
      "grad_norm": 1.032921792376323,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9455
    },
    {
      "epoch": 0.09456,
      "grad_norm": 1.0398131988505905,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 9456
    },
    {
      "epoch": 0.09457,
      "grad_norm": 1.0491898439800345,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 9457
    },
    {
      "epoch": 0.09458,
      "grad_norm": 0.9354580793426419,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 9458
    },
    {
      "epoch": 0.09459,
      "grad_norm": 0.9716067137585749,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 9459
    },
    {
      "epoch": 0.0946,
      "grad_norm": 1.0690755013786215,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 9460
    },
    {
      "epoch": 0.09461,
      "grad_norm": 1.1380058613035746,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 9461
    },
    {
      "epoch": 0.09462,
      "grad_norm": 0.9795890325721478,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 9462
    },
    {
      "epoch": 0.09463,
      "grad_norm": 1.0112480646782764,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 9463
    },
    {
      "epoch": 0.09464,
      "grad_norm": 1.4174770915576664,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9464
    },
    {
      "epoch": 0.09465,
      "grad_norm": 0.7118195304889618,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 9465
    },
    {
      "epoch": 0.09466,
      "grad_norm": 0.7397324947507956,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 9466
    },
    {
      "epoch": 0.09467,
      "grad_norm": 0.9228651076473079,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 9467
    },
    {
      "epoch": 0.09468,
      "grad_norm": 1.053278439976445,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 9468
    },
    {
      "epoch": 0.09469,
      "grad_norm": 1.160147117243451,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 9469
    },
    {
      "epoch": 0.0947,
      "grad_norm": 1.026133164845867,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9470
    },
    {
      "epoch": 0.09471,
      "grad_norm": 1.164591722645013,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 9471
    },
    {
      "epoch": 0.09472,
      "grad_norm": 0.7985197662831892,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9472
    },
    {
      "epoch": 0.09473,
      "grad_norm": 0.763553189156105,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 9473
    },
    {
      "epoch": 0.09474,
      "grad_norm": 0.8175418581057333,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9474
    },
    {
      "epoch": 0.09475,
      "grad_norm": 0.9100168795925652,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 9475
    },
    {
      "epoch": 0.09476,
      "grad_norm": 0.9026725830353436,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 9476
    },
    {
      "epoch": 0.09477,
      "grad_norm": 0.9624617938673868,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 9477
    },
    {
      "epoch": 0.09478,
      "grad_norm": 1.081533379210801,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 9478
    },
    {
      "epoch": 0.09479,
      "grad_norm": 1.0540513595530958,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 9479
    },
    {
      "epoch": 0.0948,
      "grad_norm": 1.0198438271547687,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 9480
    },
    {
      "epoch": 0.09481,
      "grad_norm": 1.2125887044399974,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 9481
    },
    {
      "epoch": 0.09482,
      "grad_norm": 1.1188988335754426,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 9482
    },
    {
      "epoch": 0.09483,
      "grad_norm": 0.9414423522388563,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 9483
    },
    {
      "epoch": 0.09484,
      "grad_norm": 1.0501757386247388,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 9484
    },
    {
      "epoch": 0.09485,
      "grad_norm": 1.0797081595739608,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9485
    },
    {
      "epoch": 0.09486,
      "grad_norm": 1.1593524413712497,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 9486
    },
    {
      "epoch": 0.09487,
      "grad_norm": 0.9008992419862395,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 9487
    },
    {
      "epoch": 0.09488,
      "grad_norm": 0.928425213219695,
      "learning_rate": 0.003,
      "loss": 4.1322,
      "step": 9488
    },
    {
      "epoch": 0.09489,
      "grad_norm": 1.1255438362788508,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9489
    },
    {
      "epoch": 0.0949,
      "grad_norm": 0.9565865756324435,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 9490
    },
    {
      "epoch": 0.09491,
      "grad_norm": 1.091082485215994,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 9491
    },
    {
      "epoch": 0.09492,
      "grad_norm": 1.0432359393866104,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9492
    },
    {
      "epoch": 0.09493,
      "grad_norm": 0.9947831762502642,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 9493
    },
    {
      "epoch": 0.09494,
      "grad_norm": 1.170287104954015,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 9494
    },
    {
      "epoch": 0.09495,
      "grad_norm": 1.0499066150365186,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9495
    },
    {
      "epoch": 0.09496,
      "grad_norm": 1.4045418939748382,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 9496
    },
    {
      "epoch": 0.09497,
      "grad_norm": 0.8509044383443939,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 9497
    },
    {
      "epoch": 0.09498,
      "grad_norm": 0.8693676163780726,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 9498
    },
    {
      "epoch": 0.09499,
      "grad_norm": 0.8019980727919567,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 9499
    },
    {
      "epoch": 0.095,
      "grad_norm": 0.8765503155986619,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 9500
    },
    {
      "epoch": 0.09501,
      "grad_norm": 0.8949783455821654,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 9501
    },
    {
      "epoch": 0.09502,
      "grad_norm": 0.85473282604943,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 9502
    },
    {
      "epoch": 0.09503,
      "grad_norm": 0.8962207462857377,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 9503
    },
    {
      "epoch": 0.09504,
      "grad_norm": 0.8523909551635448,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9504
    },
    {
      "epoch": 0.09505,
      "grad_norm": 0.9576305558046839,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 9505
    },
    {
      "epoch": 0.09506,
      "grad_norm": 1.2342117179715257,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 9506
    },
    {
      "epoch": 0.09507,
      "grad_norm": 1.2976078459163851,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 9507
    },
    {
      "epoch": 0.09508,
      "grad_norm": 1.185691690338481,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 9508
    },
    {
      "epoch": 0.09509,
      "grad_norm": 0.9674516966851515,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9509
    },
    {
      "epoch": 0.0951,
      "grad_norm": 1.014442271229994,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9510
    },
    {
      "epoch": 0.09511,
      "grad_norm": 1.0385934154900638,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 9511
    },
    {
      "epoch": 0.09512,
      "grad_norm": 0.9361766204688062,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9512
    },
    {
      "epoch": 0.09513,
      "grad_norm": 1.0033999785344312,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9513
    },
    {
      "epoch": 0.09514,
      "grad_norm": 1.05390650992189,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9514
    },
    {
      "epoch": 0.09515,
      "grad_norm": 1.116848612024721,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9515
    },
    {
      "epoch": 0.09516,
      "grad_norm": 0.9674724683645552,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 9516
    },
    {
      "epoch": 0.09517,
      "grad_norm": 1.0957320303666789,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 9517
    },
    {
      "epoch": 0.09518,
      "grad_norm": 1.0257060215691023,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 9518
    },
    {
      "epoch": 0.09519,
      "grad_norm": 1.0744943454127902,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 9519
    },
    {
      "epoch": 0.0952,
      "grad_norm": 1.2248690019464181,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 9520
    },
    {
      "epoch": 0.09521,
      "grad_norm": 0.9237408992684386,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 9521
    },
    {
      "epoch": 0.09522,
      "grad_norm": 0.8081645199251014,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9522
    },
    {
      "epoch": 0.09523,
      "grad_norm": 0.8001415036331109,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9523
    },
    {
      "epoch": 0.09524,
      "grad_norm": 0.8721020743119626,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9524
    },
    {
      "epoch": 0.09525,
      "grad_norm": 1.0070519804673295,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 9525
    },
    {
      "epoch": 0.09526,
      "grad_norm": 1.0874508141227783,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 9526
    },
    {
      "epoch": 0.09527,
      "grad_norm": 0.9064561128680414,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 9527
    },
    {
      "epoch": 0.09528,
      "grad_norm": 0.8819852497888345,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 9528
    },
    {
      "epoch": 0.09529,
      "grad_norm": 1.0155582815835493,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 9529
    },
    {
      "epoch": 0.0953,
      "grad_norm": 1.2335728455983175,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9530
    },
    {
      "epoch": 0.09531,
      "grad_norm": 0.9890751026219354,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 9531
    },
    {
      "epoch": 0.09532,
      "grad_norm": 1.0139299938673432,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 9532
    },
    {
      "epoch": 0.09533,
      "grad_norm": 1.0406685870722243,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 9533
    },
    {
      "epoch": 0.09534,
      "grad_norm": 1.0415786211613325,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 9534
    },
    {
      "epoch": 0.09535,
      "grad_norm": 1.0266775971668027,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 9535
    },
    {
      "epoch": 0.09536,
      "grad_norm": 1.0680875423201488,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 9536
    },
    {
      "epoch": 0.09537,
      "grad_norm": 0.9532000734499076,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 9537
    },
    {
      "epoch": 0.09538,
      "grad_norm": 1.0875678049842619,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 9538
    },
    {
      "epoch": 0.09539,
      "grad_norm": 1.204316570645322,
      "learning_rate": 0.003,
      "loss": 4.1112,
      "step": 9539
    },
    {
      "epoch": 0.0954,
      "grad_norm": 1.030083701060855,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 9540
    },
    {
      "epoch": 0.09541,
      "grad_norm": 1.1297064889435355,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 9541
    },
    {
      "epoch": 0.09542,
      "grad_norm": 0.9173234611129824,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 9542
    },
    {
      "epoch": 0.09543,
      "grad_norm": 0.9536043595896253,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 9543
    },
    {
      "epoch": 0.09544,
      "grad_norm": 1.111358012851114,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 9544
    },
    {
      "epoch": 0.09545,
      "grad_norm": 1.0573210212427635,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 9545
    },
    {
      "epoch": 0.09546,
      "grad_norm": 1.0671113439958861,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 9546
    },
    {
      "epoch": 0.09547,
      "grad_norm": 0.9551135025314722,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9547
    },
    {
      "epoch": 0.09548,
      "grad_norm": 1.089615403951393,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 9548
    },
    {
      "epoch": 0.09549,
      "grad_norm": 1.1596112981509654,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 9549
    },
    {
      "epoch": 0.0955,
      "grad_norm": 0.9088861727644186,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 9550
    },
    {
      "epoch": 0.09551,
      "grad_norm": 1.2391520142985264,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 9551
    },
    {
      "epoch": 0.09552,
      "grad_norm": 1.0362472449674538,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 9552
    },
    {
      "epoch": 0.09553,
      "grad_norm": 0.9059425211543064,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9553
    },
    {
      "epoch": 0.09554,
      "grad_norm": 0.7685716396395997,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 9554
    },
    {
      "epoch": 0.09555,
      "grad_norm": 0.7760521570571651,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 9555
    },
    {
      "epoch": 0.09556,
      "grad_norm": 0.915451851496684,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 9556
    },
    {
      "epoch": 0.09557,
      "grad_norm": 1.0725359354533892,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 9557
    },
    {
      "epoch": 0.09558,
      "grad_norm": 1.1165179826378548,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 9558
    },
    {
      "epoch": 0.09559,
      "grad_norm": 1.0260042229050033,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 9559
    },
    {
      "epoch": 0.0956,
      "grad_norm": 1.1128206813614063,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 9560
    },
    {
      "epoch": 0.09561,
      "grad_norm": 0.9561200539995762,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 9561
    },
    {
      "epoch": 0.09562,
      "grad_norm": 0.7921512806544463,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 9562
    },
    {
      "epoch": 0.09563,
      "grad_norm": 0.9041768945308537,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 9563
    },
    {
      "epoch": 0.09564,
      "grad_norm": 1.1681721170614285,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 9564
    },
    {
      "epoch": 0.09565,
      "grad_norm": 1.229817627860069,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 9565
    },
    {
      "epoch": 0.09566,
      "grad_norm": 0.8408199819338241,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9566
    },
    {
      "epoch": 0.09567,
      "grad_norm": 1.0091863337567137,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 9567
    },
    {
      "epoch": 0.09568,
      "grad_norm": 1.3094571250437879,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9568
    },
    {
      "epoch": 0.09569,
      "grad_norm": 0.8623005542878108,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9569
    },
    {
      "epoch": 0.0957,
      "grad_norm": 0.9413601755374859,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 9570
    },
    {
      "epoch": 0.09571,
      "grad_norm": 0.8957027980528774,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 9571
    },
    {
      "epoch": 0.09572,
      "grad_norm": 0.9187975967965829,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 9572
    },
    {
      "epoch": 0.09573,
      "grad_norm": 1.137687044149432,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 9573
    },
    {
      "epoch": 0.09574,
      "grad_norm": 1.1691163828761597,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 9574
    },
    {
      "epoch": 0.09575,
      "grad_norm": 0.9206397514109596,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 9575
    },
    {
      "epoch": 0.09576,
      "grad_norm": 0.9884030120613543,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9576
    },
    {
      "epoch": 0.09577,
      "grad_norm": 1.151909672565129,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 9577
    },
    {
      "epoch": 0.09578,
      "grad_norm": 0.943817382689356,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 9578
    },
    {
      "epoch": 0.09579,
      "grad_norm": 0.9266875861613927,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 9579
    },
    {
      "epoch": 0.0958,
      "grad_norm": 1.1200656888418847,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9580
    },
    {
      "epoch": 0.09581,
      "grad_norm": 1.0924396133419563,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 9581
    },
    {
      "epoch": 0.09582,
      "grad_norm": 0.8011259972597407,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 9582
    },
    {
      "epoch": 0.09583,
      "grad_norm": 0.9706166643361759,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 9583
    },
    {
      "epoch": 0.09584,
      "grad_norm": 1.0375471914237828,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 9584
    },
    {
      "epoch": 0.09585,
      "grad_norm": 1.162671870104563,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9585
    },
    {
      "epoch": 0.09586,
      "grad_norm": 1.1269620104984417,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 9586
    },
    {
      "epoch": 0.09587,
      "grad_norm": 1.1157717297270022,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 9587
    },
    {
      "epoch": 0.09588,
      "grad_norm": 1.1329330107551132,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 9588
    },
    {
      "epoch": 0.09589,
      "grad_norm": 1.1375709999149357,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 9589
    },
    {
      "epoch": 0.0959,
      "grad_norm": 1.0008878090061901,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 9590
    },
    {
      "epoch": 0.09591,
      "grad_norm": 1.0935601459276771,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 9591
    },
    {
      "epoch": 0.09592,
      "grad_norm": 0.9286570503636103,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9592
    },
    {
      "epoch": 0.09593,
      "grad_norm": 0.9832018964530126,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 9593
    },
    {
      "epoch": 0.09594,
      "grad_norm": 0.9244421706166079,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 9594
    },
    {
      "epoch": 0.09595,
      "grad_norm": 0.9104310402580331,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9595
    },
    {
      "epoch": 0.09596,
      "grad_norm": 0.9548164533755923,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9596
    },
    {
      "epoch": 0.09597,
      "grad_norm": 1.141997527843847,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 9597
    },
    {
      "epoch": 0.09598,
      "grad_norm": 1.245711271585723,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9598
    },
    {
      "epoch": 0.09599,
      "grad_norm": 0.8088955787411081,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 9599
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.8952527018243382,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 9600
    },
    {
      "epoch": 0.09601,
      "grad_norm": 1.0349856846675602,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 9601
    },
    {
      "epoch": 0.09602,
      "grad_norm": 1.0594507709330971,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 9602
    },
    {
      "epoch": 0.09603,
      "grad_norm": 1.1028365715897552,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9603
    },
    {
      "epoch": 0.09604,
      "grad_norm": 0.8849548250049927,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 9604
    },
    {
      "epoch": 0.09605,
      "grad_norm": 0.922563197820632,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 9605
    },
    {
      "epoch": 0.09606,
      "grad_norm": 1.105340202233273,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9606
    },
    {
      "epoch": 0.09607,
      "grad_norm": 1.2220515830119882,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 9607
    },
    {
      "epoch": 0.09608,
      "grad_norm": 0.9494734780740829,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 9608
    },
    {
      "epoch": 0.09609,
      "grad_norm": 1.0816550584500007,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 9609
    },
    {
      "epoch": 0.0961,
      "grad_norm": 0.9492548605654175,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9610
    },
    {
      "epoch": 0.09611,
      "grad_norm": 0.9587253967158587,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9611
    },
    {
      "epoch": 0.09612,
      "grad_norm": 1.1146970448641045,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 9612
    },
    {
      "epoch": 0.09613,
      "grad_norm": 1.0580108312474177,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 9613
    },
    {
      "epoch": 0.09614,
      "grad_norm": 1.0472858692305076,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 9614
    },
    {
      "epoch": 0.09615,
      "grad_norm": 1.2366350300452777,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 9615
    },
    {
      "epoch": 0.09616,
      "grad_norm": 0.8816270210973588,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9616
    },
    {
      "epoch": 0.09617,
      "grad_norm": 0.8722350756954568,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9617
    },
    {
      "epoch": 0.09618,
      "grad_norm": 0.9067366940665751,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 9618
    },
    {
      "epoch": 0.09619,
      "grad_norm": 1.0481783137642813,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 9619
    },
    {
      "epoch": 0.0962,
      "grad_norm": 1.015015664502483,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 9620
    },
    {
      "epoch": 0.09621,
      "grad_norm": 1.057917176675139,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 9621
    },
    {
      "epoch": 0.09622,
      "grad_norm": 1.062674626405321,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 9622
    },
    {
      "epoch": 0.09623,
      "grad_norm": 1.0840254275116858,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 9623
    },
    {
      "epoch": 0.09624,
      "grad_norm": 0.921050877590142,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9624
    },
    {
      "epoch": 0.09625,
      "grad_norm": 0.9745687452474494,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9625
    },
    {
      "epoch": 0.09626,
      "grad_norm": 1.0761942857238649,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 9626
    },
    {
      "epoch": 0.09627,
      "grad_norm": 1.2836132658639128,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 9627
    },
    {
      "epoch": 0.09628,
      "grad_norm": 1.2249241628868797,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 9628
    },
    {
      "epoch": 0.09629,
      "grad_norm": 0.8425877511796831,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 9629
    },
    {
      "epoch": 0.0963,
      "grad_norm": 0.8336489764607532,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 9630
    },
    {
      "epoch": 0.09631,
      "grad_norm": 0.9683817459934598,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 9631
    },
    {
      "epoch": 0.09632,
      "grad_norm": 1.1120343603001173,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 9632
    },
    {
      "epoch": 0.09633,
      "grad_norm": 0.9976490542448117,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 9633
    },
    {
      "epoch": 0.09634,
      "grad_norm": 1.0472125777692542,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 9634
    },
    {
      "epoch": 0.09635,
      "grad_norm": 1.046705158707178,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 9635
    },
    {
      "epoch": 0.09636,
      "grad_norm": 1.1240079021749696,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 9636
    },
    {
      "epoch": 0.09637,
      "grad_norm": 0.9390395253897545,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 9637
    },
    {
      "epoch": 0.09638,
      "grad_norm": 0.8986737735751393,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 9638
    },
    {
      "epoch": 0.09639,
      "grad_norm": 0.9270217680975166,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 9639
    },
    {
      "epoch": 0.0964,
      "grad_norm": 0.9594265746905929,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 9640
    },
    {
      "epoch": 0.09641,
      "grad_norm": 1.0246638935142558,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 9641
    },
    {
      "epoch": 0.09642,
      "grad_norm": 1.1350695971418943,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 9642
    },
    {
      "epoch": 0.09643,
      "grad_norm": 1.058296721974063,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 9643
    },
    {
      "epoch": 0.09644,
      "grad_norm": 1.1412333449738055,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 9644
    },
    {
      "epoch": 0.09645,
      "grad_norm": 1.0255174315928441,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 9645
    },
    {
      "epoch": 0.09646,
      "grad_norm": 1.1981325820421778,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9646
    },
    {
      "epoch": 0.09647,
      "grad_norm": 0.96198785938754,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 9647
    },
    {
      "epoch": 0.09648,
      "grad_norm": 0.9103300780359126,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 9648
    },
    {
      "epoch": 0.09649,
      "grad_norm": 1.0146850485441834,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 9649
    },
    {
      "epoch": 0.0965,
      "grad_norm": 1.1014496175687365,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 9650
    },
    {
      "epoch": 0.09651,
      "grad_norm": 0.9799556586854834,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 9651
    },
    {
      "epoch": 0.09652,
      "grad_norm": 0.8859085759414587,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9652
    },
    {
      "epoch": 0.09653,
      "grad_norm": 0.9436219326106128,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 9653
    },
    {
      "epoch": 0.09654,
      "grad_norm": 1.049922096432769,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 9654
    },
    {
      "epoch": 0.09655,
      "grad_norm": 1.0512634494244577,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9655
    },
    {
      "epoch": 0.09656,
      "grad_norm": 1.2561052980696712,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 9656
    },
    {
      "epoch": 0.09657,
      "grad_norm": 0.9464113750920866,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9657
    },
    {
      "epoch": 0.09658,
      "grad_norm": 0.8936277941328696,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9658
    },
    {
      "epoch": 0.09659,
      "grad_norm": 0.9074412530038821,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 9659
    },
    {
      "epoch": 0.0966,
      "grad_norm": 0.959500742463755,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 9660
    },
    {
      "epoch": 0.09661,
      "grad_norm": 1.025665764705048,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9661
    },
    {
      "epoch": 0.09662,
      "grad_norm": 0.9870720584477939,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 9662
    },
    {
      "epoch": 0.09663,
      "grad_norm": 1.1394465371192255,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 9663
    },
    {
      "epoch": 0.09664,
      "grad_norm": 1.0049834681547085,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 9664
    },
    {
      "epoch": 0.09665,
      "grad_norm": 1.1993551495956545,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9665
    },
    {
      "epoch": 0.09666,
      "grad_norm": 1.2099942084141782,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9666
    },
    {
      "epoch": 0.09667,
      "grad_norm": 0.9917211003714878,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 9667
    },
    {
      "epoch": 0.09668,
      "grad_norm": 1.1434647616000775,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 9668
    },
    {
      "epoch": 0.09669,
      "grad_norm": 0.8296204462737752,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9669
    },
    {
      "epoch": 0.0967,
      "grad_norm": 0.8774506033092024,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9670
    },
    {
      "epoch": 0.09671,
      "grad_norm": 0.9598356684861605,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 9671
    },
    {
      "epoch": 0.09672,
      "grad_norm": 1.1282693171893508,
      "learning_rate": 0.003,
      "loss": 4.1169,
      "step": 9672
    },
    {
      "epoch": 0.09673,
      "grad_norm": 0.9782521798986386,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 9673
    },
    {
      "epoch": 0.09674,
      "grad_norm": 1.0774337255597977,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 9674
    },
    {
      "epoch": 0.09675,
      "grad_norm": 1.1642571154544263,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 9675
    },
    {
      "epoch": 0.09676,
      "grad_norm": 1.163816620331313,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 9676
    },
    {
      "epoch": 0.09677,
      "grad_norm": 0.9786912961895912,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 9677
    },
    {
      "epoch": 0.09678,
      "grad_norm": 1.073702735587886,
      "learning_rate": 0.003,
      "loss": 4.1213,
      "step": 9678
    },
    {
      "epoch": 0.09679,
      "grad_norm": 1.060204601252453,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 9679
    },
    {
      "epoch": 0.0968,
      "grad_norm": 1.0203738671671754,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 9680
    },
    {
      "epoch": 0.09681,
      "grad_norm": 1.047329246570657,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9681
    },
    {
      "epoch": 0.09682,
      "grad_norm": 1.0233270765342837,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 9682
    },
    {
      "epoch": 0.09683,
      "grad_norm": 1.1862154773782052,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 9683
    },
    {
      "epoch": 0.09684,
      "grad_norm": 0.9846506225870897,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 9684
    },
    {
      "epoch": 0.09685,
      "grad_norm": 0.9516168976016748,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 9685
    },
    {
      "epoch": 0.09686,
      "grad_norm": 1.003400150329955,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 9686
    },
    {
      "epoch": 0.09687,
      "grad_norm": 1.231150381428762,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 9687
    },
    {
      "epoch": 0.09688,
      "grad_norm": 0.8015979155657765,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 9688
    },
    {
      "epoch": 0.09689,
      "grad_norm": 0.7299334181220091,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9689
    },
    {
      "epoch": 0.0969,
      "grad_norm": 0.7282724727736201,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 9690
    },
    {
      "epoch": 0.09691,
      "grad_norm": 0.7638905566157688,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 9691
    },
    {
      "epoch": 0.09692,
      "grad_norm": 0.96832199777829,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 9692
    },
    {
      "epoch": 0.09693,
      "grad_norm": 1.1902987491844759,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 9693
    },
    {
      "epoch": 0.09694,
      "grad_norm": 0.8606385726141452,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 9694
    },
    {
      "epoch": 0.09695,
      "grad_norm": 0.8380888636245558,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9695
    },
    {
      "epoch": 0.09696,
      "grad_norm": 0.8309556717207915,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 9696
    },
    {
      "epoch": 0.09697,
      "grad_norm": 1.1345771693009117,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 9697
    },
    {
      "epoch": 0.09698,
      "grad_norm": 1.397021456563483,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 9698
    },
    {
      "epoch": 0.09699,
      "grad_norm": 0.8546336497231758,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 9699
    },
    {
      "epoch": 0.097,
      "grad_norm": 0.8327596035211329,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 9700
    },
    {
      "epoch": 0.09701,
      "grad_norm": 1.1146501724689344,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 9701
    },
    {
      "epoch": 0.09702,
      "grad_norm": 0.9998326345834255,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 9702
    },
    {
      "epoch": 0.09703,
      "grad_norm": 1.1440436859408962,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 9703
    },
    {
      "epoch": 0.09704,
      "grad_norm": 1.1173001326683785,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 9704
    },
    {
      "epoch": 0.09705,
      "grad_norm": 1.0523970352233503,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 9705
    },
    {
      "epoch": 0.09706,
      "grad_norm": 1.1713762482483892,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 9706
    },
    {
      "epoch": 0.09707,
      "grad_norm": 0.8804097816544441,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 9707
    },
    {
      "epoch": 0.09708,
      "grad_norm": 1.0194966549341173,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 9708
    },
    {
      "epoch": 0.09709,
      "grad_norm": 0.9694301047009535,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 9709
    },
    {
      "epoch": 0.0971,
      "grad_norm": 0.9481709323161568,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 9710
    },
    {
      "epoch": 0.09711,
      "grad_norm": 1.172731065283466,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 9711
    },
    {
      "epoch": 0.09712,
      "grad_norm": 0.8011684036455335,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 9712
    },
    {
      "epoch": 0.09713,
      "grad_norm": 0.9173643441590453,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 9713
    },
    {
      "epoch": 0.09714,
      "grad_norm": 1.0566896957902046,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 9714
    },
    {
      "epoch": 0.09715,
      "grad_norm": 1.0004516075644887,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 9715
    },
    {
      "epoch": 0.09716,
      "grad_norm": 1.1696605958430277,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 9716
    },
    {
      "epoch": 0.09717,
      "grad_norm": 1.1191433852342278,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 9717
    },
    {
      "epoch": 0.09718,
      "grad_norm": 1.1011799234185506,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 9718
    },
    {
      "epoch": 0.09719,
      "grad_norm": 1.1663962151665566,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9719
    },
    {
      "epoch": 0.0972,
      "grad_norm": 1.1267499527720306,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 9720
    },
    {
      "epoch": 0.09721,
      "grad_norm": 0.9631878849174689,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 9721
    },
    {
      "epoch": 0.09722,
      "grad_norm": 1.1131135286770764,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 9722
    },
    {
      "epoch": 0.09723,
      "grad_norm": 1.1406866316885176,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 9723
    },
    {
      "epoch": 0.09724,
      "grad_norm": 0.9896761239836782,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 9724
    },
    {
      "epoch": 0.09725,
      "grad_norm": 1.0608334963018875,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 9725
    },
    {
      "epoch": 0.09726,
      "grad_norm": 1.0223160222808552,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 9726
    },
    {
      "epoch": 0.09727,
      "grad_norm": 1.1275241052880784,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9727
    },
    {
      "epoch": 0.09728,
      "grad_norm": 0.8895537961162207,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 9728
    },
    {
      "epoch": 0.09729,
      "grad_norm": 0.9705906071167186,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9729
    },
    {
      "epoch": 0.0973,
      "grad_norm": 0.9678791008979821,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9730
    },
    {
      "epoch": 0.09731,
      "grad_norm": 0.9789358792375261,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 9731
    },
    {
      "epoch": 0.09732,
      "grad_norm": 1.1262165489255724,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9732
    },
    {
      "epoch": 0.09733,
      "grad_norm": 0.9935366558924056,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 9733
    },
    {
      "epoch": 0.09734,
      "grad_norm": 0.9093887102489477,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 9734
    },
    {
      "epoch": 0.09735,
      "grad_norm": 0.8527078328559802,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9735
    },
    {
      "epoch": 0.09736,
      "grad_norm": 0.9159904912201342,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 9736
    },
    {
      "epoch": 0.09737,
      "grad_norm": 1.0697597475828968,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 9737
    },
    {
      "epoch": 0.09738,
      "grad_norm": 1.1285651318486103,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 9738
    },
    {
      "epoch": 0.09739,
      "grad_norm": 0.9693482906116275,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 9739
    },
    {
      "epoch": 0.0974,
      "grad_norm": 0.8804291449230246,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9740
    },
    {
      "epoch": 0.09741,
      "grad_norm": 0.8841645343019247,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 9741
    },
    {
      "epoch": 0.09742,
      "grad_norm": 0.9553532279141466,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9742
    },
    {
      "epoch": 0.09743,
      "grad_norm": 0.9782366425572494,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9743
    },
    {
      "epoch": 0.09744,
      "grad_norm": 0.9869511307484383,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 9744
    },
    {
      "epoch": 0.09745,
      "grad_norm": 1.2099299560104868,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 9745
    },
    {
      "epoch": 0.09746,
      "grad_norm": 1.4209495817651363,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 9746
    },
    {
      "epoch": 0.09747,
      "grad_norm": 0.8357217470404129,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 9747
    },
    {
      "epoch": 0.09748,
      "grad_norm": 0.8752427718006248,
      "learning_rate": 0.003,
      "loss": 4.1209,
      "step": 9748
    },
    {
      "epoch": 0.09749,
      "grad_norm": 1.1935288313099783,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 9749
    },
    {
      "epoch": 0.0975,
      "grad_norm": 0.9694278343181028,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 9750
    },
    {
      "epoch": 0.09751,
      "grad_norm": 1.1606139746617636,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 9751
    },
    {
      "epoch": 0.09752,
      "grad_norm": 1.06007198810202,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9752
    },
    {
      "epoch": 0.09753,
      "grad_norm": 1.2442118361414638,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 9753
    },
    {
      "epoch": 0.09754,
      "grad_norm": 0.8708103331423904,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 9754
    },
    {
      "epoch": 0.09755,
      "grad_norm": 0.8902103561683131,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 9755
    },
    {
      "epoch": 0.09756,
      "grad_norm": 0.8312788620338242,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 9756
    },
    {
      "epoch": 0.09757,
      "grad_norm": 0.9879969036527848,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9757
    },
    {
      "epoch": 0.09758,
      "grad_norm": 1.1228401769540215,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 9758
    },
    {
      "epoch": 0.09759,
      "grad_norm": 1.0242499113055676,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9759
    },
    {
      "epoch": 0.0976,
      "grad_norm": 1.2189767248747545,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9760
    },
    {
      "epoch": 0.09761,
      "grad_norm": 1.0854332954217905,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 9761
    },
    {
      "epoch": 0.09762,
      "grad_norm": 0.9142472209749145,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 9762
    },
    {
      "epoch": 0.09763,
      "grad_norm": 0.8999514242684113,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 9763
    },
    {
      "epoch": 0.09764,
      "grad_norm": 0.9671425605430154,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 9764
    },
    {
      "epoch": 0.09765,
      "grad_norm": 1.0849879456729752,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 9765
    },
    {
      "epoch": 0.09766,
      "grad_norm": 1.0329063206869054,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 9766
    },
    {
      "epoch": 0.09767,
      "grad_norm": 1.0178074898813942,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 9767
    },
    {
      "epoch": 0.09768,
      "grad_norm": 1.2302824794252587,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 9768
    },
    {
      "epoch": 0.09769,
      "grad_norm": 0.8998354813027947,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 9769
    },
    {
      "epoch": 0.0977,
      "grad_norm": 0.9191096965901105,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 9770
    },
    {
      "epoch": 0.09771,
      "grad_norm": 1.2023215373833045,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 9771
    },
    {
      "epoch": 0.09772,
      "grad_norm": 1.221541578031842,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 9772
    },
    {
      "epoch": 0.09773,
      "grad_norm": 1.0360835407587528,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 9773
    },
    {
      "epoch": 0.09774,
      "grad_norm": 0.9446876117619419,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 9774
    },
    {
      "epoch": 0.09775,
      "grad_norm": 1.0303074439810562,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 9775
    },
    {
      "epoch": 0.09776,
      "grad_norm": 1.1253618313034983,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 9776
    },
    {
      "epoch": 0.09777,
      "grad_norm": 0.9439094392869328,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9777
    },
    {
      "epoch": 0.09778,
      "grad_norm": 0.9618684632253788,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 9778
    },
    {
      "epoch": 0.09779,
      "grad_norm": 0.9811748296593858,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 9779
    },
    {
      "epoch": 0.0978,
      "grad_norm": 1.1858634020984171,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 9780
    },
    {
      "epoch": 0.09781,
      "grad_norm": 0.9338661709362392,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9781
    },
    {
      "epoch": 0.09782,
      "grad_norm": 0.9332022515801041,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 9782
    },
    {
      "epoch": 0.09783,
      "grad_norm": 1.1071223925652374,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 9783
    },
    {
      "epoch": 0.09784,
      "grad_norm": 1.2093941427447275,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 9784
    },
    {
      "epoch": 0.09785,
      "grad_norm": 1.102823929587096,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 9785
    },
    {
      "epoch": 0.09786,
      "grad_norm": 1.0035787179191338,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 9786
    },
    {
      "epoch": 0.09787,
      "grad_norm": 1.078102955585403,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 9787
    },
    {
      "epoch": 0.09788,
      "grad_norm": 1.0629142159355294,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9788
    },
    {
      "epoch": 0.09789,
      "grad_norm": 1.0804726075419981,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 9789
    },
    {
      "epoch": 0.0979,
      "grad_norm": 1.1720178170225048,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 9790
    },
    {
      "epoch": 0.09791,
      "grad_norm": 1.1235469017472899,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 9791
    },
    {
      "epoch": 0.09792,
      "grad_norm": 1.2149830845028162,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 9792
    },
    {
      "epoch": 0.09793,
      "grad_norm": 0.9420274703508541,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 9793
    },
    {
      "epoch": 0.09794,
      "grad_norm": 1.033046463245827,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 9794
    },
    {
      "epoch": 0.09795,
      "grad_norm": 1.1348495739940019,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 9795
    },
    {
      "epoch": 0.09796,
      "grad_norm": 0.9757339572958833,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 9796
    },
    {
      "epoch": 0.09797,
      "grad_norm": 1.0351452973168425,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 9797
    },
    {
      "epoch": 0.09798,
      "grad_norm": 0.9782960688592283,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9798
    },
    {
      "epoch": 0.09799,
      "grad_norm": 1.0164438910147562,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 9799
    },
    {
      "epoch": 0.098,
      "grad_norm": 1.1163964919065634,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9800
    },
    {
      "epoch": 0.09801,
      "grad_norm": 1.0433207564204117,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 9801
    },
    {
      "epoch": 0.09802,
      "grad_norm": 1.006414968625939,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 9802
    },
    {
      "epoch": 0.09803,
      "grad_norm": 1.139600456125198,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 9803
    },
    {
      "epoch": 0.09804,
      "grad_norm": 1.0284067990368186,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 9804
    },
    {
      "epoch": 0.09805,
      "grad_norm": 1.1861908924400755,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 9805
    },
    {
      "epoch": 0.09806,
      "grad_norm": 0.8349801162214475,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 9806
    },
    {
      "epoch": 0.09807,
      "grad_norm": 0.8131825023363899,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9807
    },
    {
      "epoch": 0.09808,
      "grad_norm": 0.8785674410167105,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 9808
    },
    {
      "epoch": 0.09809,
      "grad_norm": 1.134448089593032,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 9809
    },
    {
      "epoch": 0.0981,
      "grad_norm": 0.9440022631359163,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 9810
    },
    {
      "epoch": 0.09811,
      "grad_norm": 0.8747418766506534,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 9811
    },
    {
      "epoch": 0.09812,
      "grad_norm": 1.0271563855822783,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 9812
    },
    {
      "epoch": 0.09813,
      "grad_norm": 1.268250855234208,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 9813
    },
    {
      "epoch": 0.09814,
      "grad_norm": 0.9423284371731849,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 9814
    },
    {
      "epoch": 0.09815,
      "grad_norm": 0.8587272806085712,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 9815
    },
    {
      "epoch": 0.09816,
      "grad_norm": 0.8898000553630154,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 9816
    },
    {
      "epoch": 0.09817,
      "grad_norm": 0.8967370159090363,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 9817
    },
    {
      "epoch": 0.09818,
      "grad_norm": 0.9683907679710231,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 9818
    },
    {
      "epoch": 0.09819,
      "grad_norm": 1.1611928616933378,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 9819
    },
    {
      "epoch": 0.0982,
      "grad_norm": 0.9627567145629911,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9820
    },
    {
      "epoch": 0.09821,
      "grad_norm": 1.0864907954219507,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 9821
    },
    {
      "epoch": 0.09822,
      "grad_norm": 1.3249140720038677,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 9822
    },
    {
      "epoch": 0.09823,
      "grad_norm": 0.8605074850339292,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 9823
    },
    {
      "epoch": 0.09824,
      "grad_norm": 0.9566425600412122,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 9824
    },
    {
      "epoch": 0.09825,
      "grad_norm": 1.0233355222981693,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9825
    },
    {
      "epoch": 0.09826,
      "grad_norm": 1.0786561064635678,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 9826
    },
    {
      "epoch": 0.09827,
      "grad_norm": 1.0514971996142433,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 9827
    },
    {
      "epoch": 0.09828,
      "grad_norm": 0.8623547190602233,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 9828
    },
    {
      "epoch": 0.09829,
      "grad_norm": 0.8781618673749649,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9829
    },
    {
      "epoch": 0.0983,
      "grad_norm": 0.8682088320906876,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 9830
    },
    {
      "epoch": 0.09831,
      "grad_norm": 0.8891553003739594,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 9831
    },
    {
      "epoch": 0.09832,
      "grad_norm": 1.0895544072278256,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 9832
    },
    {
      "epoch": 0.09833,
      "grad_norm": 1.1615633672113128,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 9833
    },
    {
      "epoch": 0.09834,
      "grad_norm": 0.9838275297396515,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 9834
    },
    {
      "epoch": 0.09835,
      "grad_norm": 1.0531808921035346,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 9835
    },
    {
      "epoch": 0.09836,
      "grad_norm": 1.1225625590156698,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 9836
    },
    {
      "epoch": 0.09837,
      "grad_norm": 1.047559273171705,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 9837
    },
    {
      "epoch": 0.09838,
      "grad_norm": 1.0625120495027733,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9838
    },
    {
      "epoch": 0.09839,
      "grad_norm": 1.1447218613408165,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 9839
    },
    {
      "epoch": 0.0984,
      "grad_norm": 0.9466230883186256,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 9840
    },
    {
      "epoch": 0.09841,
      "grad_norm": 0.9948433758958204,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 9841
    },
    {
      "epoch": 0.09842,
      "grad_norm": 1.0916342093786202,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9842
    },
    {
      "epoch": 0.09843,
      "grad_norm": 1.0810547158105994,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 9843
    },
    {
      "epoch": 0.09844,
      "grad_norm": 1.0666836813512826,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 9844
    },
    {
      "epoch": 0.09845,
      "grad_norm": 0.9364969036871016,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 9845
    },
    {
      "epoch": 0.09846,
      "grad_norm": 1.247977198047358,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 9846
    },
    {
      "epoch": 0.09847,
      "grad_norm": 1.0983305121758822,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 9847
    },
    {
      "epoch": 0.09848,
      "grad_norm": 0.9241457854629477,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 9848
    },
    {
      "epoch": 0.09849,
      "grad_norm": 0.8899279219685088,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 9849
    },
    {
      "epoch": 0.0985,
      "grad_norm": 0.9422816629698554,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 9850
    },
    {
      "epoch": 0.09851,
      "grad_norm": 0.9730853740755129,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9851
    },
    {
      "epoch": 0.09852,
      "grad_norm": 1.0334597856537537,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9852
    },
    {
      "epoch": 0.09853,
      "grad_norm": 1.026542924409616,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 9853
    },
    {
      "epoch": 0.09854,
      "grad_norm": 1.191401021749489,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 9854
    },
    {
      "epoch": 0.09855,
      "grad_norm": 1.059077578402062,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 9855
    },
    {
      "epoch": 0.09856,
      "grad_norm": 1.0498244668875298,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 9856
    },
    {
      "epoch": 0.09857,
      "grad_norm": 1.091092729660013,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9857
    },
    {
      "epoch": 0.09858,
      "grad_norm": 0.9250542311162983,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 9858
    },
    {
      "epoch": 0.09859,
      "grad_norm": 1.0800039787638633,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 9859
    },
    {
      "epoch": 0.0986,
      "grad_norm": 1.041419214223572,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 9860
    },
    {
      "epoch": 0.09861,
      "grad_norm": 0.9982888817464664,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 9861
    },
    {
      "epoch": 0.09862,
      "grad_norm": 1.1239145862062716,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 9862
    },
    {
      "epoch": 0.09863,
      "grad_norm": 0.9592019467214191,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9863
    },
    {
      "epoch": 0.09864,
      "grad_norm": 1.0352169689669506,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 9864
    },
    {
      "epoch": 0.09865,
      "grad_norm": 1.181841188508446,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 9865
    },
    {
      "epoch": 0.09866,
      "grad_norm": 0.9389016144596009,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 9866
    },
    {
      "epoch": 0.09867,
      "grad_norm": 0.8893717783577848,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 9867
    },
    {
      "epoch": 0.09868,
      "grad_norm": 0.9999511334783509,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 9868
    },
    {
      "epoch": 0.09869,
      "grad_norm": 1.3766579673159203,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 9869
    },
    {
      "epoch": 0.0987,
      "grad_norm": 0.8413732379628507,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 9870
    },
    {
      "epoch": 0.09871,
      "grad_norm": 0.7967590725793811,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 9871
    },
    {
      "epoch": 0.09872,
      "grad_norm": 0.941707830097863,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 9872
    },
    {
      "epoch": 0.09873,
      "grad_norm": 0.9855273751841129,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 9873
    },
    {
      "epoch": 0.09874,
      "grad_norm": 0.8989862551825634,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 9874
    },
    {
      "epoch": 0.09875,
      "grad_norm": 0.968402761982066,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 9875
    },
    {
      "epoch": 0.09876,
      "grad_norm": 1.0298264328815836,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 9876
    },
    {
      "epoch": 0.09877,
      "grad_norm": 1.0715370435837641,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 9877
    },
    {
      "epoch": 0.09878,
      "grad_norm": 0.9061930517874249,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 9878
    },
    {
      "epoch": 0.09879,
      "grad_norm": 1.031786492000977,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 9879
    },
    {
      "epoch": 0.0988,
      "grad_norm": 1.3228565671620516,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 9880
    },
    {
      "epoch": 0.09881,
      "grad_norm": 1.035790465396129,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 9881
    },
    {
      "epoch": 0.09882,
      "grad_norm": 1.1699610602457922,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 9882
    },
    {
      "epoch": 0.09883,
      "grad_norm": 1.0555690693909043,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 9883
    },
    {
      "epoch": 0.09884,
      "grad_norm": 1.0713445946006626,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 9884
    },
    {
      "epoch": 0.09885,
      "grad_norm": 1.0685798383445284,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 9885
    },
    {
      "epoch": 0.09886,
      "grad_norm": 1.3401385584546417,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 9886
    },
    {
      "epoch": 0.09887,
      "grad_norm": 0.9107037359332328,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 9887
    },
    {
      "epoch": 0.09888,
      "grad_norm": 1.1599530666035878,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 9888
    },
    {
      "epoch": 0.09889,
      "grad_norm": 1.0285469001260248,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 9889
    },
    {
      "epoch": 0.0989,
      "grad_norm": 0.9967838845583898,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9890
    },
    {
      "epoch": 0.09891,
      "grad_norm": 1.2711079083653072,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 9891
    },
    {
      "epoch": 0.09892,
      "grad_norm": 0.9072541972447289,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 9892
    },
    {
      "epoch": 0.09893,
      "grad_norm": 0.9675167406986386,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 9893
    },
    {
      "epoch": 0.09894,
      "grad_norm": 0.9826844070764257,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 9894
    },
    {
      "epoch": 0.09895,
      "grad_norm": 1.1555014469207152,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 9895
    },
    {
      "epoch": 0.09896,
      "grad_norm": 0.8705076015969703,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 9896
    },
    {
      "epoch": 0.09897,
      "grad_norm": 1.0606028357144373,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 9897
    },
    {
      "epoch": 0.09898,
      "grad_norm": 1.0431107749585196,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 9898
    },
    {
      "epoch": 0.09899,
      "grad_norm": 1.1940626392877045,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 9899
    },
    {
      "epoch": 0.099,
      "grad_norm": 1.0231246381632437,
      "learning_rate": 0.003,
      "loss": 4.1123,
      "step": 9900
    },
    {
      "epoch": 0.09901,
      "grad_norm": 1.1622210994726128,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9901
    },
    {
      "epoch": 0.09902,
      "grad_norm": 0.9443388666248479,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 9902
    },
    {
      "epoch": 0.09903,
      "grad_norm": 1.0384709982323672,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 9903
    },
    {
      "epoch": 0.09904,
      "grad_norm": 0.9723894387638847,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 9904
    },
    {
      "epoch": 0.09905,
      "grad_norm": 1.1523154218331038,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 9905
    },
    {
      "epoch": 0.09906,
      "grad_norm": 1.0547854344378345,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 9906
    },
    {
      "epoch": 0.09907,
      "grad_norm": 1.182257869541173,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 9907
    },
    {
      "epoch": 0.09908,
      "grad_norm": 1.161665767642822,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 9908
    },
    {
      "epoch": 0.09909,
      "grad_norm": 0.9087642677557614,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 9909
    },
    {
      "epoch": 0.0991,
      "grad_norm": 1.0313878368391898,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 9910
    },
    {
      "epoch": 0.09911,
      "grad_norm": 1.2847136513646835,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 9911
    },
    {
      "epoch": 0.09912,
      "grad_norm": 0.8904708929002996,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 9912
    },
    {
      "epoch": 0.09913,
      "grad_norm": 1.0112171424842842,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9913
    },
    {
      "epoch": 0.09914,
      "grad_norm": 0.9794967717604602,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 9914
    },
    {
      "epoch": 0.09915,
      "grad_norm": 0.9423290437307489,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 9915
    },
    {
      "epoch": 0.09916,
      "grad_norm": 0.950907824636009,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 9916
    },
    {
      "epoch": 0.09917,
      "grad_norm": 1.0219884227601401,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 9917
    },
    {
      "epoch": 0.09918,
      "grad_norm": 1.032340531956216,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 9918
    },
    {
      "epoch": 0.09919,
      "grad_norm": 0.9004643268011698,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 9919
    },
    {
      "epoch": 0.0992,
      "grad_norm": 1.0606211698418486,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 9920
    },
    {
      "epoch": 0.09921,
      "grad_norm": 1.3653476752568316,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 9921
    },
    {
      "epoch": 0.09922,
      "grad_norm": 0.8159267949493182,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9922
    },
    {
      "epoch": 0.09923,
      "grad_norm": 0.8093245515380362,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 9923
    },
    {
      "epoch": 0.09924,
      "grad_norm": 0.9136880506548397,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 9924
    },
    {
      "epoch": 0.09925,
      "grad_norm": 0.8854387914223482,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 9925
    },
    {
      "epoch": 0.09926,
      "grad_norm": 1.0055098961573377,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 9926
    },
    {
      "epoch": 0.09927,
      "grad_norm": 0.9799629554615119,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 9927
    },
    {
      "epoch": 0.09928,
      "grad_norm": 1.097232317085835,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 9928
    },
    {
      "epoch": 0.09929,
      "grad_norm": 1.2181303331479505,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 9929
    },
    {
      "epoch": 0.0993,
      "grad_norm": 1.41744952211023,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 9930
    },
    {
      "epoch": 0.09931,
      "grad_norm": 0.8845168557408566,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 9931
    },
    {
      "epoch": 0.09932,
      "grad_norm": 0.9049290312336111,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 9932
    },
    {
      "epoch": 0.09933,
      "grad_norm": 1.1600752655306843,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 9933
    },
    {
      "epoch": 0.09934,
      "grad_norm": 1.0319033481213138,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 9934
    },
    {
      "epoch": 0.09935,
      "grad_norm": 1.2341422592136662,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 9935
    },
    {
      "epoch": 0.09936,
      "grad_norm": 1.0628728338770057,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 9936
    },
    {
      "epoch": 0.09937,
      "grad_norm": 0.8801535749313547,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 9937
    },
    {
      "epoch": 0.09938,
      "grad_norm": 0.8196753503087768,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 9938
    },
    {
      "epoch": 0.09939,
      "grad_norm": 0.9592522974642151,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 9939
    },
    {
      "epoch": 0.0994,
      "grad_norm": 1.0721297761126327,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 9940
    },
    {
      "epoch": 0.09941,
      "grad_norm": 1.0763492581657323,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 9941
    },
    {
      "epoch": 0.09942,
      "grad_norm": 1.019164330036407,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 9942
    },
    {
      "epoch": 0.09943,
      "grad_norm": 1.2248432437361976,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 9943
    },
    {
      "epoch": 0.09944,
      "grad_norm": 1.0747625128630949,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 9944
    },
    {
      "epoch": 0.09945,
      "grad_norm": 0.9570097418411981,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 9945
    },
    {
      "epoch": 0.09946,
      "grad_norm": 0.9943121246735858,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 9946
    },
    {
      "epoch": 0.09947,
      "grad_norm": 1.2372355073906356,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 9947
    },
    {
      "epoch": 0.09948,
      "grad_norm": 0.8502611811001046,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 9948
    },
    {
      "epoch": 0.09949,
      "grad_norm": 0.9231335871633486,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 9949
    },
    {
      "epoch": 0.0995,
      "grad_norm": 1.0621020837735249,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 9950
    },
    {
      "epoch": 0.09951,
      "grad_norm": 1.1092075040539509,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 9951
    },
    {
      "epoch": 0.09952,
      "grad_norm": 1.1180875245705675,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 9952
    },
    {
      "epoch": 0.09953,
      "grad_norm": 1.0583089223558193,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 9953
    },
    {
      "epoch": 0.09954,
      "grad_norm": 1.3656489755235695,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 9954
    },
    {
      "epoch": 0.09955,
      "grad_norm": 0.9216137223262066,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 9955
    },
    {
      "epoch": 0.09956,
      "grad_norm": 0.8331207661096087,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 9956
    },
    {
      "epoch": 0.09957,
      "grad_norm": 0.9411322885803222,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 9957
    },
    {
      "epoch": 0.09958,
      "grad_norm": 0.9765577007820736,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 9958
    },
    {
      "epoch": 0.09959,
      "grad_norm": 0.9833086407810087,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 9959
    },
    {
      "epoch": 0.0996,
      "grad_norm": 1.208131243797863,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 9960
    },
    {
      "epoch": 0.09961,
      "grad_norm": 0.9410870086120185,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 9961
    },
    {
      "epoch": 0.09962,
      "grad_norm": 0.9901662898897093,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 9962
    },
    {
      "epoch": 0.09963,
      "grad_norm": 1.0595283366933754,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 9963
    },
    {
      "epoch": 0.09964,
      "grad_norm": 1.1230131079391967,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 9964
    },
    {
      "epoch": 0.09965,
      "grad_norm": 1.408656633378828,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 9965
    },
    {
      "epoch": 0.09966,
      "grad_norm": 1.0494066026723656,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 9966
    },
    {
      "epoch": 0.09967,
      "grad_norm": 1.0510949632983806,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 9967
    },
    {
      "epoch": 0.09968,
      "grad_norm": 0.9184349300356729,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9968
    },
    {
      "epoch": 0.09969,
      "grad_norm": 0.912619317530374,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 9969
    },
    {
      "epoch": 0.0997,
      "grad_norm": 0.9253745452194618,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 9970
    },
    {
      "epoch": 0.09971,
      "grad_norm": 0.9414528249595521,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 9971
    },
    {
      "epoch": 0.09972,
      "grad_norm": 1.212415753181934,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 9972
    },
    {
      "epoch": 0.09973,
      "grad_norm": 1.3032259078570259,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 9973
    },
    {
      "epoch": 0.09974,
      "grad_norm": 0.8834224407852096,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 9974
    },
    {
      "epoch": 0.09975,
      "grad_norm": 1.0257775503141435,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 9975
    },
    {
      "epoch": 0.09976,
      "grad_norm": 1.0027851181438205,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 9976
    },
    {
      "epoch": 0.09977,
      "grad_norm": 1.1054062839337309,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 9977
    },
    {
      "epoch": 0.09978,
      "grad_norm": 0.9983871843931804,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 9978
    },
    {
      "epoch": 0.09979,
      "grad_norm": 1.1469658774556046,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 9979
    },
    {
      "epoch": 0.0998,
      "grad_norm": 1.0417510639490304,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 9980
    },
    {
      "epoch": 0.09981,
      "grad_norm": 0.8997050282680549,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 9981
    },
    {
      "epoch": 0.09982,
      "grad_norm": 0.9795564166461124,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 9982
    },
    {
      "epoch": 0.09983,
      "grad_norm": 1.1265795885230583,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 9983
    },
    {
      "epoch": 0.09984,
      "grad_norm": 1.075755261601754,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 9984
    },
    {
      "epoch": 0.09985,
      "grad_norm": 1.1117966143075422,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 9985
    },
    {
      "epoch": 0.09986,
      "grad_norm": 0.8947693590206178,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 9986
    },
    {
      "epoch": 0.09987,
      "grad_norm": 0.964118687177187,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 9987
    },
    {
      "epoch": 0.09988,
      "grad_norm": 1.1466654575825213,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 9988
    },
    {
      "epoch": 0.09989,
      "grad_norm": 0.9371466526373202,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 9989
    },
    {
      "epoch": 0.0999,
      "grad_norm": 1.1021242856154043,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 9990
    },
    {
      "epoch": 0.09991,
      "grad_norm": 1.4409488657540674,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 9991
    },
    {
      "epoch": 0.09992,
      "grad_norm": 0.9081140092271789,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 9992
    },
    {
      "epoch": 0.09993,
      "grad_norm": 1.0604218435107378,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 9993
    },
    {
      "epoch": 0.09994,
      "grad_norm": 1.2125286721128687,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 9994
    },
    {
      "epoch": 0.09995,
      "grad_norm": 1.0829501150938283,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 9995
    },
    {
      "epoch": 0.09996,
      "grad_norm": 1.0733185017791256,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 9996
    },
    {
      "epoch": 0.09997,
      "grad_norm": 1.021879556888008,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 9997
    },
    {
      "epoch": 0.09998,
      "grad_norm": 1.097873921598905,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 9998
    },
    {
      "epoch": 0.09999,
      "grad_norm": 1.0928565866659528,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 9999
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9146824902563694,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 10000
    },
    {
      "epoch": 0.10001,
      "grad_norm": 1.0268828767480918,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 10001
    },
    {
      "epoch": 0.10002,
      "grad_norm": 0.9456355767793398,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10002
    },
    {
      "epoch": 0.10003,
      "grad_norm": 0.95383283914356,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10003
    },
    {
      "epoch": 0.10004,
      "grad_norm": 1.045290937365292,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 10004
    },
    {
      "epoch": 0.10005,
      "grad_norm": 0.9788174490820064,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10005
    },
    {
      "epoch": 0.10006,
      "grad_norm": 1.1152617835523273,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10006
    },
    {
      "epoch": 0.10007,
      "grad_norm": 1.0774775903166232,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 10007
    },
    {
      "epoch": 0.10008,
      "grad_norm": 1.2010923165312688,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 10008
    },
    {
      "epoch": 0.10009,
      "grad_norm": 0.9615549697604012,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 10009
    },
    {
      "epoch": 0.1001,
      "grad_norm": 0.8606186327961719,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 10010
    },
    {
      "epoch": 0.10011,
      "grad_norm": 0.9641733347424303,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 10011
    },
    {
      "epoch": 0.10012,
      "grad_norm": 1.1026647643286422,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10012
    },
    {
      "epoch": 0.10013,
      "grad_norm": 1.0701035577711429,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 10013
    },
    {
      "epoch": 0.10014,
      "grad_norm": 1.0817395215435701,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 10014
    },
    {
      "epoch": 0.10015,
      "grad_norm": 1.4023794761788895,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 10015
    },
    {
      "epoch": 0.10016,
      "grad_norm": 0.8702730093122781,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 10016
    },
    {
      "epoch": 0.10017,
      "grad_norm": 0.7570220498972725,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 10017
    },
    {
      "epoch": 0.10018,
      "grad_norm": 0.8707038500254759,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 10018
    },
    {
      "epoch": 0.10019,
      "grad_norm": 1.010728470425738,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 10019
    },
    {
      "epoch": 0.1002,
      "grad_norm": 0.9979899408508959,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 10020
    },
    {
      "epoch": 0.10021,
      "grad_norm": 1.0674101631325568,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 10021
    },
    {
      "epoch": 0.10022,
      "grad_norm": 1.0266913018745583,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 10022
    },
    {
      "epoch": 0.10023,
      "grad_norm": 1.0638117707816785,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 10023
    },
    {
      "epoch": 0.10024,
      "grad_norm": 1.133369886133261,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 10024
    },
    {
      "epoch": 0.10025,
      "grad_norm": 1.102307846355811,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10025
    },
    {
      "epoch": 0.10026,
      "grad_norm": 1.3400225966662347,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 10026
    },
    {
      "epoch": 0.10027,
      "grad_norm": 1.0211547903184859,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 10027
    },
    {
      "epoch": 0.10028,
      "grad_norm": 1.1603850478475943,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 10028
    },
    {
      "epoch": 0.10029,
      "grad_norm": 1.148079705243166,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 10029
    },
    {
      "epoch": 0.1003,
      "grad_norm": 0.9117292209524029,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 10030
    },
    {
      "epoch": 0.10031,
      "grad_norm": 0.9445865002921869,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 10031
    },
    {
      "epoch": 0.10032,
      "grad_norm": 0.9161171568886268,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10032
    },
    {
      "epoch": 0.10033,
      "grad_norm": 0.85950044194575,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10033
    },
    {
      "epoch": 0.10034,
      "grad_norm": 0.9072001134697321,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 10034
    },
    {
      "epoch": 0.10035,
      "grad_norm": 1.1474126252804075,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 10035
    },
    {
      "epoch": 0.10036,
      "grad_norm": 0.9761134534605453,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 10036
    },
    {
      "epoch": 0.10037,
      "grad_norm": 1.0237415513918258,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 10037
    },
    {
      "epoch": 0.10038,
      "grad_norm": 1.148438737648086,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 10038
    },
    {
      "epoch": 0.10039,
      "grad_norm": 1.0514373112344126,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10039
    },
    {
      "epoch": 0.1004,
      "grad_norm": 1.1847286970538577,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 10040
    },
    {
      "epoch": 0.10041,
      "grad_norm": 0.8922451988552241,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10041
    },
    {
      "epoch": 0.10042,
      "grad_norm": 0.8397773653104834,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 10042
    },
    {
      "epoch": 0.10043,
      "grad_norm": 0.9089542086627763,
      "learning_rate": 0.003,
      "loss": 4.1133,
      "step": 10043
    },
    {
      "epoch": 0.10044,
      "grad_norm": 1.0073280068294317,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 10044
    },
    {
      "epoch": 0.10045,
      "grad_norm": 1.0440598241093555,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 10045
    },
    {
      "epoch": 0.10046,
      "grad_norm": 1.0485056600329703,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 10046
    },
    {
      "epoch": 0.10047,
      "grad_norm": 1.2429836270576817,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 10047
    },
    {
      "epoch": 0.10048,
      "grad_norm": 1.2106150367464596,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 10048
    },
    {
      "epoch": 0.10049,
      "grad_norm": 1.2553698920040894,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 10049
    },
    {
      "epoch": 0.1005,
      "grad_norm": 1.0468634756369095,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10050
    },
    {
      "epoch": 0.10051,
      "grad_norm": 1.118778052143475,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10051
    },
    {
      "epoch": 0.10052,
      "grad_norm": 1.179218879697393,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 10052
    },
    {
      "epoch": 0.10053,
      "grad_norm": 0.9200813106865441,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 10053
    },
    {
      "epoch": 0.10054,
      "grad_norm": 0.8562476127183147,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 10054
    },
    {
      "epoch": 0.10055,
      "grad_norm": 0.9908634064538296,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 10055
    },
    {
      "epoch": 0.10056,
      "grad_norm": 1.3614880394829219,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 10056
    },
    {
      "epoch": 0.10057,
      "grad_norm": 0.6795058628261522,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 10057
    },
    {
      "epoch": 0.10058,
      "grad_norm": 0.6725766270507677,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10058
    },
    {
      "epoch": 0.10059,
      "grad_norm": 0.7021007029637584,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 10059
    },
    {
      "epoch": 0.1006,
      "grad_norm": 0.7029672751184473,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 10060
    },
    {
      "epoch": 0.10061,
      "grad_norm": 0.86274187466931,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 10061
    },
    {
      "epoch": 0.10062,
      "grad_norm": 1.1476605169788783,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 10062
    },
    {
      "epoch": 0.10063,
      "grad_norm": 1.1736377656589307,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10063
    },
    {
      "epoch": 0.10064,
      "grad_norm": 1.0192130436130324,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10064
    },
    {
      "epoch": 0.10065,
      "grad_norm": 1.2950673831269726,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 10065
    },
    {
      "epoch": 0.10066,
      "grad_norm": 0.8946009368675681,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 10066
    },
    {
      "epoch": 0.10067,
      "grad_norm": 0.9034084243233744,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 10067
    },
    {
      "epoch": 0.10068,
      "grad_norm": 0.9279795526086888,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10068
    },
    {
      "epoch": 0.10069,
      "grad_norm": 0.941069604433777,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 10069
    },
    {
      "epoch": 0.1007,
      "grad_norm": 0.9419306165269437,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 10070
    },
    {
      "epoch": 0.10071,
      "grad_norm": 0.9249349363580369,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10071
    },
    {
      "epoch": 0.10072,
      "grad_norm": 1.0579196675587663,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10072
    },
    {
      "epoch": 0.10073,
      "grad_norm": 1.0641190190848893,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 10073
    },
    {
      "epoch": 0.10074,
      "grad_norm": 1.2438602971074477,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 10074
    },
    {
      "epoch": 0.10075,
      "grad_norm": 1.0564497520727791,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 10075
    },
    {
      "epoch": 0.10076,
      "grad_norm": 1.3090603530576794,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 10076
    },
    {
      "epoch": 0.10077,
      "grad_norm": 0.7533461669269756,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 10077
    },
    {
      "epoch": 0.10078,
      "grad_norm": 0.7061355548387933,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 10078
    },
    {
      "epoch": 0.10079,
      "grad_norm": 0.8644525263853029,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 10079
    },
    {
      "epoch": 0.1008,
      "grad_norm": 1.10452779458274,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 10080
    },
    {
      "epoch": 0.10081,
      "grad_norm": 1.3756511095540078,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 10081
    },
    {
      "epoch": 0.10082,
      "grad_norm": 1.038088737997971,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 10082
    },
    {
      "epoch": 0.10083,
      "grad_norm": 1.1207040034663152,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 10083
    },
    {
      "epoch": 0.10084,
      "grad_norm": 0.850821620028518,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 10084
    },
    {
      "epoch": 0.10085,
      "grad_norm": 1.0748025908828802,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 10085
    },
    {
      "epoch": 0.10086,
      "grad_norm": 1.3549025239319474,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 10086
    },
    {
      "epoch": 0.10087,
      "grad_norm": 1.0747448453376909,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 10087
    },
    {
      "epoch": 0.10088,
      "grad_norm": 1.2456306464923101,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10088
    },
    {
      "epoch": 0.10089,
      "grad_norm": 0.9472976337664214,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 10089
    },
    {
      "epoch": 0.1009,
      "grad_norm": 0.9815403359612819,
      "learning_rate": 0.003,
      "loss": 4.1124,
      "step": 10090
    },
    {
      "epoch": 0.10091,
      "grad_norm": 0.9741684116233433,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 10091
    },
    {
      "epoch": 0.10092,
      "grad_norm": 0.9476855315324259,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 10092
    },
    {
      "epoch": 0.10093,
      "grad_norm": 0.9811508699302622,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10093
    },
    {
      "epoch": 0.10094,
      "grad_norm": 1.1541788163650089,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 10094
    },
    {
      "epoch": 0.10095,
      "grad_norm": 1.1472639269086242,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 10095
    },
    {
      "epoch": 0.10096,
      "grad_norm": 1.0555714689128464,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10096
    },
    {
      "epoch": 0.10097,
      "grad_norm": 0.9083517008773805,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 10097
    },
    {
      "epoch": 0.10098,
      "grad_norm": 0.931319184757005,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 10098
    },
    {
      "epoch": 0.10099,
      "grad_norm": 1.2061065602641496,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 10099
    },
    {
      "epoch": 0.101,
      "grad_norm": 1.0790936601780008,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 10100
    },
    {
      "epoch": 0.10101,
      "grad_norm": 1.049700687920486,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 10101
    },
    {
      "epoch": 0.10102,
      "grad_norm": 1.0588165543231847,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 10102
    },
    {
      "epoch": 0.10103,
      "grad_norm": 0.8785533801322097,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 10103
    },
    {
      "epoch": 0.10104,
      "grad_norm": 0.947148926018248,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10104
    },
    {
      "epoch": 0.10105,
      "grad_norm": 1.0594563152689682,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10105
    },
    {
      "epoch": 0.10106,
      "grad_norm": 1.0225956027823544,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 10106
    },
    {
      "epoch": 0.10107,
      "grad_norm": 1.0308601830350073,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 10107
    },
    {
      "epoch": 0.10108,
      "grad_norm": 1.0350078878450226,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 10108
    },
    {
      "epoch": 0.10109,
      "grad_norm": 1.1686695733740247,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 10109
    },
    {
      "epoch": 0.1011,
      "grad_norm": 1.2115978839317425,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 10110
    },
    {
      "epoch": 0.10111,
      "grad_norm": 0.978402576765035,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 10111
    },
    {
      "epoch": 0.10112,
      "grad_norm": 1.0623646734983812,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 10112
    },
    {
      "epoch": 0.10113,
      "grad_norm": 0.9614458949786193,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 10113
    },
    {
      "epoch": 0.10114,
      "grad_norm": 1.179118562016018,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 10114
    },
    {
      "epoch": 0.10115,
      "grad_norm": 1.0233642500878612,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 10115
    },
    {
      "epoch": 0.10116,
      "grad_norm": 1.00109985164738,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 10116
    },
    {
      "epoch": 0.10117,
      "grad_norm": 1.2153279913183614,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10117
    },
    {
      "epoch": 0.10118,
      "grad_norm": 0.9253105426894055,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 10118
    },
    {
      "epoch": 0.10119,
      "grad_norm": 1.201988642511759,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10119
    },
    {
      "epoch": 0.1012,
      "grad_norm": 0.8945884294792387,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 10120
    },
    {
      "epoch": 0.10121,
      "grad_norm": 0.8959801151367283,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 10121
    },
    {
      "epoch": 0.10122,
      "grad_norm": 1.0389605252243181,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 10122
    },
    {
      "epoch": 0.10123,
      "grad_norm": 1.1853235896574628,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 10123
    },
    {
      "epoch": 0.10124,
      "grad_norm": 1.1175599651017967,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 10124
    },
    {
      "epoch": 0.10125,
      "grad_norm": 1.1730052020802815,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 10125
    },
    {
      "epoch": 0.10126,
      "grad_norm": 1.0540782103922477,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 10126
    },
    {
      "epoch": 0.10127,
      "grad_norm": 1.1471297550451278,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 10127
    },
    {
      "epoch": 0.10128,
      "grad_norm": 1.113435608277784,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10128
    },
    {
      "epoch": 0.10129,
      "grad_norm": 1.1059156980127567,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 10129
    },
    {
      "epoch": 0.1013,
      "grad_norm": 0.8882002679244408,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 10130
    },
    {
      "epoch": 0.10131,
      "grad_norm": 0.9154448816742513,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 10131
    },
    {
      "epoch": 0.10132,
      "grad_norm": 0.9450551054713623,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 10132
    },
    {
      "epoch": 0.10133,
      "grad_norm": 0.9595142116108409,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 10133
    },
    {
      "epoch": 0.10134,
      "grad_norm": 0.9736572446471938,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10134
    },
    {
      "epoch": 0.10135,
      "grad_norm": 1.0390522302798173,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10135
    },
    {
      "epoch": 0.10136,
      "grad_norm": 1.086694019268405,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 10136
    },
    {
      "epoch": 0.10137,
      "grad_norm": 1.4802280932553884,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10137
    },
    {
      "epoch": 0.10138,
      "grad_norm": 0.8935572863107328,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 10138
    },
    {
      "epoch": 0.10139,
      "grad_norm": 0.9129477796554756,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 10139
    },
    {
      "epoch": 0.1014,
      "grad_norm": 1.0712559348135158,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 10140
    },
    {
      "epoch": 0.10141,
      "grad_norm": 1.129298993945068,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 10141
    },
    {
      "epoch": 0.10142,
      "grad_norm": 1.0063431597085504,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 10142
    },
    {
      "epoch": 0.10143,
      "grad_norm": 1.0164759574916866,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 10143
    },
    {
      "epoch": 0.10144,
      "grad_norm": 1.102176942327319,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 10144
    },
    {
      "epoch": 0.10145,
      "grad_norm": 0.8846991416216525,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 10145
    },
    {
      "epoch": 0.10146,
      "grad_norm": 0.7954991760065576,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 10146
    },
    {
      "epoch": 0.10147,
      "grad_norm": 0.9389225672902068,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 10147
    },
    {
      "epoch": 0.10148,
      "grad_norm": 0.9979062071789803,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 10148
    },
    {
      "epoch": 0.10149,
      "grad_norm": 0.9358394152637304,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 10149
    },
    {
      "epoch": 0.1015,
      "grad_norm": 1.2030745401236895,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 10150
    },
    {
      "epoch": 0.10151,
      "grad_norm": 1.19363323179694,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10151
    },
    {
      "epoch": 0.10152,
      "grad_norm": 1.051166326109995,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 10152
    },
    {
      "epoch": 0.10153,
      "grad_norm": 1.0612612415288574,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 10153
    },
    {
      "epoch": 0.10154,
      "grad_norm": 1.0366017630808055,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10154
    },
    {
      "epoch": 0.10155,
      "grad_norm": 1.0829838583007372,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 10155
    },
    {
      "epoch": 0.10156,
      "grad_norm": 1.2320653205953953,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 10156
    },
    {
      "epoch": 0.10157,
      "grad_norm": 1.1889966899388862,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10157
    },
    {
      "epoch": 0.10158,
      "grad_norm": 1.0696254422253109,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10158
    },
    {
      "epoch": 0.10159,
      "grad_norm": 1.0566907846701998,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 10159
    },
    {
      "epoch": 0.1016,
      "grad_norm": 1.0525638580660244,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 10160
    },
    {
      "epoch": 0.10161,
      "grad_norm": 1.046979807820574,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 10161
    },
    {
      "epoch": 0.10162,
      "grad_norm": 0.9848404406583341,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10162
    },
    {
      "epoch": 0.10163,
      "grad_norm": 1.0285044268005161,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 10163
    },
    {
      "epoch": 0.10164,
      "grad_norm": 1.0192066947627048,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10164
    },
    {
      "epoch": 0.10165,
      "grad_norm": 1.1070606086478363,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 10165
    },
    {
      "epoch": 0.10166,
      "grad_norm": 1.0624502625108392,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10166
    },
    {
      "epoch": 0.10167,
      "grad_norm": 1.1263826776460912,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10167
    },
    {
      "epoch": 0.10168,
      "grad_norm": 0.886296216319193,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 10168
    },
    {
      "epoch": 0.10169,
      "grad_norm": 1.0451723259522945,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10169
    },
    {
      "epoch": 0.1017,
      "grad_norm": 1.2995896070350659,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10170
    },
    {
      "epoch": 0.10171,
      "grad_norm": 1.033886351063779,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 10171
    },
    {
      "epoch": 0.10172,
      "grad_norm": 1.1153685517458167,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 10172
    },
    {
      "epoch": 0.10173,
      "grad_norm": 1.0772592426847833,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10173
    },
    {
      "epoch": 0.10174,
      "grad_norm": 1.13970670521581,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10174
    },
    {
      "epoch": 0.10175,
      "grad_norm": 0.8756802427578665,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10175
    },
    {
      "epoch": 0.10176,
      "grad_norm": 0.8661317521414144,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 10176
    },
    {
      "epoch": 0.10177,
      "grad_norm": 0.9875363098650536,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 10177
    },
    {
      "epoch": 0.10178,
      "grad_norm": 1.1709545498623426,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 10178
    },
    {
      "epoch": 0.10179,
      "grad_norm": 1.0144311108168578,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 10179
    },
    {
      "epoch": 0.1018,
      "grad_norm": 1.2533092220049373,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10180
    },
    {
      "epoch": 0.10181,
      "grad_norm": 0.8942442154370184,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 10181
    },
    {
      "epoch": 0.10182,
      "grad_norm": 1.0521845568594967,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 10182
    },
    {
      "epoch": 0.10183,
      "grad_norm": 1.1139008203198577,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 10183
    },
    {
      "epoch": 0.10184,
      "grad_norm": 0.955052265077462,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 10184
    },
    {
      "epoch": 0.10185,
      "grad_norm": 1.2294865872901575,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 10185
    },
    {
      "epoch": 0.10186,
      "grad_norm": 1.0405948014225823,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 10186
    },
    {
      "epoch": 0.10187,
      "grad_norm": 1.068606777390599,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 10187
    },
    {
      "epoch": 0.10188,
      "grad_norm": 1.0324240707550445,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 10188
    },
    {
      "epoch": 0.10189,
      "grad_norm": 1.0255280020860909,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 10189
    },
    {
      "epoch": 0.1019,
      "grad_norm": 0.9186025646322838,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 10190
    },
    {
      "epoch": 0.10191,
      "grad_norm": 0.88989047653086,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 10191
    },
    {
      "epoch": 0.10192,
      "grad_norm": 1.1633171294690876,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 10192
    },
    {
      "epoch": 0.10193,
      "grad_norm": 1.2427486358359723,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 10193
    },
    {
      "epoch": 0.10194,
      "grad_norm": 1.0586470094561826,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 10194
    },
    {
      "epoch": 0.10195,
      "grad_norm": 1.0036822636108969,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 10195
    },
    {
      "epoch": 0.10196,
      "grad_norm": 0.9466215049233101,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 10196
    },
    {
      "epoch": 0.10197,
      "grad_norm": 0.8112351254656962,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 10197
    },
    {
      "epoch": 0.10198,
      "grad_norm": 0.819974365220033,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 10198
    },
    {
      "epoch": 0.10199,
      "grad_norm": 0.9933425025429821,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 10199
    },
    {
      "epoch": 0.102,
      "grad_norm": 1.0531125515747866,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 10200
    },
    {
      "epoch": 0.10201,
      "grad_norm": 1.0550511348110323,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10201
    },
    {
      "epoch": 0.10202,
      "grad_norm": 1.0661497923251,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 10202
    },
    {
      "epoch": 0.10203,
      "grad_norm": 1.220413862454087,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 10203
    },
    {
      "epoch": 0.10204,
      "grad_norm": 1.195159818899411,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 10204
    },
    {
      "epoch": 0.10205,
      "grad_norm": 1.0453481471485864,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 10205
    },
    {
      "epoch": 0.10206,
      "grad_norm": 1.2132010685629997,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 10206
    },
    {
      "epoch": 0.10207,
      "grad_norm": 0.8484088484915983,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10207
    },
    {
      "epoch": 0.10208,
      "grad_norm": 0.7164134447580448,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 10208
    },
    {
      "epoch": 0.10209,
      "grad_norm": 0.8905702473288374,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 10209
    },
    {
      "epoch": 0.1021,
      "grad_norm": 1.0567582830849451,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 10210
    },
    {
      "epoch": 0.10211,
      "grad_norm": 1.2831260130563622,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 10211
    },
    {
      "epoch": 0.10212,
      "grad_norm": 0.9130536063331267,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10212
    },
    {
      "epoch": 0.10213,
      "grad_norm": 0.9842667361773123,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 10213
    },
    {
      "epoch": 0.10214,
      "grad_norm": 1.2213307026626754,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 10214
    },
    {
      "epoch": 0.10215,
      "grad_norm": 1.09106396917812,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 10215
    },
    {
      "epoch": 0.10216,
      "grad_norm": 0.8701974848383143,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 10216
    },
    {
      "epoch": 0.10217,
      "grad_norm": 0.9028785450684711,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 10217
    },
    {
      "epoch": 0.10218,
      "grad_norm": 0.9895049497286713,
      "learning_rate": 0.003,
      "loss": 4.1137,
      "step": 10218
    },
    {
      "epoch": 0.10219,
      "grad_norm": 0.9976561368167319,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 10219
    },
    {
      "epoch": 0.1022,
      "grad_norm": 1.1566379033564083,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 10220
    },
    {
      "epoch": 0.10221,
      "grad_norm": 0.9436125666687795,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 10221
    },
    {
      "epoch": 0.10222,
      "grad_norm": 1.1745810965985861,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 10222
    },
    {
      "epoch": 0.10223,
      "grad_norm": 1.1161877928907953,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 10223
    },
    {
      "epoch": 0.10224,
      "grad_norm": 1.1101784497581406,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 10224
    },
    {
      "epoch": 0.10225,
      "grad_norm": 1.166479137680796,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 10225
    },
    {
      "epoch": 0.10226,
      "grad_norm": 1.0999501771547044,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 10226
    },
    {
      "epoch": 0.10227,
      "grad_norm": 1.1405389566462554,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 10227
    },
    {
      "epoch": 0.10228,
      "grad_norm": 1.237424322483817,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 10228
    },
    {
      "epoch": 0.10229,
      "grad_norm": 0.9639553122447139,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 10229
    },
    {
      "epoch": 0.1023,
      "grad_norm": 1.0161290893618329,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10230
    },
    {
      "epoch": 0.10231,
      "grad_norm": 1.157806573153118,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 10231
    },
    {
      "epoch": 0.10232,
      "grad_norm": 1.1552991257125518,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 10232
    },
    {
      "epoch": 0.10233,
      "grad_norm": 1.0844941301760342,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 10233
    },
    {
      "epoch": 0.10234,
      "grad_norm": 0.9783976398030328,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 10234
    },
    {
      "epoch": 0.10235,
      "grad_norm": 1.0167830602919732,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10235
    },
    {
      "epoch": 0.10236,
      "grad_norm": 1.2164144746685364,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 10236
    },
    {
      "epoch": 0.10237,
      "grad_norm": 0.9709684203383907,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10237
    },
    {
      "epoch": 0.10238,
      "grad_norm": 1.0206937521127843,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 10238
    },
    {
      "epoch": 0.10239,
      "grad_norm": 1.115053000183982,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 10239
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.9632605456430765,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 10240
    },
    {
      "epoch": 0.10241,
      "grad_norm": 0.9223306760546657,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 10241
    },
    {
      "epoch": 0.10242,
      "grad_norm": 1.0350416435917151,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10242
    },
    {
      "epoch": 0.10243,
      "grad_norm": 0.9740341906220983,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 10243
    },
    {
      "epoch": 0.10244,
      "grad_norm": 1.0309607354313888,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10244
    },
    {
      "epoch": 0.10245,
      "grad_norm": 1.2098078378544046,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 10245
    },
    {
      "epoch": 0.10246,
      "grad_norm": 1.181999436226006,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 10246
    },
    {
      "epoch": 0.10247,
      "grad_norm": 1.0574827260952755,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 10247
    },
    {
      "epoch": 0.10248,
      "grad_norm": 1.0257693998664565,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 10248
    },
    {
      "epoch": 0.10249,
      "grad_norm": 0.9554869482772075,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 10249
    },
    {
      "epoch": 0.1025,
      "grad_norm": 1.0132475399789136,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 10250
    },
    {
      "epoch": 0.10251,
      "grad_norm": 1.0895426483345603,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 10251
    },
    {
      "epoch": 0.10252,
      "grad_norm": 1.1015345937455574,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 10252
    },
    {
      "epoch": 0.10253,
      "grad_norm": 0.9037748854628284,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 10253
    },
    {
      "epoch": 0.10254,
      "grad_norm": 1.0414946280019493,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 10254
    },
    {
      "epoch": 0.10255,
      "grad_norm": 1.3181367319322463,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10255
    },
    {
      "epoch": 0.10256,
      "grad_norm": 0.8784215981738813,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 10256
    },
    {
      "epoch": 0.10257,
      "grad_norm": 0.794653621913254,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 10257
    },
    {
      "epoch": 0.10258,
      "grad_norm": 0.9681340816585842,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 10258
    },
    {
      "epoch": 0.10259,
      "grad_norm": 1.1841380702726292,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 10259
    },
    {
      "epoch": 0.1026,
      "grad_norm": 0.9894474812507661,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 10260
    },
    {
      "epoch": 0.10261,
      "grad_norm": 0.8997159163161771,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10261
    },
    {
      "epoch": 0.10262,
      "grad_norm": 1.088375858779392,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10262
    },
    {
      "epoch": 0.10263,
      "grad_norm": 1.0827529991090061,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10263
    },
    {
      "epoch": 0.10264,
      "grad_norm": 1.099911495151273,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 10264
    },
    {
      "epoch": 0.10265,
      "grad_norm": 0.9649937187292063,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10265
    },
    {
      "epoch": 0.10266,
      "grad_norm": 0.9430820433940578,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 10266
    },
    {
      "epoch": 0.10267,
      "grad_norm": 1.0166591619014769,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 10267
    },
    {
      "epoch": 0.10268,
      "grad_norm": 1.0981876655466734,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 10268
    },
    {
      "epoch": 0.10269,
      "grad_norm": 1.1636560229250192,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10269
    },
    {
      "epoch": 0.1027,
      "grad_norm": 1.147237455230559,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 10270
    },
    {
      "epoch": 0.10271,
      "grad_norm": 1.2545858285112752,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 10271
    },
    {
      "epoch": 0.10272,
      "grad_norm": 1.2305947119786966,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 10272
    },
    {
      "epoch": 0.10273,
      "grad_norm": 1.2638269237402653,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 10273
    },
    {
      "epoch": 0.10274,
      "grad_norm": 0.786046900482246,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 10274
    },
    {
      "epoch": 0.10275,
      "grad_norm": 0.7962616812711191,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 10275
    },
    {
      "epoch": 0.10276,
      "grad_norm": 0.9054468958301576,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 10276
    },
    {
      "epoch": 0.10277,
      "grad_norm": 1.1773560503821552,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10277
    },
    {
      "epoch": 0.10278,
      "grad_norm": 0.8437933387987852,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 10278
    },
    {
      "epoch": 0.10279,
      "grad_norm": 1.0099305707822057,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 10279
    },
    {
      "epoch": 0.1028,
      "grad_norm": 1.172542815070957,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 10280
    },
    {
      "epoch": 0.10281,
      "grad_norm": 0.9226538744503163,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 10281
    },
    {
      "epoch": 0.10282,
      "grad_norm": 1.0087487540894944,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 10282
    },
    {
      "epoch": 0.10283,
      "grad_norm": 1.0526662717264217,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10283
    },
    {
      "epoch": 0.10284,
      "grad_norm": 1.2763147693807468,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 10284
    },
    {
      "epoch": 0.10285,
      "grad_norm": 1.1523737559077365,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 10285
    },
    {
      "epoch": 0.10286,
      "grad_norm": 0.9428247281357661,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 10286
    },
    {
      "epoch": 0.10287,
      "grad_norm": 1.0532765499789432,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10287
    },
    {
      "epoch": 0.10288,
      "grad_norm": 1.0871678672283156,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 10288
    },
    {
      "epoch": 0.10289,
      "grad_norm": 1.1107409124743397,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 10289
    },
    {
      "epoch": 0.1029,
      "grad_norm": 0.9157068380119726,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 10290
    },
    {
      "epoch": 0.10291,
      "grad_norm": 0.9147142055941883,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10291
    },
    {
      "epoch": 0.10292,
      "grad_norm": 1.015781317673214,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 10292
    },
    {
      "epoch": 0.10293,
      "grad_norm": 1.0357861344949726,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 10293
    },
    {
      "epoch": 0.10294,
      "grad_norm": 1.0614880964315667,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 10294
    },
    {
      "epoch": 0.10295,
      "grad_norm": 1.1224618732216065,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 10295
    },
    {
      "epoch": 0.10296,
      "grad_norm": 1.1959008142012484,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 10296
    },
    {
      "epoch": 0.10297,
      "grad_norm": 1.3089339465817695,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10297
    },
    {
      "epoch": 0.10298,
      "grad_norm": 1.2566589944135738,
      "learning_rate": 0.003,
      "loss": 4.1278,
      "step": 10298
    },
    {
      "epoch": 0.10299,
      "grad_norm": 0.8853939220827068,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 10299
    },
    {
      "epoch": 0.103,
      "grad_norm": 0.9357792578389142,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10300
    },
    {
      "epoch": 0.10301,
      "grad_norm": 1.1368678165849317,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 10301
    },
    {
      "epoch": 0.10302,
      "grad_norm": 1.0614029064175152,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 10302
    },
    {
      "epoch": 0.10303,
      "grad_norm": 0.9536131022368639,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10303
    },
    {
      "epoch": 0.10304,
      "grad_norm": 1.0489700654643546,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 10304
    },
    {
      "epoch": 0.10305,
      "grad_norm": 1.173340722230068,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 10305
    },
    {
      "epoch": 0.10306,
      "grad_norm": 0.9482417193908688,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 10306
    },
    {
      "epoch": 0.10307,
      "grad_norm": 0.8614319455407266,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 10307
    },
    {
      "epoch": 0.10308,
      "grad_norm": 0.8944477264218518,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 10308
    },
    {
      "epoch": 0.10309,
      "grad_norm": 1.0220567330268846,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 10309
    },
    {
      "epoch": 0.1031,
      "grad_norm": 1.223436846411137,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10310
    },
    {
      "epoch": 0.10311,
      "grad_norm": 0.9505791636624836,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 10311
    },
    {
      "epoch": 0.10312,
      "grad_norm": 0.8911177782963302,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 10312
    },
    {
      "epoch": 0.10313,
      "grad_norm": 0.9244202750779421,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 10313
    },
    {
      "epoch": 0.10314,
      "grad_norm": 0.9999763464957447,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 10314
    },
    {
      "epoch": 0.10315,
      "grad_norm": 1.3083868898257345,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10315
    },
    {
      "epoch": 0.10316,
      "grad_norm": 0.9300920090393802,
      "learning_rate": 0.003,
      "loss": 4.1149,
      "step": 10316
    },
    {
      "epoch": 0.10317,
      "grad_norm": 1.1215040272892525,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 10317
    },
    {
      "epoch": 0.10318,
      "grad_norm": 1.1317715673516657,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 10318
    },
    {
      "epoch": 0.10319,
      "grad_norm": 1.0742094095501016,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10319
    },
    {
      "epoch": 0.1032,
      "grad_norm": 1.1313280016207532,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 10320
    },
    {
      "epoch": 0.10321,
      "grad_norm": 1.0785567501408917,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 10321
    },
    {
      "epoch": 0.10322,
      "grad_norm": 1.116372226706656,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 10322
    },
    {
      "epoch": 0.10323,
      "grad_norm": 1.1381641966602283,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 10323
    },
    {
      "epoch": 0.10324,
      "grad_norm": 1.1469641020232408,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 10324
    },
    {
      "epoch": 0.10325,
      "grad_norm": 1.020356233262012,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 10325
    },
    {
      "epoch": 0.10326,
      "grad_norm": 1.09961944340241,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 10326
    },
    {
      "epoch": 0.10327,
      "grad_norm": 1.35925570455499,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 10327
    },
    {
      "epoch": 0.10328,
      "grad_norm": 0.9169067271168425,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 10328
    },
    {
      "epoch": 0.10329,
      "grad_norm": 1.0577568254856755,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 10329
    },
    {
      "epoch": 0.1033,
      "grad_norm": 1.134734217462752,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 10330
    },
    {
      "epoch": 0.10331,
      "grad_norm": 1.0651170802221943,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10331
    },
    {
      "epoch": 0.10332,
      "grad_norm": 1.006008548176185,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10332
    },
    {
      "epoch": 0.10333,
      "grad_norm": 1.1589315309415948,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 10333
    },
    {
      "epoch": 0.10334,
      "grad_norm": 1.1452212070860508,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 10334
    },
    {
      "epoch": 0.10335,
      "grad_norm": 0.8896619267885125,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 10335
    },
    {
      "epoch": 0.10336,
      "grad_norm": 0.985250176255216,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 10336
    },
    {
      "epoch": 0.10337,
      "grad_norm": 1.2128001129054007,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10337
    },
    {
      "epoch": 0.10338,
      "grad_norm": 0.9030879185357017,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 10338
    },
    {
      "epoch": 0.10339,
      "grad_norm": 1.0505055949856439,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 10339
    },
    {
      "epoch": 0.1034,
      "grad_norm": 1.1890035409892301,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 10340
    },
    {
      "epoch": 0.10341,
      "grad_norm": 0.8832065619363271,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 10341
    },
    {
      "epoch": 0.10342,
      "grad_norm": 0.9181271214030564,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 10342
    },
    {
      "epoch": 0.10343,
      "grad_norm": 1.0026233570871133,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 10343
    },
    {
      "epoch": 0.10344,
      "grad_norm": 1.195425512592561,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 10344
    },
    {
      "epoch": 0.10345,
      "grad_norm": 0.9124977352765757,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 10345
    },
    {
      "epoch": 0.10346,
      "grad_norm": 1.0417341030487266,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 10346
    },
    {
      "epoch": 0.10347,
      "grad_norm": 1.125139684288503,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 10347
    },
    {
      "epoch": 0.10348,
      "grad_norm": 0.9637619839862117,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 10348
    },
    {
      "epoch": 0.10349,
      "grad_norm": 1.0000001930540041,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 10349
    },
    {
      "epoch": 0.1035,
      "grad_norm": 1.3264629530651777,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 10350
    },
    {
      "epoch": 0.10351,
      "grad_norm": 0.9410407767703273,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 10351
    },
    {
      "epoch": 0.10352,
      "grad_norm": 1.2064264493745134,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 10352
    },
    {
      "epoch": 0.10353,
      "grad_norm": 0.9581655627934581,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 10353
    },
    {
      "epoch": 0.10354,
      "grad_norm": 1.0813766464475838,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 10354
    },
    {
      "epoch": 0.10355,
      "grad_norm": 1.0105125418694436,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 10355
    },
    {
      "epoch": 0.10356,
      "grad_norm": 1.190959028236916,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 10356
    },
    {
      "epoch": 0.10357,
      "grad_norm": 1.255268509152371,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 10357
    },
    {
      "epoch": 0.10358,
      "grad_norm": 1.1990872530553214,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 10358
    },
    {
      "epoch": 0.10359,
      "grad_norm": 0.9754039308040845,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 10359
    },
    {
      "epoch": 0.1036,
      "grad_norm": 1.0169279962302413,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 10360
    },
    {
      "epoch": 0.10361,
      "grad_norm": 1.095290235333622,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 10361
    },
    {
      "epoch": 0.10362,
      "grad_norm": 1.0889728298226304,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 10362
    },
    {
      "epoch": 0.10363,
      "grad_norm": 0.9704203825231105,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10363
    },
    {
      "epoch": 0.10364,
      "grad_norm": 1.1099160526895995,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 10364
    },
    {
      "epoch": 0.10365,
      "grad_norm": 1.1233714244351318,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 10365
    },
    {
      "epoch": 0.10366,
      "grad_norm": 0.9225455273609658,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10366
    },
    {
      "epoch": 0.10367,
      "grad_norm": 1.0325933199333683,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 10367
    },
    {
      "epoch": 0.10368,
      "grad_norm": 1.1791979754809137,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 10368
    },
    {
      "epoch": 0.10369,
      "grad_norm": 0.8022118836361686,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 10369
    },
    {
      "epoch": 0.1037,
      "grad_norm": 0.9774590280143128,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 10370
    },
    {
      "epoch": 0.10371,
      "grad_norm": 1.3320865254161802,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 10371
    },
    {
      "epoch": 0.10372,
      "grad_norm": 1.1731206938597318,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 10372
    },
    {
      "epoch": 0.10373,
      "grad_norm": 0.8445497416221235,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10373
    },
    {
      "epoch": 0.10374,
      "grad_norm": 0.8558340467037168,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 10374
    },
    {
      "epoch": 0.10375,
      "grad_norm": 1.0298854077289796,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 10375
    },
    {
      "epoch": 0.10376,
      "grad_norm": 1.3300837557126834,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10376
    },
    {
      "epoch": 0.10377,
      "grad_norm": 0.9249173973773479,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10377
    },
    {
      "epoch": 0.10378,
      "grad_norm": 1.0239269983800974,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 10378
    },
    {
      "epoch": 0.10379,
      "grad_norm": 1.121667224761993,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 10379
    },
    {
      "epoch": 0.1038,
      "grad_norm": 1.0149496858844194,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 10380
    },
    {
      "epoch": 0.10381,
      "grad_norm": 1.0559874429370242,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10381
    },
    {
      "epoch": 0.10382,
      "grad_norm": 0.9631985714419412,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 10382
    },
    {
      "epoch": 0.10383,
      "grad_norm": 0.9514945697162748,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 10383
    },
    {
      "epoch": 0.10384,
      "grad_norm": 1.1301104439298963,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 10384
    },
    {
      "epoch": 0.10385,
      "grad_norm": 1.0145183022082147,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 10385
    },
    {
      "epoch": 0.10386,
      "grad_norm": 1.2581553816472262,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 10386
    },
    {
      "epoch": 0.10387,
      "grad_norm": 0.9484969624909415,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10387
    },
    {
      "epoch": 0.10388,
      "grad_norm": 0.9469902349525261,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 10388
    },
    {
      "epoch": 0.10389,
      "grad_norm": 0.9773642974595819,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 10389
    },
    {
      "epoch": 0.1039,
      "grad_norm": 1.0578749602427104,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 10390
    },
    {
      "epoch": 0.10391,
      "grad_norm": 1.0152407829650854,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 10391
    },
    {
      "epoch": 0.10392,
      "grad_norm": 1.1687985660961098,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10392
    },
    {
      "epoch": 0.10393,
      "grad_norm": 1.274162871161583,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 10393
    },
    {
      "epoch": 0.10394,
      "grad_norm": 1.0376815910121797,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10394
    },
    {
      "epoch": 0.10395,
      "grad_norm": 1.1455649817000828,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 10395
    },
    {
      "epoch": 0.10396,
      "grad_norm": 1.048450933526513,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10396
    },
    {
      "epoch": 0.10397,
      "grad_norm": 1.0835103485182267,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 10397
    },
    {
      "epoch": 0.10398,
      "grad_norm": 1.1028674181870997,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 10398
    },
    {
      "epoch": 0.10399,
      "grad_norm": 1.1462902932061454,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 10399
    },
    {
      "epoch": 0.104,
      "grad_norm": 1.070594058981979,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 10400
    },
    {
      "epoch": 0.10401,
      "grad_norm": 1.041054683828324,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 10401
    },
    {
      "epoch": 0.10402,
      "grad_norm": 1.1525497854276674,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 10402
    },
    {
      "epoch": 0.10403,
      "grad_norm": 1.0190917702762687,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 10403
    },
    {
      "epoch": 0.10404,
      "grad_norm": 1.1285203858625503,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 10404
    },
    {
      "epoch": 0.10405,
      "grad_norm": 1.0516484986800476,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 10405
    },
    {
      "epoch": 0.10406,
      "grad_norm": 1.1198310400726428,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 10406
    },
    {
      "epoch": 0.10407,
      "grad_norm": 0.9507450061808831,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 10407
    },
    {
      "epoch": 0.10408,
      "grad_norm": 0.9195044187230283,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 10408
    },
    {
      "epoch": 0.10409,
      "grad_norm": 0.9570767287416899,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 10409
    },
    {
      "epoch": 0.1041,
      "grad_norm": 1.0343173813267,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 10410
    },
    {
      "epoch": 0.10411,
      "grad_norm": 1.1150088548083112,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 10411
    },
    {
      "epoch": 0.10412,
      "grad_norm": 1.2434197115786008,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 10412
    },
    {
      "epoch": 0.10413,
      "grad_norm": 0.9686766565831221,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 10413
    },
    {
      "epoch": 0.10414,
      "grad_norm": 1.0870609051427256,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 10414
    },
    {
      "epoch": 0.10415,
      "grad_norm": 1.1242148496718316,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10415
    },
    {
      "epoch": 0.10416,
      "grad_norm": 1.0292327009210644,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 10416
    },
    {
      "epoch": 0.10417,
      "grad_norm": 1.0822743668308503,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 10417
    },
    {
      "epoch": 0.10418,
      "grad_norm": 1.0278876159616779,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 10418
    },
    {
      "epoch": 0.10419,
      "grad_norm": 1.0491246301200836,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 10419
    },
    {
      "epoch": 0.1042,
      "grad_norm": 0.9477998200019015,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 10420
    },
    {
      "epoch": 0.10421,
      "grad_norm": 1.0788587406474686,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 10421
    },
    {
      "epoch": 0.10422,
      "grad_norm": 0.9632100820284665,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 10422
    },
    {
      "epoch": 0.10423,
      "grad_norm": 1.0019319499065928,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 10423
    },
    {
      "epoch": 0.10424,
      "grad_norm": 1.143735387277337,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 10424
    },
    {
      "epoch": 0.10425,
      "grad_norm": 0.7908128783667394,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 10425
    },
    {
      "epoch": 0.10426,
      "grad_norm": 0.9313781838068929,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 10426
    },
    {
      "epoch": 0.10427,
      "grad_norm": 1.1984211817477621,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 10427
    },
    {
      "epoch": 0.10428,
      "grad_norm": 0.9431446163495729,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 10428
    },
    {
      "epoch": 0.10429,
      "grad_norm": 1.168965044587708,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 10429
    },
    {
      "epoch": 0.1043,
      "grad_norm": 1.3908380610858844,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 10430
    },
    {
      "epoch": 0.10431,
      "grad_norm": 1.0676013928236892,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 10431
    },
    {
      "epoch": 0.10432,
      "grad_norm": 1.066980716023993,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 10432
    },
    {
      "epoch": 0.10433,
      "grad_norm": 0.9948132055941239,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10433
    },
    {
      "epoch": 0.10434,
      "grad_norm": 1.0508222592142207,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 10434
    },
    {
      "epoch": 0.10435,
      "grad_norm": 1.0748934991259524,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 10435
    },
    {
      "epoch": 0.10436,
      "grad_norm": 0.9961213953784716,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 10436
    },
    {
      "epoch": 0.10437,
      "grad_norm": 1.2265320822152506,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 10437
    },
    {
      "epoch": 0.10438,
      "grad_norm": 0.9727388902004728,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 10438
    },
    {
      "epoch": 0.10439,
      "grad_norm": 0.9461742697962658,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10439
    },
    {
      "epoch": 0.1044,
      "grad_norm": 0.9885970310180309,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 10440
    },
    {
      "epoch": 0.10441,
      "grad_norm": 1.0654469009945895,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 10441
    },
    {
      "epoch": 0.10442,
      "grad_norm": 0.9305298263072211,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 10442
    },
    {
      "epoch": 0.10443,
      "grad_norm": 1.0404155491200193,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 10443
    },
    {
      "epoch": 0.10444,
      "grad_norm": 1.1653476936243858,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 10444
    },
    {
      "epoch": 0.10445,
      "grad_norm": 1.2134056802647015,
      "learning_rate": 0.003,
      "loss": 4.1121,
      "step": 10445
    },
    {
      "epoch": 0.10446,
      "grad_norm": 0.8997490311080218,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 10446
    },
    {
      "epoch": 0.10447,
      "grad_norm": 0.8352844337856086,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 10447
    },
    {
      "epoch": 0.10448,
      "grad_norm": 0.8792325987966317,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 10448
    },
    {
      "epoch": 0.10449,
      "grad_norm": 1.1230065414769836,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10449
    },
    {
      "epoch": 0.1045,
      "grad_norm": 1.0569256969404204,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 10450
    },
    {
      "epoch": 0.10451,
      "grad_norm": 0.9741322946515599,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 10451
    },
    {
      "epoch": 0.10452,
      "grad_norm": 1.12094286850372,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 10452
    },
    {
      "epoch": 0.10453,
      "grad_norm": 1.0634000951508014,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10453
    },
    {
      "epoch": 0.10454,
      "grad_norm": 1.0398692030033183,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 10454
    },
    {
      "epoch": 0.10455,
      "grad_norm": 1.0697825048342626,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10455
    },
    {
      "epoch": 0.10456,
      "grad_norm": 0.9369648357510157,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 10456
    },
    {
      "epoch": 0.10457,
      "grad_norm": 1.1077828883876948,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 10457
    },
    {
      "epoch": 0.10458,
      "grad_norm": 1.2486942415113471,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 10458
    },
    {
      "epoch": 0.10459,
      "grad_norm": 1.0800758371094688,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 10459
    },
    {
      "epoch": 0.1046,
      "grad_norm": 1.1580347302708973,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 10460
    },
    {
      "epoch": 0.10461,
      "grad_norm": 1.1725452690555769,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 10461
    },
    {
      "epoch": 0.10462,
      "grad_norm": 1.0908592370020997,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 10462
    },
    {
      "epoch": 0.10463,
      "grad_norm": 0.991908326018903,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 10463
    },
    {
      "epoch": 0.10464,
      "grad_norm": 1.2301306291338194,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 10464
    },
    {
      "epoch": 0.10465,
      "grad_norm": 0.9472521418396895,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10465
    },
    {
      "epoch": 0.10466,
      "grad_norm": 1.0771140483899684,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10466
    },
    {
      "epoch": 0.10467,
      "grad_norm": 1.0045428457733532,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10467
    },
    {
      "epoch": 0.10468,
      "grad_norm": 1.063329814375369,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10468
    },
    {
      "epoch": 0.10469,
      "grad_norm": 0.8044646549832464,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10469
    },
    {
      "epoch": 0.1047,
      "grad_norm": 0.8705710123890662,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10470
    },
    {
      "epoch": 0.10471,
      "grad_norm": 0.9026927449559127,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 10471
    },
    {
      "epoch": 0.10472,
      "grad_norm": 0.9956602765823049,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10472
    },
    {
      "epoch": 0.10473,
      "grad_norm": 1.1589637831106856,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 10473
    },
    {
      "epoch": 0.10474,
      "grad_norm": 1.184785189846938,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 10474
    },
    {
      "epoch": 0.10475,
      "grad_norm": 1.3550631194540537,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 10475
    },
    {
      "epoch": 0.10476,
      "grad_norm": 1.0320795978049657,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 10476
    },
    {
      "epoch": 0.10477,
      "grad_norm": 1.0604195369959852,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 10477
    },
    {
      "epoch": 0.10478,
      "grad_norm": 1.0637193655219612,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 10478
    },
    {
      "epoch": 0.10479,
      "grad_norm": 0.9630771797139417,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 10479
    },
    {
      "epoch": 0.1048,
      "grad_norm": 0.9484006284165586,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 10480
    },
    {
      "epoch": 0.10481,
      "grad_norm": 1.054567576660935,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 10481
    },
    {
      "epoch": 0.10482,
      "grad_norm": 1.3360352292856175,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 10482
    },
    {
      "epoch": 0.10483,
      "grad_norm": 0.9019567481761226,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 10483
    },
    {
      "epoch": 0.10484,
      "grad_norm": 0.991285568753618,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10484
    },
    {
      "epoch": 0.10485,
      "grad_norm": 1.1370528625194423,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10485
    },
    {
      "epoch": 0.10486,
      "grad_norm": 0.9127707353956426,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 10486
    },
    {
      "epoch": 0.10487,
      "grad_norm": 0.9285607355356765,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10487
    },
    {
      "epoch": 0.10488,
      "grad_norm": 1.0231832309662972,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 10488
    },
    {
      "epoch": 0.10489,
      "grad_norm": 1.1181475303743535,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 10489
    },
    {
      "epoch": 0.1049,
      "grad_norm": 1.058957141573963,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 10490
    },
    {
      "epoch": 0.10491,
      "grad_norm": 1.4398699183245742,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 10491
    },
    {
      "epoch": 0.10492,
      "grad_norm": 0.7504173202433805,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 10492
    },
    {
      "epoch": 0.10493,
      "grad_norm": 0.8117939810615702,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 10493
    },
    {
      "epoch": 0.10494,
      "grad_norm": 1.0189444416069287,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 10494
    },
    {
      "epoch": 0.10495,
      "grad_norm": 1.0926650227724901,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 10495
    },
    {
      "epoch": 0.10496,
      "grad_norm": 0.9245404161543932,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 10496
    },
    {
      "epoch": 0.10497,
      "grad_norm": 0.8690029751405167,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 10497
    },
    {
      "epoch": 0.10498,
      "grad_norm": 0.9260719581793728,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10498
    },
    {
      "epoch": 0.10499,
      "grad_norm": 0.9504218723749869,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 10499
    },
    {
      "epoch": 0.105,
      "grad_norm": 1.049731320459712,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 10500
    },
    {
      "epoch": 0.10501,
      "grad_norm": 1.2194994953775773,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 10501
    },
    {
      "epoch": 0.10502,
      "grad_norm": 1.3901211020384214,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 10502
    },
    {
      "epoch": 0.10503,
      "grad_norm": 0.8847049875069319,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 10503
    },
    {
      "epoch": 0.10504,
      "grad_norm": 0.9053464516632651,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 10504
    },
    {
      "epoch": 0.10505,
      "grad_norm": 1.0078443390204859,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 10505
    },
    {
      "epoch": 0.10506,
      "grad_norm": 1.0299360239934658,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 10506
    },
    {
      "epoch": 0.10507,
      "grad_norm": 1.0873339450933095,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10507
    },
    {
      "epoch": 0.10508,
      "grad_norm": 0.9753848523226265,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 10508
    },
    {
      "epoch": 0.10509,
      "grad_norm": 1.1237575917023788,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 10509
    },
    {
      "epoch": 0.1051,
      "grad_norm": 1.24958584132952,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 10510
    },
    {
      "epoch": 0.10511,
      "grad_norm": 1.2571755536635105,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 10511
    },
    {
      "epoch": 0.10512,
      "grad_norm": 0.935825340827118,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 10512
    },
    {
      "epoch": 0.10513,
      "grad_norm": 0.954581530206918,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 10513
    },
    {
      "epoch": 0.10514,
      "grad_norm": 1.0318931238482127,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 10514
    },
    {
      "epoch": 0.10515,
      "grad_norm": 1.1905130618079427,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 10515
    },
    {
      "epoch": 0.10516,
      "grad_norm": 0.9218196739218961,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 10516
    },
    {
      "epoch": 0.10517,
      "grad_norm": 1.0812257850070317,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 10517
    },
    {
      "epoch": 0.10518,
      "grad_norm": 1.0289312096687557,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 10518
    },
    {
      "epoch": 0.10519,
      "grad_norm": 1.0840110264499334,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 10519
    },
    {
      "epoch": 0.1052,
      "grad_norm": 1.0163700110837486,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 10520
    },
    {
      "epoch": 0.10521,
      "grad_norm": 1.168120560099558,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10521
    },
    {
      "epoch": 0.10522,
      "grad_norm": 0.941883900004278,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 10522
    },
    {
      "epoch": 0.10523,
      "grad_norm": 0.9365433648184146,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 10523
    },
    {
      "epoch": 0.10524,
      "grad_norm": 1.0607376077744424,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 10524
    },
    {
      "epoch": 0.10525,
      "grad_norm": 1.1053780367918074,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 10525
    },
    {
      "epoch": 0.10526,
      "grad_norm": 1.1732250128516668,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 10526
    },
    {
      "epoch": 0.10527,
      "grad_norm": 1.0742015400018985,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 10527
    },
    {
      "epoch": 0.10528,
      "grad_norm": 1.366765506902064,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 10528
    },
    {
      "epoch": 0.10529,
      "grad_norm": 0.8202184376308155,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 10529
    },
    {
      "epoch": 0.1053,
      "grad_norm": 0.8208632856067515,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 10530
    },
    {
      "epoch": 0.10531,
      "grad_norm": 0.8352417106297955,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 10531
    },
    {
      "epoch": 0.10532,
      "grad_norm": 0.9160049441948251,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 10532
    },
    {
      "epoch": 0.10533,
      "grad_norm": 1.0631425057807433,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 10533
    },
    {
      "epoch": 0.10534,
      "grad_norm": 1.1827574920147312,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 10534
    },
    {
      "epoch": 0.10535,
      "grad_norm": 0.8811514403995269,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 10535
    },
    {
      "epoch": 0.10536,
      "grad_norm": 1.3452482194853557,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 10536
    },
    {
      "epoch": 0.10537,
      "grad_norm": 1.2025873890846424,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 10537
    },
    {
      "epoch": 0.10538,
      "grad_norm": 0.9467288343288578,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 10538
    },
    {
      "epoch": 0.10539,
      "grad_norm": 1.1008525955886186,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 10539
    },
    {
      "epoch": 0.1054,
      "grad_norm": 1.17784567298581,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 10540
    },
    {
      "epoch": 0.10541,
      "grad_norm": 0.9005444820875294,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10541
    },
    {
      "epoch": 0.10542,
      "grad_norm": 1.077049482321095,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 10542
    },
    {
      "epoch": 0.10543,
      "grad_norm": 1.0664201972433869,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 10543
    },
    {
      "epoch": 0.10544,
      "grad_norm": 1.0681106820442525,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 10544
    },
    {
      "epoch": 0.10545,
      "grad_norm": 1.2213402778855749,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 10545
    },
    {
      "epoch": 0.10546,
      "grad_norm": 0.9978122256357836,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 10546
    },
    {
      "epoch": 0.10547,
      "grad_norm": 1.2943816452235526,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 10547
    },
    {
      "epoch": 0.10548,
      "grad_norm": 0.8865941146378675,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 10548
    },
    {
      "epoch": 0.10549,
      "grad_norm": 1.0650208342194967,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 10549
    },
    {
      "epoch": 0.1055,
      "grad_norm": 1.133121694024822,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 10550
    },
    {
      "epoch": 0.10551,
      "grad_norm": 1.0356315787198624,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 10551
    },
    {
      "epoch": 0.10552,
      "grad_norm": 1.0470920531533148,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 10552
    },
    {
      "epoch": 0.10553,
      "grad_norm": 0.932425325526456,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 10553
    },
    {
      "epoch": 0.10554,
      "grad_norm": 1.0524772486523728,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10554
    },
    {
      "epoch": 0.10555,
      "grad_norm": 1.09832900977572,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 10555
    },
    {
      "epoch": 0.10556,
      "grad_norm": 0.9863036466375944,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 10556
    },
    {
      "epoch": 0.10557,
      "grad_norm": 0.992725409801414,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 10557
    },
    {
      "epoch": 0.10558,
      "grad_norm": 1.0608804710461675,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 10558
    },
    {
      "epoch": 0.10559,
      "grad_norm": 1.0901193752068266,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 10559
    },
    {
      "epoch": 0.1056,
      "grad_norm": 1.2892535190614614,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10560
    },
    {
      "epoch": 0.10561,
      "grad_norm": 0.9940038126547566,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 10561
    },
    {
      "epoch": 0.10562,
      "grad_norm": 0.9826732604285782,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 10562
    },
    {
      "epoch": 0.10563,
      "grad_norm": 1.1983939579421365,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 10563
    },
    {
      "epoch": 0.10564,
      "grad_norm": 1.0231507937173892,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 10564
    },
    {
      "epoch": 0.10565,
      "grad_norm": 1.125203308541098,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10565
    },
    {
      "epoch": 0.10566,
      "grad_norm": 0.8906519373180206,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 10566
    },
    {
      "epoch": 0.10567,
      "grad_norm": 0.962568584030796,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 10567
    },
    {
      "epoch": 0.10568,
      "grad_norm": 1.2402788749569134,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 10568
    },
    {
      "epoch": 0.10569,
      "grad_norm": 1.1134136146515001,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 10569
    },
    {
      "epoch": 0.1057,
      "grad_norm": 1.3133871676098183,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 10570
    },
    {
      "epoch": 0.10571,
      "grad_norm": 0.951198930506272,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10571
    },
    {
      "epoch": 0.10572,
      "grad_norm": 0.9761781766255831,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10572
    },
    {
      "epoch": 0.10573,
      "grad_norm": 1.0133150920197171,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 10573
    },
    {
      "epoch": 0.10574,
      "grad_norm": 1.013912210371583,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 10574
    },
    {
      "epoch": 0.10575,
      "grad_norm": 1.05313647300968,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 10575
    },
    {
      "epoch": 0.10576,
      "grad_norm": 0.9509674082920994,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 10576
    },
    {
      "epoch": 0.10577,
      "grad_norm": 1.0053376130317002,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 10577
    },
    {
      "epoch": 0.10578,
      "grad_norm": 1.0795367145271022,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 10578
    },
    {
      "epoch": 0.10579,
      "grad_norm": 1.1298331639731727,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 10579
    },
    {
      "epoch": 0.1058,
      "grad_norm": 1.0884931502898405,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 10580
    },
    {
      "epoch": 0.10581,
      "grad_norm": 1.0468541072064026,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 10581
    },
    {
      "epoch": 0.10582,
      "grad_norm": 1.1974492928780258,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 10582
    },
    {
      "epoch": 0.10583,
      "grad_norm": 0.882313206303918,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 10583
    },
    {
      "epoch": 0.10584,
      "grad_norm": 0.9033505969353639,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 10584
    },
    {
      "epoch": 0.10585,
      "grad_norm": 1.0056878191730187,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 10585
    },
    {
      "epoch": 0.10586,
      "grad_norm": 1.237777886872006,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 10586
    },
    {
      "epoch": 0.10587,
      "grad_norm": 0.9248181845601777,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 10587
    },
    {
      "epoch": 0.10588,
      "grad_norm": 1.0882857074285057,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 10588
    },
    {
      "epoch": 0.10589,
      "grad_norm": 1.1046332241765098,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10589
    },
    {
      "epoch": 0.1059,
      "grad_norm": 1.0054015979587632,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 10590
    },
    {
      "epoch": 0.10591,
      "grad_norm": 1.1622872539084141,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 10591
    },
    {
      "epoch": 0.10592,
      "grad_norm": 0.8654066003416311,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 10592
    },
    {
      "epoch": 0.10593,
      "grad_norm": 0.9936129959819774,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 10593
    },
    {
      "epoch": 0.10594,
      "grad_norm": 1.1212371509536319,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 10594
    },
    {
      "epoch": 0.10595,
      "grad_norm": 1.1599243180097774,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10595
    },
    {
      "epoch": 0.10596,
      "grad_norm": 1.0242071396552408,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 10596
    },
    {
      "epoch": 0.10597,
      "grad_norm": 0.9678502785231397,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 10597
    },
    {
      "epoch": 0.10598,
      "grad_norm": 1.030853541727889,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10598
    },
    {
      "epoch": 0.10599,
      "grad_norm": 1.4289073689256773,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10599
    },
    {
      "epoch": 0.106,
      "grad_norm": 1.0707379947995017,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10600
    },
    {
      "epoch": 0.10601,
      "grad_norm": 1.0119993011968624,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10601
    },
    {
      "epoch": 0.10602,
      "grad_norm": 1.1556786577647435,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 10602
    },
    {
      "epoch": 0.10603,
      "grad_norm": 0.9062233720174303,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 10603
    },
    {
      "epoch": 0.10604,
      "grad_norm": 0.7573877794731967,
      "learning_rate": 0.003,
      "loss": 4.1117,
      "step": 10604
    },
    {
      "epoch": 0.10605,
      "grad_norm": 0.8217624074209846,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 10605
    },
    {
      "epoch": 0.10606,
      "grad_norm": 0.9714939673985139,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10606
    },
    {
      "epoch": 0.10607,
      "grad_norm": 1.0141974591485123,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 10607
    },
    {
      "epoch": 0.10608,
      "grad_norm": 1.0963103808271801,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 10608
    },
    {
      "epoch": 0.10609,
      "grad_norm": 1.1431306058396495,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 10609
    },
    {
      "epoch": 0.1061,
      "grad_norm": 1.4008154829290844,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 10610
    },
    {
      "epoch": 0.10611,
      "grad_norm": 0.9470325380773612,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 10611
    },
    {
      "epoch": 0.10612,
      "grad_norm": 1.0311831527113837,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 10612
    },
    {
      "epoch": 0.10613,
      "grad_norm": 1.164743448303555,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 10613
    },
    {
      "epoch": 0.10614,
      "grad_norm": 0.9502120855552426,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 10614
    },
    {
      "epoch": 0.10615,
      "grad_norm": 1.1998095664589368,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 10615
    },
    {
      "epoch": 0.10616,
      "grad_norm": 1.0083901716861463,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 10616
    },
    {
      "epoch": 0.10617,
      "grad_norm": 1.0907852145771553,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 10617
    },
    {
      "epoch": 0.10618,
      "grad_norm": 1.0124541580039028,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 10618
    },
    {
      "epoch": 0.10619,
      "grad_norm": 1.113581821108827,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 10619
    },
    {
      "epoch": 0.1062,
      "grad_norm": 1.1026787669661373,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 10620
    },
    {
      "epoch": 0.10621,
      "grad_norm": 0.9751223940659827,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 10621
    },
    {
      "epoch": 0.10622,
      "grad_norm": 1.0749021101818106,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 10622
    },
    {
      "epoch": 0.10623,
      "grad_norm": 0.945349520748991,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 10623
    },
    {
      "epoch": 0.10624,
      "grad_norm": 1.0556364929047508,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10624
    },
    {
      "epoch": 0.10625,
      "grad_norm": 1.2238541346788792,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10625
    },
    {
      "epoch": 0.10626,
      "grad_norm": 0.9013129245547551,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 10626
    },
    {
      "epoch": 0.10627,
      "grad_norm": 0.864027919043482,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10627
    },
    {
      "epoch": 0.10628,
      "grad_norm": 0.9094857462874787,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 10628
    },
    {
      "epoch": 0.10629,
      "grad_norm": 1.0335557146674315,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 10629
    },
    {
      "epoch": 0.1063,
      "grad_norm": 1.292500999524564,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 10630
    },
    {
      "epoch": 0.10631,
      "grad_norm": 0.9835133011239242,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 10631
    },
    {
      "epoch": 0.10632,
      "grad_norm": 1.0666622832328627,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 10632
    },
    {
      "epoch": 0.10633,
      "grad_norm": 1.0826186803880389,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 10633
    },
    {
      "epoch": 0.10634,
      "grad_norm": 1.0641990091605356,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10634
    },
    {
      "epoch": 0.10635,
      "grad_norm": 1.1612087316359914,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 10635
    },
    {
      "epoch": 0.10636,
      "grad_norm": 0.8748537891089967,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 10636
    },
    {
      "epoch": 0.10637,
      "grad_norm": 1.0481532412413115,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 10637
    },
    {
      "epoch": 0.10638,
      "grad_norm": 1.2606203743062692,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 10638
    },
    {
      "epoch": 0.10639,
      "grad_norm": 1.1805082052116393,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 10639
    },
    {
      "epoch": 0.1064,
      "grad_norm": 1.2398870931992865,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 10640
    },
    {
      "epoch": 0.10641,
      "grad_norm": 0.9218771485309251,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 10641
    },
    {
      "epoch": 0.10642,
      "grad_norm": 1.1272822505994036,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 10642
    },
    {
      "epoch": 0.10643,
      "grad_norm": 1.175175782769236,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 10643
    },
    {
      "epoch": 0.10644,
      "grad_norm": 0.8929122030535886,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 10644
    },
    {
      "epoch": 0.10645,
      "grad_norm": 1.0037622742587453,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 10645
    },
    {
      "epoch": 0.10646,
      "grad_norm": 1.1610195694424994,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 10646
    },
    {
      "epoch": 0.10647,
      "grad_norm": 0.852454983713457,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 10647
    },
    {
      "epoch": 0.10648,
      "grad_norm": 0.7401630263924073,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 10648
    },
    {
      "epoch": 0.10649,
      "grad_norm": 0.8311609147359353,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 10649
    },
    {
      "epoch": 0.1065,
      "grad_norm": 0.9504931584008245,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 10650
    },
    {
      "epoch": 0.10651,
      "grad_norm": 1.122066042533435,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 10651
    },
    {
      "epoch": 0.10652,
      "grad_norm": 1.2329797804688072,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 10652
    },
    {
      "epoch": 0.10653,
      "grad_norm": 0.8425687985433964,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 10653
    },
    {
      "epoch": 0.10654,
      "grad_norm": 0.9139911827111628,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 10654
    },
    {
      "epoch": 0.10655,
      "grad_norm": 1.011999739520214,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 10655
    },
    {
      "epoch": 0.10656,
      "grad_norm": 1.2012862999427212,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 10656
    },
    {
      "epoch": 0.10657,
      "grad_norm": 1.0352708658419536,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 10657
    },
    {
      "epoch": 0.10658,
      "grad_norm": 1.2516087672765586,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 10658
    },
    {
      "epoch": 0.10659,
      "grad_norm": 0.8772559044208297,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 10659
    },
    {
      "epoch": 0.1066,
      "grad_norm": 0.9125719571822856,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 10660
    },
    {
      "epoch": 0.10661,
      "grad_norm": 0.8865786014446685,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 10661
    },
    {
      "epoch": 0.10662,
      "grad_norm": 1.0850987643365564,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10662
    },
    {
      "epoch": 0.10663,
      "grad_norm": 1.2748185200581434,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10663
    },
    {
      "epoch": 0.10664,
      "grad_norm": 1.1162026613036358,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 10664
    },
    {
      "epoch": 0.10665,
      "grad_norm": 0.9494012979211818,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 10665
    },
    {
      "epoch": 0.10666,
      "grad_norm": 1.02397050134285,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 10666
    },
    {
      "epoch": 0.10667,
      "grad_norm": 1.140121878572695,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 10667
    },
    {
      "epoch": 0.10668,
      "grad_norm": 1.2059412023031264,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10668
    },
    {
      "epoch": 0.10669,
      "grad_norm": 1.1076246091621595,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 10669
    },
    {
      "epoch": 0.1067,
      "grad_norm": 1.0177092098793927,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 10670
    },
    {
      "epoch": 0.10671,
      "grad_norm": 1.1723966711259484,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 10671
    },
    {
      "epoch": 0.10672,
      "grad_norm": 1.0135262749335747,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 10672
    },
    {
      "epoch": 0.10673,
      "grad_norm": 1.2974838543788962,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 10673
    },
    {
      "epoch": 0.10674,
      "grad_norm": 1.025359321526118,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 10674
    },
    {
      "epoch": 0.10675,
      "grad_norm": 1.1358099502014838,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 10675
    },
    {
      "epoch": 0.10676,
      "grad_norm": 0.9642777344120331,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 10676
    },
    {
      "epoch": 0.10677,
      "grad_norm": 1.1745840304279103,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 10677
    },
    {
      "epoch": 0.10678,
      "grad_norm": 0.9325862438493242,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 10678
    },
    {
      "epoch": 0.10679,
      "grad_norm": 0.9072654723239565,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 10679
    },
    {
      "epoch": 0.1068,
      "grad_norm": 1.2030425278652832,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 10680
    },
    {
      "epoch": 0.10681,
      "grad_norm": 1.1857166606883949,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 10681
    },
    {
      "epoch": 0.10682,
      "grad_norm": 0.9635367808951002,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 10682
    },
    {
      "epoch": 0.10683,
      "grad_norm": 1.0311474575422612,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10683
    },
    {
      "epoch": 0.10684,
      "grad_norm": 1.0449055839802592,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10684
    },
    {
      "epoch": 0.10685,
      "grad_norm": 0.9408793445391519,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 10685
    },
    {
      "epoch": 0.10686,
      "grad_norm": 1.0709142575491677,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10686
    },
    {
      "epoch": 0.10687,
      "grad_norm": 1.118246780411596,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 10687
    },
    {
      "epoch": 0.10688,
      "grad_norm": 1.0804038641938123,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 10688
    },
    {
      "epoch": 0.10689,
      "grad_norm": 1.0753929491965348,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 10689
    },
    {
      "epoch": 0.1069,
      "grad_norm": 1.0631896177874685,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10690
    },
    {
      "epoch": 0.10691,
      "grad_norm": 1.0556411062232636,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 10691
    },
    {
      "epoch": 0.10692,
      "grad_norm": 1.0547716927758737,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 10692
    },
    {
      "epoch": 0.10693,
      "grad_norm": 1.1858072902357788,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 10693
    },
    {
      "epoch": 0.10694,
      "grad_norm": 0.9053792623061231,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 10694
    },
    {
      "epoch": 0.10695,
      "grad_norm": 1.1363465670513433,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 10695
    },
    {
      "epoch": 0.10696,
      "grad_norm": 1.1855669176217434,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 10696
    },
    {
      "epoch": 0.10697,
      "grad_norm": 1.1003628381077988,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10697
    },
    {
      "epoch": 0.10698,
      "grad_norm": 1.1351058188402894,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 10698
    },
    {
      "epoch": 0.10699,
      "grad_norm": 0.9768170554346906,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 10699
    },
    {
      "epoch": 0.107,
      "grad_norm": 1.0710799140402476,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 10700
    },
    {
      "epoch": 0.10701,
      "grad_norm": 1.0344209083339428,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10701
    },
    {
      "epoch": 0.10702,
      "grad_norm": 1.061616138491857,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 10702
    },
    {
      "epoch": 0.10703,
      "grad_norm": 0.9847321842818494,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 10703
    },
    {
      "epoch": 0.10704,
      "grad_norm": 1.0378569487826044,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 10704
    },
    {
      "epoch": 0.10705,
      "grad_norm": 1.227085863193695,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 10705
    },
    {
      "epoch": 0.10706,
      "grad_norm": 0.9995137410984207,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 10706
    },
    {
      "epoch": 0.10707,
      "grad_norm": 1.1986759073174438,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 10707
    },
    {
      "epoch": 0.10708,
      "grad_norm": 1.1415366654865533,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 10708
    },
    {
      "epoch": 0.10709,
      "grad_norm": 1.5136241575025582,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 10709
    },
    {
      "epoch": 0.1071,
      "grad_norm": 0.8341804004870889,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 10710
    },
    {
      "epoch": 0.10711,
      "grad_norm": 0.926630320291128,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 10711
    },
    {
      "epoch": 0.10712,
      "grad_norm": 0.9663161163591082,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 10712
    },
    {
      "epoch": 0.10713,
      "grad_norm": 1.2205745710480322,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 10713
    },
    {
      "epoch": 0.10714,
      "grad_norm": 1.0390508678036858,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 10714
    },
    {
      "epoch": 0.10715,
      "grad_norm": 1.0528320057494256,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 10715
    },
    {
      "epoch": 0.10716,
      "grad_norm": 1.2072562792099912,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 10716
    },
    {
      "epoch": 0.10717,
      "grad_norm": 0.8790537191222857,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10717
    },
    {
      "epoch": 0.10718,
      "grad_norm": 0.8808640271468053,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 10718
    },
    {
      "epoch": 0.10719,
      "grad_norm": 1.0126623071945835,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 10719
    },
    {
      "epoch": 0.1072,
      "grad_norm": 1.319727283236711,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 10720
    },
    {
      "epoch": 0.10721,
      "grad_norm": 1.2466570549525533,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 10721
    },
    {
      "epoch": 0.10722,
      "grad_norm": 0.97840046732779,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 10722
    },
    {
      "epoch": 0.10723,
      "grad_norm": 1.0704044212362178,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10723
    },
    {
      "epoch": 0.10724,
      "grad_norm": 1.0468299829861158,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 10724
    },
    {
      "epoch": 0.10725,
      "grad_norm": 1.1732056432117681,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 10725
    },
    {
      "epoch": 0.10726,
      "grad_norm": 0.9618675203781902,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 10726
    },
    {
      "epoch": 0.10727,
      "grad_norm": 1.056036020432587,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10727
    },
    {
      "epoch": 0.10728,
      "grad_norm": 0.9599269707382877,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 10728
    },
    {
      "epoch": 0.10729,
      "grad_norm": 1.0564091003019227,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 10729
    },
    {
      "epoch": 0.1073,
      "grad_norm": 0.9973209788927452,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 10730
    },
    {
      "epoch": 0.10731,
      "grad_norm": 1.0212852406413646,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 10731
    },
    {
      "epoch": 0.10732,
      "grad_norm": 1.1209440659622514,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 10732
    },
    {
      "epoch": 0.10733,
      "grad_norm": 1.0618758913171396,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 10733
    },
    {
      "epoch": 0.10734,
      "grad_norm": 1.1488201145415744,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 10734
    },
    {
      "epoch": 0.10735,
      "grad_norm": 0.9043046551792133,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10735
    },
    {
      "epoch": 0.10736,
      "grad_norm": 0.892103611120565,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 10736
    },
    {
      "epoch": 0.10737,
      "grad_norm": 0.970984267995846,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 10737
    },
    {
      "epoch": 0.10738,
      "grad_norm": 1.1262141754422643,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 10738
    },
    {
      "epoch": 0.10739,
      "grad_norm": 0.9769718344245536,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 10739
    },
    {
      "epoch": 0.1074,
      "grad_norm": 1.4962807603874,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 10740
    },
    {
      "epoch": 0.10741,
      "grad_norm": 1.155833820559741,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 10741
    },
    {
      "epoch": 0.10742,
      "grad_norm": 1.174159756749479,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 10742
    },
    {
      "epoch": 0.10743,
      "grad_norm": 1.012862742000115,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 10743
    },
    {
      "epoch": 0.10744,
      "grad_norm": 1.1370193003708244,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 10744
    },
    {
      "epoch": 0.10745,
      "grad_norm": 0.9464204645242252,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 10745
    },
    {
      "epoch": 0.10746,
      "grad_norm": 1.3033475431034287,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 10746
    },
    {
      "epoch": 0.10747,
      "grad_norm": 1.06969235369634,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 10747
    },
    {
      "epoch": 0.10748,
      "grad_norm": 1.0142439992763128,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 10748
    },
    {
      "epoch": 0.10749,
      "grad_norm": 1.0685245746096876,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 10749
    },
    {
      "epoch": 0.1075,
      "grad_norm": 0.8449126664648987,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 10750
    },
    {
      "epoch": 0.10751,
      "grad_norm": 0.9743842398056648,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 10751
    },
    {
      "epoch": 0.10752,
      "grad_norm": 1.1098706461499115,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10752
    },
    {
      "epoch": 0.10753,
      "grad_norm": 1.1485536672705534,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 10753
    },
    {
      "epoch": 0.10754,
      "grad_norm": 0.9857930113106791,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 10754
    },
    {
      "epoch": 0.10755,
      "grad_norm": 1.2508493100595617,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 10755
    },
    {
      "epoch": 0.10756,
      "grad_norm": 0.9426440254849656,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 10756
    },
    {
      "epoch": 0.10757,
      "grad_norm": 1.2917432581035484,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 10757
    },
    {
      "epoch": 0.10758,
      "grad_norm": 1.0086154909276714,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 10758
    },
    {
      "epoch": 0.10759,
      "grad_norm": 1.3374586895422267,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 10759
    },
    {
      "epoch": 0.1076,
      "grad_norm": 0.8816798826741424,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 10760
    },
    {
      "epoch": 0.10761,
      "grad_norm": 0.7655410742706316,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 10761
    },
    {
      "epoch": 0.10762,
      "grad_norm": 0.9097709470196474,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 10762
    },
    {
      "epoch": 0.10763,
      "grad_norm": 1.1233632654664232,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 10763
    },
    {
      "epoch": 0.10764,
      "grad_norm": 0.8479098631341053,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10764
    },
    {
      "epoch": 0.10765,
      "grad_norm": 1.0699190076068852,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 10765
    },
    {
      "epoch": 0.10766,
      "grad_norm": 1.2865992907381223,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 10766
    },
    {
      "epoch": 0.10767,
      "grad_norm": 0.9373164003694404,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10767
    },
    {
      "epoch": 0.10768,
      "grad_norm": 1.0709980796464582,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 10768
    },
    {
      "epoch": 0.10769,
      "grad_norm": 1.3406893808408222,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10769
    },
    {
      "epoch": 0.1077,
      "grad_norm": 1.0224000701141334,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 10770
    },
    {
      "epoch": 0.10771,
      "grad_norm": 1.0526142707839026,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 10771
    },
    {
      "epoch": 0.10772,
      "grad_norm": 0.9443675613044936,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 10772
    },
    {
      "epoch": 0.10773,
      "grad_norm": 0.9631051318117351,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 10773
    },
    {
      "epoch": 0.10774,
      "grad_norm": 1.0549186900463854,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 10774
    },
    {
      "epoch": 0.10775,
      "grad_norm": 1.2375124249342755,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 10775
    },
    {
      "epoch": 0.10776,
      "grad_norm": 0.9660424900300653,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 10776
    },
    {
      "epoch": 0.10777,
      "grad_norm": 1.0703994783409576,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10777
    },
    {
      "epoch": 0.10778,
      "grad_norm": 1.0264385844846082,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10778
    },
    {
      "epoch": 0.10779,
      "grad_norm": 1.3643862274983778,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 10779
    },
    {
      "epoch": 0.1078,
      "grad_norm": 0.8924594651101884,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 10780
    },
    {
      "epoch": 0.10781,
      "grad_norm": 0.9716988610336873,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 10781
    },
    {
      "epoch": 0.10782,
      "grad_norm": 1.2408412829715514,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 10782
    },
    {
      "epoch": 0.10783,
      "grad_norm": 1.115281436405084,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 10783
    },
    {
      "epoch": 0.10784,
      "grad_norm": 0.9579470079302335,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 10784
    },
    {
      "epoch": 0.10785,
      "grad_norm": 0.9541684831885465,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 10785
    },
    {
      "epoch": 0.10786,
      "grad_norm": 1.0880236074356275,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 10786
    },
    {
      "epoch": 0.10787,
      "grad_norm": 0.86760847992642,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 10787
    },
    {
      "epoch": 0.10788,
      "grad_norm": 0.9992650939886262,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 10788
    },
    {
      "epoch": 0.10789,
      "grad_norm": 1.2753021365318398,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 10789
    },
    {
      "epoch": 0.1079,
      "grad_norm": 1.0105579588137055,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 10790
    },
    {
      "epoch": 0.10791,
      "grad_norm": 1.243559766886193,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 10791
    },
    {
      "epoch": 0.10792,
      "grad_norm": 1.1291284966148707,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 10792
    },
    {
      "epoch": 0.10793,
      "grad_norm": 1.1994369192885233,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 10793
    },
    {
      "epoch": 0.10794,
      "grad_norm": 0.9428615077798324,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 10794
    },
    {
      "epoch": 0.10795,
      "grad_norm": 0.9872260672261326,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 10795
    },
    {
      "epoch": 0.10796,
      "grad_norm": 0.9935783558284699,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 10796
    },
    {
      "epoch": 0.10797,
      "grad_norm": 1.0810295695783516,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 10797
    },
    {
      "epoch": 0.10798,
      "grad_norm": 1.0770398752097612,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 10798
    },
    {
      "epoch": 0.10799,
      "grad_norm": 1.1063950750874518,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 10799
    },
    {
      "epoch": 0.108,
      "grad_norm": 1.0326094754145043,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 10800
    },
    {
      "epoch": 0.10801,
      "grad_norm": 1.0588200103733418,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 10801
    },
    {
      "epoch": 0.10802,
      "grad_norm": 1.1974743622830182,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 10802
    },
    {
      "epoch": 0.10803,
      "grad_norm": 1.0601372020160826,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 10803
    },
    {
      "epoch": 0.10804,
      "grad_norm": 1.1471872600753268,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 10804
    },
    {
      "epoch": 0.10805,
      "grad_norm": 1.2004049726774384,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 10805
    },
    {
      "epoch": 0.10806,
      "grad_norm": 0.9872725686808971,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 10806
    },
    {
      "epoch": 0.10807,
      "grad_norm": 1.1116803409225564,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 10807
    },
    {
      "epoch": 0.10808,
      "grad_norm": 1.0580006606096264,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 10808
    },
    {
      "epoch": 0.10809,
      "grad_norm": 1.102724868900808,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10809
    },
    {
      "epoch": 0.1081,
      "grad_norm": 1.073448446686264,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 10810
    },
    {
      "epoch": 0.10811,
      "grad_norm": 0.9747513001123251,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 10811
    },
    {
      "epoch": 0.10812,
      "grad_norm": 0.967118322196395,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 10812
    },
    {
      "epoch": 0.10813,
      "grad_norm": 0.9459854684115462,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 10813
    },
    {
      "epoch": 0.10814,
      "grad_norm": 1.1274264516553678,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 10814
    },
    {
      "epoch": 0.10815,
      "grad_norm": 1.0099841714367115,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 10815
    },
    {
      "epoch": 0.10816,
      "grad_norm": 1.1528605178413807,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 10816
    },
    {
      "epoch": 0.10817,
      "grad_norm": 1.3025341293550339,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 10817
    },
    {
      "epoch": 0.10818,
      "grad_norm": 1.1569271056419455,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 10818
    },
    {
      "epoch": 0.10819,
      "grad_norm": 1.013481567794959,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 10819
    },
    {
      "epoch": 0.1082,
      "grad_norm": 1.007612982502211,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 10820
    },
    {
      "epoch": 0.10821,
      "grad_norm": 0.9647199516942806,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 10821
    },
    {
      "epoch": 0.10822,
      "grad_norm": 0.9587875659941222,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 10822
    },
    {
      "epoch": 0.10823,
      "grad_norm": 0.8512682493653407,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 10823
    },
    {
      "epoch": 0.10824,
      "grad_norm": 0.8911143440247272,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 10824
    },
    {
      "epoch": 0.10825,
      "grad_norm": 1.0269152554885081,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 10825
    },
    {
      "epoch": 0.10826,
      "grad_norm": 1.0423619545724405,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 10826
    },
    {
      "epoch": 0.10827,
      "grad_norm": 1.1771294384202926,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 10827
    },
    {
      "epoch": 0.10828,
      "grad_norm": 1.2582773741455715,
      "learning_rate": 0.003,
      "loss": 4.1221,
      "step": 10828
    },
    {
      "epoch": 0.10829,
      "grad_norm": 1.0786632104433649,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 10829
    },
    {
      "epoch": 0.1083,
      "grad_norm": 1.1644041071796276,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 10830
    },
    {
      "epoch": 0.10831,
      "grad_norm": 1.023753602537439,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 10831
    },
    {
      "epoch": 0.10832,
      "grad_norm": 1.0261082103416639,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 10832
    },
    {
      "epoch": 0.10833,
      "grad_norm": 1.1215552384510263,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10833
    },
    {
      "epoch": 0.10834,
      "grad_norm": 1.065027071141165,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 10834
    },
    {
      "epoch": 0.10835,
      "grad_norm": 1.0766952436089778,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10835
    },
    {
      "epoch": 0.10836,
      "grad_norm": 1.0098706745625174,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 10836
    },
    {
      "epoch": 0.10837,
      "grad_norm": 1.1013680159530501,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 10837
    },
    {
      "epoch": 0.10838,
      "grad_norm": 1.1065831185757904,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 10838
    },
    {
      "epoch": 0.10839,
      "grad_norm": 1.1637812774623542,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10839
    },
    {
      "epoch": 0.1084,
      "grad_norm": 1.148431636090298,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 10840
    },
    {
      "epoch": 0.10841,
      "grad_norm": 1.2366796360721557,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 10841
    },
    {
      "epoch": 0.10842,
      "grad_norm": 1.0577104738593404,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 10842
    },
    {
      "epoch": 0.10843,
      "grad_norm": 1.0484337010776401,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 10843
    },
    {
      "epoch": 0.10844,
      "grad_norm": 1.212192275320524,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 10844
    },
    {
      "epoch": 0.10845,
      "grad_norm": 0.9957500087946555,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 10845
    },
    {
      "epoch": 0.10846,
      "grad_norm": 1.0979094349107996,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 10846
    },
    {
      "epoch": 0.10847,
      "grad_norm": 0.9969902766004411,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 10847
    },
    {
      "epoch": 0.10848,
      "grad_norm": 1.252783740484669,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 10848
    },
    {
      "epoch": 0.10849,
      "grad_norm": 1.139595362233358,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 10849
    },
    {
      "epoch": 0.1085,
      "grad_norm": 1.1030600445586416,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10850
    },
    {
      "epoch": 0.10851,
      "grad_norm": 1.0384848869505103,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 10851
    },
    {
      "epoch": 0.10852,
      "grad_norm": 0.9121348149264054,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 10852
    },
    {
      "epoch": 0.10853,
      "grad_norm": 0.8855881354922143,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 10853
    },
    {
      "epoch": 0.10854,
      "grad_norm": 1.1175936911572262,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 10854
    },
    {
      "epoch": 0.10855,
      "grad_norm": 1.0245129092214602,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 10855
    },
    {
      "epoch": 0.10856,
      "grad_norm": 1.119836326344118,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 10856
    },
    {
      "epoch": 0.10857,
      "grad_norm": 0.9836679344334276,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 10857
    },
    {
      "epoch": 0.10858,
      "grad_norm": 0.9511844758669555,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 10858
    },
    {
      "epoch": 0.10859,
      "grad_norm": 0.9264557593108973,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 10859
    },
    {
      "epoch": 0.1086,
      "grad_norm": 1.1703961872501316,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 10860
    },
    {
      "epoch": 0.10861,
      "grad_norm": 1.2581401538656547,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 10861
    },
    {
      "epoch": 0.10862,
      "grad_norm": 1.2444168962017241,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 10862
    },
    {
      "epoch": 0.10863,
      "grad_norm": 0.9504300595021578,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 10863
    },
    {
      "epoch": 0.10864,
      "grad_norm": 0.9554197526092899,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 10864
    },
    {
      "epoch": 0.10865,
      "grad_norm": 1.2014661969325289,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 10865
    },
    {
      "epoch": 0.10866,
      "grad_norm": 0.952712085935606,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 10866
    },
    {
      "epoch": 0.10867,
      "grad_norm": 0.8805538915855081,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 10867
    },
    {
      "epoch": 0.10868,
      "grad_norm": 1.1935095308976902,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 10868
    },
    {
      "epoch": 0.10869,
      "grad_norm": 1.0856390824449236,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 10869
    },
    {
      "epoch": 0.1087,
      "grad_norm": 1.1514811577286823,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 10870
    },
    {
      "epoch": 0.10871,
      "grad_norm": 1.0182709548176088,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 10871
    },
    {
      "epoch": 0.10872,
      "grad_norm": 1.0620238002252582,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 10872
    },
    {
      "epoch": 0.10873,
      "grad_norm": 1.045749830039589,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 10873
    },
    {
      "epoch": 0.10874,
      "grad_norm": 1.2175684337392043,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 10874
    },
    {
      "epoch": 0.10875,
      "grad_norm": 0.9961669916986353,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 10875
    },
    {
      "epoch": 0.10876,
      "grad_norm": 1.1138750412136187,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 10876
    },
    {
      "epoch": 0.10877,
      "grad_norm": 0.9856995941163944,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 10877
    },
    {
      "epoch": 0.10878,
      "grad_norm": 1.1563223710627515,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 10878
    },
    {
      "epoch": 0.10879,
      "grad_norm": 0.8985461595260961,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10879
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.9295949582223914,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 10880
    },
    {
      "epoch": 0.10881,
      "grad_norm": 1.0211027521470517,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 10881
    },
    {
      "epoch": 0.10882,
      "grad_norm": 1.0248546264914715,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 10882
    },
    {
      "epoch": 0.10883,
      "grad_norm": 1.0416183187131618,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 10883
    },
    {
      "epoch": 0.10884,
      "grad_norm": 1.164528405126475,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 10884
    },
    {
      "epoch": 0.10885,
      "grad_norm": 0.9203263062179105,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 10885
    },
    {
      "epoch": 0.10886,
      "grad_norm": 1.3593110655231138,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 10886
    },
    {
      "epoch": 0.10887,
      "grad_norm": 1.2827560760894003,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 10887
    },
    {
      "epoch": 0.10888,
      "grad_norm": 0.8780693627092314,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 10888
    },
    {
      "epoch": 0.10889,
      "grad_norm": 0.8007569440304311,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 10889
    },
    {
      "epoch": 0.1089,
      "grad_norm": 0.89074198244628,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 10890
    },
    {
      "epoch": 0.10891,
      "grad_norm": 0.9352229500355524,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 10891
    },
    {
      "epoch": 0.10892,
      "grad_norm": 1.0574403109866946,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 10892
    },
    {
      "epoch": 0.10893,
      "grad_norm": 1.213448491555586,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 10893
    },
    {
      "epoch": 0.10894,
      "grad_norm": 1.0633428552392232,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 10894
    },
    {
      "epoch": 0.10895,
      "grad_norm": 1.2172544183203573,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 10895
    },
    {
      "epoch": 0.10896,
      "grad_norm": 1.0344575054322562,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10896
    },
    {
      "epoch": 0.10897,
      "grad_norm": 1.2888555732863958,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 10897
    },
    {
      "epoch": 0.10898,
      "grad_norm": 0.9468165429324117,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 10898
    },
    {
      "epoch": 0.10899,
      "grad_norm": 0.9937944702570864,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 10899
    },
    {
      "epoch": 0.109,
      "grad_norm": 1.0784998893580096,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 10900
    },
    {
      "epoch": 0.10901,
      "grad_norm": 0.9586474239659144,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 10901
    },
    {
      "epoch": 0.10902,
      "grad_norm": 1.2612176071375183,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 10902
    },
    {
      "epoch": 0.10903,
      "grad_norm": 1.2552020722155313,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 10903
    },
    {
      "epoch": 0.10904,
      "grad_norm": 1.1986973542170098,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 10904
    },
    {
      "epoch": 0.10905,
      "grad_norm": 1.1362325615742652,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 10905
    },
    {
      "epoch": 0.10906,
      "grad_norm": 0.9680044227786018,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 10906
    },
    {
      "epoch": 0.10907,
      "grad_norm": 1.266038411305887,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 10907
    },
    {
      "epoch": 0.10908,
      "grad_norm": 0.8194233694583234,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 10908
    },
    {
      "epoch": 0.10909,
      "grad_norm": 0.9516173062517561,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 10909
    },
    {
      "epoch": 0.1091,
      "grad_norm": 1.190532497934995,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 10910
    },
    {
      "epoch": 0.10911,
      "grad_norm": 1.062507805821197,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 10911
    },
    {
      "epoch": 0.10912,
      "grad_norm": 0.935468487016387,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 10912
    },
    {
      "epoch": 0.10913,
      "grad_norm": 1.0332631560712264,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 10913
    },
    {
      "epoch": 0.10914,
      "grad_norm": 1.2241198062781187,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 10914
    },
    {
      "epoch": 0.10915,
      "grad_norm": 0.9200270302646114,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 10915
    },
    {
      "epoch": 0.10916,
      "grad_norm": 1.020811811570247,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 10916
    },
    {
      "epoch": 0.10917,
      "grad_norm": 1.1022325333127732,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 10917
    },
    {
      "epoch": 0.10918,
      "grad_norm": 1.1836549733574424,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 10918
    },
    {
      "epoch": 0.10919,
      "grad_norm": 1.170550896066117,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 10919
    },
    {
      "epoch": 0.1092,
      "grad_norm": 0.9028965097833227,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 10920
    },
    {
      "epoch": 0.10921,
      "grad_norm": 1.0067120812866543,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 10921
    },
    {
      "epoch": 0.10922,
      "grad_norm": 1.1613890341872082,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 10922
    },
    {
      "epoch": 0.10923,
      "grad_norm": 1.0303548671060356,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 10923
    },
    {
      "epoch": 0.10924,
      "grad_norm": 1.058935100561485,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 10924
    },
    {
      "epoch": 0.10925,
      "grad_norm": 0.982292907773461,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 10925
    },
    {
      "epoch": 0.10926,
      "grad_norm": 1.2578268516771849,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 10926
    },
    {
      "epoch": 0.10927,
      "grad_norm": 1.0731552123121946,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 10927
    },
    {
      "epoch": 0.10928,
      "grad_norm": 1.0945521448743394,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 10928
    },
    {
      "epoch": 0.10929,
      "grad_norm": 1.200417936854523,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 10929
    },
    {
      "epoch": 0.1093,
      "grad_norm": 0.958607621161436,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 10930
    },
    {
      "epoch": 0.10931,
      "grad_norm": 1.066110489708051,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 10931
    },
    {
      "epoch": 0.10932,
      "grad_norm": 0.9762745373990046,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10932
    },
    {
      "epoch": 0.10933,
      "grad_norm": 1.2896825869107031,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 10933
    },
    {
      "epoch": 0.10934,
      "grad_norm": 0.9507052260715008,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 10934
    },
    {
      "epoch": 0.10935,
      "grad_norm": 1.3124453372335498,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 10935
    },
    {
      "epoch": 0.10936,
      "grad_norm": 0.852998866567921,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 10936
    },
    {
      "epoch": 0.10937,
      "grad_norm": 0.9231917675292967,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 10937
    },
    {
      "epoch": 0.10938,
      "grad_norm": 0.9979776375461109,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 10938
    },
    {
      "epoch": 0.10939,
      "grad_norm": 1.2108718834019867,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 10939
    },
    {
      "epoch": 0.1094,
      "grad_norm": 1.0380696008785413,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 10940
    },
    {
      "epoch": 0.10941,
      "grad_norm": 1.2623260185642127,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 10941
    },
    {
      "epoch": 0.10942,
      "grad_norm": 0.9668178565519607,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 10942
    },
    {
      "epoch": 0.10943,
      "grad_norm": 1.0862618944194453,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 10943
    },
    {
      "epoch": 0.10944,
      "grad_norm": 0.8509290193448615,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 10944
    },
    {
      "epoch": 0.10945,
      "grad_norm": 0.9300723483707174,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 10945
    },
    {
      "epoch": 0.10946,
      "grad_norm": 1.1516366707370755,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 10946
    },
    {
      "epoch": 0.10947,
      "grad_norm": 1.0628248021466853,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 10947
    },
    {
      "epoch": 0.10948,
      "grad_norm": 1.1222961873727684,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 10948
    },
    {
      "epoch": 0.10949,
      "grad_norm": 1.1146700599169557,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 10949
    },
    {
      "epoch": 0.1095,
      "grad_norm": 0.9290653295123669,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 10950
    },
    {
      "epoch": 0.10951,
      "grad_norm": 1.0203598045295497,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 10951
    },
    {
      "epoch": 0.10952,
      "grad_norm": 1.3264874444048882,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10952
    },
    {
      "epoch": 0.10953,
      "grad_norm": 0.97658568631152,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 10953
    },
    {
      "epoch": 0.10954,
      "grad_norm": 1.3420224388646687,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 10954
    },
    {
      "epoch": 0.10955,
      "grad_norm": 1.0692922240105442,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 10955
    },
    {
      "epoch": 0.10956,
      "grad_norm": 1.1967597984232787,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 10956
    },
    {
      "epoch": 0.10957,
      "grad_norm": 1.0369332211726097,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 10957
    },
    {
      "epoch": 0.10958,
      "grad_norm": 1.0667165567030539,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 10958
    },
    {
      "epoch": 0.10959,
      "grad_norm": 1.0467307328354403,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 10959
    },
    {
      "epoch": 0.1096,
      "grad_norm": 1.121746796410831,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 10960
    },
    {
      "epoch": 0.10961,
      "grad_norm": 1.1066837634594826,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 10961
    },
    {
      "epoch": 0.10962,
      "grad_norm": 1.0270607977048283,
      "learning_rate": 0.003,
      "loss": 4.1035,
      "step": 10962
    },
    {
      "epoch": 0.10963,
      "grad_norm": 0.9476690702771576,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 10963
    },
    {
      "epoch": 0.10964,
      "grad_norm": 1.0618940504526093,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 10964
    },
    {
      "epoch": 0.10965,
      "grad_norm": 1.2518627931195352,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 10965
    },
    {
      "epoch": 0.10966,
      "grad_norm": 1.1576505895330684,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 10966
    },
    {
      "epoch": 0.10967,
      "grad_norm": 0.8989888307735112,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 10967
    },
    {
      "epoch": 0.10968,
      "grad_norm": 0.9840336646590621,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 10968
    },
    {
      "epoch": 0.10969,
      "grad_norm": 1.0738411321450865,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 10969
    },
    {
      "epoch": 0.1097,
      "grad_norm": 1.1647834960677774,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 10970
    },
    {
      "epoch": 0.10971,
      "grad_norm": 1.0416427485833692,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 10971
    },
    {
      "epoch": 0.10972,
      "grad_norm": 1.1140608228352962,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 10972
    },
    {
      "epoch": 0.10973,
      "grad_norm": 1.0655226246249958,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 10973
    },
    {
      "epoch": 0.10974,
      "grad_norm": 0.9914871581858634,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 10974
    },
    {
      "epoch": 0.10975,
      "grad_norm": 0.9698908639834418,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 10975
    },
    {
      "epoch": 0.10976,
      "grad_norm": 1.241829245843147,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 10976
    },
    {
      "epoch": 0.10977,
      "grad_norm": 1.0944763139029807,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 10977
    },
    {
      "epoch": 0.10978,
      "grad_norm": 1.0969877929366048,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 10978
    },
    {
      "epoch": 0.10979,
      "grad_norm": 1.1257370588440034,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 10979
    },
    {
      "epoch": 0.1098,
      "grad_norm": 0.9569312789633604,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 10980
    },
    {
      "epoch": 0.10981,
      "grad_norm": 1.1566047577430134,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 10981
    },
    {
      "epoch": 0.10982,
      "grad_norm": 1.0417786216342104,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 10982
    },
    {
      "epoch": 0.10983,
      "grad_norm": 1.212286634724521,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 10983
    },
    {
      "epoch": 0.10984,
      "grad_norm": 1.0337908704581455,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 10984
    },
    {
      "epoch": 0.10985,
      "grad_norm": 1.1192563238020419,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 10985
    },
    {
      "epoch": 0.10986,
      "grad_norm": 0.965393768299253,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 10986
    },
    {
      "epoch": 0.10987,
      "grad_norm": 1.2578089988837124,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 10987
    },
    {
      "epoch": 0.10988,
      "grad_norm": 0.8914465824720181,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 10988
    },
    {
      "epoch": 0.10989,
      "grad_norm": 1.0285487793885861,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 10989
    },
    {
      "epoch": 0.1099,
      "grad_norm": 1.453015158533003,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 10990
    },
    {
      "epoch": 0.10991,
      "grad_norm": 1.0961809557509807,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 10991
    },
    {
      "epoch": 0.10992,
      "grad_norm": 1.2656872601856095,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 10992
    },
    {
      "epoch": 0.10993,
      "grad_norm": 1.0578502295128602,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 10993
    },
    {
      "epoch": 0.10994,
      "grad_norm": 0.9979520192054689,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 10994
    },
    {
      "epoch": 0.10995,
      "grad_norm": 1.1815914008752326,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 10995
    },
    {
      "epoch": 0.10996,
      "grad_norm": 1.0807357231606585,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 10996
    },
    {
      "epoch": 0.10997,
      "grad_norm": 1.012982864418392,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 10997
    },
    {
      "epoch": 0.10998,
      "grad_norm": 1.151115580492071,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 10998
    },
    {
      "epoch": 0.10999,
      "grad_norm": 0.9517386329626287,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 10999
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9617818605873265,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 11000
    },
    {
      "epoch": 0.11001,
      "grad_norm": 1.0372420260839046,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 11001
    },
    {
      "epoch": 0.11002,
      "grad_norm": 1.130538854053489,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 11002
    },
    {
      "epoch": 0.11003,
      "grad_norm": 1.281765199831236,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 11003
    },
    {
      "epoch": 0.11004,
      "grad_norm": 1.0930629510140741,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 11004
    },
    {
      "epoch": 0.11005,
      "grad_norm": 1.0734309697476472,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 11005
    },
    {
      "epoch": 0.11006,
      "grad_norm": 0.9376530387047713,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 11006
    },
    {
      "epoch": 0.11007,
      "grad_norm": 1.0045813713815963,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 11007
    },
    {
      "epoch": 0.11008,
      "grad_norm": 1.1196806585546806,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 11008
    },
    {
      "epoch": 0.11009,
      "grad_norm": 1.0021801710470826,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 11009
    },
    {
      "epoch": 0.1101,
      "grad_norm": 1.1943219579368831,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 11010
    },
    {
      "epoch": 0.11011,
      "grad_norm": 0.9349293317995805,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 11011
    },
    {
      "epoch": 0.11012,
      "grad_norm": 1.0816389241323243,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 11012
    },
    {
      "epoch": 0.11013,
      "grad_norm": 0.8440526839080905,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 11013
    },
    {
      "epoch": 0.11014,
      "grad_norm": 0.9622663832307063,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 11014
    },
    {
      "epoch": 0.11015,
      "grad_norm": 1.0314368411196604,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 11015
    },
    {
      "epoch": 0.11016,
      "grad_norm": 1.0418344216916757,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 11016
    },
    {
      "epoch": 0.11017,
      "grad_norm": 1.109456113280434,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 11017
    },
    {
      "epoch": 0.11018,
      "grad_norm": 1.1317173100404638,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 11018
    },
    {
      "epoch": 0.11019,
      "grad_norm": 1.3365549493223279,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 11019
    },
    {
      "epoch": 0.1102,
      "grad_norm": 1.140450402764878,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 11020
    },
    {
      "epoch": 0.11021,
      "grad_norm": 1.089641104327388,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 11021
    },
    {
      "epoch": 0.11022,
      "grad_norm": 1.0571029585661946,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 11022
    },
    {
      "epoch": 0.11023,
      "grad_norm": 0.9112995162650028,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 11023
    },
    {
      "epoch": 0.11024,
      "grad_norm": 1.0823248116168254,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 11024
    },
    {
      "epoch": 0.11025,
      "grad_norm": 1.269798404786744,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 11025
    },
    {
      "epoch": 0.11026,
      "grad_norm": 1.0179991697387427,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 11026
    },
    {
      "epoch": 0.11027,
      "grad_norm": 1.283354079509163,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 11027
    },
    {
      "epoch": 0.11028,
      "grad_norm": 0.9881442149331964,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 11028
    },
    {
      "epoch": 0.11029,
      "grad_norm": 1.1506448604245498,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 11029
    },
    {
      "epoch": 0.1103,
      "grad_norm": 1.0701109805534972,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 11030
    },
    {
      "epoch": 0.11031,
      "grad_norm": 1.045008921347305,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 11031
    },
    {
      "epoch": 0.11032,
      "grad_norm": 0.9940343193835519,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 11032
    },
    {
      "epoch": 0.11033,
      "grad_norm": 1.2452726277820456,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 11033
    },
    {
      "epoch": 0.11034,
      "grad_norm": 0.8438970698928558,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 11034
    },
    {
      "epoch": 0.11035,
      "grad_norm": 1.08785159976017,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 11035
    },
    {
      "epoch": 0.11036,
      "grad_norm": 1.3447184537992016,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 11036
    },
    {
      "epoch": 0.11037,
      "grad_norm": 0.879399892817881,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 11037
    },
    {
      "epoch": 0.11038,
      "grad_norm": 0.9955476735433527,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 11038
    },
    {
      "epoch": 0.11039,
      "grad_norm": 1.3410411619156724,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 11039
    },
    {
      "epoch": 0.1104,
      "grad_norm": 0.823843991929917,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 11040
    },
    {
      "epoch": 0.11041,
      "grad_norm": 0.957799568021374,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 11041
    },
    {
      "epoch": 0.11042,
      "grad_norm": 1.2027831600136363,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 11042
    },
    {
      "epoch": 0.11043,
      "grad_norm": 1.2711340978021695,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 11043
    },
    {
      "epoch": 0.11044,
      "grad_norm": 0.9926544740714466,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 11044
    },
    {
      "epoch": 0.11045,
      "grad_norm": 0.9737001715578901,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 11045
    },
    {
      "epoch": 0.11046,
      "grad_norm": 1.025534429126934,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 11046
    },
    {
      "epoch": 0.11047,
      "grad_norm": 1.0178943889938226,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 11047
    },
    {
      "epoch": 0.11048,
      "grad_norm": 1.1132590043193709,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 11048
    },
    {
      "epoch": 0.11049,
      "grad_norm": 1.1688632704206605,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 11049
    },
    {
      "epoch": 0.1105,
      "grad_norm": 0.9935437751692451,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 11050
    },
    {
      "epoch": 0.11051,
      "grad_norm": 0.9625585983183765,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 11051
    },
    {
      "epoch": 0.11052,
      "grad_norm": 1.0492638687163323,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 11052
    },
    {
      "epoch": 0.11053,
      "grad_norm": 1.1494360439828093,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 11053
    },
    {
      "epoch": 0.11054,
      "grad_norm": 1.1748136441795494,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 11054
    },
    {
      "epoch": 0.11055,
      "grad_norm": 1.1836800651405188,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 11055
    },
    {
      "epoch": 0.11056,
      "grad_norm": 1.0175350898679119,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 11056
    },
    {
      "epoch": 0.11057,
      "grad_norm": 1.1406708256575615,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 11057
    },
    {
      "epoch": 0.11058,
      "grad_norm": 0.9332127881921526,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 11058
    },
    {
      "epoch": 0.11059,
      "grad_norm": 0.8981176131082979,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 11059
    },
    {
      "epoch": 0.1106,
      "grad_norm": 1.067628186409739,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 11060
    },
    {
      "epoch": 0.11061,
      "grad_norm": 1.2663751330065962,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 11061
    },
    {
      "epoch": 0.11062,
      "grad_norm": 0.9825686470798557,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 11062
    },
    {
      "epoch": 0.11063,
      "grad_norm": 1.339545896310836,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11063
    },
    {
      "epoch": 0.11064,
      "grad_norm": 0.8470417821578471,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 11064
    },
    {
      "epoch": 0.11065,
      "grad_norm": 1.0311320495760354,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 11065
    },
    {
      "epoch": 0.11066,
      "grad_norm": 1.275046723295672,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 11066
    },
    {
      "epoch": 0.11067,
      "grad_norm": 0.9743425862414491,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 11067
    },
    {
      "epoch": 0.11068,
      "grad_norm": 1.1291392189378937,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 11068
    },
    {
      "epoch": 0.11069,
      "grad_norm": 1.0381149065784203,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 11069
    },
    {
      "epoch": 0.1107,
      "grad_norm": 1.0261726634476829,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 11070
    },
    {
      "epoch": 0.11071,
      "grad_norm": 1.2432042063184994,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 11071
    },
    {
      "epoch": 0.11072,
      "grad_norm": 1.00271066051892,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 11072
    },
    {
      "epoch": 0.11073,
      "grad_norm": 1.2876725887592981,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 11073
    },
    {
      "epoch": 0.11074,
      "grad_norm": 1.0237671279976865,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 11074
    },
    {
      "epoch": 0.11075,
      "grad_norm": 1.1387702287135297,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 11075
    },
    {
      "epoch": 0.11076,
      "grad_norm": 1.082095577956758,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 11076
    },
    {
      "epoch": 0.11077,
      "grad_norm": 1.075721944065049,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 11077
    },
    {
      "epoch": 0.11078,
      "grad_norm": 1.2229269824042177,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 11078
    },
    {
      "epoch": 0.11079,
      "grad_norm": 0.9430305345548962,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 11079
    },
    {
      "epoch": 0.1108,
      "grad_norm": 1.048335653430088,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 11080
    },
    {
      "epoch": 0.11081,
      "grad_norm": 1.0976125542943163,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 11081
    },
    {
      "epoch": 0.11082,
      "grad_norm": 1.1782408105639972,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 11082
    },
    {
      "epoch": 0.11083,
      "grad_norm": 1.0710585939603285,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 11083
    },
    {
      "epoch": 0.11084,
      "grad_norm": 1.294803945413708,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 11084
    },
    {
      "epoch": 0.11085,
      "grad_norm": 1.078729240870453,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 11085
    },
    {
      "epoch": 0.11086,
      "grad_norm": 1.1632020820573332,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 11086
    },
    {
      "epoch": 0.11087,
      "grad_norm": 0.9669066808716378,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 11087
    },
    {
      "epoch": 0.11088,
      "grad_norm": 1.0850182096865149,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 11088
    },
    {
      "epoch": 0.11089,
      "grad_norm": 1.2549161161581106,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 11089
    },
    {
      "epoch": 0.1109,
      "grad_norm": 1.0164081938030312,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 11090
    },
    {
      "epoch": 0.11091,
      "grad_norm": 1.1071292805706272,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 11091
    },
    {
      "epoch": 0.11092,
      "grad_norm": 1.1062151657057917,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 11092
    },
    {
      "epoch": 0.11093,
      "grad_norm": 1.1232047830134266,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 11093
    },
    {
      "epoch": 0.11094,
      "grad_norm": 0.8778638234580115,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 11094
    },
    {
      "epoch": 0.11095,
      "grad_norm": 0.930227028930981,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 11095
    },
    {
      "epoch": 0.11096,
      "grad_norm": 1.0920410338682343,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 11096
    },
    {
      "epoch": 0.11097,
      "grad_norm": 1.2024142449099442,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 11097
    },
    {
      "epoch": 0.11098,
      "grad_norm": 1.1127225562622876,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 11098
    },
    {
      "epoch": 0.11099,
      "grad_norm": 1.0323801131506445,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 11099
    },
    {
      "epoch": 0.111,
      "grad_norm": 1.1710136677384504,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 11100
    },
    {
      "epoch": 0.11101,
      "grad_norm": 0.8840824981939511,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 11101
    },
    {
      "epoch": 0.11102,
      "grad_norm": 0.8862206579405567,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 11102
    },
    {
      "epoch": 0.11103,
      "grad_norm": 1.1293426398135202,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 11103
    },
    {
      "epoch": 0.11104,
      "grad_norm": 1.1343343190329958,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 11104
    },
    {
      "epoch": 0.11105,
      "grad_norm": 1.073432893320314,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 11105
    },
    {
      "epoch": 0.11106,
      "grad_norm": 1.0142594610929732,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 11106
    },
    {
      "epoch": 0.11107,
      "grad_norm": 1.154167579934993,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 11107
    },
    {
      "epoch": 0.11108,
      "grad_norm": 1.0955201894180502,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 11108
    },
    {
      "epoch": 0.11109,
      "grad_norm": 1.0748307909472112,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 11109
    },
    {
      "epoch": 0.1111,
      "grad_norm": 1.1984134922116756,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 11110
    },
    {
      "epoch": 0.11111,
      "grad_norm": 1.023062750460469,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 11111
    },
    {
      "epoch": 0.11112,
      "grad_norm": 1.148035200097653,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 11112
    },
    {
      "epoch": 0.11113,
      "grad_norm": 0.9643801573215892,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 11113
    },
    {
      "epoch": 0.11114,
      "grad_norm": 1.1451404200430764,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 11114
    },
    {
      "epoch": 0.11115,
      "grad_norm": 1.1395071268509465,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 11115
    },
    {
      "epoch": 0.11116,
      "grad_norm": 1.3169506565149234,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 11116
    },
    {
      "epoch": 0.11117,
      "grad_norm": 1.2272020840995974,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 11117
    },
    {
      "epoch": 0.11118,
      "grad_norm": 0.9150172409767166,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 11118
    },
    {
      "epoch": 0.11119,
      "grad_norm": 1.0060344474151803,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 11119
    },
    {
      "epoch": 0.1112,
      "grad_norm": 1.1316813330593096,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 11120
    },
    {
      "epoch": 0.11121,
      "grad_norm": 1.1888005089421085,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 11121
    },
    {
      "epoch": 0.11122,
      "grad_norm": 1.1305955226981466,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 11122
    },
    {
      "epoch": 0.11123,
      "grad_norm": 1.0833462030744596,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 11123
    },
    {
      "epoch": 0.11124,
      "grad_norm": 1.0638925302665134,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 11124
    },
    {
      "epoch": 0.11125,
      "grad_norm": 1.0069967290408879,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 11125
    },
    {
      "epoch": 0.11126,
      "grad_norm": 1.2261766349487866,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 11126
    },
    {
      "epoch": 0.11127,
      "grad_norm": 1.0028029989186422,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 11127
    },
    {
      "epoch": 0.11128,
      "grad_norm": 1.1780495201828052,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 11128
    },
    {
      "epoch": 0.11129,
      "grad_norm": 0.8299165332619102,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 11129
    },
    {
      "epoch": 0.1113,
      "grad_norm": 0.8202902758038285,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 11130
    },
    {
      "epoch": 0.11131,
      "grad_norm": 0.8321204036949327,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 11131
    },
    {
      "epoch": 0.11132,
      "grad_norm": 0.9729737967538941,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 11132
    },
    {
      "epoch": 0.11133,
      "grad_norm": 1.3548737367613601,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 11133
    },
    {
      "epoch": 0.11134,
      "grad_norm": 1.1233322420616223,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 11134
    },
    {
      "epoch": 0.11135,
      "grad_norm": 1.2626200980787057,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 11135
    },
    {
      "epoch": 0.11136,
      "grad_norm": 0.9766533789117673,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 11136
    },
    {
      "epoch": 0.11137,
      "grad_norm": 1.021474893266102,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 11137
    },
    {
      "epoch": 0.11138,
      "grad_norm": 1.1274478767407865,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11138
    },
    {
      "epoch": 0.11139,
      "grad_norm": 1.1267855782303196,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 11139
    },
    {
      "epoch": 0.1114,
      "grad_norm": 1.0560581171106265,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 11140
    },
    {
      "epoch": 0.11141,
      "grad_norm": 1.1071428100121916,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 11141
    },
    {
      "epoch": 0.11142,
      "grad_norm": 1.049316649242421,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 11142
    },
    {
      "epoch": 0.11143,
      "grad_norm": 1.1684822588023638,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 11143
    },
    {
      "epoch": 0.11144,
      "grad_norm": 1.0905646220362137,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 11144
    },
    {
      "epoch": 0.11145,
      "grad_norm": 1.1383713278971077,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 11145
    },
    {
      "epoch": 0.11146,
      "grad_norm": 1.0662303902298917,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 11146
    },
    {
      "epoch": 0.11147,
      "grad_norm": 1.136790129915702,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 11147
    },
    {
      "epoch": 0.11148,
      "grad_norm": 1.1811233013860687,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 11148
    },
    {
      "epoch": 0.11149,
      "grad_norm": 1.0329754282368866,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 11149
    },
    {
      "epoch": 0.1115,
      "grad_norm": 1.1162248343049477,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 11150
    },
    {
      "epoch": 0.11151,
      "grad_norm": 1.0803414568954743,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 11151
    },
    {
      "epoch": 0.11152,
      "grad_norm": 0.9621437109509381,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 11152
    },
    {
      "epoch": 0.11153,
      "grad_norm": 1.0074645368691362,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 11153
    },
    {
      "epoch": 0.11154,
      "grad_norm": 1.3297829766491356,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 11154
    },
    {
      "epoch": 0.11155,
      "grad_norm": 0.917150674753661,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 11155
    },
    {
      "epoch": 0.11156,
      "grad_norm": 1.0716683342142723,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 11156
    },
    {
      "epoch": 0.11157,
      "grad_norm": 0.9908296621704735,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 11157
    },
    {
      "epoch": 0.11158,
      "grad_norm": 1.195847729550385,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 11158
    },
    {
      "epoch": 0.11159,
      "grad_norm": 0.9140159093620859,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 11159
    },
    {
      "epoch": 0.1116,
      "grad_norm": 1.0693262870879578,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 11160
    },
    {
      "epoch": 0.11161,
      "grad_norm": 1.3174320581418724,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 11161
    },
    {
      "epoch": 0.11162,
      "grad_norm": 0.8934094400104304,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 11162
    },
    {
      "epoch": 0.11163,
      "grad_norm": 0.9190953694678776,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 11163
    },
    {
      "epoch": 0.11164,
      "grad_norm": 0.9907049440370059,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 11164
    },
    {
      "epoch": 0.11165,
      "grad_norm": 1.1572973589976452,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 11165
    },
    {
      "epoch": 0.11166,
      "grad_norm": 1.2329249741723933,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 11166
    },
    {
      "epoch": 0.11167,
      "grad_norm": 1.0903320835115502,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 11167
    },
    {
      "epoch": 0.11168,
      "grad_norm": 1.2283881015667557,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 11168
    },
    {
      "epoch": 0.11169,
      "grad_norm": 1.052811405237688,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 11169
    },
    {
      "epoch": 0.1117,
      "grad_norm": 1.1879599635362192,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 11170
    },
    {
      "epoch": 0.11171,
      "grad_norm": 1.1562509517323158,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 11171
    },
    {
      "epoch": 0.11172,
      "grad_norm": 1.0746628555807403,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 11172
    },
    {
      "epoch": 0.11173,
      "grad_norm": 1.1465154807026283,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 11173
    },
    {
      "epoch": 0.11174,
      "grad_norm": 1.1107132259817063,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 11174
    },
    {
      "epoch": 0.11175,
      "grad_norm": 1.4460414667703407,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 11175
    },
    {
      "epoch": 0.11176,
      "grad_norm": 0.8687107072072395,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 11176
    },
    {
      "epoch": 0.11177,
      "grad_norm": 0.9451656942827614,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 11177
    },
    {
      "epoch": 0.11178,
      "grad_norm": 1.273691659752799,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 11178
    },
    {
      "epoch": 0.11179,
      "grad_norm": 1.1851241645193193,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 11179
    },
    {
      "epoch": 0.1118,
      "grad_norm": 0.9995561306224513,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 11180
    },
    {
      "epoch": 0.11181,
      "grad_norm": 1.0561104495198952,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 11181
    },
    {
      "epoch": 0.11182,
      "grad_norm": 1.2094738379122607,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 11182
    },
    {
      "epoch": 0.11183,
      "grad_norm": 0.9389470450058727,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 11183
    },
    {
      "epoch": 0.11184,
      "grad_norm": 0.8960144214081149,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 11184
    },
    {
      "epoch": 0.11185,
      "grad_norm": 0.9677063470994468,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 11185
    },
    {
      "epoch": 0.11186,
      "grad_norm": 1.0636641010940282,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 11186
    },
    {
      "epoch": 0.11187,
      "grad_norm": 1.0042556361054829,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 11187
    },
    {
      "epoch": 0.11188,
      "grad_norm": 1.1804908575937623,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 11188
    },
    {
      "epoch": 0.11189,
      "grad_norm": 0.9760623848126584,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 11189
    },
    {
      "epoch": 0.1119,
      "grad_norm": 1.0655847280266737,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 11190
    },
    {
      "epoch": 0.11191,
      "grad_norm": 1.1804808531390547,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 11191
    },
    {
      "epoch": 0.11192,
      "grad_norm": 1.0417522219563624,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 11192
    },
    {
      "epoch": 0.11193,
      "grad_norm": 1.2328046581965018,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 11193
    },
    {
      "epoch": 0.11194,
      "grad_norm": 0.9678481952515546,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 11194
    },
    {
      "epoch": 0.11195,
      "grad_norm": 1.0849728191438266,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 11195
    },
    {
      "epoch": 0.11196,
      "grad_norm": 1.137568879220585,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 11196
    },
    {
      "epoch": 0.11197,
      "grad_norm": 1.0788625907543243,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 11197
    },
    {
      "epoch": 0.11198,
      "grad_norm": 1.200279594942393,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 11198
    },
    {
      "epoch": 0.11199,
      "grad_norm": 1.075283856632629,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11199
    },
    {
      "epoch": 0.112,
      "grad_norm": 1.3496552532591504,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 11200
    },
    {
      "epoch": 0.11201,
      "grad_norm": 0.915277435554514,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 11201
    },
    {
      "epoch": 0.11202,
      "grad_norm": 1.023416164318843,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 11202
    },
    {
      "epoch": 0.11203,
      "grad_norm": 1.0929941854292053,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 11203
    },
    {
      "epoch": 0.11204,
      "grad_norm": 0.9980228419712516,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 11204
    },
    {
      "epoch": 0.11205,
      "grad_norm": 1.1050330304505716,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 11205
    },
    {
      "epoch": 0.11206,
      "grad_norm": 1.031157908059373,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 11206
    },
    {
      "epoch": 0.11207,
      "grad_norm": 0.9567238626654728,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 11207
    },
    {
      "epoch": 0.11208,
      "grad_norm": 1.0268115166473828,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 11208
    },
    {
      "epoch": 0.11209,
      "grad_norm": 1.0830399806039088,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 11209
    },
    {
      "epoch": 0.1121,
      "grad_norm": 1.0100548761394696,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 11210
    },
    {
      "epoch": 0.11211,
      "grad_norm": 1.3731084883013778,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 11211
    },
    {
      "epoch": 0.11212,
      "grad_norm": 0.9029837035173026,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 11212
    },
    {
      "epoch": 0.11213,
      "grad_norm": 0.8742947566558211,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 11213
    },
    {
      "epoch": 0.11214,
      "grad_norm": 1.093073657307132,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 11214
    },
    {
      "epoch": 0.11215,
      "grad_norm": 1.150468961400977,
      "learning_rate": 0.003,
      "loss": 4.1138,
      "step": 11215
    },
    {
      "epoch": 0.11216,
      "grad_norm": 1.1623110083022907,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 11216
    },
    {
      "epoch": 0.11217,
      "grad_norm": 1.0136109306734828,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 11217
    },
    {
      "epoch": 0.11218,
      "grad_norm": 1.2225786094760291,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 11218
    },
    {
      "epoch": 0.11219,
      "grad_norm": 1.0818760771784373,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 11219
    },
    {
      "epoch": 0.1122,
      "grad_norm": 1.0582777227739197,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 11220
    },
    {
      "epoch": 0.11221,
      "grad_norm": 1.0362983069658394,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 11221
    },
    {
      "epoch": 0.11222,
      "grad_norm": 1.2275266816383967,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 11222
    },
    {
      "epoch": 0.11223,
      "grad_norm": 1.0692292653315416,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 11223
    },
    {
      "epoch": 0.11224,
      "grad_norm": 1.1917186048611152,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 11224
    },
    {
      "epoch": 0.11225,
      "grad_norm": 1.0387982212954594,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 11225
    },
    {
      "epoch": 0.11226,
      "grad_norm": 1.3224755641339132,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 11226
    },
    {
      "epoch": 0.11227,
      "grad_norm": 1.119941539039678,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 11227
    },
    {
      "epoch": 0.11228,
      "grad_norm": 1.2446054902205859,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 11228
    },
    {
      "epoch": 0.11229,
      "grad_norm": 1.159551158039648,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 11229
    },
    {
      "epoch": 0.1123,
      "grad_norm": 0.9846382272809041,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 11230
    },
    {
      "epoch": 0.11231,
      "grad_norm": 1.0618869653269813,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 11231
    },
    {
      "epoch": 0.11232,
      "grad_norm": 1.0924187874072424,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 11232
    },
    {
      "epoch": 0.11233,
      "grad_norm": 1.043653612630689,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 11233
    },
    {
      "epoch": 0.11234,
      "grad_norm": 1.2055624769452127,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 11234
    },
    {
      "epoch": 0.11235,
      "grad_norm": 1.0189350713213847,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 11235
    },
    {
      "epoch": 0.11236,
      "grad_norm": 1.1370600949836156,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 11236
    },
    {
      "epoch": 0.11237,
      "grad_norm": 1.0502229610345457,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 11237
    },
    {
      "epoch": 0.11238,
      "grad_norm": 1.3088946159549792,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 11238
    },
    {
      "epoch": 0.11239,
      "grad_norm": 0.9674213958822012,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 11239
    },
    {
      "epoch": 0.1124,
      "grad_norm": 1.1108252118839896,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 11240
    },
    {
      "epoch": 0.11241,
      "grad_norm": 1.0788904121254563,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 11241
    },
    {
      "epoch": 0.11242,
      "grad_norm": 0.8805638731782014,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 11242
    },
    {
      "epoch": 0.11243,
      "grad_norm": 0.953293260942598,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 11243
    },
    {
      "epoch": 0.11244,
      "grad_norm": 1.0944982541263293,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 11244
    },
    {
      "epoch": 0.11245,
      "grad_norm": 0.9878904899811639,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 11245
    },
    {
      "epoch": 0.11246,
      "grad_norm": 1.0169671093460908,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 11246
    },
    {
      "epoch": 0.11247,
      "grad_norm": 0.9576195622409958,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 11247
    },
    {
      "epoch": 0.11248,
      "grad_norm": 1.1663956402495321,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 11248
    },
    {
      "epoch": 0.11249,
      "grad_norm": 1.0707941267852739,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 11249
    },
    {
      "epoch": 0.1125,
      "grad_norm": 1.459014431371671,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 11250
    },
    {
      "epoch": 0.11251,
      "grad_norm": 1.1274606042467634,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 11251
    },
    {
      "epoch": 0.11252,
      "grad_norm": 1.110095946102636,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 11252
    },
    {
      "epoch": 0.11253,
      "grad_norm": 1.1658531601254527,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 11253
    },
    {
      "epoch": 0.11254,
      "grad_norm": 0.9050266401818737,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 11254
    },
    {
      "epoch": 0.11255,
      "grad_norm": 0.9758153208182531,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 11255
    },
    {
      "epoch": 0.11256,
      "grad_norm": 1.0385188029198487,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 11256
    },
    {
      "epoch": 0.11257,
      "grad_norm": 1.1873576303217848,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 11257
    },
    {
      "epoch": 0.11258,
      "grad_norm": 1.0026884103018077,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 11258
    },
    {
      "epoch": 0.11259,
      "grad_norm": 1.1656263767341002,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 11259
    },
    {
      "epoch": 0.1126,
      "grad_norm": 1.2347238493290456,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 11260
    },
    {
      "epoch": 0.11261,
      "grad_norm": 1.1284817796438356,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 11261
    },
    {
      "epoch": 0.11262,
      "grad_norm": 1.1365957031399199,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 11262
    },
    {
      "epoch": 0.11263,
      "grad_norm": 0.9204181849384226,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 11263
    },
    {
      "epoch": 0.11264,
      "grad_norm": 0.8164884905461206,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 11264
    },
    {
      "epoch": 0.11265,
      "grad_norm": 0.8990747848873674,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 11265
    },
    {
      "epoch": 0.11266,
      "grad_norm": 1.0489106348607906,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 11266
    },
    {
      "epoch": 0.11267,
      "grad_norm": 1.1865102633489928,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 11267
    },
    {
      "epoch": 0.11268,
      "grad_norm": 1.1467484081323438,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 11268
    },
    {
      "epoch": 0.11269,
      "grad_norm": 1.1552841477401594,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 11269
    },
    {
      "epoch": 0.1127,
      "grad_norm": 1.0772199439954333,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 11270
    },
    {
      "epoch": 0.11271,
      "grad_norm": 1.0118300121513137,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11271
    },
    {
      "epoch": 0.11272,
      "grad_norm": 1.1063044542278664,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 11272
    },
    {
      "epoch": 0.11273,
      "grad_norm": 0.9312828951421663,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 11273
    },
    {
      "epoch": 0.11274,
      "grad_norm": 0.9940260808226513,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 11274
    },
    {
      "epoch": 0.11275,
      "grad_norm": 1.1446577070018433,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 11275
    },
    {
      "epoch": 0.11276,
      "grad_norm": 1.0018077554843705,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 11276
    },
    {
      "epoch": 0.11277,
      "grad_norm": 1.1677079749269628,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 11277
    },
    {
      "epoch": 0.11278,
      "grad_norm": 1.1237391502408127,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 11278
    },
    {
      "epoch": 0.11279,
      "grad_norm": 1.2019139370235028,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 11279
    },
    {
      "epoch": 0.1128,
      "grad_norm": 1.2419978997736385,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 11280
    },
    {
      "epoch": 0.11281,
      "grad_norm": 1.0872558401323524,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 11281
    },
    {
      "epoch": 0.11282,
      "grad_norm": 1.209488425363245,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 11282
    },
    {
      "epoch": 0.11283,
      "grad_norm": 0.9858160666512163,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 11283
    },
    {
      "epoch": 0.11284,
      "grad_norm": 1.00042028888944,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 11284
    },
    {
      "epoch": 0.11285,
      "grad_norm": 1.206729879952282,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 11285
    },
    {
      "epoch": 0.11286,
      "grad_norm": 1.058137393919995,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 11286
    },
    {
      "epoch": 0.11287,
      "grad_norm": 1.1303700238289203,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 11287
    },
    {
      "epoch": 0.11288,
      "grad_norm": 1.0001149604553041,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 11288
    },
    {
      "epoch": 0.11289,
      "grad_norm": 0.8541549247427613,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 11289
    },
    {
      "epoch": 0.1129,
      "grad_norm": 1.0899293245321309,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 11290
    },
    {
      "epoch": 0.11291,
      "grad_norm": 1.2312123995841915,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 11291
    },
    {
      "epoch": 0.11292,
      "grad_norm": 1.0199608911110976,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 11292
    },
    {
      "epoch": 0.11293,
      "grad_norm": 1.1065648579263287,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 11293
    },
    {
      "epoch": 0.11294,
      "grad_norm": 0.9875106628926394,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 11294
    },
    {
      "epoch": 0.11295,
      "grad_norm": 1.0024741305766218,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 11295
    },
    {
      "epoch": 0.11296,
      "grad_norm": 1.0033426301135893,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 11296
    },
    {
      "epoch": 0.11297,
      "grad_norm": 1.1664712252997975,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 11297
    },
    {
      "epoch": 0.11298,
      "grad_norm": 1.034552101124066,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 11298
    },
    {
      "epoch": 0.11299,
      "grad_norm": 1.149005602372779,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 11299
    },
    {
      "epoch": 0.113,
      "grad_norm": 0.9280762195278215,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 11300
    },
    {
      "epoch": 0.11301,
      "grad_norm": 0.9174481792983767,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 11301
    },
    {
      "epoch": 0.11302,
      "grad_norm": 1.355403034506496,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 11302
    },
    {
      "epoch": 0.11303,
      "grad_norm": 1.2515777074789904,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 11303
    },
    {
      "epoch": 0.11304,
      "grad_norm": 1.026679034534568,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 11304
    },
    {
      "epoch": 0.11305,
      "grad_norm": 1.3542331189203112,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 11305
    },
    {
      "epoch": 0.11306,
      "grad_norm": 1.094094766775474,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 11306
    },
    {
      "epoch": 0.11307,
      "grad_norm": 1.149755997638481,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 11307
    },
    {
      "epoch": 0.11308,
      "grad_norm": 0.8949282992535843,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 11308
    },
    {
      "epoch": 0.11309,
      "grad_norm": 0.8965306355096659,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 11309
    },
    {
      "epoch": 0.1131,
      "grad_norm": 0.8035470699791126,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 11310
    },
    {
      "epoch": 0.11311,
      "grad_norm": 0.901000387358812,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 11311
    },
    {
      "epoch": 0.11312,
      "grad_norm": 1.1716063743041008,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 11312
    },
    {
      "epoch": 0.11313,
      "grad_norm": 1.2117160260271929,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 11313
    },
    {
      "epoch": 0.11314,
      "grad_norm": 1.1322856823309642,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 11314
    },
    {
      "epoch": 0.11315,
      "grad_norm": 1.1405877982475148,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11315
    },
    {
      "epoch": 0.11316,
      "grad_norm": 0.9413939056674395,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 11316
    },
    {
      "epoch": 0.11317,
      "grad_norm": 1.1013015545182125,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 11317
    },
    {
      "epoch": 0.11318,
      "grad_norm": 1.0467991658776217,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 11318
    },
    {
      "epoch": 0.11319,
      "grad_norm": 1.1550280753143543,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 11319
    },
    {
      "epoch": 0.1132,
      "grad_norm": 1.1479696936429864,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 11320
    },
    {
      "epoch": 0.11321,
      "grad_norm": 1.0152156602974067,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 11321
    },
    {
      "epoch": 0.11322,
      "grad_norm": 1.075778583656459,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 11322
    },
    {
      "epoch": 0.11323,
      "grad_norm": 1.0869881631034657,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 11323
    },
    {
      "epoch": 0.11324,
      "grad_norm": 1.1804229227385865,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 11324
    },
    {
      "epoch": 0.11325,
      "grad_norm": 0.8810791102719396,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 11325
    },
    {
      "epoch": 0.11326,
      "grad_norm": 1.072467901629413,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 11326
    },
    {
      "epoch": 0.11327,
      "grad_norm": 1.2032702548110166,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 11327
    },
    {
      "epoch": 0.11328,
      "grad_norm": 1.2904192484723571,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 11328
    },
    {
      "epoch": 0.11329,
      "grad_norm": 1.0583280692957269,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 11329
    },
    {
      "epoch": 0.1133,
      "grad_norm": 0.9699677436263262,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 11330
    },
    {
      "epoch": 0.11331,
      "grad_norm": 1.1451909643330345,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 11331
    },
    {
      "epoch": 0.11332,
      "grad_norm": 1.046111033258354,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 11332
    },
    {
      "epoch": 0.11333,
      "grad_norm": 0.9462258269015973,
      "learning_rate": 0.003,
      "loss": 4.0298,
      "step": 11333
    },
    {
      "epoch": 0.11334,
      "grad_norm": 0.9957039438891643,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 11334
    },
    {
      "epoch": 0.11335,
      "grad_norm": 1.2195131489129374,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 11335
    },
    {
      "epoch": 0.11336,
      "grad_norm": 1.033509006350732,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 11336
    },
    {
      "epoch": 0.11337,
      "grad_norm": 1.2648851398653458,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 11337
    },
    {
      "epoch": 0.11338,
      "grad_norm": 1.177703282524506,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 11338
    },
    {
      "epoch": 0.11339,
      "grad_norm": 1.0056354528106204,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 11339
    },
    {
      "epoch": 0.1134,
      "grad_norm": 1.3867214949166673,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 11340
    },
    {
      "epoch": 0.11341,
      "grad_norm": 0.9906858050160079,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 11341
    },
    {
      "epoch": 0.11342,
      "grad_norm": 1.0706164693016034,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 11342
    },
    {
      "epoch": 0.11343,
      "grad_norm": 1.0292726784936213,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 11343
    },
    {
      "epoch": 0.11344,
      "grad_norm": 1.1232646849413133,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 11344
    },
    {
      "epoch": 0.11345,
      "grad_norm": 1.1124167099644198,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 11345
    },
    {
      "epoch": 0.11346,
      "grad_norm": 0.9798550547485815,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 11346
    },
    {
      "epoch": 0.11347,
      "grad_norm": 0.9799428380823298,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 11347
    },
    {
      "epoch": 0.11348,
      "grad_norm": 1.0124447590781391,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 11348
    },
    {
      "epoch": 0.11349,
      "grad_norm": 1.318597220274999,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 11349
    },
    {
      "epoch": 0.1135,
      "grad_norm": 1.0490907851050608,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 11350
    },
    {
      "epoch": 0.11351,
      "grad_norm": 1.1275921563276423,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 11351
    },
    {
      "epoch": 0.11352,
      "grad_norm": 1.2906503939683447,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 11352
    },
    {
      "epoch": 0.11353,
      "grad_norm": 1.1088910910099248,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 11353
    },
    {
      "epoch": 0.11354,
      "grad_norm": 1.1908101825995385,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 11354
    },
    {
      "epoch": 0.11355,
      "grad_norm": 0.9800870108001858,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 11355
    },
    {
      "epoch": 0.11356,
      "grad_norm": 1.2127513148741016,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 11356
    },
    {
      "epoch": 0.11357,
      "grad_norm": 1.0616934197142416,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 11357
    },
    {
      "epoch": 0.11358,
      "grad_norm": 1.1793154754632509,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 11358
    },
    {
      "epoch": 0.11359,
      "grad_norm": 1.0791143615314955,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 11359
    },
    {
      "epoch": 0.1136,
      "grad_norm": 1.0992938215616361,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 11360
    },
    {
      "epoch": 0.11361,
      "grad_norm": 0.9884345665722764,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 11361
    },
    {
      "epoch": 0.11362,
      "grad_norm": 1.1988548038821205,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 11362
    },
    {
      "epoch": 0.11363,
      "grad_norm": 0.9759992044198456,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 11363
    },
    {
      "epoch": 0.11364,
      "grad_norm": 1.2599692044663442,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 11364
    },
    {
      "epoch": 0.11365,
      "grad_norm": 0.8745803209168422,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 11365
    },
    {
      "epoch": 0.11366,
      "grad_norm": 0.8593192580434853,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 11366
    },
    {
      "epoch": 0.11367,
      "grad_norm": 1.030383110518904,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 11367
    },
    {
      "epoch": 0.11368,
      "grad_norm": 1.2702615917194533,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 11368
    },
    {
      "epoch": 0.11369,
      "grad_norm": 1.3157323355023252,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 11369
    },
    {
      "epoch": 0.1137,
      "grad_norm": 0.9479350053364931,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 11370
    },
    {
      "epoch": 0.11371,
      "grad_norm": 1.0001937118939637,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 11371
    },
    {
      "epoch": 0.11372,
      "grad_norm": 1.2520454896918638,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 11372
    },
    {
      "epoch": 0.11373,
      "grad_norm": 1.0850205495773182,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 11373
    },
    {
      "epoch": 0.11374,
      "grad_norm": 1.1670106510434204,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 11374
    },
    {
      "epoch": 0.11375,
      "grad_norm": 0.8561088091995893,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 11375
    },
    {
      "epoch": 0.11376,
      "grad_norm": 0.7637915719569975,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 11376
    },
    {
      "epoch": 0.11377,
      "grad_norm": 0.9313796841061992,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 11377
    },
    {
      "epoch": 0.11378,
      "grad_norm": 1.1937782116002877,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 11378
    },
    {
      "epoch": 0.11379,
      "grad_norm": 1.01019759802451,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 11379
    },
    {
      "epoch": 0.1138,
      "grad_norm": 1.2969923102509442,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 11380
    },
    {
      "epoch": 0.11381,
      "grad_norm": 1.1287939651559524,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 11381
    },
    {
      "epoch": 0.11382,
      "grad_norm": 1.2076230352917354,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 11382
    },
    {
      "epoch": 0.11383,
      "grad_norm": 1.0185750251810939,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 11383
    },
    {
      "epoch": 0.11384,
      "grad_norm": 1.0371687278136845,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 11384
    },
    {
      "epoch": 0.11385,
      "grad_norm": 1.2192193474269375,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 11385
    },
    {
      "epoch": 0.11386,
      "grad_norm": 1.1899210254042143,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 11386
    },
    {
      "epoch": 0.11387,
      "grad_norm": 1.1919981601440022,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 11387
    },
    {
      "epoch": 0.11388,
      "grad_norm": 1.0815046104847168,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 11388
    },
    {
      "epoch": 0.11389,
      "grad_norm": 1.1584442503144354,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 11389
    },
    {
      "epoch": 0.1139,
      "grad_norm": 1.1663439723281048,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11390
    },
    {
      "epoch": 0.11391,
      "grad_norm": 1.0722133583119948,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 11391
    },
    {
      "epoch": 0.11392,
      "grad_norm": 1.2653674305987967,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 11392
    },
    {
      "epoch": 0.11393,
      "grad_norm": 0.9632936012758943,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 11393
    },
    {
      "epoch": 0.11394,
      "grad_norm": 1.2086534538177434,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 11394
    },
    {
      "epoch": 0.11395,
      "grad_norm": 1.2404795628469127,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 11395
    },
    {
      "epoch": 0.11396,
      "grad_norm": 1.1143481503393529,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 11396
    },
    {
      "epoch": 0.11397,
      "grad_norm": 1.2209668324766623,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 11397
    },
    {
      "epoch": 0.11398,
      "grad_norm": 0.9180476890565811,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 11398
    },
    {
      "epoch": 0.11399,
      "grad_norm": 1.0341543566407236,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 11399
    },
    {
      "epoch": 0.114,
      "grad_norm": 1.2550533601100675,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 11400
    },
    {
      "epoch": 0.11401,
      "grad_norm": 0.9408742974294715,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 11401
    },
    {
      "epoch": 0.11402,
      "grad_norm": 1.089989319062244,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 11402
    },
    {
      "epoch": 0.11403,
      "grad_norm": 0.9617519579250814,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 11403
    },
    {
      "epoch": 0.11404,
      "grad_norm": 1.094611527464326,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 11404
    },
    {
      "epoch": 0.11405,
      "grad_norm": 1.2924390248074644,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 11405
    },
    {
      "epoch": 0.11406,
      "grad_norm": 0.8613651877594999,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 11406
    },
    {
      "epoch": 0.11407,
      "grad_norm": 0.9020507368404247,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 11407
    },
    {
      "epoch": 0.11408,
      "grad_norm": 1.0194826586040544,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 11408
    },
    {
      "epoch": 0.11409,
      "grad_norm": 1.2170302680890877,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 11409
    },
    {
      "epoch": 0.1141,
      "grad_norm": 0.9748439760302462,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 11410
    },
    {
      "epoch": 0.11411,
      "grad_norm": 1.1438379739008626,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 11411
    },
    {
      "epoch": 0.11412,
      "grad_norm": 1.1914607143578906,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 11412
    },
    {
      "epoch": 0.11413,
      "grad_norm": 0.9667367633697801,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 11413
    },
    {
      "epoch": 0.11414,
      "grad_norm": 1.0806930724974129,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 11414
    },
    {
      "epoch": 0.11415,
      "grad_norm": 1.0476544469337348,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 11415
    },
    {
      "epoch": 0.11416,
      "grad_norm": 1.375644241403238,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 11416
    },
    {
      "epoch": 0.11417,
      "grad_norm": 0.9556383642152451,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 11417
    },
    {
      "epoch": 0.11418,
      "grad_norm": 1.3551574310415802,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 11418
    },
    {
      "epoch": 0.11419,
      "grad_norm": 1.1236593786751847,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 11419
    },
    {
      "epoch": 0.1142,
      "grad_norm": 1.0487403125417163,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 11420
    },
    {
      "epoch": 0.11421,
      "grad_norm": 1.1149362265682679,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 11421
    },
    {
      "epoch": 0.11422,
      "grad_norm": 0.9316950413130666,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11422
    },
    {
      "epoch": 0.11423,
      "grad_norm": 1.112559363387894,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 11423
    },
    {
      "epoch": 0.11424,
      "grad_norm": 1.046947547930098,
      "learning_rate": 0.003,
      "loss": 4.1052,
      "step": 11424
    },
    {
      "epoch": 0.11425,
      "grad_norm": 1.025164897373827,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 11425
    },
    {
      "epoch": 0.11426,
      "grad_norm": 1.1000273644854461,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 11426
    },
    {
      "epoch": 0.11427,
      "grad_norm": 0.967175793628095,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 11427
    },
    {
      "epoch": 0.11428,
      "grad_norm": 1.1723637978459762,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 11428
    },
    {
      "epoch": 0.11429,
      "grad_norm": 1.0287812781648369,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 11429
    },
    {
      "epoch": 0.1143,
      "grad_norm": 1.3493177052407335,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 11430
    },
    {
      "epoch": 0.11431,
      "grad_norm": 1.0285149025991758,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 11431
    },
    {
      "epoch": 0.11432,
      "grad_norm": 1.3023078221513005,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 11432
    },
    {
      "epoch": 0.11433,
      "grad_norm": 1.2940193777377493,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 11433
    },
    {
      "epoch": 0.11434,
      "grad_norm": 1.1521329915424683,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 11434
    },
    {
      "epoch": 0.11435,
      "grad_norm": 1.09046750871652,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 11435
    },
    {
      "epoch": 0.11436,
      "grad_norm": 1.239387425708267,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 11436
    },
    {
      "epoch": 0.11437,
      "grad_norm": 0.9839702313725753,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 11437
    },
    {
      "epoch": 0.11438,
      "grad_norm": 1.0105437037979108,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 11438
    },
    {
      "epoch": 0.11439,
      "grad_norm": 1.2805117104272348,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 11439
    },
    {
      "epoch": 0.1144,
      "grad_norm": 1.3017141223339386,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 11440
    },
    {
      "epoch": 0.11441,
      "grad_norm": 0.8997940091230753,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 11441
    },
    {
      "epoch": 0.11442,
      "grad_norm": 0.9081136329240335,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 11442
    },
    {
      "epoch": 0.11443,
      "grad_norm": 0.9592890689351932,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 11443
    },
    {
      "epoch": 0.11444,
      "grad_norm": 1.0751507227866715,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 11444
    },
    {
      "epoch": 0.11445,
      "grad_norm": 1.0817511739453243,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 11445
    },
    {
      "epoch": 0.11446,
      "grad_norm": 0.985678283326643,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 11446
    },
    {
      "epoch": 0.11447,
      "grad_norm": 1.1095480987200779,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 11447
    },
    {
      "epoch": 0.11448,
      "grad_norm": 1.1999458551915834,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 11448
    },
    {
      "epoch": 0.11449,
      "grad_norm": 1.2755194245240447,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 11449
    },
    {
      "epoch": 0.1145,
      "grad_norm": 0.9554237876405631,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 11450
    },
    {
      "epoch": 0.11451,
      "grad_norm": 0.9921100619442162,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 11451
    },
    {
      "epoch": 0.11452,
      "grad_norm": 1.0983081877471728,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 11452
    },
    {
      "epoch": 0.11453,
      "grad_norm": 1.1593883506577896,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 11453
    },
    {
      "epoch": 0.11454,
      "grad_norm": 0.9269990739942203,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 11454
    },
    {
      "epoch": 0.11455,
      "grad_norm": 1.1980303015914466,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 11455
    },
    {
      "epoch": 0.11456,
      "grad_norm": 1.1480252892729894,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 11456
    },
    {
      "epoch": 0.11457,
      "grad_norm": 1.1836346713452197,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11457
    },
    {
      "epoch": 0.11458,
      "grad_norm": 0.9351218802077039,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 11458
    },
    {
      "epoch": 0.11459,
      "grad_norm": 1.1033123819477257,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 11459
    },
    {
      "epoch": 0.1146,
      "grad_norm": 1.012445753388458,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 11460
    },
    {
      "epoch": 0.11461,
      "grad_norm": 1.3576147209988854,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 11461
    },
    {
      "epoch": 0.11462,
      "grad_norm": 1.0883600335321508,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 11462
    },
    {
      "epoch": 0.11463,
      "grad_norm": 1.2657634520019636,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 11463
    },
    {
      "epoch": 0.11464,
      "grad_norm": 1.2311124665624869,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 11464
    },
    {
      "epoch": 0.11465,
      "grad_norm": 1.1257407859233135,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 11465
    },
    {
      "epoch": 0.11466,
      "grad_norm": 0.944138764965183,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 11466
    },
    {
      "epoch": 0.11467,
      "grad_norm": 1.1813424893876345,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 11467
    },
    {
      "epoch": 0.11468,
      "grad_norm": 1.0271819896606211,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 11468
    },
    {
      "epoch": 0.11469,
      "grad_norm": 1.0710135800874787,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 11469
    },
    {
      "epoch": 0.1147,
      "grad_norm": 1.3291275440585504,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 11470
    },
    {
      "epoch": 0.11471,
      "grad_norm": 0.8813798372990844,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 11471
    },
    {
      "epoch": 0.11472,
      "grad_norm": 0.9608440586409432,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 11472
    },
    {
      "epoch": 0.11473,
      "grad_norm": 1.159326505594525,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 11473
    },
    {
      "epoch": 0.11474,
      "grad_norm": 0.982827792187837,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 11474
    },
    {
      "epoch": 0.11475,
      "grad_norm": 1.2625211478329064,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 11475
    },
    {
      "epoch": 0.11476,
      "grad_norm": 0.9599156828526338,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 11476
    },
    {
      "epoch": 0.11477,
      "grad_norm": 1.15290384723048,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 11477
    },
    {
      "epoch": 0.11478,
      "grad_norm": 1.1254972374254937,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 11478
    },
    {
      "epoch": 0.11479,
      "grad_norm": 1.039084825812637,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 11479
    },
    {
      "epoch": 0.1148,
      "grad_norm": 1.010880890786295,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 11480
    },
    {
      "epoch": 0.11481,
      "grad_norm": 1.125666168910612,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 11481
    },
    {
      "epoch": 0.11482,
      "grad_norm": 1.069273471709598,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 11482
    },
    {
      "epoch": 0.11483,
      "grad_norm": 1.0996797520912263,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 11483
    },
    {
      "epoch": 0.11484,
      "grad_norm": 1.053058869147153,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 11484
    },
    {
      "epoch": 0.11485,
      "grad_norm": 0.95953616963765,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 11485
    },
    {
      "epoch": 0.11486,
      "grad_norm": 1.0673135991705158,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 11486
    },
    {
      "epoch": 0.11487,
      "grad_norm": 1.2347727658118037,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 11487
    },
    {
      "epoch": 0.11488,
      "grad_norm": 1.1323197031207368,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 11488
    },
    {
      "epoch": 0.11489,
      "grad_norm": 1.233987085538438,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 11489
    },
    {
      "epoch": 0.1149,
      "grad_norm": 1.437730399768374,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 11490
    },
    {
      "epoch": 0.11491,
      "grad_norm": 1.0812904240647934,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 11491
    },
    {
      "epoch": 0.11492,
      "grad_norm": 1.2382203916879087,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 11492
    },
    {
      "epoch": 0.11493,
      "grad_norm": 1.1868789758176679,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 11493
    },
    {
      "epoch": 0.11494,
      "grad_norm": 1.0905374174299731,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 11494
    },
    {
      "epoch": 0.11495,
      "grad_norm": 1.0025408948887682,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 11495
    },
    {
      "epoch": 0.11496,
      "grad_norm": 1.059501013039697,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 11496
    },
    {
      "epoch": 0.11497,
      "grad_norm": 0.97928784748038,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 11497
    },
    {
      "epoch": 0.11498,
      "grad_norm": 0.9912519652552354,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 11498
    },
    {
      "epoch": 0.11499,
      "grad_norm": 1.2182826126289492,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 11499
    },
    {
      "epoch": 0.115,
      "grad_norm": 1.0575900148374628,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 11500
    },
    {
      "epoch": 0.11501,
      "grad_norm": 1.042259946596932,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 11501
    },
    {
      "epoch": 0.11502,
      "grad_norm": 1.1491360742259606,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 11502
    },
    {
      "epoch": 0.11503,
      "grad_norm": 1.1515986591053582,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 11503
    },
    {
      "epoch": 0.11504,
      "grad_norm": 1.0082890004795189,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 11504
    },
    {
      "epoch": 0.11505,
      "grad_norm": 1.2398006653371065,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 11505
    },
    {
      "epoch": 0.11506,
      "grad_norm": 0.9506533343856579,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 11506
    },
    {
      "epoch": 0.11507,
      "grad_norm": 1.2049014161424605,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 11507
    },
    {
      "epoch": 0.11508,
      "grad_norm": 1.016229736336884,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 11508
    },
    {
      "epoch": 0.11509,
      "grad_norm": 1.059800157746038,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 11509
    },
    {
      "epoch": 0.1151,
      "grad_norm": 1.2187976213839506,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 11510
    },
    {
      "epoch": 0.11511,
      "grad_norm": 1.2526644591744729,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 11511
    },
    {
      "epoch": 0.11512,
      "grad_norm": 1.346175519230743,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 11512
    },
    {
      "epoch": 0.11513,
      "grad_norm": 1.0251209000081711,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 11513
    },
    {
      "epoch": 0.11514,
      "grad_norm": 1.1098866603814985,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 11514
    },
    {
      "epoch": 0.11515,
      "grad_norm": 1.1205216374867848,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 11515
    },
    {
      "epoch": 0.11516,
      "grad_norm": 1.133872234739474,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 11516
    },
    {
      "epoch": 0.11517,
      "grad_norm": 0.9546557729774939,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 11517
    },
    {
      "epoch": 0.11518,
      "grad_norm": 1.0000084540499257,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 11518
    },
    {
      "epoch": 0.11519,
      "grad_norm": 1.1378977666905323,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 11519
    },
    {
      "epoch": 0.1152,
      "grad_norm": 1.066034099269205,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 11520
    },
    {
      "epoch": 0.11521,
      "grad_norm": 1.4441324390232206,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 11521
    },
    {
      "epoch": 0.11522,
      "grad_norm": 0.9289400412549517,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 11522
    },
    {
      "epoch": 0.11523,
      "grad_norm": 1.0067482412606141,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 11523
    },
    {
      "epoch": 0.11524,
      "grad_norm": 1.17887672911843,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 11524
    },
    {
      "epoch": 0.11525,
      "grad_norm": 1.0209397605299606,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 11525
    },
    {
      "epoch": 0.11526,
      "grad_norm": 1.1822412291720787,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 11526
    },
    {
      "epoch": 0.11527,
      "grad_norm": 0.9907113033206698,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 11527
    },
    {
      "epoch": 0.11528,
      "grad_norm": 1.004037870295607,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 11528
    },
    {
      "epoch": 0.11529,
      "grad_norm": 1.27345356444765,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 11529
    },
    {
      "epoch": 0.1153,
      "grad_norm": 0.9091681488318872,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 11530
    },
    {
      "epoch": 0.11531,
      "grad_norm": 1.1485271799948789,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 11531
    },
    {
      "epoch": 0.11532,
      "grad_norm": 1.2286560991322542,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 11532
    },
    {
      "epoch": 0.11533,
      "grad_norm": 0.957779523859025,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 11533
    },
    {
      "epoch": 0.11534,
      "grad_norm": 1.0703209588477642,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 11534
    },
    {
      "epoch": 0.11535,
      "grad_norm": 1.4076865000833358,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 11535
    },
    {
      "epoch": 0.11536,
      "grad_norm": 0.8453178792943659,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11536
    },
    {
      "epoch": 0.11537,
      "grad_norm": 0.8850418916267287,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 11537
    },
    {
      "epoch": 0.11538,
      "grad_norm": 1.0115039027362196,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 11538
    },
    {
      "epoch": 0.11539,
      "grad_norm": 1.296796036565217,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 11539
    },
    {
      "epoch": 0.1154,
      "grad_norm": 0.9679605146867507,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 11540
    },
    {
      "epoch": 0.11541,
      "grad_norm": 1.1842908793146563,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 11541
    },
    {
      "epoch": 0.11542,
      "grad_norm": 0.9950285761038729,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 11542
    },
    {
      "epoch": 0.11543,
      "grad_norm": 1.1300853554103478,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 11543
    },
    {
      "epoch": 0.11544,
      "grad_norm": 1.1354787892059146,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 11544
    },
    {
      "epoch": 0.11545,
      "grad_norm": 1.0187525640548183,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 11545
    },
    {
      "epoch": 0.11546,
      "grad_norm": 1.3098113199196204,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 11546
    },
    {
      "epoch": 0.11547,
      "grad_norm": 1.0110391512040715,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 11547
    },
    {
      "epoch": 0.11548,
      "grad_norm": 1.3003353713494008,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 11548
    },
    {
      "epoch": 0.11549,
      "grad_norm": 0.9518470943238035,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 11549
    },
    {
      "epoch": 0.1155,
      "grad_norm": 1.1331531114660882,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 11550
    },
    {
      "epoch": 0.11551,
      "grad_norm": 1.1341814792195717,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 11551
    },
    {
      "epoch": 0.11552,
      "grad_norm": 1.1642435526875725,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 11552
    },
    {
      "epoch": 0.11553,
      "grad_norm": 1.1915747696382795,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 11553
    },
    {
      "epoch": 0.11554,
      "grad_norm": 0.9805179418386152,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 11554
    },
    {
      "epoch": 0.11555,
      "grad_norm": 1.1284658454331822,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 11555
    },
    {
      "epoch": 0.11556,
      "grad_norm": 1.1631076906779456,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 11556
    },
    {
      "epoch": 0.11557,
      "grad_norm": 1.0466827929340412,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 11557
    },
    {
      "epoch": 0.11558,
      "grad_norm": 1.1745724748654662,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 11558
    },
    {
      "epoch": 0.11559,
      "grad_norm": 0.9459708236772346,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 11559
    },
    {
      "epoch": 0.1156,
      "grad_norm": 1.0021916398425177,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 11560
    },
    {
      "epoch": 0.11561,
      "grad_norm": 1.4207016748395007,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 11561
    },
    {
      "epoch": 0.11562,
      "grad_norm": 0.9699229174078287,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 11562
    },
    {
      "epoch": 0.11563,
      "grad_norm": 1.1311822024867861,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 11563
    },
    {
      "epoch": 0.11564,
      "grad_norm": 0.9679692818516854,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 11564
    },
    {
      "epoch": 0.11565,
      "grad_norm": 1.1922138762146304,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 11565
    },
    {
      "epoch": 0.11566,
      "grad_norm": 1.237058073396377,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 11566
    },
    {
      "epoch": 0.11567,
      "grad_norm": 1.3887929548971565,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 11567
    },
    {
      "epoch": 0.11568,
      "grad_norm": 0.958572461549701,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 11568
    },
    {
      "epoch": 0.11569,
      "grad_norm": 1.0671355046838664,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 11569
    },
    {
      "epoch": 0.1157,
      "grad_norm": 1.2790464207002419,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 11570
    },
    {
      "epoch": 0.11571,
      "grad_norm": 0.9377850639923463,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 11571
    },
    {
      "epoch": 0.11572,
      "grad_norm": 0.8880366639899385,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 11572
    },
    {
      "epoch": 0.11573,
      "grad_norm": 1.0787963181925324,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 11573
    },
    {
      "epoch": 0.11574,
      "grad_norm": 1.123504886496314,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 11574
    },
    {
      "epoch": 0.11575,
      "grad_norm": 0.9183337157079674,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 11575
    },
    {
      "epoch": 0.11576,
      "grad_norm": 1.2491884156219748,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 11576
    },
    {
      "epoch": 0.11577,
      "grad_norm": 1.1547263596011585,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 11577
    },
    {
      "epoch": 0.11578,
      "grad_norm": 1.1201197406092993,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 11578
    },
    {
      "epoch": 0.11579,
      "grad_norm": 1.2033338134892475,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 11579
    },
    {
      "epoch": 0.1158,
      "grad_norm": 1.0470671889660987,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 11580
    },
    {
      "epoch": 0.11581,
      "grad_norm": 1.2185434900378873,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 11581
    },
    {
      "epoch": 0.11582,
      "grad_norm": 1.041824094279636,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 11582
    },
    {
      "epoch": 0.11583,
      "grad_norm": 1.3025002104024346,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 11583
    },
    {
      "epoch": 0.11584,
      "grad_norm": 0.8944428556353944,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 11584
    },
    {
      "epoch": 0.11585,
      "grad_norm": 0.9221289377596882,
      "learning_rate": 0.003,
      "loss": 4.1051,
      "step": 11585
    },
    {
      "epoch": 0.11586,
      "grad_norm": 1.1158151867038537,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 11586
    },
    {
      "epoch": 0.11587,
      "grad_norm": 1.0633801503841447,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 11587
    },
    {
      "epoch": 0.11588,
      "grad_norm": 1.2175728246475466,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 11588
    },
    {
      "epoch": 0.11589,
      "grad_norm": 0.9755229650012226,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 11589
    },
    {
      "epoch": 0.1159,
      "grad_norm": 1.2474696834282595,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 11590
    },
    {
      "epoch": 0.11591,
      "grad_norm": 0.9857271628762859,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 11591
    },
    {
      "epoch": 0.11592,
      "grad_norm": 1.1544073912447943,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 11592
    },
    {
      "epoch": 0.11593,
      "grad_norm": 1.1913649125562014,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 11593
    },
    {
      "epoch": 0.11594,
      "grad_norm": 1.2360990835362289,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 11594
    },
    {
      "epoch": 0.11595,
      "grad_norm": 1.1668662086330877,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 11595
    },
    {
      "epoch": 0.11596,
      "grad_norm": 1.2566056003469033,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 11596
    },
    {
      "epoch": 0.11597,
      "grad_norm": 1.0629373859203337,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 11597
    },
    {
      "epoch": 0.11598,
      "grad_norm": 1.1663361648055266,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 11598
    },
    {
      "epoch": 0.11599,
      "grad_norm": 1.2961480043008549,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 11599
    },
    {
      "epoch": 0.116,
      "grad_norm": 1.014250628737564,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 11600
    },
    {
      "epoch": 0.11601,
      "grad_norm": 1.0304229179205031,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 11601
    },
    {
      "epoch": 0.11602,
      "grad_norm": 1.2069097235923103,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 11602
    },
    {
      "epoch": 0.11603,
      "grad_norm": 1.0367709582145608,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 11603
    },
    {
      "epoch": 0.11604,
      "grad_norm": 1.1276340617378737,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 11604
    },
    {
      "epoch": 0.11605,
      "grad_norm": 0.9578098940648896,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 11605
    },
    {
      "epoch": 0.11606,
      "grad_norm": 1.0784692357920973,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 11606
    },
    {
      "epoch": 0.11607,
      "grad_norm": 1.0911690675049772,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 11607
    },
    {
      "epoch": 0.11608,
      "grad_norm": 1.0784819343728043,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 11608
    },
    {
      "epoch": 0.11609,
      "grad_norm": 1.1460958124906055,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 11609
    },
    {
      "epoch": 0.1161,
      "grad_norm": 0.9943464723111869,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 11610
    },
    {
      "epoch": 0.11611,
      "grad_norm": 1.098850299349954,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 11611
    },
    {
      "epoch": 0.11612,
      "grad_norm": 1.1362782642396552,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 11612
    },
    {
      "epoch": 0.11613,
      "grad_norm": 1.1387512980756966,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 11613
    },
    {
      "epoch": 0.11614,
      "grad_norm": 1.345923900697499,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 11614
    },
    {
      "epoch": 0.11615,
      "grad_norm": 1.130939584436006,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 11615
    },
    {
      "epoch": 0.11616,
      "grad_norm": 0.886731255863067,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 11616
    },
    {
      "epoch": 0.11617,
      "grad_norm": 1.1544801702721095,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 11617
    },
    {
      "epoch": 0.11618,
      "grad_norm": 1.1811782827774127,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 11618
    },
    {
      "epoch": 0.11619,
      "grad_norm": 1.0060331338988424,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 11619
    },
    {
      "epoch": 0.1162,
      "grad_norm": 1.1521508346964646,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 11620
    },
    {
      "epoch": 0.11621,
      "grad_norm": 1.0257932213801997,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 11621
    },
    {
      "epoch": 0.11622,
      "grad_norm": 1.0642025893798226,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 11622
    },
    {
      "epoch": 0.11623,
      "grad_norm": 1.0541204723690525,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 11623
    },
    {
      "epoch": 0.11624,
      "grad_norm": 0.963137136298594,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 11624
    },
    {
      "epoch": 0.11625,
      "grad_norm": 1.2253091846333735,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 11625
    },
    {
      "epoch": 0.11626,
      "grad_norm": 1.0179136549959853,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 11626
    },
    {
      "epoch": 0.11627,
      "grad_norm": 1.284748849879381,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 11627
    },
    {
      "epoch": 0.11628,
      "grad_norm": 1.207376669095626,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 11628
    },
    {
      "epoch": 0.11629,
      "grad_norm": 1.201903728509045,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 11629
    },
    {
      "epoch": 0.1163,
      "grad_norm": 1.1275076903728707,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 11630
    },
    {
      "epoch": 0.11631,
      "grad_norm": 0.8270761377773259,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 11631
    },
    {
      "epoch": 0.11632,
      "grad_norm": 1.0228543931299996,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 11632
    },
    {
      "epoch": 0.11633,
      "grad_norm": 1.1302523592518674,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 11633
    },
    {
      "epoch": 0.11634,
      "grad_norm": 0.8638865702028005,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 11634
    },
    {
      "epoch": 0.11635,
      "grad_norm": 1.0027464596812203,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 11635
    },
    {
      "epoch": 0.11636,
      "grad_norm": 1.3485058979634321,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 11636
    },
    {
      "epoch": 0.11637,
      "grad_norm": 1.154943472996014,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 11637
    },
    {
      "epoch": 0.11638,
      "grad_norm": 1.0927699715554624,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 11638
    },
    {
      "epoch": 0.11639,
      "grad_norm": 0.9883426317196221,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 11639
    },
    {
      "epoch": 0.1164,
      "grad_norm": 1.1813503322922911,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 11640
    },
    {
      "epoch": 0.11641,
      "grad_norm": 0.9833000059546451,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 11641
    },
    {
      "epoch": 0.11642,
      "grad_norm": 1.017905346925556,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 11642
    },
    {
      "epoch": 0.11643,
      "grad_norm": 1.274217063915014,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 11643
    },
    {
      "epoch": 0.11644,
      "grad_norm": 1.0301408484730104,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 11644
    },
    {
      "epoch": 0.11645,
      "grad_norm": 1.009275001586971,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 11645
    },
    {
      "epoch": 0.11646,
      "grad_norm": 1.2709057781981588,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 11646
    },
    {
      "epoch": 0.11647,
      "grad_norm": 1.1255501598073867,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 11647
    },
    {
      "epoch": 0.11648,
      "grad_norm": 0.984766142304668,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 11648
    },
    {
      "epoch": 0.11649,
      "grad_norm": 1.0563767234375925,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 11649
    },
    {
      "epoch": 0.1165,
      "grad_norm": 1.1606860639127619,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 11650
    },
    {
      "epoch": 0.11651,
      "grad_norm": 1.1015639702221316,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 11651
    },
    {
      "epoch": 0.11652,
      "grad_norm": 1.143719872576751,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 11652
    },
    {
      "epoch": 0.11653,
      "grad_norm": 1.2263249319837026,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 11653
    },
    {
      "epoch": 0.11654,
      "grad_norm": 0.8919799265058325,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 11654
    },
    {
      "epoch": 0.11655,
      "grad_norm": 1.1735764182230106,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 11655
    },
    {
      "epoch": 0.11656,
      "grad_norm": 1.033198482124397,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 11656
    },
    {
      "epoch": 0.11657,
      "grad_norm": 1.3681255518075837,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 11657
    },
    {
      "epoch": 0.11658,
      "grad_norm": 1.2738790837565772,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 11658
    },
    {
      "epoch": 0.11659,
      "grad_norm": 1.3607779431837075,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 11659
    },
    {
      "epoch": 0.1166,
      "grad_norm": 0.986669792330658,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 11660
    },
    {
      "epoch": 0.11661,
      "grad_norm": 1.152518378500052,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 11661
    },
    {
      "epoch": 0.11662,
      "grad_norm": 1.1196506107063688,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 11662
    },
    {
      "epoch": 0.11663,
      "grad_norm": 0.995810438761169,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 11663
    },
    {
      "epoch": 0.11664,
      "grad_norm": 1.1216949623356607,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 11664
    },
    {
      "epoch": 0.11665,
      "grad_norm": 1.0930625371595246,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 11665
    },
    {
      "epoch": 0.11666,
      "grad_norm": 1.1430523888668103,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 11666
    },
    {
      "epoch": 0.11667,
      "grad_norm": 1.4096528741555352,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 11667
    },
    {
      "epoch": 0.11668,
      "grad_norm": 0.8572070174035606,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 11668
    },
    {
      "epoch": 0.11669,
      "grad_norm": 0.9520455561841698,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 11669
    },
    {
      "epoch": 0.1167,
      "grad_norm": 1.1344283234440606,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 11670
    },
    {
      "epoch": 0.11671,
      "grad_norm": 1.1367286001531283,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 11671
    },
    {
      "epoch": 0.11672,
      "grad_norm": 1.1019504330512884,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 11672
    },
    {
      "epoch": 0.11673,
      "grad_norm": 1.2257422864371508,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 11673
    },
    {
      "epoch": 0.11674,
      "grad_norm": 1.0657527751866165,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 11674
    },
    {
      "epoch": 0.11675,
      "grad_norm": 1.1985652793128885,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 11675
    },
    {
      "epoch": 0.11676,
      "grad_norm": 0.9332198870740341,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 11676
    },
    {
      "epoch": 0.11677,
      "grad_norm": 1.0545279590864574,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 11677
    },
    {
      "epoch": 0.11678,
      "grad_norm": 1.3340048562889086,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 11678
    },
    {
      "epoch": 0.11679,
      "grad_norm": 0.9379675744284625,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 11679
    },
    {
      "epoch": 0.1168,
      "grad_norm": 1.1141130817934959,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 11680
    },
    {
      "epoch": 0.11681,
      "grad_norm": 1.177277877794882,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 11681
    },
    {
      "epoch": 0.11682,
      "grad_norm": 1.1277834165780665,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 11682
    },
    {
      "epoch": 0.11683,
      "grad_norm": 1.281372903846131,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 11683
    },
    {
      "epoch": 0.11684,
      "grad_norm": 0.9181486510427459,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 11684
    },
    {
      "epoch": 0.11685,
      "grad_norm": 0.9626362886189451,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 11685
    },
    {
      "epoch": 0.11686,
      "grad_norm": 1.0406179327979335,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 11686
    },
    {
      "epoch": 0.11687,
      "grad_norm": 1.1605950057136238,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 11687
    },
    {
      "epoch": 0.11688,
      "grad_norm": 1.1183597832884749,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 11688
    },
    {
      "epoch": 0.11689,
      "grad_norm": 1.3177810263811713,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 11689
    },
    {
      "epoch": 0.1169,
      "grad_norm": 1.0339702155942279,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11690
    },
    {
      "epoch": 0.11691,
      "grad_norm": 1.2249081132818636,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 11691
    },
    {
      "epoch": 0.11692,
      "grad_norm": 1.266927910285917,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 11692
    },
    {
      "epoch": 0.11693,
      "grad_norm": 1.0213845058392041,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 11693
    },
    {
      "epoch": 0.11694,
      "grad_norm": 1.0362441103808235,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 11694
    },
    {
      "epoch": 0.11695,
      "grad_norm": 1.100736599721524,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 11695
    },
    {
      "epoch": 0.11696,
      "grad_norm": 1.0627450412294686,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 11696
    },
    {
      "epoch": 0.11697,
      "grad_norm": 1.2418770546939046,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 11697
    },
    {
      "epoch": 0.11698,
      "grad_norm": 0.8793593011265435,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 11698
    },
    {
      "epoch": 0.11699,
      "grad_norm": 0.9356580319854434,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 11699
    },
    {
      "epoch": 0.117,
      "grad_norm": 1.0565849927073654,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 11700
    },
    {
      "epoch": 0.11701,
      "grad_norm": 1.0321570661745199,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 11701
    },
    {
      "epoch": 0.11702,
      "grad_norm": 1.1302279305254166,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 11702
    },
    {
      "epoch": 0.11703,
      "grad_norm": 1.1276832996289374,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 11703
    },
    {
      "epoch": 0.11704,
      "grad_norm": 1.643091586238816,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 11704
    },
    {
      "epoch": 0.11705,
      "grad_norm": 0.8483221837791483,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 11705
    },
    {
      "epoch": 0.11706,
      "grad_norm": 1.118465992742277,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 11706
    },
    {
      "epoch": 0.11707,
      "grad_norm": 1.0063144729463553,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 11707
    },
    {
      "epoch": 0.11708,
      "grad_norm": 1.0977058727516777,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 11708
    },
    {
      "epoch": 0.11709,
      "grad_norm": 1.3282663240246637,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 11709
    },
    {
      "epoch": 0.1171,
      "grad_norm": 0.948967010561446,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 11710
    },
    {
      "epoch": 0.11711,
      "grad_norm": 1.1042229472219267,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 11711
    },
    {
      "epoch": 0.11712,
      "grad_norm": 1.1451972308255882,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 11712
    },
    {
      "epoch": 0.11713,
      "grad_norm": 1.1229877422007808,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 11713
    },
    {
      "epoch": 0.11714,
      "grad_norm": 1.3025407860311622,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 11714
    },
    {
      "epoch": 0.11715,
      "grad_norm": 0.9538833016975586,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 11715
    },
    {
      "epoch": 0.11716,
      "grad_norm": 1.1855847461520062,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 11716
    },
    {
      "epoch": 0.11717,
      "grad_norm": 1.1245699404440197,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 11717
    },
    {
      "epoch": 0.11718,
      "grad_norm": 1.1777591935526606,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 11718
    },
    {
      "epoch": 0.11719,
      "grad_norm": 1.1755063251134579,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 11719
    },
    {
      "epoch": 0.1172,
      "grad_norm": 1.1096620840332625,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 11720
    },
    {
      "epoch": 0.11721,
      "grad_norm": 1.0515005669697468,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 11721
    },
    {
      "epoch": 0.11722,
      "grad_norm": 1.1629975308450062,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 11722
    },
    {
      "epoch": 0.11723,
      "grad_norm": 0.975517625425596,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 11723
    },
    {
      "epoch": 0.11724,
      "grad_norm": 1.1223667286703718,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 11724
    },
    {
      "epoch": 0.11725,
      "grad_norm": 0.9892581009311286,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 11725
    },
    {
      "epoch": 0.11726,
      "grad_norm": 1.0867359120534805,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 11726
    },
    {
      "epoch": 0.11727,
      "grad_norm": 1.0824253250195415,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 11727
    },
    {
      "epoch": 0.11728,
      "grad_norm": 1.1775735354805041,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 11728
    },
    {
      "epoch": 0.11729,
      "grad_norm": 1.218274123653387,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 11729
    },
    {
      "epoch": 0.1173,
      "grad_norm": 1.1566066956016399,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 11730
    },
    {
      "epoch": 0.11731,
      "grad_norm": 1.088035629745312,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 11731
    },
    {
      "epoch": 0.11732,
      "grad_norm": 1.165765137910104,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 11732
    },
    {
      "epoch": 0.11733,
      "grad_norm": 1.223175992941106,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 11733
    },
    {
      "epoch": 0.11734,
      "grad_norm": 0.9196427956388414,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 11734
    },
    {
      "epoch": 0.11735,
      "grad_norm": 0.8899377969449798,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 11735
    },
    {
      "epoch": 0.11736,
      "grad_norm": 0.8856112456450769,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 11736
    },
    {
      "epoch": 0.11737,
      "grad_norm": 0.8304790484767605,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 11737
    },
    {
      "epoch": 0.11738,
      "grad_norm": 0.8856599944783037,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 11738
    },
    {
      "epoch": 0.11739,
      "grad_norm": 1.2201352791875548,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 11739
    },
    {
      "epoch": 0.1174,
      "grad_norm": 1.1462293051204187,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 11740
    },
    {
      "epoch": 0.11741,
      "grad_norm": 1.1459578959083037,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 11741
    },
    {
      "epoch": 0.11742,
      "grad_norm": 1.0533399880453962,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 11742
    },
    {
      "epoch": 0.11743,
      "grad_norm": 1.121969173337314,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 11743
    },
    {
      "epoch": 0.11744,
      "grad_norm": 1.4689998599569376,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 11744
    },
    {
      "epoch": 0.11745,
      "grad_norm": 1.3274205401294934,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 11745
    },
    {
      "epoch": 0.11746,
      "grad_norm": 0.8918388851683189,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 11746
    },
    {
      "epoch": 0.11747,
      "grad_norm": 1.0022823662035623,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 11747
    },
    {
      "epoch": 0.11748,
      "grad_norm": 1.2740630639286075,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 11748
    },
    {
      "epoch": 0.11749,
      "grad_norm": 1.1144566017947264,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 11749
    },
    {
      "epoch": 0.1175,
      "grad_norm": 1.2794518402465218,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 11750
    },
    {
      "epoch": 0.11751,
      "grad_norm": 0.9239685969957789,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 11751
    },
    {
      "epoch": 0.11752,
      "grad_norm": 0.9848298008138613,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 11752
    },
    {
      "epoch": 0.11753,
      "grad_norm": 1.2900820404962074,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 11753
    },
    {
      "epoch": 0.11754,
      "grad_norm": 1.06059796209915,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 11754
    },
    {
      "epoch": 0.11755,
      "grad_norm": 1.060399615268824,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 11755
    },
    {
      "epoch": 0.11756,
      "grad_norm": 1.2775892797667003,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 11756
    },
    {
      "epoch": 0.11757,
      "grad_norm": 0.972795804791523,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 11757
    },
    {
      "epoch": 0.11758,
      "grad_norm": 1.0490010562233407,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 11758
    },
    {
      "epoch": 0.11759,
      "grad_norm": 1.2564337969810733,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 11759
    },
    {
      "epoch": 0.1176,
      "grad_norm": 0.9491679826420742,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 11760
    },
    {
      "epoch": 0.11761,
      "grad_norm": 1.1248877574419467,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 11761
    },
    {
      "epoch": 0.11762,
      "grad_norm": 1.0079745325917062,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 11762
    },
    {
      "epoch": 0.11763,
      "grad_norm": 1.3124075277396952,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 11763
    },
    {
      "epoch": 0.11764,
      "grad_norm": 0.9367086914805486,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 11764
    },
    {
      "epoch": 0.11765,
      "grad_norm": 1.1444616985411946,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 11765
    },
    {
      "epoch": 0.11766,
      "grad_norm": 1.3212645040441942,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 11766
    },
    {
      "epoch": 0.11767,
      "grad_norm": 1.1244238476096746,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 11767
    },
    {
      "epoch": 0.11768,
      "grad_norm": 1.1773234175338592,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 11768
    },
    {
      "epoch": 0.11769,
      "grad_norm": 0.9581425424892575,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 11769
    },
    {
      "epoch": 0.1177,
      "grad_norm": 1.0621317193895994,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 11770
    },
    {
      "epoch": 0.11771,
      "grad_norm": 1.0546416995775967,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 11771
    },
    {
      "epoch": 0.11772,
      "grad_norm": 1.4850667429157458,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 11772
    },
    {
      "epoch": 0.11773,
      "grad_norm": 1.1016435020141717,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 11773
    },
    {
      "epoch": 0.11774,
      "grad_norm": 1.356521945100985,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 11774
    },
    {
      "epoch": 0.11775,
      "grad_norm": 1.0785970233598485,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 11775
    },
    {
      "epoch": 0.11776,
      "grad_norm": 1.0520217338363091,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 11776
    },
    {
      "epoch": 0.11777,
      "grad_norm": 1.1807726280590305,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 11777
    },
    {
      "epoch": 0.11778,
      "grad_norm": 0.9514944018351177,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 11778
    },
    {
      "epoch": 0.11779,
      "grad_norm": 1.2589173671064429,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 11779
    },
    {
      "epoch": 0.1178,
      "grad_norm": 0.9943740001334537,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 11780
    },
    {
      "epoch": 0.11781,
      "grad_norm": 1.1756358853971083,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 11781
    },
    {
      "epoch": 0.11782,
      "grad_norm": 1.2088112431840263,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 11782
    },
    {
      "epoch": 0.11783,
      "grad_norm": 1.3874050801177034,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 11783
    },
    {
      "epoch": 0.11784,
      "grad_norm": 0.9053806520882297,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 11784
    },
    {
      "epoch": 0.11785,
      "grad_norm": 0.9013846202308751,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 11785
    },
    {
      "epoch": 0.11786,
      "grad_norm": 0.9788339236866542,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 11786
    },
    {
      "epoch": 0.11787,
      "grad_norm": 1.0873653791790059,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 11787
    },
    {
      "epoch": 0.11788,
      "grad_norm": 1.1421214463569709,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 11788
    },
    {
      "epoch": 0.11789,
      "grad_norm": 1.1779963108684894,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 11789
    },
    {
      "epoch": 0.1179,
      "grad_norm": 0.982025514320481,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 11790
    },
    {
      "epoch": 0.11791,
      "grad_norm": 1.0313215188232239,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 11791
    },
    {
      "epoch": 0.11792,
      "grad_norm": 1.0818733705067438,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 11792
    },
    {
      "epoch": 0.11793,
      "grad_norm": 1.0254561930890065,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 11793
    },
    {
      "epoch": 0.11794,
      "grad_norm": 1.0942218467450253,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 11794
    },
    {
      "epoch": 0.11795,
      "grad_norm": 1.3739924318419376,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 11795
    },
    {
      "epoch": 0.11796,
      "grad_norm": 1.0386170689594838,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 11796
    },
    {
      "epoch": 0.11797,
      "grad_norm": 1.468240521388495,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11797
    },
    {
      "epoch": 0.11798,
      "grad_norm": 0.8881530246130017,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 11798
    },
    {
      "epoch": 0.11799,
      "grad_norm": 0.8667721143037403,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 11799
    },
    {
      "epoch": 0.118,
      "grad_norm": 0.9828151445089266,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 11800
    },
    {
      "epoch": 0.11801,
      "grad_norm": 1.0962943506473461,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 11801
    },
    {
      "epoch": 0.11802,
      "grad_norm": 1.032329474607015,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 11802
    },
    {
      "epoch": 0.11803,
      "grad_norm": 1.0895359905818427,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 11803
    },
    {
      "epoch": 0.11804,
      "grad_norm": 1.2567454842956758,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 11804
    },
    {
      "epoch": 0.11805,
      "grad_norm": 1.1423825854362604,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 11805
    },
    {
      "epoch": 0.11806,
      "grad_norm": 1.289721276916898,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 11806
    },
    {
      "epoch": 0.11807,
      "grad_norm": 1.0671857466036752,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 11807
    },
    {
      "epoch": 0.11808,
      "grad_norm": 1.343505970047031,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 11808
    },
    {
      "epoch": 0.11809,
      "grad_norm": 1.139936318061021,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 11809
    },
    {
      "epoch": 0.1181,
      "grad_norm": 1.244710701696588,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 11810
    },
    {
      "epoch": 0.11811,
      "grad_norm": 1.0503231155416128,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 11811
    },
    {
      "epoch": 0.11812,
      "grad_norm": 1.0926414145035097,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 11812
    },
    {
      "epoch": 0.11813,
      "grad_norm": 1.023841354122939,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 11813
    },
    {
      "epoch": 0.11814,
      "grad_norm": 1.1156410068020937,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 11814
    },
    {
      "epoch": 0.11815,
      "grad_norm": 1.1491708173222321,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 11815
    },
    {
      "epoch": 0.11816,
      "grad_norm": 1.284086474032784,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 11816
    },
    {
      "epoch": 0.11817,
      "grad_norm": 1.0748777372173741,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 11817
    },
    {
      "epoch": 0.11818,
      "grad_norm": 1.029728756967552,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 11818
    },
    {
      "epoch": 0.11819,
      "grad_norm": 1.1223699498022326,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 11819
    },
    {
      "epoch": 0.1182,
      "grad_norm": 1.2151216877560418,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 11820
    },
    {
      "epoch": 0.11821,
      "grad_norm": 0.9447500324212605,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 11821
    },
    {
      "epoch": 0.11822,
      "grad_norm": 1.10373474981526,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 11822
    },
    {
      "epoch": 0.11823,
      "grad_norm": 1.1021794004120868,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 11823
    },
    {
      "epoch": 0.11824,
      "grad_norm": 1.1387450997635156,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 11824
    },
    {
      "epoch": 0.11825,
      "grad_norm": 1.2274566209269968,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 11825
    },
    {
      "epoch": 0.11826,
      "grad_norm": 0.9287450532497696,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 11826
    },
    {
      "epoch": 0.11827,
      "grad_norm": 1.1155557049195648,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 11827
    },
    {
      "epoch": 0.11828,
      "grad_norm": 1.1489952053902721,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 11828
    },
    {
      "epoch": 0.11829,
      "grad_norm": 1.0876359379671903,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 11829
    },
    {
      "epoch": 0.1183,
      "grad_norm": 1.4525289352556954,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 11830
    },
    {
      "epoch": 0.11831,
      "grad_norm": 1.0619833667413887,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 11831
    },
    {
      "epoch": 0.11832,
      "grad_norm": 1.3638115504838992,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 11832
    },
    {
      "epoch": 0.11833,
      "grad_norm": 1.1812529631150601,
      "learning_rate": 0.003,
      "loss": 4.1135,
      "step": 11833
    },
    {
      "epoch": 0.11834,
      "grad_norm": 1.375271052465075,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 11834
    },
    {
      "epoch": 0.11835,
      "grad_norm": 1.1164555597617594,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 11835
    },
    {
      "epoch": 0.11836,
      "grad_norm": 1.009193328635914,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 11836
    },
    {
      "epoch": 0.11837,
      "grad_norm": 1.1768578245175483,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 11837
    },
    {
      "epoch": 0.11838,
      "grad_norm": 1.0258693957941911,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 11838
    },
    {
      "epoch": 0.11839,
      "grad_norm": 1.3065155218168176,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 11839
    },
    {
      "epoch": 0.1184,
      "grad_norm": 0.8799608760122808,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 11840
    },
    {
      "epoch": 0.11841,
      "grad_norm": 0.98901493208036,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 11841
    },
    {
      "epoch": 0.11842,
      "grad_norm": 1.25068801265814,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 11842
    },
    {
      "epoch": 0.11843,
      "grad_norm": 0.8406258499606613,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 11843
    },
    {
      "epoch": 0.11844,
      "grad_norm": 0.8570725315156429,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 11844
    },
    {
      "epoch": 0.11845,
      "grad_norm": 1.0311542246289087,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 11845
    },
    {
      "epoch": 0.11846,
      "grad_norm": 1.3311754584449043,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11846
    },
    {
      "epoch": 0.11847,
      "grad_norm": 1.0010647914601536,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 11847
    },
    {
      "epoch": 0.11848,
      "grad_norm": 1.485318441648701,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 11848
    },
    {
      "epoch": 0.11849,
      "grad_norm": 1.1214138373683373,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 11849
    },
    {
      "epoch": 0.1185,
      "grad_norm": 1.0196511782777165,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 11850
    },
    {
      "epoch": 0.11851,
      "grad_norm": 1.1687837566468635,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 11851
    },
    {
      "epoch": 0.11852,
      "grad_norm": 1.0554697586410853,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 11852
    },
    {
      "epoch": 0.11853,
      "grad_norm": 1.0916125410803077,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 11853
    },
    {
      "epoch": 0.11854,
      "grad_norm": 1.0562018909120507,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 11854
    },
    {
      "epoch": 0.11855,
      "grad_norm": 1.150141586864,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 11855
    },
    {
      "epoch": 0.11856,
      "grad_norm": 1.133521226026218,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 11856
    },
    {
      "epoch": 0.11857,
      "grad_norm": 0.8634651948359986,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 11857
    },
    {
      "epoch": 0.11858,
      "grad_norm": 0.9642019874347069,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 11858
    },
    {
      "epoch": 0.11859,
      "grad_norm": 1.1319671295432074,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 11859
    },
    {
      "epoch": 0.1186,
      "grad_norm": 1.5953846007576176,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 11860
    },
    {
      "epoch": 0.11861,
      "grad_norm": 1.024932034441909,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 11861
    },
    {
      "epoch": 0.11862,
      "grad_norm": 1.2986812785569133,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 11862
    },
    {
      "epoch": 0.11863,
      "grad_norm": 1.0985985673307168,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 11863
    },
    {
      "epoch": 0.11864,
      "grad_norm": 1.310174333179123,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 11864
    },
    {
      "epoch": 0.11865,
      "grad_norm": 0.899614133293594,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 11865
    },
    {
      "epoch": 0.11866,
      "grad_norm": 1.2292245748747577,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 11866
    },
    {
      "epoch": 0.11867,
      "grad_norm": 1.1814766245058643,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 11867
    },
    {
      "epoch": 0.11868,
      "grad_norm": 1.1547264563679283,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 11868
    },
    {
      "epoch": 0.11869,
      "grad_norm": 1.135468755245357,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 11869
    },
    {
      "epoch": 0.1187,
      "grad_norm": 1.1283808115880198,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 11870
    },
    {
      "epoch": 0.11871,
      "grad_norm": 1.1651147439593341,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 11871
    },
    {
      "epoch": 0.11872,
      "grad_norm": 0.952732737427695,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 11872
    },
    {
      "epoch": 0.11873,
      "grad_norm": 1.0044383418061162,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 11873
    },
    {
      "epoch": 0.11874,
      "grad_norm": 1.2403290296527454,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 11874
    },
    {
      "epoch": 0.11875,
      "grad_norm": 1.051254612260916,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 11875
    },
    {
      "epoch": 0.11876,
      "grad_norm": 1.0815830505822788,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 11876
    },
    {
      "epoch": 0.11877,
      "grad_norm": 1.0620445404429997,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 11877
    },
    {
      "epoch": 0.11878,
      "grad_norm": 1.389493017030221,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 11878
    },
    {
      "epoch": 0.11879,
      "grad_norm": 1.085364632408075,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 11879
    },
    {
      "epoch": 0.1188,
      "grad_norm": 1.0560306403664153,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 11880
    },
    {
      "epoch": 0.11881,
      "grad_norm": 1.2325914466530217,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 11881
    },
    {
      "epoch": 0.11882,
      "grad_norm": 0.9430171665821426,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 11882
    },
    {
      "epoch": 0.11883,
      "grad_norm": 0.9922866907169047,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 11883
    },
    {
      "epoch": 0.11884,
      "grad_norm": 1.1876342527608525,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 11884
    },
    {
      "epoch": 0.11885,
      "grad_norm": 1.127731596475419,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 11885
    },
    {
      "epoch": 0.11886,
      "grad_norm": 1.219941205295011,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 11886
    },
    {
      "epoch": 0.11887,
      "grad_norm": 1.1402853107785413,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 11887
    },
    {
      "epoch": 0.11888,
      "grad_norm": 1.2308931402462295,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 11888
    },
    {
      "epoch": 0.11889,
      "grad_norm": 1.0604024665201628,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 11889
    },
    {
      "epoch": 0.1189,
      "grad_norm": 1.221421625109362,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 11890
    },
    {
      "epoch": 0.11891,
      "grad_norm": 1.073227547779595,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 11891
    },
    {
      "epoch": 0.11892,
      "grad_norm": 1.091635809784193,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 11892
    },
    {
      "epoch": 0.11893,
      "grad_norm": 1.0494296892537798,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 11893
    },
    {
      "epoch": 0.11894,
      "grad_norm": 1.0319729573369925,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 11894
    },
    {
      "epoch": 0.11895,
      "grad_norm": 1.0088727056452342,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 11895
    },
    {
      "epoch": 0.11896,
      "grad_norm": 1.2211882582437528,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 11896
    },
    {
      "epoch": 0.11897,
      "grad_norm": 1.22459237408397,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 11897
    },
    {
      "epoch": 0.11898,
      "grad_norm": 1.2964883702512249,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 11898
    },
    {
      "epoch": 0.11899,
      "grad_norm": 1.1006710611829373,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 11899
    },
    {
      "epoch": 0.119,
      "grad_norm": 1.1283526124007004,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 11900
    },
    {
      "epoch": 0.11901,
      "grad_norm": 1.0477828176444812,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 11901
    },
    {
      "epoch": 0.11902,
      "grad_norm": 1.1105312518883845,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 11902
    },
    {
      "epoch": 0.11903,
      "grad_norm": 0.884943929413305,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 11903
    },
    {
      "epoch": 0.11904,
      "grad_norm": 0.9284368125352005,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 11904
    },
    {
      "epoch": 0.11905,
      "grad_norm": 1.0589265482717447,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 11905
    },
    {
      "epoch": 0.11906,
      "grad_norm": 1.1092228565605042,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 11906
    },
    {
      "epoch": 0.11907,
      "grad_norm": 1.1941986399865445,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 11907
    },
    {
      "epoch": 0.11908,
      "grad_norm": 1.046439168125344,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 11908
    },
    {
      "epoch": 0.11909,
      "grad_norm": 1.3284280153633887,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 11909
    },
    {
      "epoch": 0.1191,
      "grad_norm": 1.1162343111004016,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 11910
    },
    {
      "epoch": 0.11911,
      "grad_norm": 1.2286716614076243,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 11911
    },
    {
      "epoch": 0.11912,
      "grad_norm": 1.0845810445242683,
      "learning_rate": 0.003,
      "loss": 4.1171,
      "step": 11912
    },
    {
      "epoch": 0.11913,
      "grad_norm": 1.1757700274556278,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 11913
    },
    {
      "epoch": 0.11914,
      "grad_norm": 1.0720002142634966,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 11914
    },
    {
      "epoch": 0.11915,
      "grad_norm": 0.9585568712803956,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 11915
    },
    {
      "epoch": 0.11916,
      "grad_norm": 1.118410361514395,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 11916
    },
    {
      "epoch": 0.11917,
      "grad_norm": 1.1439516461494486,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 11917
    },
    {
      "epoch": 0.11918,
      "grad_norm": 0.8774537417656473,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 11918
    },
    {
      "epoch": 0.11919,
      "grad_norm": 0.9912767987498607,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 11919
    },
    {
      "epoch": 0.1192,
      "grad_norm": 1.402160777194763,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 11920
    },
    {
      "epoch": 0.11921,
      "grad_norm": 1.068468107904343,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 11921
    },
    {
      "epoch": 0.11922,
      "grad_norm": 1.3659602551264145,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 11922
    },
    {
      "epoch": 0.11923,
      "grad_norm": 1.373546498881002,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 11923
    },
    {
      "epoch": 0.11924,
      "grad_norm": 1.2060840228601442,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 11924
    },
    {
      "epoch": 0.11925,
      "grad_norm": 1.127806727390189,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 11925
    },
    {
      "epoch": 0.11926,
      "grad_norm": 1.0365064543575828,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 11926
    },
    {
      "epoch": 0.11927,
      "grad_norm": 1.1116861558348419,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 11927
    },
    {
      "epoch": 0.11928,
      "grad_norm": 0.8824401374304045,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 11928
    },
    {
      "epoch": 0.11929,
      "grad_norm": 0.91805897099131,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 11929
    },
    {
      "epoch": 0.1193,
      "grad_norm": 1.1680131795885302,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 11930
    },
    {
      "epoch": 0.11931,
      "grad_norm": 1.1886823191495501,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 11931
    },
    {
      "epoch": 0.11932,
      "grad_norm": 1.1887751402826594,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 11932
    },
    {
      "epoch": 0.11933,
      "grad_norm": 1.1179307530288558,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 11933
    },
    {
      "epoch": 0.11934,
      "grad_norm": 1.0489444539389505,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 11934
    },
    {
      "epoch": 0.11935,
      "grad_norm": 1.237415215488098,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 11935
    },
    {
      "epoch": 0.11936,
      "grad_norm": 0.9957532742252633,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 11936
    },
    {
      "epoch": 0.11937,
      "grad_norm": 1.1891332584840746,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 11937
    },
    {
      "epoch": 0.11938,
      "grad_norm": 1.0463104442432303,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 11938
    },
    {
      "epoch": 0.11939,
      "grad_norm": 1.2373487906517007,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 11939
    },
    {
      "epoch": 0.1194,
      "grad_norm": 0.9982842826684045,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 11940
    },
    {
      "epoch": 0.11941,
      "grad_norm": 1.4640902331855827,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 11941
    },
    {
      "epoch": 0.11942,
      "grad_norm": 1.0480877070110093,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 11942
    },
    {
      "epoch": 0.11943,
      "grad_norm": 1.2282928382729854,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 11943
    },
    {
      "epoch": 0.11944,
      "grad_norm": 1.2343243580325551,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 11944
    },
    {
      "epoch": 0.11945,
      "grad_norm": 1.2027915615430216,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 11945
    },
    {
      "epoch": 0.11946,
      "grad_norm": 0.8917739824369894,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 11946
    },
    {
      "epoch": 0.11947,
      "grad_norm": 1.0447613041945965,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 11947
    },
    {
      "epoch": 0.11948,
      "grad_norm": 1.2178185052397315,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 11948
    },
    {
      "epoch": 0.11949,
      "grad_norm": 0.9378135352565747,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 11949
    },
    {
      "epoch": 0.1195,
      "grad_norm": 1.092216526784142,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 11950
    },
    {
      "epoch": 0.11951,
      "grad_norm": 1.100524960745319,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 11951
    },
    {
      "epoch": 0.11952,
      "grad_norm": 1.1444651267711,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 11952
    },
    {
      "epoch": 0.11953,
      "grad_norm": 1.203498741669757,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 11953
    },
    {
      "epoch": 0.11954,
      "grad_norm": 1.1427459985323243,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 11954
    },
    {
      "epoch": 0.11955,
      "grad_norm": 1.1361070526546504,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 11955
    },
    {
      "epoch": 0.11956,
      "grad_norm": 1.1205161898186382,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 11956
    },
    {
      "epoch": 0.11957,
      "grad_norm": 0.9815044846386909,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 11957
    },
    {
      "epoch": 0.11958,
      "grad_norm": 1.0013378538157867,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 11958
    },
    {
      "epoch": 0.11959,
      "grad_norm": 1.000998732925407,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 11959
    },
    {
      "epoch": 0.1196,
      "grad_norm": 1.0893670772900628,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 11960
    },
    {
      "epoch": 0.11961,
      "grad_norm": 1.0591646896060922,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 11961
    },
    {
      "epoch": 0.11962,
      "grad_norm": 1.093398185067271,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 11962
    },
    {
      "epoch": 0.11963,
      "grad_norm": 1.2085490823454053,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 11963
    },
    {
      "epoch": 0.11964,
      "grad_norm": 1.1371502930714275,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 11964
    },
    {
      "epoch": 0.11965,
      "grad_norm": 1.1255892314647256,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 11965
    },
    {
      "epoch": 0.11966,
      "grad_norm": 1.0853962886107014,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 11966
    },
    {
      "epoch": 0.11967,
      "grad_norm": 1.3747539499194141,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 11967
    },
    {
      "epoch": 0.11968,
      "grad_norm": 1.2647185472649858,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 11968
    },
    {
      "epoch": 0.11969,
      "grad_norm": 1.0741775586091273,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 11969
    },
    {
      "epoch": 0.1197,
      "grad_norm": 1.0350752138665191,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 11970
    },
    {
      "epoch": 0.11971,
      "grad_norm": 1.191119641426522,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 11971
    },
    {
      "epoch": 0.11972,
      "grad_norm": 0.9437137639131957,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 11972
    },
    {
      "epoch": 0.11973,
      "grad_norm": 1.283613714036086,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 11973
    },
    {
      "epoch": 0.11974,
      "grad_norm": 1.279899355911932,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 11974
    },
    {
      "epoch": 0.11975,
      "grad_norm": 1.0249399599556033,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 11975
    },
    {
      "epoch": 0.11976,
      "grad_norm": 1.321212623947207,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 11976
    },
    {
      "epoch": 0.11977,
      "grad_norm": 1.0165950102473211,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 11977
    },
    {
      "epoch": 0.11978,
      "grad_norm": 0.9277193711873348,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 11978
    },
    {
      "epoch": 0.11979,
      "grad_norm": 0.9280891520464905,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 11979
    },
    {
      "epoch": 0.1198,
      "grad_norm": 0.9840854182938272,
      "learning_rate": 0.003,
      "loss": 4.1049,
      "step": 11980
    },
    {
      "epoch": 0.11981,
      "grad_norm": 1.3684303366293669,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 11981
    },
    {
      "epoch": 0.11982,
      "grad_norm": 0.9945901016820448,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 11982
    },
    {
      "epoch": 0.11983,
      "grad_norm": 1.2752089237002326,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 11983
    },
    {
      "epoch": 0.11984,
      "grad_norm": 1.0771984928694656,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 11984
    },
    {
      "epoch": 0.11985,
      "grad_norm": 1.3299901938391445,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 11985
    },
    {
      "epoch": 0.11986,
      "grad_norm": 0.9586548273984268,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 11986
    },
    {
      "epoch": 0.11987,
      "grad_norm": 0.9948402579369039,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 11987
    },
    {
      "epoch": 0.11988,
      "grad_norm": 1.3296600841391015,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 11988
    },
    {
      "epoch": 0.11989,
      "grad_norm": 0.8762782098697669,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 11989
    },
    {
      "epoch": 0.1199,
      "grad_norm": 1.076562501910153,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 11990
    },
    {
      "epoch": 0.11991,
      "grad_norm": 1.4625601275009004,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 11991
    },
    {
      "epoch": 0.11992,
      "grad_norm": 1.00751868644314,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 11992
    },
    {
      "epoch": 0.11993,
      "grad_norm": 1.3989214406531236,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 11993
    },
    {
      "epoch": 0.11994,
      "grad_norm": 0.7532071646779008,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 11994
    },
    {
      "epoch": 0.11995,
      "grad_norm": 0.9108244533553659,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 11995
    },
    {
      "epoch": 0.11996,
      "grad_norm": 1.037693604414362,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 11996
    },
    {
      "epoch": 0.11997,
      "grad_norm": 1.510249381824229,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 11997
    },
    {
      "epoch": 0.11998,
      "grad_norm": 0.9025590524571965,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 11998
    },
    {
      "epoch": 0.11999,
      "grad_norm": 1.189616049192571,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 11999
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3350930490406336,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 12000
    },
    {
      "epoch": 0.12001,
      "grad_norm": 1.1180567164673338,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 12001
    },
    {
      "epoch": 0.12002,
      "grad_norm": 1.1746116800621182,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 12002
    },
    {
      "epoch": 0.12003,
      "grad_norm": 1.0368064271859383,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 12003
    },
    {
      "epoch": 0.12004,
      "grad_norm": 1.0681096635948948,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 12004
    },
    {
      "epoch": 0.12005,
      "grad_norm": 1.1099414967667618,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 12005
    },
    {
      "epoch": 0.12006,
      "grad_norm": 0.9803751601666846,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 12006
    },
    {
      "epoch": 0.12007,
      "grad_norm": 1.2223238024022165,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 12007
    },
    {
      "epoch": 0.12008,
      "grad_norm": 1.0685166133012711,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 12008
    },
    {
      "epoch": 0.12009,
      "grad_norm": 1.1599099213556685,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 12009
    },
    {
      "epoch": 0.1201,
      "grad_norm": 1.021768365708924,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 12010
    },
    {
      "epoch": 0.12011,
      "grad_norm": 1.1087618201802516,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 12011
    },
    {
      "epoch": 0.12012,
      "grad_norm": 1.3091376574754292,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 12012
    },
    {
      "epoch": 0.12013,
      "grad_norm": 1.3389948200754302,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 12013
    },
    {
      "epoch": 0.12014,
      "grad_norm": 1.0677259298014297,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 12014
    },
    {
      "epoch": 0.12015,
      "grad_norm": 1.3527692324615435,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12015
    },
    {
      "epoch": 0.12016,
      "grad_norm": 1.1878208635574032,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 12016
    },
    {
      "epoch": 0.12017,
      "grad_norm": 1.1556818900091732,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 12017
    },
    {
      "epoch": 0.12018,
      "grad_norm": 1.2874267600368892,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 12018
    },
    {
      "epoch": 0.12019,
      "grad_norm": 0.9336107902349933,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 12019
    },
    {
      "epoch": 0.1202,
      "grad_norm": 1.2932714597916453,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 12020
    },
    {
      "epoch": 0.12021,
      "grad_norm": 1.0130309654760399,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 12021
    },
    {
      "epoch": 0.12022,
      "grad_norm": 1.2879254166353613,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 12022
    },
    {
      "epoch": 0.12023,
      "grad_norm": 1.0818443860488969,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 12023
    },
    {
      "epoch": 0.12024,
      "grad_norm": 1.0860566591981275,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 12024
    },
    {
      "epoch": 0.12025,
      "grad_norm": 1.2135435164557746,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 12025
    },
    {
      "epoch": 0.12026,
      "grad_norm": 0.9841989092312695,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 12026
    },
    {
      "epoch": 0.12027,
      "grad_norm": 1.0266977635183048,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 12027
    },
    {
      "epoch": 0.12028,
      "grad_norm": 1.2117018689721994,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 12028
    },
    {
      "epoch": 0.12029,
      "grad_norm": 1.2276826573957662,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 12029
    },
    {
      "epoch": 0.1203,
      "grad_norm": 1.1518195603304868,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 12030
    },
    {
      "epoch": 0.12031,
      "grad_norm": 1.2467861618259808,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 12031
    },
    {
      "epoch": 0.12032,
      "grad_norm": 1.16415301239379,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 12032
    },
    {
      "epoch": 0.12033,
      "grad_norm": 1.0590101051657046,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 12033
    },
    {
      "epoch": 0.12034,
      "grad_norm": 1.2402649179536591,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 12034
    },
    {
      "epoch": 0.12035,
      "grad_norm": 1.0642608565899094,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 12035
    },
    {
      "epoch": 0.12036,
      "grad_norm": 1.1255827242322958,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 12036
    },
    {
      "epoch": 0.12037,
      "grad_norm": 1.1384628048440417,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 12037
    },
    {
      "epoch": 0.12038,
      "grad_norm": 1.1274178417869605,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 12038
    },
    {
      "epoch": 0.12039,
      "grad_norm": 0.9995922903631983,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 12039
    },
    {
      "epoch": 0.1204,
      "grad_norm": 1.026389899399592,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 12040
    },
    {
      "epoch": 0.12041,
      "grad_norm": 1.1044160233714642,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 12041
    },
    {
      "epoch": 0.12042,
      "grad_norm": 1.080147358818234,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 12042
    },
    {
      "epoch": 0.12043,
      "grad_norm": 1.121781114814016,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 12043
    },
    {
      "epoch": 0.12044,
      "grad_norm": 1.0909551039016203,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 12044
    },
    {
      "epoch": 0.12045,
      "grad_norm": 1.191158141010099,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 12045
    },
    {
      "epoch": 0.12046,
      "grad_norm": 1.066915824442074,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 12046
    },
    {
      "epoch": 0.12047,
      "grad_norm": 0.9596003185844033,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 12047
    },
    {
      "epoch": 0.12048,
      "grad_norm": 1.1701885942631547,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12048
    },
    {
      "epoch": 0.12049,
      "grad_norm": 1.0547752412751898,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 12049
    },
    {
      "epoch": 0.1205,
      "grad_norm": 1.410530280129304,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 12050
    },
    {
      "epoch": 0.12051,
      "grad_norm": 1.2104800826176374,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 12051
    },
    {
      "epoch": 0.12052,
      "grad_norm": 1.259195544404022,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 12052
    },
    {
      "epoch": 0.12053,
      "grad_norm": 1.0120480121923963,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 12053
    },
    {
      "epoch": 0.12054,
      "grad_norm": 1.2133091190993877,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 12054
    },
    {
      "epoch": 0.12055,
      "grad_norm": 1.086573324709519,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 12055
    },
    {
      "epoch": 0.12056,
      "grad_norm": 0.9366357796969548,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 12056
    },
    {
      "epoch": 0.12057,
      "grad_norm": 0.9413207797417531,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12057
    },
    {
      "epoch": 0.12058,
      "grad_norm": 1.054970577763672,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 12058
    },
    {
      "epoch": 0.12059,
      "grad_norm": 1.241755057586524,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 12059
    },
    {
      "epoch": 0.1206,
      "grad_norm": 1.178759157791092,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 12060
    },
    {
      "epoch": 0.12061,
      "grad_norm": 1.072634607621007,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 12061
    },
    {
      "epoch": 0.12062,
      "grad_norm": 1.2335975090778577,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 12062
    },
    {
      "epoch": 0.12063,
      "grad_norm": 1.0043724859492582,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 12063
    },
    {
      "epoch": 0.12064,
      "grad_norm": 1.1268181667654482,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 12064
    },
    {
      "epoch": 0.12065,
      "grad_norm": 1.3467584212302015,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 12065
    },
    {
      "epoch": 0.12066,
      "grad_norm": 1.1544538001555558,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 12066
    },
    {
      "epoch": 0.12067,
      "grad_norm": 1.255977573156133,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 12067
    },
    {
      "epoch": 0.12068,
      "grad_norm": 0.9350775949046741,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 12068
    },
    {
      "epoch": 0.12069,
      "grad_norm": 1.088070082817422,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 12069
    },
    {
      "epoch": 0.1207,
      "grad_norm": 1.0486078245374844,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 12070
    },
    {
      "epoch": 0.12071,
      "grad_norm": 1.107745717985881,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 12071
    },
    {
      "epoch": 0.12072,
      "grad_norm": 1.376209870161379,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 12072
    },
    {
      "epoch": 0.12073,
      "grad_norm": 1.0983080244094117,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 12073
    },
    {
      "epoch": 0.12074,
      "grad_norm": 1.2145115045698978,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 12074
    },
    {
      "epoch": 0.12075,
      "grad_norm": 1.0792027485780036,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 12075
    },
    {
      "epoch": 0.12076,
      "grad_norm": 0.9724230941494237,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 12076
    },
    {
      "epoch": 0.12077,
      "grad_norm": 1.0309662283880128,
      "learning_rate": 0.003,
      "loss": 4.0275,
      "step": 12077
    },
    {
      "epoch": 0.12078,
      "grad_norm": 1.4142456890578097,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 12078
    },
    {
      "epoch": 0.12079,
      "grad_norm": 1.0604933955299787,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 12079
    },
    {
      "epoch": 0.1208,
      "grad_norm": 1.1335698224805126,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 12080
    },
    {
      "epoch": 0.12081,
      "grad_norm": 1.015134918616046,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 12081
    },
    {
      "epoch": 0.12082,
      "grad_norm": 1.3097320253417457,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 12082
    },
    {
      "epoch": 0.12083,
      "grad_norm": 1.0066362360319434,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 12083
    },
    {
      "epoch": 0.12084,
      "grad_norm": 1.0235728852892743,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 12084
    },
    {
      "epoch": 0.12085,
      "grad_norm": 1.0630954451681172,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 12085
    },
    {
      "epoch": 0.12086,
      "grad_norm": 1.218539724165097,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 12086
    },
    {
      "epoch": 0.12087,
      "grad_norm": 1.1620137919872433,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 12087
    },
    {
      "epoch": 0.12088,
      "grad_norm": 1.177282673119695,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 12088
    },
    {
      "epoch": 0.12089,
      "grad_norm": 0.9972352639834798,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12089
    },
    {
      "epoch": 0.1209,
      "grad_norm": 1.0959043321646311,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 12090
    },
    {
      "epoch": 0.12091,
      "grad_norm": 1.1874303616047068,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 12091
    },
    {
      "epoch": 0.12092,
      "grad_norm": 1.101021855644516,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 12092
    },
    {
      "epoch": 0.12093,
      "grad_norm": 1.2633017678088525,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 12093
    },
    {
      "epoch": 0.12094,
      "grad_norm": 1.0519698321311366,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 12094
    },
    {
      "epoch": 0.12095,
      "grad_norm": 1.1665993943190327,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 12095
    },
    {
      "epoch": 0.12096,
      "grad_norm": 1.1096000107872868,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 12096
    },
    {
      "epoch": 0.12097,
      "grad_norm": 1.0635730265142278,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 12097
    },
    {
      "epoch": 0.12098,
      "grad_norm": 1.2764783466361678,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 12098
    },
    {
      "epoch": 0.12099,
      "grad_norm": 0.9051264914681529,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 12099
    },
    {
      "epoch": 0.121,
      "grad_norm": 0.9249608490415164,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 12100
    },
    {
      "epoch": 0.12101,
      "grad_norm": 1.0835718975720605,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 12101
    },
    {
      "epoch": 0.12102,
      "grad_norm": 1.1642303094091533,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 12102
    },
    {
      "epoch": 0.12103,
      "grad_norm": 1.1982430786318965,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 12103
    },
    {
      "epoch": 0.12104,
      "grad_norm": 0.9021525338319302,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 12104
    },
    {
      "epoch": 0.12105,
      "grad_norm": 1.025465093748409,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 12105
    },
    {
      "epoch": 0.12106,
      "grad_norm": 1.0287080964025033,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 12106
    },
    {
      "epoch": 0.12107,
      "grad_norm": 1.1893433762623093,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 12107
    },
    {
      "epoch": 0.12108,
      "grad_norm": 1.1346629077604669,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 12108
    },
    {
      "epoch": 0.12109,
      "grad_norm": 1.384241655173453,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 12109
    },
    {
      "epoch": 0.1211,
      "grad_norm": 1.1929830857172083,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 12110
    },
    {
      "epoch": 0.12111,
      "grad_norm": 1.2157542584926002,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 12111
    },
    {
      "epoch": 0.12112,
      "grad_norm": 1.1414637982634577,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 12112
    },
    {
      "epoch": 0.12113,
      "grad_norm": 1.0356391274044419,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 12113
    },
    {
      "epoch": 0.12114,
      "grad_norm": 1.182813601046115,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 12114
    },
    {
      "epoch": 0.12115,
      "grad_norm": 0.9762033320326391,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 12115
    },
    {
      "epoch": 0.12116,
      "grad_norm": 1.1921654583478314,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 12116
    },
    {
      "epoch": 0.12117,
      "grad_norm": 1.341606992951073,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 12117
    },
    {
      "epoch": 0.12118,
      "grad_norm": 0.8182518816856665,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 12118
    },
    {
      "epoch": 0.12119,
      "grad_norm": 0.8704177926517407,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 12119
    },
    {
      "epoch": 0.1212,
      "grad_norm": 1.0058197349809026,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 12120
    },
    {
      "epoch": 0.12121,
      "grad_norm": 1.069283234495121,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 12121
    },
    {
      "epoch": 0.12122,
      "grad_norm": 0.9201542200711905,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 12122
    },
    {
      "epoch": 0.12123,
      "grad_norm": 1.110694878714642,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 12123
    },
    {
      "epoch": 0.12124,
      "grad_norm": 1.1291370286216942,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 12124
    },
    {
      "epoch": 0.12125,
      "grad_norm": 1.161710409599256,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 12125
    },
    {
      "epoch": 0.12126,
      "grad_norm": 1.2729178933350158,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 12126
    },
    {
      "epoch": 0.12127,
      "grad_norm": 1.1261437334380782,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 12127
    },
    {
      "epoch": 0.12128,
      "grad_norm": 1.093746833791882,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 12128
    },
    {
      "epoch": 0.12129,
      "grad_norm": 1.2284949310268252,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 12129
    },
    {
      "epoch": 0.1213,
      "grad_norm": 0.9984993555404615,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 12130
    },
    {
      "epoch": 0.12131,
      "grad_norm": 1.0508566744500767,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 12131
    },
    {
      "epoch": 0.12132,
      "grad_norm": 1.140820512254794,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 12132
    },
    {
      "epoch": 0.12133,
      "grad_norm": 1.0677992380438892,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 12133
    },
    {
      "epoch": 0.12134,
      "grad_norm": 1.041338898545635,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 12134
    },
    {
      "epoch": 0.12135,
      "grad_norm": 1.3946407208815519,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 12135
    },
    {
      "epoch": 0.12136,
      "grad_norm": 1.1006237334365854,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 12136
    },
    {
      "epoch": 0.12137,
      "grad_norm": 1.458823136129887,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 12137
    },
    {
      "epoch": 0.12138,
      "grad_norm": 1.0559635113382873,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 12138
    },
    {
      "epoch": 0.12139,
      "grad_norm": 1.1285272904820276,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 12139
    },
    {
      "epoch": 0.1214,
      "grad_norm": 1.0897185720280367,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 12140
    },
    {
      "epoch": 0.12141,
      "grad_norm": 1.1955016181778377,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 12141
    },
    {
      "epoch": 0.12142,
      "grad_norm": 1.325169066507099,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 12142
    },
    {
      "epoch": 0.12143,
      "grad_norm": 1.035485453079449,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 12143
    },
    {
      "epoch": 0.12144,
      "grad_norm": 1.226647122460927,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 12144
    },
    {
      "epoch": 0.12145,
      "grad_norm": 0.9832385513042932,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 12145
    },
    {
      "epoch": 0.12146,
      "grad_norm": 1.3042070469595108,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 12146
    },
    {
      "epoch": 0.12147,
      "grad_norm": 1.2735050426487993,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 12147
    },
    {
      "epoch": 0.12148,
      "grad_norm": 1.011082663592818,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 12148
    },
    {
      "epoch": 0.12149,
      "grad_norm": 1.2474343009869946,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 12149
    },
    {
      "epoch": 0.1215,
      "grad_norm": 1.079953110653683,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 12150
    },
    {
      "epoch": 0.12151,
      "grad_norm": 1.1442982278151164,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 12151
    },
    {
      "epoch": 0.12152,
      "grad_norm": 0.9983388572149424,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 12152
    },
    {
      "epoch": 0.12153,
      "grad_norm": 1.2742688036210805,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 12153
    },
    {
      "epoch": 0.12154,
      "grad_norm": 1.1950400478983967,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 12154
    },
    {
      "epoch": 0.12155,
      "grad_norm": 1.19847125467165,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 12155
    },
    {
      "epoch": 0.12156,
      "grad_norm": 1.0274718673323897,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 12156
    },
    {
      "epoch": 0.12157,
      "grad_norm": 1.100133073873854,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 12157
    },
    {
      "epoch": 0.12158,
      "grad_norm": 1.1818660159283922,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 12158
    },
    {
      "epoch": 0.12159,
      "grad_norm": 1.13858706810257,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 12159
    },
    {
      "epoch": 0.1216,
      "grad_norm": 0.9275078508631227,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 12160
    },
    {
      "epoch": 0.12161,
      "grad_norm": 1.1328726130452051,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 12161
    },
    {
      "epoch": 0.12162,
      "grad_norm": 1.2164427451782454,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 12162
    },
    {
      "epoch": 0.12163,
      "grad_norm": 1.231293188312221,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 12163
    },
    {
      "epoch": 0.12164,
      "grad_norm": 1.000554865386488,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 12164
    },
    {
      "epoch": 0.12165,
      "grad_norm": 1.2202696872836503,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 12165
    },
    {
      "epoch": 0.12166,
      "grad_norm": 1.0390095032635105,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12166
    },
    {
      "epoch": 0.12167,
      "grad_norm": 1.0585462345095171,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 12167
    },
    {
      "epoch": 0.12168,
      "grad_norm": 1.2386723240033908,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 12168
    },
    {
      "epoch": 0.12169,
      "grad_norm": 0.9660827919308314,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 12169
    },
    {
      "epoch": 0.1217,
      "grad_norm": 1.1345186857579614,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 12170
    },
    {
      "epoch": 0.12171,
      "grad_norm": 1.0463712239949596,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12171
    },
    {
      "epoch": 0.12172,
      "grad_norm": 1.1581387142451605,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 12172
    },
    {
      "epoch": 0.12173,
      "grad_norm": 1.0109749774734238,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12173
    },
    {
      "epoch": 0.12174,
      "grad_norm": 1.2056594441816655,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 12174
    },
    {
      "epoch": 0.12175,
      "grad_norm": 1.1351586399822327,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 12175
    },
    {
      "epoch": 0.12176,
      "grad_norm": 1.1631779505207185,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 12176
    },
    {
      "epoch": 0.12177,
      "grad_norm": 1.1536486927661425,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 12177
    },
    {
      "epoch": 0.12178,
      "grad_norm": 1.0692593338617544,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 12178
    },
    {
      "epoch": 0.12179,
      "grad_norm": 1.1624174038135675,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 12179
    },
    {
      "epoch": 0.1218,
      "grad_norm": 0.9717071055481967,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 12180
    },
    {
      "epoch": 0.12181,
      "grad_norm": 1.098385052809246,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 12181
    },
    {
      "epoch": 0.12182,
      "grad_norm": 1.1531436274372961,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 12182
    },
    {
      "epoch": 0.12183,
      "grad_norm": 0.9570030729100146,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 12183
    },
    {
      "epoch": 0.12184,
      "grad_norm": 1.0412301813920144,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 12184
    },
    {
      "epoch": 0.12185,
      "grad_norm": 1.3219954197115575,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 12185
    },
    {
      "epoch": 0.12186,
      "grad_norm": 1.2067563045208698,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 12186
    },
    {
      "epoch": 0.12187,
      "grad_norm": 1.3637614562032707,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 12187
    },
    {
      "epoch": 0.12188,
      "grad_norm": 1.082708529547863,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 12188
    },
    {
      "epoch": 0.12189,
      "grad_norm": 1.2208834261957746,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 12189
    },
    {
      "epoch": 0.1219,
      "grad_norm": 1.086438339445444,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12190
    },
    {
      "epoch": 0.12191,
      "grad_norm": 0.9994395476535417,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 12191
    },
    {
      "epoch": 0.12192,
      "grad_norm": 1.1637447492294328,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 12192
    },
    {
      "epoch": 0.12193,
      "grad_norm": 1.0482426916290108,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 12193
    },
    {
      "epoch": 0.12194,
      "grad_norm": 1.19499071880003,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 12194
    },
    {
      "epoch": 0.12195,
      "grad_norm": 1.0344994185916214,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 12195
    },
    {
      "epoch": 0.12196,
      "grad_norm": 1.2946632998055925,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 12196
    },
    {
      "epoch": 0.12197,
      "grad_norm": 1.1635059304700333,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 12197
    },
    {
      "epoch": 0.12198,
      "grad_norm": 0.927098380296469,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 12198
    },
    {
      "epoch": 0.12199,
      "grad_norm": 1.057694142435552,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 12199
    },
    {
      "epoch": 0.122,
      "grad_norm": 1.3472658537272546,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 12200
    },
    {
      "epoch": 0.12201,
      "grad_norm": 1.0889234944776072,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 12201
    },
    {
      "epoch": 0.12202,
      "grad_norm": 1.3188432472041998,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 12202
    },
    {
      "epoch": 0.12203,
      "grad_norm": 0.9595579361138602,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 12203
    },
    {
      "epoch": 0.12204,
      "grad_norm": 1.1922547447780105,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 12204
    },
    {
      "epoch": 0.12205,
      "grad_norm": 1.2992635103281398,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 12205
    },
    {
      "epoch": 0.12206,
      "grad_norm": 0.9557362537371793,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 12206
    },
    {
      "epoch": 0.12207,
      "grad_norm": 0.9184107752252704,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 12207
    },
    {
      "epoch": 0.12208,
      "grad_norm": 1.178789407610477,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 12208
    },
    {
      "epoch": 0.12209,
      "grad_norm": 1.1231869598587525,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12209
    },
    {
      "epoch": 0.1221,
      "grad_norm": 1.2585732335258468,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 12210
    },
    {
      "epoch": 0.12211,
      "grad_norm": 1.309785344594289,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 12211
    },
    {
      "epoch": 0.12212,
      "grad_norm": 1.1308828923601764,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 12212
    },
    {
      "epoch": 0.12213,
      "grad_norm": 1.472297745706311,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 12213
    },
    {
      "epoch": 0.12214,
      "grad_norm": 1.1166151733609975,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 12214
    },
    {
      "epoch": 0.12215,
      "grad_norm": 1.0804515266161436,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 12215
    },
    {
      "epoch": 0.12216,
      "grad_norm": 1.2400711237317346,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 12216
    },
    {
      "epoch": 0.12217,
      "grad_norm": 0.9418513988414021,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 12217
    },
    {
      "epoch": 0.12218,
      "grad_norm": 1.1394363437989878,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 12218
    },
    {
      "epoch": 0.12219,
      "grad_norm": 1.1372327507115731,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 12219
    },
    {
      "epoch": 0.1222,
      "grad_norm": 1.161677959167776,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12220
    },
    {
      "epoch": 0.12221,
      "grad_norm": 1.0789544842404515,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 12221
    },
    {
      "epoch": 0.12222,
      "grad_norm": 1.0398693221757789,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12222
    },
    {
      "epoch": 0.12223,
      "grad_norm": 1.1267339048732326,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 12223
    },
    {
      "epoch": 0.12224,
      "grad_norm": 1.2760752189140523,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 12224
    },
    {
      "epoch": 0.12225,
      "grad_norm": 0.9262198806329611,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 12225
    },
    {
      "epoch": 0.12226,
      "grad_norm": 1.043866646116881,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 12226
    },
    {
      "epoch": 0.12227,
      "grad_norm": 1.26294589943615,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 12227
    },
    {
      "epoch": 0.12228,
      "grad_norm": 0.7806838445270528,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 12228
    },
    {
      "epoch": 0.12229,
      "grad_norm": 1.1547423675011483,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 12229
    },
    {
      "epoch": 0.1223,
      "grad_norm": 1.1320036145420767,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 12230
    },
    {
      "epoch": 0.12231,
      "grad_norm": 1.3764638273278833,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 12231
    },
    {
      "epoch": 0.12232,
      "grad_norm": 1.0128482401645624,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 12232
    },
    {
      "epoch": 0.12233,
      "grad_norm": 1.1726363376664437,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 12233
    },
    {
      "epoch": 0.12234,
      "grad_norm": 1.1040794996492702,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 12234
    },
    {
      "epoch": 0.12235,
      "grad_norm": 1.0358138471123548,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 12235
    },
    {
      "epoch": 0.12236,
      "grad_norm": 1.2145747382949974,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 12236
    },
    {
      "epoch": 0.12237,
      "grad_norm": 1.2700041322440572,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 12237
    },
    {
      "epoch": 0.12238,
      "grad_norm": 1.1154508261314582,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 12238
    },
    {
      "epoch": 0.12239,
      "grad_norm": 1.0109089876091282,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 12239
    },
    {
      "epoch": 0.1224,
      "grad_norm": 1.185544974621423,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 12240
    },
    {
      "epoch": 0.12241,
      "grad_norm": 1.1933483328999805,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 12241
    },
    {
      "epoch": 0.12242,
      "grad_norm": 1.2439943885465512,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 12242
    },
    {
      "epoch": 0.12243,
      "grad_norm": 1.2486428182377738,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 12243
    },
    {
      "epoch": 0.12244,
      "grad_norm": 1.2058088997505476,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 12244
    },
    {
      "epoch": 0.12245,
      "grad_norm": 1.0118112593781434,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 12245
    },
    {
      "epoch": 0.12246,
      "grad_norm": 1.2149480279494373,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 12246
    },
    {
      "epoch": 0.12247,
      "grad_norm": 1.178407916487301,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 12247
    },
    {
      "epoch": 0.12248,
      "grad_norm": 1.0458635406153554,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 12248
    },
    {
      "epoch": 0.12249,
      "grad_norm": 1.1773227629424923,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 12249
    },
    {
      "epoch": 0.1225,
      "grad_norm": 1.1971688988027553,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 12250
    },
    {
      "epoch": 0.12251,
      "grad_norm": 1.130549267275754,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 12251
    },
    {
      "epoch": 0.12252,
      "grad_norm": 0.9347182431246374,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 12252
    },
    {
      "epoch": 0.12253,
      "grad_norm": 1.020618982790968,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 12253
    },
    {
      "epoch": 0.12254,
      "grad_norm": 1.2336898416551805,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 12254
    },
    {
      "epoch": 0.12255,
      "grad_norm": 0.9198700961085998,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 12255
    },
    {
      "epoch": 0.12256,
      "grad_norm": 1.2550574554890428,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 12256
    },
    {
      "epoch": 0.12257,
      "grad_norm": 1.1934871340723554,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 12257
    },
    {
      "epoch": 0.12258,
      "grad_norm": 1.3489157246382726,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 12258
    },
    {
      "epoch": 0.12259,
      "grad_norm": 1.0125660002793162,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 12259
    },
    {
      "epoch": 0.1226,
      "grad_norm": 1.2472895843558622,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 12260
    },
    {
      "epoch": 0.12261,
      "grad_norm": 0.954366178907707,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 12261
    },
    {
      "epoch": 0.12262,
      "grad_norm": 1.1015996587580237,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 12262
    },
    {
      "epoch": 0.12263,
      "grad_norm": 1.1436633592224317,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 12263
    },
    {
      "epoch": 0.12264,
      "grad_norm": 1.1346288912154268,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 12264
    },
    {
      "epoch": 0.12265,
      "grad_norm": 1.0618662450209828,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 12265
    },
    {
      "epoch": 0.12266,
      "grad_norm": 1.178050447666747,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 12266
    },
    {
      "epoch": 0.12267,
      "grad_norm": 1.0553507490994913,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 12267
    },
    {
      "epoch": 0.12268,
      "grad_norm": 1.2296144758772443,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 12268
    },
    {
      "epoch": 0.12269,
      "grad_norm": 1.0457920930485292,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 12269
    },
    {
      "epoch": 0.1227,
      "grad_norm": 1.0840324335435874,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 12270
    },
    {
      "epoch": 0.12271,
      "grad_norm": 1.1445528256408435,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 12271
    },
    {
      "epoch": 0.12272,
      "grad_norm": 1.0789090092167062,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 12272
    },
    {
      "epoch": 0.12273,
      "grad_norm": 1.2904987590364192,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 12273
    },
    {
      "epoch": 0.12274,
      "grad_norm": 1.2998917837299295,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 12274
    },
    {
      "epoch": 0.12275,
      "grad_norm": 0.9954805985642087,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 12275
    },
    {
      "epoch": 0.12276,
      "grad_norm": 1.29013688740999,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 12276
    },
    {
      "epoch": 0.12277,
      "grad_norm": 0.8047671585111673,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 12277
    },
    {
      "epoch": 0.12278,
      "grad_norm": 0.986527265960683,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 12278
    },
    {
      "epoch": 0.12279,
      "grad_norm": 1.2484795259575783,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 12279
    },
    {
      "epoch": 0.1228,
      "grad_norm": 1.221383876281741,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 12280
    },
    {
      "epoch": 0.12281,
      "grad_norm": 1.0856061468785485,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 12281
    },
    {
      "epoch": 0.12282,
      "grad_norm": 1.3831497350750577,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 12282
    },
    {
      "epoch": 0.12283,
      "grad_norm": 1.035678610826925,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 12283
    },
    {
      "epoch": 0.12284,
      "grad_norm": 1.0961280068062669,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 12284
    },
    {
      "epoch": 0.12285,
      "grad_norm": 1.2558527303666362,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 12285
    },
    {
      "epoch": 0.12286,
      "grad_norm": 0.972650104375496,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 12286
    },
    {
      "epoch": 0.12287,
      "grad_norm": 1.0085264354475671,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 12287
    },
    {
      "epoch": 0.12288,
      "grad_norm": 1.2253160289482012,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 12288
    },
    {
      "epoch": 0.12289,
      "grad_norm": 1.2852323689267404,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 12289
    },
    {
      "epoch": 0.1229,
      "grad_norm": 1.3072183218190938,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 12290
    },
    {
      "epoch": 0.12291,
      "grad_norm": 1.0026332220982133,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 12291
    },
    {
      "epoch": 0.12292,
      "grad_norm": 1.1161617368412051,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 12292
    },
    {
      "epoch": 0.12293,
      "grad_norm": 1.217967475014931,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 12293
    },
    {
      "epoch": 0.12294,
      "grad_norm": 1.2381766283955784,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 12294
    },
    {
      "epoch": 0.12295,
      "grad_norm": 0.9408012807630449,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 12295
    },
    {
      "epoch": 0.12296,
      "grad_norm": 1.189999796632258,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 12296
    },
    {
      "epoch": 0.12297,
      "grad_norm": 1.030538072411565,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 12297
    },
    {
      "epoch": 0.12298,
      "grad_norm": 1.253456405281475,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 12298
    },
    {
      "epoch": 0.12299,
      "grad_norm": 0.9763589760714394,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 12299
    },
    {
      "epoch": 0.123,
      "grad_norm": 1.1196444799820449,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 12300
    },
    {
      "epoch": 0.12301,
      "grad_norm": 1.3099114731176258,
      "learning_rate": 0.003,
      "loss": 4.1057,
      "step": 12301
    },
    {
      "epoch": 0.12302,
      "grad_norm": 1.1333606762203758,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 12302
    },
    {
      "epoch": 0.12303,
      "grad_norm": 1.1145907132725608,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 12303
    },
    {
      "epoch": 0.12304,
      "grad_norm": 1.236368919516794,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 12304
    },
    {
      "epoch": 0.12305,
      "grad_norm": 1.1082195413575602,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 12305
    },
    {
      "epoch": 0.12306,
      "grad_norm": 1.0898656695938542,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 12306
    },
    {
      "epoch": 0.12307,
      "grad_norm": 1.408135156441792,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12307
    },
    {
      "epoch": 0.12308,
      "grad_norm": 1.1627256450794383,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 12308
    },
    {
      "epoch": 0.12309,
      "grad_norm": 1.0927700416865638,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 12309
    },
    {
      "epoch": 0.1231,
      "grad_norm": 1.2506427098071364,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 12310
    },
    {
      "epoch": 0.12311,
      "grad_norm": 1.036971362561517,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 12311
    },
    {
      "epoch": 0.12312,
      "grad_norm": 1.151090599954385,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 12312
    },
    {
      "epoch": 0.12313,
      "grad_norm": 1.08749534761496,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 12313
    },
    {
      "epoch": 0.12314,
      "grad_norm": 1.0919178585163267,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 12314
    },
    {
      "epoch": 0.12315,
      "grad_norm": 1.0458368945872953,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 12315
    },
    {
      "epoch": 0.12316,
      "grad_norm": 1.1467188804604123,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 12316
    },
    {
      "epoch": 0.12317,
      "grad_norm": 1.1735741971873328,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 12317
    },
    {
      "epoch": 0.12318,
      "grad_norm": 1.1233946426565826,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 12318
    },
    {
      "epoch": 0.12319,
      "grad_norm": 1.0783419878905267,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 12319
    },
    {
      "epoch": 0.1232,
      "grad_norm": 1.3161800684804994,
      "learning_rate": 0.003,
      "loss": 4.1036,
      "step": 12320
    },
    {
      "epoch": 0.12321,
      "grad_norm": 1.2003452285321743,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 12321
    },
    {
      "epoch": 0.12322,
      "grad_norm": 1.0232241990056046,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 12322
    },
    {
      "epoch": 0.12323,
      "grad_norm": 1.4150191393922857,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 12323
    },
    {
      "epoch": 0.12324,
      "grad_norm": 0.9549941813608207,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 12324
    },
    {
      "epoch": 0.12325,
      "grad_norm": 1.2920608946841499,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 12325
    },
    {
      "epoch": 0.12326,
      "grad_norm": 1.1018558427657896,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 12326
    },
    {
      "epoch": 0.12327,
      "grad_norm": 1.052198948646688,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 12327
    },
    {
      "epoch": 0.12328,
      "grad_norm": 1.340495755595897,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 12328
    },
    {
      "epoch": 0.12329,
      "grad_norm": 1.0170314488453516,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 12329
    },
    {
      "epoch": 0.1233,
      "grad_norm": 1.1788094344761215,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 12330
    },
    {
      "epoch": 0.12331,
      "grad_norm": 1.197821830527221,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 12331
    },
    {
      "epoch": 0.12332,
      "grad_norm": 1.0098326384816398,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 12332
    },
    {
      "epoch": 0.12333,
      "grad_norm": 1.2040309792724453,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 12333
    },
    {
      "epoch": 0.12334,
      "grad_norm": 1.1531517961981639,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 12334
    },
    {
      "epoch": 0.12335,
      "grad_norm": 1.3252120350812502,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 12335
    },
    {
      "epoch": 0.12336,
      "grad_norm": 0.9357893130632275,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 12336
    },
    {
      "epoch": 0.12337,
      "grad_norm": 1.2881720330741127,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 12337
    },
    {
      "epoch": 0.12338,
      "grad_norm": 1.327319438684096,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 12338
    },
    {
      "epoch": 0.12339,
      "grad_norm": 0.7620877981202213,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 12339
    },
    {
      "epoch": 0.1234,
      "grad_norm": 0.9047840231504427,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 12340
    },
    {
      "epoch": 0.12341,
      "grad_norm": 1.366151905719102,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 12341
    },
    {
      "epoch": 0.12342,
      "grad_norm": 1.0551986957803026,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 12342
    },
    {
      "epoch": 0.12343,
      "grad_norm": 1.2070422667122573,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 12343
    },
    {
      "epoch": 0.12344,
      "grad_norm": 1.0631378446641342,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 12344
    },
    {
      "epoch": 0.12345,
      "grad_norm": 1.158407287153162,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 12345
    },
    {
      "epoch": 0.12346,
      "grad_norm": 1.012305861351394,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 12346
    },
    {
      "epoch": 0.12347,
      "grad_norm": 1.1892828019352273,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 12347
    },
    {
      "epoch": 0.12348,
      "grad_norm": 1.3626667884912291,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 12348
    },
    {
      "epoch": 0.12349,
      "grad_norm": 1.1332303731996494,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 12349
    },
    {
      "epoch": 0.1235,
      "grad_norm": 1.0076252552415714,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 12350
    },
    {
      "epoch": 0.12351,
      "grad_norm": 1.3460749776655139,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 12351
    },
    {
      "epoch": 0.12352,
      "grad_norm": 0.8901594605028754,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 12352
    },
    {
      "epoch": 0.12353,
      "grad_norm": 1.143898844720561,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 12353
    },
    {
      "epoch": 0.12354,
      "grad_norm": 1.3439182405422685,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 12354
    },
    {
      "epoch": 0.12355,
      "grad_norm": 1.0678329059929563,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 12355
    },
    {
      "epoch": 0.12356,
      "grad_norm": 1.4136298856727907,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 12356
    },
    {
      "epoch": 0.12357,
      "grad_norm": 1.2423905187146829,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 12357
    },
    {
      "epoch": 0.12358,
      "grad_norm": 0.9767041739713436,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 12358
    },
    {
      "epoch": 0.12359,
      "grad_norm": 1.1113949696387744,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 12359
    },
    {
      "epoch": 0.1236,
      "grad_norm": 1.0883575699397783,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 12360
    },
    {
      "epoch": 0.12361,
      "grad_norm": 1.0359979808099924,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 12361
    },
    {
      "epoch": 0.12362,
      "grad_norm": 1.1557713223728756,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 12362
    },
    {
      "epoch": 0.12363,
      "grad_norm": 1.14932398730516,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 12363
    },
    {
      "epoch": 0.12364,
      "grad_norm": 1.1830974866306603,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 12364
    },
    {
      "epoch": 0.12365,
      "grad_norm": 1.3246802002730393,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 12365
    },
    {
      "epoch": 0.12366,
      "grad_norm": 1.0784211269847974,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 12366
    },
    {
      "epoch": 0.12367,
      "grad_norm": 1.0898334963731273,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 12367
    },
    {
      "epoch": 0.12368,
      "grad_norm": 1.1578889814224254,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 12368
    },
    {
      "epoch": 0.12369,
      "grad_norm": 1.062330069966536,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 12369
    },
    {
      "epoch": 0.1237,
      "grad_norm": 1.2139385466118495,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 12370
    },
    {
      "epoch": 0.12371,
      "grad_norm": 1.0020124725683932,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 12371
    },
    {
      "epoch": 0.12372,
      "grad_norm": 1.1484049368261768,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 12372
    },
    {
      "epoch": 0.12373,
      "grad_norm": 1.1737217200295595,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 12373
    },
    {
      "epoch": 0.12374,
      "grad_norm": 0.9986106363869645,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 12374
    },
    {
      "epoch": 0.12375,
      "grad_norm": 1.3287108256165667,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 12375
    },
    {
      "epoch": 0.12376,
      "grad_norm": 0.9346266004574802,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 12376
    },
    {
      "epoch": 0.12377,
      "grad_norm": 1.3672500381682775,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 12377
    },
    {
      "epoch": 0.12378,
      "grad_norm": 1.1428084365292464,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 12378
    },
    {
      "epoch": 0.12379,
      "grad_norm": 1.488126705290297,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 12379
    },
    {
      "epoch": 0.1238,
      "grad_norm": 0.9572206193227556,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 12380
    },
    {
      "epoch": 0.12381,
      "grad_norm": 0.968112714151001,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 12381
    },
    {
      "epoch": 0.12382,
      "grad_norm": 1.0824430350452237,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 12382
    },
    {
      "epoch": 0.12383,
      "grad_norm": 1.1291656916632755,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 12383
    },
    {
      "epoch": 0.12384,
      "grad_norm": 1.004588532244519,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 12384
    },
    {
      "epoch": 0.12385,
      "grad_norm": 1.1335076678685279,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 12385
    },
    {
      "epoch": 0.12386,
      "grad_norm": 1.016742229821469,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 12386
    },
    {
      "epoch": 0.12387,
      "grad_norm": 1.3524202383088189,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 12387
    },
    {
      "epoch": 0.12388,
      "grad_norm": 1.0308871095461538,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 12388
    },
    {
      "epoch": 0.12389,
      "grad_norm": 1.3224180622611557,
      "learning_rate": 0.003,
      "loss": 4.105,
      "step": 12389
    },
    {
      "epoch": 0.1239,
      "grad_norm": 1.31342304381205,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 12390
    },
    {
      "epoch": 0.12391,
      "grad_norm": 1.1169817951291854,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 12391
    },
    {
      "epoch": 0.12392,
      "grad_norm": 1.2741989835211562,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 12392
    },
    {
      "epoch": 0.12393,
      "grad_norm": 1.108532000671943,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 12393
    },
    {
      "epoch": 0.12394,
      "grad_norm": 1.066676166676712,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 12394
    },
    {
      "epoch": 0.12395,
      "grad_norm": 1.1873521205310513,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 12395
    },
    {
      "epoch": 0.12396,
      "grad_norm": 1.1429627893687861,
      "learning_rate": 0.003,
      "loss": 4.0994,
      "step": 12396
    },
    {
      "epoch": 0.12397,
      "grad_norm": 1.2164184009580055,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 12397
    },
    {
      "epoch": 0.12398,
      "grad_norm": 1.0163836731887739,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 12398
    },
    {
      "epoch": 0.12399,
      "grad_norm": 1.0866617882267402,
      "learning_rate": 0.003,
      "loss": 4.1053,
      "step": 12399
    },
    {
      "epoch": 0.124,
      "grad_norm": 1.2176442736299438,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 12400
    },
    {
      "epoch": 0.12401,
      "grad_norm": 0.9756572130875887,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 12401
    },
    {
      "epoch": 0.12402,
      "grad_norm": 1.0963849133526373,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 12402
    },
    {
      "epoch": 0.12403,
      "grad_norm": 1.007708889867285,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 12403
    },
    {
      "epoch": 0.12404,
      "grad_norm": 1.4254241170094033,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 12404
    },
    {
      "epoch": 0.12405,
      "grad_norm": 1.1163735038983496,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 12405
    },
    {
      "epoch": 0.12406,
      "grad_norm": 1.208191698514626,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 12406
    },
    {
      "epoch": 0.12407,
      "grad_norm": 1.3588304955679549,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 12407
    },
    {
      "epoch": 0.12408,
      "grad_norm": 1.050585779633214,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 12408
    },
    {
      "epoch": 0.12409,
      "grad_norm": 1.1268501042795123,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 12409
    },
    {
      "epoch": 0.1241,
      "grad_norm": 1.2536825066597725,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 12410
    },
    {
      "epoch": 0.12411,
      "grad_norm": 1.031264785916332,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 12411
    },
    {
      "epoch": 0.12412,
      "grad_norm": 1.184343588918058,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 12412
    },
    {
      "epoch": 0.12413,
      "grad_norm": 0.9025710769482064,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 12413
    },
    {
      "epoch": 0.12414,
      "grad_norm": 0.9955728246628766,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 12414
    },
    {
      "epoch": 0.12415,
      "grad_norm": 1.2692295368173465,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 12415
    },
    {
      "epoch": 0.12416,
      "grad_norm": 0.9685638434293641,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 12416
    },
    {
      "epoch": 0.12417,
      "grad_norm": 1.1440919602921278,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 12417
    },
    {
      "epoch": 0.12418,
      "grad_norm": 1.1641712288650334,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 12418
    },
    {
      "epoch": 0.12419,
      "grad_norm": 1.0038079520291594,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 12419
    },
    {
      "epoch": 0.1242,
      "grad_norm": 1.3473679768905762,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 12420
    },
    {
      "epoch": 0.12421,
      "grad_norm": 1.065665013632022,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 12421
    },
    {
      "epoch": 0.12422,
      "grad_norm": 1.3072996720657586,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 12422
    },
    {
      "epoch": 0.12423,
      "grad_norm": 1.1561999326808896,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 12423
    },
    {
      "epoch": 0.12424,
      "grad_norm": 1.1921792539983773,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 12424
    },
    {
      "epoch": 0.12425,
      "grad_norm": 1.1873182303310676,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 12425
    },
    {
      "epoch": 0.12426,
      "grad_norm": 1.0056230801509678,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 12426
    },
    {
      "epoch": 0.12427,
      "grad_norm": 1.1040079016766144,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 12427
    },
    {
      "epoch": 0.12428,
      "grad_norm": 1.2743386374313739,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 12428
    },
    {
      "epoch": 0.12429,
      "grad_norm": 1.0874576902520072,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 12429
    },
    {
      "epoch": 0.1243,
      "grad_norm": 1.123082312308546,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 12430
    },
    {
      "epoch": 0.12431,
      "grad_norm": 1.388091941410794,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 12431
    },
    {
      "epoch": 0.12432,
      "grad_norm": 1.0354864246784283,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 12432
    },
    {
      "epoch": 0.12433,
      "grad_norm": 1.3912030063770227,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 12433
    },
    {
      "epoch": 0.12434,
      "grad_norm": 0.9577813790528974,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 12434
    },
    {
      "epoch": 0.12435,
      "grad_norm": 1.0523873799438321,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 12435
    },
    {
      "epoch": 0.12436,
      "grad_norm": 1.211875416464265,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 12436
    },
    {
      "epoch": 0.12437,
      "grad_norm": 1.0160095230818849,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 12437
    },
    {
      "epoch": 0.12438,
      "grad_norm": 1.1346874971281438,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 12438
    },
    {
      "epoch": 0.12439,
      "grad_norm": 1.062510295796496,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 12439
    },
    {
      "epoch": 0.1244,
      "grad_norm": 1.3567923004962648,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 12440
    },
    {
      "epoch": 0.12441,
      "grad_norm": 1.2016760115144474,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 12441
    },
    {
      "epoch": 0.12442,
      "grad_norm": 0.9627995060103828,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 12442
    },
    {
      "epoch": 0.12443,
      "grad_norm": 1.2294875695669985,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 12443
    },
    {
      "epoch": 0.12444,
      "grad_norm": 1.0422975531725298,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 12444
    },
    {
      "epoch": 0.12445,
      "grad_norm": 1.058240202018281,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 12445
    },
    {
      "epoch": 0.12446,
      "grad_norm": 1.0860192893087985,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 12446
    },
    {
      "epoch": 0.12447,
      "grad_norm": 1.0425529671685176,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 12447
    },
    {
      "epoch": 0.12448,
      "grad_norm": 1.153564456565648,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 12448
    },
    {
      "epoch": 0.12449,
      "grad_norm": 1.1238750683883074,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 12449
    },
    {
      "epoch": 0.1245,
      "grad_norm": 1.1043722679708285,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 12450
    },
    {
      "epoch": 0.12451,
      "grad_norm": 1.0674229419538308,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 12451
    },
    {
      "epoch": 0.12452,
      "grad_norm": 1.1411490342650177,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 12452
    },
    {
      "epoch": 0.12453,
      "grad_norm": 1.0323760242004496,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 12453
    },
    {
      "epoch": 0.12454,
      "grad_norm": 1.3536660491480184,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 12454
    },
    {
      "epoch": 0.12455,
      "grad_norm": 1.1029696954685784,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 12455
    },
    {
      "epoch": 0.12456,
      "grad_norm": 1.6733393686585805,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12456
    },
    {
      "epoch": 0.12457,
      "grad_norm": 1.1447453786927126,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 12457
    },
    {
      "epoch": 0.12458,
      "grad_norm": 0.9986680711717424,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 12458
    },
    {
      "epoch": 0.12459,
      "grad_norm": 1.1317782308250313,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 12459
    },
    {
      "epoch": 0.1246,
      "grad_norm": 1.361354226191513,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 12460
    },
    {
      "epoch": 0.12461,
      "grad_norm": 1.0677093299915452,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 12461
    },
    {
      "epoch": 0.12462,
      "grad_norm": 1.3928548303478745,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 12462
    },
    {
      "epoch": 0.12463,
      "grad_norm": 0.9641189072985846,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 12463
    },
    {
      "epoch": 0.12464,
      "grad_norm": 1.0960604844788246,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 12464
    },
    {
      "epoch": 0.12465,
      "grad_norm": 1.1815436300582207,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 12465
    },
    {
      "epoch": 0.12466,
      "grad_norm": 1.2131619466594814,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 12466
    },
    {
      "epoch": 0.12467,
      "grad_norm": 1.0331125309199667,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 12467
    },
    {
      "epoch": 0.12468,
      "grad_norm": 1.163751673352337,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 12468
    },
    {
      "epoch": 0.12469,
      "grad_norm": 1.0031831435973153,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 12469
    },
    {
      "epoch": 0.1247,
      "grad_norm": 1.0910642628570784,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 12470
    },
    {
      "epoch": 0.12471,
      "grad_norm": 1.0918899476724453,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 12471
    },
    {
      "epoch": 0.12472,
      "grad_norm": 1.1964778879913656,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 12472
    },
    {
      "epoch": 0.12473,
      "grad_norm": 1.0079789042183729,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 12473
    },
    {
      "epoch": 0.12474,
      "grad_norm": 1.386218901058728,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 12474
    },
    {
      "epoch": 0.12475,
      "grad_norm": 0.9267867871365039,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 12475
    },
    {
      "epoch": 0.12476,
      "grad_norm": 1.2327000804525328,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 12476
    },
    {
      "epoch": 0.12477,
      "grad_norm": 1.2382234956781983,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 12477
    },
    {
      "epoch": 0.12478,
      "grad_norm": 1.051207711453641,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 12478
    },
    {
      "epoch": 0.12479,
      "grad_norm": 1.4140646728284696,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 12479
    },
    {
      "epoch": 0.1248,
      "grad_norm": 0.9227957116822071,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 12480
    },
    {
      "epoch": 0.12481,
      "grad_norm": 1.0383738395470508,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 12481
    },
    {
      "epoch": 0.12482,
      "grad_norm": 1.1713914515935469,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 12482
    },
    {
      "epoch": 0.12483,
      "grad_norm": 1.1476901482525905,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 12483
    },
    {
      "epoch": 0.12484,
      "grad_norm": 1.070386462226285,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 12484
    },
    {
      "epoch": 0.12485,
      "grad_norm": 1.3221204377348341,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 12485
    },
    {
      "epoch": 0.12486,
      "grad_norm": 0.9644922374383039,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 12486
    },
    {
      "epoch": 0.12487,
      "grad_norm": 1.3121241168194553,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 12487
    },
    {
      "epoch": 0.12488,
      "grad_norm": 1.3450318680136746,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 12488
    },
    {
      "epoch": 0.12489,
      "grad_norm": 0.9534508800418893,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 12489
    },
    {
      "epoch": 0.1249,
      "grad_norm": 1.1599208964245398,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 12490
    },
    {
      "epoch": 0.12491,
      "grad_norm": 1.147692269571279,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 12491
    },
    {
      "epoch": 0.12492,
      "grad_norm": 1.1523079863182972,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 12492
    },
    {
      "epoch": 0.12493,
      "grad_norm": 1.1634536981359016,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 12493
    },
    {
      "epoch": 0.12494,
      "grad_norm": 0.9874636309717719,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 12494
    },
    {
      "epoch": 0.12495,
      "grad_norm": 1.1353746209396673,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 12495
    },
    {
      "epoch": 0.12496,
      "grad_norm": 1.172186272262417,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 12496
    },
    {
      "epoch": 0.12497,
      "grad_norm": 1.1083322572035998,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 12497
    },
    {
      "epoch": 0.12498,
      "grad_norm": 1.1604690432657063,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 12498
    },
    {
      "epoch": 0.12499,
      "grad_norm": 1.264096108808146,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 12499
    },
    {
      "epoch": 0.125,
      "grad_norm": 1.075946107959381,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 12500
    },
    {
      "epoch": 0.12501,
      "grad_norm": 1.1792882356955536,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 12501
    },
    {
      "epoch": 0.12502,
      "grad_norm": 1.0507188272099386,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 12502
    },
    {
      "epoch": 0.12503,
      "grad_norm": 1.19495514255372,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 12503
    },
    {
      "epoch": 0.12504,
      "grad_norm": 1.1288868298706587,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 12504
    },
    {
      "epoch": 0.12505,
      "grad_norm": 1.249344322822003,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 12505
    },
    {
      "epoch": 0.12506,
      "grad_norm": 1.08839290817881,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 12506
    },
    {
      "epoch": 0.12507,
      "grad_norm": 1.2123320378563451,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 12507
    },
    {
      "epoch": 0.12508,
      "grad_norm": 1.0259058058605786,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 12508
    },
    {
      "epoch": 0.12509,
      "grad_norm": 1.0902172001824963,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 12509
    },
    {
      "epoch": 0.1251,
      "grad_norm": 1.3051601433884132,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 12510
    },
    {
      "epoch": 0.12511,
      "grad_norm": 1.0944379703777736,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 12511
    },
    {
      "epoch": 0.12512,
      "grad_norm": 1.3104568659078275,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 12512
    },
    {
      "epoch": 0.12513,
      "grad_norm": 1.1745329664550175,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 12513
    },
    {
      "epoch": 0.12514,
      "grad_norm": 0.9551689514624544,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 12514
    },
    {
      "epoch": 0.12515,
      "grad_norm": 1.0262550517878668,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 12515
    },
    {
      "epoch": 0.12516,
      "grad_norm": 1.1407767421024977,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 12516
    },
    {
      "epoch": 0.12517,
      "grad_norm": 0.9962063019603865,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 12517
    },
    {
      "epoch": 0.12518,
      "grad_norm": 1.1970608765873851,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 12518
    },
    {
      "epoch": 0.12519,
      "grad_norm": 0.959190386912845,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 12519
    },
    {
      "epoch": 0.1252,
      "grad_norm": 1.1584921160079846,
      "learning_rate": 0.003,
      "loss": 4.0328,
      "step": 12520
    },
    {
      "epoch": 0.12521,
      "grad_norm": 1.4550002952801369,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 12521
    },
    {
      "epoch": 0.12522,
      "grad_norm": 1.1136148399771353,
      "learning_rate": 0.003,
      "loss": 4.1152,
      "step": 12522
    },
    {
      "epoch": 0.12523,
      "grad_norm": 1.0842256316001873,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 12523
    },
    {
      "epoch": 0.12524,
      "grad_norm": 1.134675623921096,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 12524
    },
    {
      "epoch": 0.12525,
      "grad_norm": 1.1593697013463147,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 12525
    },
    {
      "epoch": 0.12526,
      "grad_norm": 1.0788774229439801,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 12526
    },
    {
      "epoch": 0.12527,
      "grad_norm": 1.2292604042409196,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 12527
    },
    {
      "epoch": 0.12528,
      "grad_norm": 1.1657288001713335,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 12528
    },
    {
      "epoch": 0.12529,
      "grad_norm": 1.0254204316259463,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 12529
    },
    {
      "epoch": 0.1253,
      "grad_norm": 1.1168542385335107,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 12530
    },
    {
      "epoch": 0.12531,
      "grad_norm": 1.3174987372246656,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 12531
    },
    {
      "epoch": 0.12532,
      "grad_norm": 1.072377108813044,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 12532
    },
    {
      "epoch": 0.12533,
      "grad_norm": 1.168048971647958,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 12533
    },
    {
      "epoch": 0.12534,
      "grad_norm": 0.9203558918321404,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 12534
    },
    {
      "epoch": 0.12535,
      "grad_norm": 1.0349617881965072,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 12535
    },
    {
      "epoch": 0.12536,
      "grad_norm": 1.2139465679889232,
      "learning_rate": 0.003,
      "loss": 4.0205,
      "step": 12536
    },
    {
      "epoch": 0.12537,
      "grad_norm": 1.0625145068279702,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12537
    },
    {
      "epoch": 0.12538,
      "grad_norm": 1.1817856187660896,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 12538
    },
    {
      "epoch": 0.12539,
      "grad_norm": 1.0838819731098341,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 12539
    },
    {
      "epoch": 0.1254,
      "grad_norm": 1.2713228931051814,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 12540
    },
    {
      "epoch": 0.12541,
      "grad_norm": 0.959739675663071,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 12541
    },
    {
      "epoch": 0.12542,
      "grad_norm": 1.0528085424308269,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 12542
    },
    {
      "epoch": 0.12543,
      "grad_norm": 1.1154825938444732,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 12543
    },
    {
      "epoch": 0.12544,
      "grad_norm": 1.1691545603039764,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 12544
    },
    {
      "epoch": 0.12545,
      "grad_norm": 1.1275427125623834,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 12545
    },
    {
      "epoch": 0.12546,
      "grad_norm": 1.2910133967954116,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 12546
    },
    {
      "epoch": 0.12547,
      "grad_norm": 1.2408064762982007,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 12547
    },
    {
      "epoch": 0.12548,
      "grad_norm": 1.2395442220543138,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 12548
    },
    {
      "epoch": 0.12549,
      "grad_norm": 1.018837692744063,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 12549
    },
    {
      "epoch": 0.1255,
      "grad_norm": 1.2228756800993543,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 12550
    },
    {
      "epoch": 0.12551,
      "grad_norm": 1.0041785080356689,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 12551
    },
    {
      "epoch": 0.12552,
      "grad_norm": 1.4947493162533745,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 12552
    },
    {
      "epoch": 0.12553,
      "grad_norm": 1.1094353181441732,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 12553
    },
    {
      "epoch": 0.12554,
      "grad_norm": 1.387389588948756,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 12554
    },
    {
      "epoch": 0.12555,
      "grad_norm": 0.8422219554497506,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 12555
    },
    {
      "epoch": 0.12556,
      "grad_norm": 1.0837224019829028,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 12556
    },
    {
      "epoch": 0.12557,
      "grad_norm": 1.4412117420958377,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 12557
    },
    {
      "epoch": 0.12558,
      "grad_norm": 0.8155523980416922,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12558
    },
    {
      "epoch": 0.12559,
      "grad_norm": 1.1309528187622053,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 12559
    },
    {
      "epoch": 0.1256,
      "grad_norm": 1.3344558799240693,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 12560
    },
    {
      "epoch": 0.12561,
      "grad_norm": 1.0022644446588989,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 12561
    },
    {
      "epoch": 0.12562,
      "grad_norm": 1.1951526866976123,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 12562
    },
    {
      "epoch": 0.12563,
      "grad_norm": 1.184753483762079,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 12563
    },
    {
      "epoch": 0.12564,
      "grad_norm": 1.0892418939395248,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 12564
    },
    {
      "epoch": 0.12565,
      "grad_norm": 1.0306232708896363,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 12565
    },
    {
      "epoch": 0.12566,
      "grad_norm": 1.059762840096475,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 12566
    },
    {
      "epoch": 0.12567,
      "grad_norm": 1.3303834179457787,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 12567
    },
    {
      "epoch": 0.12568,
      "grad_norm": 1.0285112629694402,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 12568
    },
    {
      "epoch": 0.12569,
      "grad_norm": 1.0888315853747654,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 12569
    },
    {
      "epoch": 0.1257,
      "grad_norm": 1.123442724607301,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 12570
    },
    {
      "epoch": 0.12571,
      "grad_norm": 1.3191518473227668,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 12571
    },
    {
      "epoch": 0.12572,
      "grad_norm": 0.9436836707054405,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 12572
    },
    {
      "epoch": 0.12573,
      "grad_norm": 1.3818106873341798,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 12573
    },
    {
      "epoch": 0.12574,
      "grad_norm": 1.335597219687906,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 12574
    },
    {
      "epoch": 0.12575,
      "grad_norm": 1.0665156220813965,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 12575
    },
    {
      "epoch": 0.12576,
      "grad_norm": 1.182121329164601,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 12576
    },
    {
      "epoch": 0.12577,
      "grad_norm": 1.1385534378456768,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 12577
    },
    {
      "epoch": 0.12578,
      "grad_norm": 1.0775480644504942,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 12578
    },
    {
      "epoch": 0.12579,
      "grad_norm": 1.0227873834379646,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 12579
    },
    {
      "epoch": 0.1258,
      "grad_norm": 1.381570570595422,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 12580
    },
    {
      "epoch": 0.12581,
      "grad_norm": 1.1654272026607977,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 12581
    },
    {
      "epoch": 0.12582,
      "grad_norm": 1.2843945997913482,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 12582
    },
    {
      "epoch": 0.12583,
      "grad_norm": 1.1052825127887618,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 12583
    },
    {
      "epoch": 0.12584,
      "grad_norm": 1.2547493392225169,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 12584
    },
    {
      "epoch": 0.12585,
      "grad_norm": 1.0500640052491619,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 12585
    },
    {
      "epoch": 0.12586,
      "grad_norm": 0.9939437347380568,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 12586
    },
    {
      "epoch": 0.12587,
      "grad_norm": 1.1801032533733091,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 12587
    },
    {
      "epoch": 0.12588,
      "grad_norm": 1.3091622637077749,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 12588
    },
    {
      "epoch": 0.12589,
      "grad_norm": 1.3128166278500608,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 12589
    },
    {
      "epoch": 0.1259,
      "grad_norm": 0.9637061554306544,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 12590
    },
    {
      "epoch": 0.12591,
      "grad_norm": 1.1711703489615166,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 12591
    },
    {
      "epoch": 0.12592,
      "grad_norm": 1.2492792198036187,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 12592
    },
    {
      "epoch": 0.12593,
      "grad_norm": 1.3356054194268845,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 12593
    },
    {
      "epoch": 0.12594,
      "grad_norm": 1.01457475403029,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 12594
    },
    {
      "epoch": 0.12595,
      "grad_norm": 1.2276265939904183,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 12595
    },
    {
      "epoch": 0.12596,
      "grad_norm": 0.9772130361552345,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 12596
    },
    {
      "epoch": 0.12597,
      "grad_norm": 1.224323778576316,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 12597
    },
    {
      "epoch": 0.12598,
      "grad_norm": 1.070538424084639,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 12598
    },
    {
      "epoch": 0.12599,
      "grad_norm": 1.1881950950666054,
      "learning_rate": 0.003,
      "loss": 4.1059,
      "step": 12599
    },
    {
      "epoch": 0.126,
      "grad_norm": 1.2033695005041851,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 12600
    },
    {
      "epoch": 0.12601,
      "grad_norm": 1.1206013859890769,
      "learning_rate": 0.003,
      "loss": 4.1168,
      "step": 12601
    },
    {
      "epoch": 0.12602,
      "grad_norm": 1.2669978901971495,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 12602
    },
    {
      "epoch": 0.12603,
      "grad_norm": 1.0868147281580245,
      "learning_rate": 0.003,
      "loss": 4.0293,
      "step": 12603
    },
    {
      "epoch": 0.12604,
      "grad_norm": 1.189777558723123,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 12604
    },
    {
      "epoch": 0.12605,
      "grad_norm": 1.2022435550486126,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 12605
    },
    {
      "epoch": 0.12606,
      "grad_norm": 1.1932237386510585,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 12606
    },
    {
      "epoch": 0.12607,
      "grad_norm": 1.1021135862889397,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 12607
    },
    {
      "epoch": 0.12608,
      "grad_norm": 1.1753432677944506,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 12608
    },
    {
      "epoch": 0.12609,
      "grad_norm": 1.1185593496277637,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 12609
    },
    {
      "epoch": 0.1261,
      "grad_norm": 1.0519538204701329,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 12610
    },
    {
      "epoch": 0.12611,
      "grad_norm": 1.035043160747932,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 12611
    },
    {
      "epoch": 0.12612,
      "grad_norm": 1.46921137359719,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 12612
    },
    {
      "epoch": 0.12613,
      "grad_norm": 1.0183327242129994,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 12613
    },
    {
      "epoch": 0.12614,
      "grad_norm": 1.1321586476834622,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 12614
    },
    {
      "epoch": 0.12615,
      "grad_norm": 0.958765388845781,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 12615
    },
    {
      "epoch": 0.12616,
      "grad_norm": 1.1385791664107119,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 12616
    },
    {
      "epoch": 0.12617,
      "grad_norm": 1.0230861459562044,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 12617
    },
    {
      "epoch": 0.12618,
      "grad_norm": 1.0869030858149058,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 12618
    },
    {
      "epoch": 0.12619,
      "grad_norm": 1.4998950561962687,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 12619
    },
    {
      "epoch": 0.1262,
      "grad_norm": 1.0522470907633974,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 12620
    },
    {
      "epoch": 0.12621,
      "grad_norm": 1.2828423035677585,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 12621
    },
    {
      "epoch": 0.12622,
      "grad_norm": 1.0900994100221286,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 12622
    },
    {
      "epoch": 0.12623,
      "grad_norm": 1.291938364401869,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 12623
    },
    {
      "epoch": 0.12624,
      "grad_norm": 0.9677638397011658,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 12624
    },
    {
      "epoch": 0.12625,
      "grad_norm": 1.1489998438922944,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 12625
    },
    {
      "epoch": 0.12626,
      "grad_norm": 1.2460095756081284,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 12626
    },
    {
      "epoch": 0.12627,
      "grad_norm": 0.9714846240349928,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 12627
    },
    {
      "epoch": 0.12628,
      "grad_norm": 1.3984461874444771,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 12628
    },
    {
      "epoch": 0.12629,
      "grad_norm": 1.1373515674926453,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 12629
    },
    {
      "epoch": 0.1263,
      "grad_norm": 1.2225090779305283,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 12630
    },
    {
      "epoch": 0.12631,
      "grad_norm": 1.1388744446609427,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 12631
    },
    {
      "epoch": 0.12632,
      "grad_norm": 1.2734851291663085,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 12632
    },
    {
      "epoch": 0.12633,
      "grad_norm": 0.9201830905127847,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 12633
    },
    {
      "epoch": 0.12634,
      "grad_norm": 1.0922115175002909,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 12634
    },
    {
      "epoch": 0.12635,
      "grad_norm": 1.145360232044216,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 12635
    },
    {
      "epoch": 0.12636,
      "grad_norm": 1.1862988393375187,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 12636
    },
    {
      "epoch": 0.12637,
      "grad_norm": 1.0348188987624127,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 12637
    },
    {
      "epoch": 0.12638,
      "grad_norm": 1.1923097559505276,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 12638
    },
    {
      "epoch": 0.12639,
      "grad_norm": 1.0703531305139602,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 12639
    },
    {
      "epoch": 0.1264,
      "grad_norm": 1.2905049113417022,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 12640
    },
    {
      "epoch": 0.12641,
      "grad_norm": 1.1902293742550116,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 12641
    },
    {
      "epoch": 0.12642,
      "grad_norm": 1.2410972077837368,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 12642
    },
    {
      "epoch": 0.12643,
      "grad_norm": 1.1406148977882309,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 12643
    },
    {
      "epoch": 0.12644,
      "grad_norm": 1.06509446797809,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 12644
    },
    {
      "epoch": 0.12645,
      "grad_norm": 1.0842259709739674,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 12645
    },
    {
      "epoch": 0.12646,
      "grad_norm": 1.2586305223074954,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 12646
    },
    {
      "epoch": 0.12647,
      "grad_norm": 1.275877459061793,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 12647
    },
    {
      "epoch": 0.12648,
      "grad_norm": 1.0925411969812069,
      "learning_rate": 0.003,
      "loss": 4.1072,
      "step": 12648
    },
    {
      "epoch": 0.12649,
      "grad_norm": 0.998476164112096,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 12649
    },
    {
      "epoch": 0.1265,
      "grad_norm": 1.0132017061136347,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 12650
    },
    {
      "epoch": 0.12651,
      "grad_norm": 1.1048870775158017,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12651
    },
    {
      "epoch": 0.12652,
      "grad_norm": 1.0423138819603341,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 12652
    },
    {
      "epoch": 0.12653,
      "grad_norm": 1.391018502997299,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 12653
    },
    {
      "epoch": 0.12654,
      "grad_norm": 1.140870115386,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 12654
    },
    {
      "epoch": 0.12655,
      "grad_norm": 1.426313352021284,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 12655
    },
    {
      "epoch": 0.12656,
      "grad_norm": 1.0375477670940312,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 12656
    },
    {
      "epoch": 0.12657,
      "grad_norm": 1.145041205013688,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 12657
    },
    {
      "epoch": 0.12658,
      "grad_norm": 1.0887290250882993,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 12658
    },
    {
      "epoch": 0.12659,
      "grad_norm": 1.1273348516279116,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 12659
    },
    {
      "epoch": 0.1266,
      "grad_norm": 1.2027122471428053,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 12660
    },
    {
      "epoch": 0.12661,
      "grad_norm": 1.0416462226946577,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 12661
    },
    {
      "epoch": 0.12662,
      "grad_norm": 1.2468210679993077,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 12662
    },
    {
      "epoch": 0.12663,
      "grad_norm": 1.0745200887126218,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 12663
    },
    {
      "epoch": 0.12664,
      "grad_norm": 1.1278433882285062,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 12664
    },
    {
      "epoch": 0.12665,
      "grad_norm": 1.0909284791055842,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 12665
    },
    {
      "epoch": 0.12666,
      "grad_norm": 1.133751811039444,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 12666
    },
    {
      "epoch": 0.12667,
      "grad_norm": 1.1802982991607618,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 12667
    },
    {
      "epoch": 0.12668,
      "grad_norm": 1.1448431857062022,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 12668
    },
    {
      "epoch": 0.12669,
      "grad_norm": 1.0836278876953978,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 12669
    },
    {
      "epoch": 0.1267,
      "grad_norm": 1.3095005531198924,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 12670
    },
    {
      "epoch": 0.12671,
      "grad_norm": 1.1863330443749722,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 12671
    },
    {
      "epoch": 0.12672,
      "grad_norm": 1.3414469570535603,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 12672
    },
    {
      "epoch": 0.12673,
      "grad_norm": 0.9841922309060664,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 12673
    },
    {
      "epoch": 0.12674,
      "grad_norm": 1.1120261446531505,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 12674
    },
    {
      "epoch": 0.12675,
      "grad_norm": 1.0960420521883425,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 12675
    },
    {
      "epoch": 0.12676,
      "grad_norm": 1.0953589001224968,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 12676
    },
    {
      "epoch": 0.12677,
      "grad_norm": 1.2852366440837981,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 12677
    },
    {
      "epoch": 0.12678,
      "grad_norm": 1.2366527749878387,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 12678
    },
    {
      "epoch": 0.12679,
      "grad_norm": 1.204052310910075,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 12679
    },
    {
      "epoch": 0.1268,
      "grad_norm": 0.9691879137113483,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 12680
    },
    {
      "epoch": 0.12681,
      "grad_norm": 1.2505083596403377,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 12681
    },
    {
      "epoch": 0.12682,
      "grad_norm": 1.1328418494563997,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 12682
    },
    {
      "epoch": 0.12683,
      "grad_norm": 1.1689224771465725,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 12683
    },
    {
      "epoch": 0.12684,
      "grad_norm": 0.9215889407945064,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 12684
    },
    {
      "epoch": 0.12685,
      "grad_norm": 1.0174059550673156,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 12685
    },
    {
      "epoch": 0.12686,
      "grad_norm": 1.0789107426790325,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 12686
    },
    {
      "epoch": 0.12687,
      "grad_norm": 1.138943276149499,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 12687
    },
    {
      "epoch": 0.12688,
      "grad_norm": 1.3491267848129291,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 12688
    },
    {
      "epoch": 0.12689,
      "grad_norm": 1.3043096443469937,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 12689
    },
    {
      "epoch": 0.1269,
      "grad_norm": 1.041888803682189,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 12690
    },
    {
      "epoch": 0.12691,
      "grad_norm": 1.1928039104322192,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 12691
    },
    {
      "epoch": 0.12692,
      "grad_norm": 0.9677463033613477,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 12692
    },
    {
      "epoch": 0.12693,
      "grad_norm": 1.297628672717164,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 12693
    },
    {
      "epoch": 0.12694,
      "grad_norm": 1.1590402253727634,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 12694
    },
    {
      "epoch": 0.12695,
      "grad_norm": 1.0929268156695968,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 12695
    },
    {
      "epoch": 0.12696,
      "grad_norm": 1.1738269618767545,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 12696
    },
    {
      "epoch": 0.12697,
      "grad_norm": 1.0869019305907497,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 12697
    },
    {
      "epoch": 0.12698,
      "grad_norm": 1.0608902265327735,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 12698
    },
    {
      "epoch": 0.12699,
      "grad_norm": 1.4341001860176978,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 12699
    },
    {
      "epoch": 0.127,
      "grad_norm": 0.9813981853354528,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12700
    },
    {
      "epoch": 0.12701,
      "grad_norm": 1.231682032088662,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 12701
    },
    {
      "epoch": 0.12702,
      "grad_norm": 1.0549722177713285,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 12702
    },
    {
      "epoch": 0.12703,
      "grad_norm": 1.158178022022144,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 12703
    },
    {
      "epoch": 0.12704,
      "grad_norm": 0.9903579204372447,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 12704
    },
    {
      "epoch": 0.12705,
      "grad_norm": 1.2901104187242485,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 12705
    },
    {
      "epoch": 0.12706,
      "grad_norm": 1.2537091170376986,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 12706
    },
    {
      "epoch": 0.12707,
      "grad_norm": 1.234455267795835,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 12707
    },
    {
      "epoch": 0.12708,
      "grad_norm": 1.1892730196421701,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 12708
    },
    {
      "epoch": 0.12709,
      "grad_norm": 1.098330523777582,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 12709
    },
    {
      "epoch": 0.1271,
      "grad_norm": 1.3091469707490966,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12710
    },
    {
      "epoch": 0.12711,
      "grad_norm": 0.8126637106244213,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 12711
    },
    {
      "epoch": 0.12712,
      "grad_norm": 0.9937269329593112,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 12712
    },
    {
      "epoch": 0.12713,
      "grad_norm": 1.4614438638500475,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 12713
    },
    {
      "epoch": 0.12714,
      "grad_norm": 1.0311941615428375,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 12714
    },
    {
      "epoch": 0.12715,
      "grad_norm": 1.368311455512496,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 12715
    },
    {
      "epoch": 0.12716,
      "grad_norm": 1.1194355453078446,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 12716
    },
    {
      "epoch": 0.12717,
      "grad_norm": 1.5343174739356966,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 12717
    },
    {
      "epoch": 0.12718,
      "grad_norm": 0.9189991945381788,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 12718
    },
    {
      "epoch": 0.12719,
      "grad_norm": 0.9429601349139406,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 12719
    },
    {
      "epoch": 0.1272,
      "grad_norm": 1.2671746093809337,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 12720
    },
    {
      "epoch": 0.12721,
      "grad_norm": 1.085669009766264,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 12721
    },
    {
      "epoch": 0.12722,
      "grad_norm": 1.2856395614673346,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 12722
    },
    {
      "epoch": 0.12723,
      "grad_norm": 1.1826019556959075,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 12723
    },
    {
      "epoch": 0.12724,
      "grad_norm": 1.2137350740175774,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 12724
    },
    {
      "epoch": 0.12725,
      "grad_norm": 0.9714405915883685,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 12725
    },
    {
      "epoch": 0.12726,
      "grad_norm": 1.233171570676695,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 12726
    },
    {
      "epoch": 0.12727,
      "grad_norm": 0.9854681450136724,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 12727
    },
    {
      "epoch": 0.12728,
      "grad_norm": 1.0247526093155872,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 12728
    },
    {
      "epoch": 0.12729,
      "grad_norm": 1.2216031150832127,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 12729
    },
    {
      "epoch": 0.1273,
      "grad_norm": 1.0900986705850872,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 12730
    },
    {
      "epoch": 0.12731,
      "grad_norm": 1.3435657482823378,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 12731
    },
    {
      "epoch": 0.12732,
      "grad_norm": 1.0939415152116774,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 12732
    },
    {
      "epoch": 0.12733,
      "grad_norm": 1.1572450146336237,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 12733
    },
    {
      "epoch": 0.12734,
      "grad_norm": 1.2666318970790864,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 12734
    },
    {
      "epoch": 0.12735,
      "grad_norm": 1.0082757442143788,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 12735
    },
    {
      "epoch": 0.12736,
      "grad_norm": 1.255147432277124,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 12736
    },
    {
      "epoch": 0.12737,
      "grad_norm": 1.1279108547109393,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 12737
    },
    {
      "epoch": 0.12738,
      "grad_norm": 1.3623171747528882,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 12738
    },
    {
      "epoch": 0.12739,
      "grad_norm": 1.1804668727447836,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 12739
    },
    {
      "epoch": 0.1274,
      "grad_norm": 1.2903582439030041,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 12740
    },
    {
      "epoch": 0.12741,
      "grad_norm": 1.1903708722838362,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 12741
    },
    {
      "epoch": 0.12742,
      "grad_norm": 0.9970718101508081,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 12742
    },
    {
      "epoch": 0.12743,
      "grad_norm": 1.228943373939233,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 12743
    },
    {
      "epoch": 0.12744,
      "grad_norm": 1.2865166226873788,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 12744
    },
    {
      "epoch": 0.12745,
      "grad_norm": 1.0020118590142613,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 12745
    },
    {
      "epoch": 0.12746,
      "grad_norm": 1.3251412767916488,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 12746
    },
    {
      "epoch": 0.12747,
      "grad_norm": 1.092232438065853,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 12747
    },
    {
      "epoch": 0.12748,
      "grad_norm": 1.214629485605157,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 12748
    },
    {
      "epoch": 0.12749,
      "grad_norm": 1.1856415955145982,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 12749
    },
    {
      "epoch": 0.1275,
      "grad_norm": 1.0913325877513864,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12750
    },
    {
      "epoch": 0.12751,
      "grad_norm": 1.4636934393582455,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 12751
    },
    {
      "epoch": 0.12752,
      "grad_norm": 1.0680594573992184,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 12752
    },
    {
      "epoch": 0.12753,
      "grad_norm": 1.1591841480104785,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 12753
    },
    {
      "epoch": 0.12754,
      "grad_norm": 1.3915949998580592,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 12754
    },
    {
      "epoch": 0.12755,
      "grad_norm": 1.100090177526697,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 12755
    },
    {
      "epoch": 0.12756,
      "grad_norm": 1.326466604650826,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 12756
    },
    {
      "epoch": 0.12757,
      "grad_norm": 0.8779866180952189,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 12757
    },
    {
      "epoch": 0.12758,
      "grad_norm": 0.9123535258907629,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 12758
    },
    {
      "epoch": 0.12759,
      "grad_norm": 1.196160302646205,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 12759
    },
    {
      "epoch": 0.1276,
      "grad_norm": 1.1557465866056045,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 12760
    },
    {
      "epoch": 0.12761,
      "grad_norm": 1.2354914565937516,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 12761
    },
    {
      "epoch": 0.12762,
      "grad_norm": 1.1764145362664384,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 12762
    },
    {
      "epoch": 0.12763,
      "grad_norm": 1.3445066594936244,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 12763
    },
    {
      "epoch": 0.12764,
      "grad_norm": 0.9739010566952191,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 12764
    },
    {
      "epoch": 0.12765,
      "grad_norm": 1.0613937411936154,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 12765
    },
    {
      "epoch": 0.12766,
      "grad_norm": 1.1363450780655793,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 12766
    },
    {
      "epoch": 0.12767,
      "grad_norm": 1.2576325751489537,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 12767
    },
    {
      "epoch": 0.12768,
      "grad_norm": 1.1040972214642275,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 12768
    },
    {
      "epoch": 0.12769,
      "grad_norm": 1.2165327989288754,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 12769
    },
    {
      "epoch": 0.1277,
      "grad_norm": 0.9812695523482138,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 12770
    },
    {
      "epoch": 0.12771,
      "grad_norm": 1.0937863210045036,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 12771
    },
    {
      "epoch": 0.12772,
      "grad_norm": 1.1843615131462144,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 12772
    },
    {
      "epoch": 0.12773,
      "grad_norm": 1.2602559036345033,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 12773
    },
    {
      "epoch": 0.12774,
      "grad_norm": 1.0963906426701393,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 12774
    },
    {
      "epoch": 0.12775,
      "grad_norm": 1.1176614549896196,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 12775
    },
    {
      "epoch": 0.12776,
      "grad_norm": 1.151217603095789,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 12776
    },
    {
      "epoch": 0.12777,
      "grad_norm": 1.238071600314445,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 12777
    },
    {
      "epoch": 0.12778,
      "grad_norm": 1.1705546677287786,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 12778
    },
    {
      "epoch": 0.12779,
      "grad_norm": 1.1871346099841646,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 12779
    },
    {
      "epoch": 0.1278,
      "grad_norm": 1.0061359267648406,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 12780
    },
    {
      "epoch": 0.12781,
      "grad_norm": 1.208483353168107,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 12781
    },
    {
      "epoch": 0.12782,
      "grad_norm": 0.9963653813126722,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 12782
    },
    {
      "epoch": 0.12783,
      "grad_norm": 1.3875762642705003,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 12783
    },
    {
      "epoch": 0.12784,
      "grad_norm": 1.0252567643115786,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 12784
    },
    {
      "epoch": 0.12785,
      "grad_norm": 1.0876953323201521,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 12785
    },
    {
      "epoch": 0.12786,
      "grad_norm": 1.1023544017582871,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 12786
    },
    {
      "epoch": 0.12787,
      "grad_norm": 1.2189812506606332,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 12787
    },
    {
      "epoch": 0.12788,
      "grad_norm": 1.1337557681724755,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 12788
    },
    {
      "epoch": 0.12789,
      "grad_norm": 1.1118245184989517,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 12789
    },
    {
      "epoch": 0.1279,
      "grad_norm": 1.1495692894460081,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 12790
    },
    {
      "epoch": 0.12791,
      "grad_norm": 1.4006130182728584,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 12791
    },
    {
      "epoch": 0.12792,
      "grad_norm": 1.149581584270721,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 12792
    },
    {
      "epoch": 0.12793,
      "grad_norm": 1.2295938062035259,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 12793
    },
    {
      "epoch": 0.12794,
      "grad_norm": 0.9870563587036222,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 12794
    },
    {
      "epoch": 0.12795,
      "grad_norm": 1.3729576291106589,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 12795
    },
    {
      "epoch": 0.12796,
      "grad_norm": 1.097502274613693,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 12796
    },
    {
      "epoch": 0.12797,
      "grad_norm": 1.2164641487977086,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 12797
    },
    {
      "epoch": 0.12798,
      "grad_norm": 0.9578732789955301,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 12798
    },
    {
      "epoch": 0.12799,
      "grad_norm": 1.0836746167399414,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 12799
    },
    {
      "epoch": 0.128,
      "grad_norm": 1.2216085200393374,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 12800
    },
    {
      "epoch": 0.12801,
      "grad_norm": 1.2345579740967534,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12801
    },
    {
      "epoch": 0.12802,
      "grad_norm": 1.2474953417169945,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 12802
    },
    {
      "epoch": 0.12803,
      "grad_norm": 1.0831676668779615,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 12803
    },
    {
      "epoch": 0.12804,
      "grad_norm": 1.193916510472046,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 12804
    },
    {
      "epoch": 0.12805,
      "grad_norm": 1.1035531417793012,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 12805
    },
    {
      "epoch": 0.12806,
      "grad_norm": 1.2292925640284165,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 12806
    },
    {
      "epoch": 0.12807,
      "grad_norm": 0.9439588937890558,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 12807
    },
    {
      "epoch": 0.12808,
      "grad_norm": 1.335736102020312,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 12808
    },
    {
      "epoch": 0.12809,
      "grad_norm": 0.9787030435226739,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12809
    },
    {
      "epoch": 0.1281,
      "grad_norm": 1.3174519343980273,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 12810
    },
    {
      "epoch": 0.12811,
      "grad_norm": 1.2254434525804792,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 12811
    },
    {
      "epoch": 0.12812,
      "grad_norm": 1.210361715843662,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 12812
    },
    {
      "epoch": 0.12813,
      "grad_norm": 1.1971184137498239,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 12813
    },
    {
      "epoch": 0.12814,
      "grad_norm": 1.082703240756199,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 12814
    },
    {
      "epoch": 0.12815,
      "grad_norm": 1.1930835496849195,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 12815
    },
    {
      "epoch": 0.12816,
      "grad_norm": 1.0318640828326053,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 12816
    },
    {
      "epoch": 0.12817,
      "grad_norm": 1.3827741801067235,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 12817
    },
    {
      "epoch": 0.12818,
      "grad_norm": 0.8675507749901307,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 12818
    },
    {
      "epoch": 0.12819,
      "grad_norm": 1.122555691577188,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 12819
    },
    {
      "epoch": 0.1282,
      "grad_norm": 1.4578571399267979,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 12820
    },
    {
      "epoch": 0.12821,
      "grad_norm": 1.2230131071711259,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 12821
    },
    {
      "epoch": 0.12822,
      "grad_norm": 1.1470390517424534,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 12822
    },
    {
      "epoch": 0.12823,
      "grad_norm": 1.21824047768869,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 12823
    },
    {
      "epoch": 0.12824,
      "grad_norm": 1.1595940248211523,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12824
    },
    {
      "epoch": 0.12825,
      "grad_norm": 1.0071679971286354,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 12825
    },
    {
      "epoch": 0.12826,
      "grad_norm": 1.0886480802636633,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 12826
    },
    {
      "epoch": 0.12827,
      "grad_norm": 0.9879075812230554,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 12827
    },
    {
      "epoch": 0.12828,
      "grad_norm": 1.25709114220997,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 12828
    },
    {
      "epoch": 0.12829,
      "grad_norm": 1.1897769499680302,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12829
    },
    {
      "epoch": 0.1283,
      "grad_norm": 1.2009363350007225,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 12830
    },
    {
      "epoch": 0.12831,
      "grad_norm": 1.0437214471375973,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 12831
    },
    {
      "epoch": 0.12832,
      "grad_norm": 1.3698860559323622,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 12832
    },
    {
      "epoch": 0.12833,
      "grad_norm": 1.0702030003794927,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 12833
    },
    {
      "epoch": 0.12834,
      "grad_norm": 1.200090242630135,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 12834
    },
    {
      "epoch": 0.12835,
      "grad_norm": 1.2746786649376214,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 12835
    },
    {
      "epoch": 0.12836,
      "grad_norm": 0.9805497035425383,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 12836
    },
    {
      "epoch": 0.12837,
      "grad_norm": 1.1633036249671689,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 12837
    },
    {
      "epoch": 0.12838,
      "grad_norm": 1.329736020879512,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 12838
    },
    {
      "epoch": 0.12839,
      "grad_norm": 1.1938318223084425,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 12839
    },
    {
      "epoch": 0.1284,
      "grad_norm": 1.120612214095928,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 12840
    },
    {
      "epoch": 0.12841,
      "grad_norm": 1.127307392558783,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 12841
    },
    {
      "epoch": 0.12842,
      "grad_norm": 1.3731566587701174,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 12842
    },
    {
      "epoch": 0.12843,
      "grad_norm": 1.0708552732989607,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 12843
    },
    {
      "epoch": 0.12844,
      "grad_norm": 1.0452607451544866,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 12844
    },
    {
      "epoch": 0.12845,
      "grad_norm": 1.299441269405997,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 12845
    },
    {
      "epoch": 0.12846,
      "grad_norm": 1.0305420303486308,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 12846
    },
    {
      "epoch": 0.12847,
      "grad_norm": 1.157120715208667,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12847
    },
    {
      "epoch": 0.12848,
      "grad_norm": 0.951748874423968,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 12848
    },
    {
      "epoch": 0.12849,
      "grad_norm": 1.2186662066141403,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 12849
    },
    {
      "epoch": 0.1285,
      "grad_norm": 1.1616550894309363,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 12850
    },
    {
      "epoch": 0.12851,
      "grad_norm": 1.3611813068636822,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 12851
    },
    {
      "epoch": 0.12852,
      "grad_norm": 1.0525954515376,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 12852
    },
    {
      "epoch": 0.12853,
      "grad_norm": 1.3130415063478154,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 12853
    },
    {
      "epoch": 0.12854,
      "grad_norm": 0.8970798116397892,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 12854
    },
    {
      "epoch": 0.12855,
      "grad_norm": 1.1852927993778317,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 12855
    },
    {
      "epoch": 0.12856,
      "grad_norm": 1.1667596413715686,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 12856
    },
    {
      "epoch": 0.12857,
      "grad_norm": 1.134386177464642,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 12857
    },
    {
      "epoch": 0.12858,
      "grad_norm": 1.3079045407666592,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 12858
    },
    {
      "epoch": 0.12859,
      "grad_norm": 0.8763922150653155,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 12859
    },
    {
      "epoch": 0.1286,
      "grad_norm": 1.011595626039733,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 12860
    },
    {
      "epoch": 0.12861,
      "grad_norm": 1.4613258416334778,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 12861
    },
    {
      "epoch": 0.12862,
      "grad_norm": 1.1180529142493827,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 12862
    },
    {
      "epoch": 0.12863,
      "grad_norm": 1.0748030892965381,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 12863
    },
    {
      "epoch": 0.12864,
      "grad_norm": 1.2663882454890067,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 12864
    },
    {
      "epoch": 0.12865,
      "grad_norm": 1.0771260362783301,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 12865
    },
    {
      "epoch": 0.12866,
      "grad_norm": 1.207007397316719,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 12866
    },
    {
      "epoch": 0.12867,
      "grad_norm": 1.251880824085332,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 12867
    },
    {
      "epoch": 0.12868,
      "grad_norm": 1.0913493230948164,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 12868
    },
    {
      "epoch": 0.12869,
      "grad_norm": 1.3775891247153929,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 12869
    },
    {
      "epoch": 0.1287,
      "grad_norm": 1.2178575125106768,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 12870
    },
    {
      "epoch": 0.12871,
      "grad_norm": 1.0777722954407951,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 12871
    },
    {
      "epoch": 0.12872,
      "grad_norm": 1.2195294753087516,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 12872
    },
    {
      "epoch": 0.12873,
      "grad_norm": 1.0988131540261645,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 12873
    },
    {
      "epoch": 0.12874,
      "grad_norm": 1.1142189015320123,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 12874
    },
    {
      "epoch": 0.12875,
      "grad_norm": 1.3077766521456173,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 12875
    },
    {
      "epoch": 0.12876,
      "grad_norm": 1.1565760435368453,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 12876
    },
    {
      "epoch": 0.12877,
      "grad_norm": 1.6149809077401527,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 12877
    },
    {
      "epoch": 0.12878,
      "grad_norm": 0.8107627956085056,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 12878
    },
    {
      "epoch": 0.12879,
      "grad_norm": 0.855341741798638,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 12879
    },
    {
      "epoch": 0.1288,
      "grad_norm": 1.0564515371668957,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 12880
    },
    {
      "epoch": 0.12881,
      "grad_norm": 1.17170624905697,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 12881
    },
    {
      "epoch": 0.12882,
      "grad_norm": 1.0267246945471074,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 12882
    },
    {
      "epoch": 0.12883,
      "grad_norm": 1.3556623878864673,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 12883
    },
    {
      "epoch": 0.12884,
      "grad_norm": 1.0475880144786536,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 12884
    },
    {
      "epoch": 0.12885,
      "grad_norm": 1.5063259284449038,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 12885
    },
    {
      "epoch": 0.12886,
      "grad_norm": 0.7899265655167541,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 12886
    },
    {
      "epoch": 0.12887,
      "grad_norm": 0.9058564323079618,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 12887
    },
    {
      "epoch": 0.12888,
      "grad_norm": 1.005475241367468,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 12888
    },
    {
      "epoch": 0.12889,
      "grad_norm": 1.2714003535983294,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 12889
    },
    {
      "epoch": 0.1289,
      "grad_norm": 0.9377231923769617,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 12890
    },
    {
      "epoch": 0.12891,
      "grad_norm": 1.2929609564956943,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 12891
    },
    {
      "epoch": 0.12892,
      "grad_norm": 1.1241978401123458,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 12892
    },
    {
      "epoch": 0.12893,
      "grad_norm": 1.4405716398513726,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 12893
    },
    {
      "epoch": 0.12894,
      "grad_norm": 1.159914398922788,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 12894
    },
    {
      "epoch": 0.12895,
      "grad_norm": 1.1015643555910721,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 12895
    },
    {
      "epoch": 0.12896,
      "grad_norm": 1.2576780878173859,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 12896
    },
    {
      "epoch": 0.12897,
      "grad_norm": 1.0783312786699863,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 12897
    },
    {
      "epoch": 0.12898,
      "grad_norm": 1.076447530213746,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 12898
    },
    {
      "epoch": 0.12899,
      "grad_norm": 1.2113329644664113,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 12899
    },
    {
      "epoch": 0.129,
      "grad_norm": 1.4884997349144662,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 12900
    },
    {
      "epoch": 0.12901,
      "grad_norm": 1.055348011982021,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 12901
    },
    {
      "epoch": 0.12902,
      "grad_norm": 1.2716009511343596,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 12902
    },
    {
      "epoch": 0.12903,
      "grad_norm": 1.310704478906888,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 12903
    },
    {
      "epoch": 0.12904,
      "grad_norm": 1.0034390547908698,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 12904
    },
    {
      "epoch": 0.12905,
      "grad_norm": 1.2912107876545866,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 12905
    },
    {
      "epoch": 0.12906,
      "grad_norm": 0.9220033689146221,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 12906
    },
    {
      "epoch": 0.12907,
      "grad_norm": 1.2545148430744932,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 12907
    },
    {
      "epoch": 0.12908,
      "grad_norm": 1.061885091173034,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 12908
    },
    {
      "epoch": 0.12909,
      "grad_norm": 1.1981459208583543,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 12909
    },
    {
      "epoch": 0.1291,
      "grad_norm": 1.4887508899796054,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 12910
    },
    {
      "epoch": 0.12911,
      "grad_norm": 1.0484698691812218,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 12911
    },
    {
      "epoch": 0.12912,
      "grad_norm": 1.1974356627595546,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 12912
    },
    {
      "epoch": 0.12913,
      "grad_norm": 1.1544594647944466,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 12913
    },
    {
      "epoch": 0.12914,
      "grad_norm": 1.06301186858534,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 12914
    },
    {
      "epoch": 0.12915,
      "grad_norm": 1.103201954615549,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 12915
    },
    {
      "epoch": 0.12916,
      "grad_norm": 1.1225680595769518,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 12916
    },
    {
      "epoch": 0.12917,
      "grad_norm": 1.216988199926607,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 12917
    },
    {
      "epoch": 0.12918,
      "grad_norm": 1.0828569235398053,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 12918
    },
    {
      "epoch": 0.12919,
      "grad_norm": 1.078791826064448,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 12919
    },
    {
      "epoch": 0.1292,
      "grad_norm": 1.1184998086032465,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 12920
    },
    {
      "epoch": 0.12921,
      "grad_norm": 1.2371745617415741,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 12921
    },
    {
      "epoch": 0.12922,
      "grad_norm": 1.231388001618202,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 12922
    },
    {
      "epoch": 0.12923,
      "grad_norm": 0.9319342231989519,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 12923
    },
    {
      "epoch": 0.12924,
      "grad_norm": 1.0626351285022981,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 12924
    },
    {
      "epoch": 0.12925,
      "grad_norm": 1.294675870585951,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 12925
    },
    {
      "epoch": 0.12926,
      "grad_norm": 1.0447166513830999,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 12926
    },
    {
      "epoch": 0.12927,
      "grad_norm": 1.3468163322260938,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 12927
    },
    {
      "epoch": 0.12928,
      "grad_norm": 1.1187109002024622,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 12928
    },
    {
      "epoch": 0.12929,
      "grad_norm": 1.341429626291632,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 12929
    },
    {
      "epoch": 0.1293,
      "grad_norm": 1.1945670634171135,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 12930
    },
    {
      "epoch": 0.12931,
      "grad_norm": 1.2265556660339738,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 12931
    },
    {
      "epoch": 0.12932,
      "grad_norm": 1.0543481447222696,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 12932
    },
    {
      "epoch": 0.12933,
      "grad_norm": 1.2250952993934159,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 12933
    },
    {
      "epoch": 0.12934,
      "grad_norm": 1.0076811874522589,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 12934
    },
    {
      "epoch": 0.12935,
      "grad_norm": 1.212985332533582,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 12935
    },
    {
      "epoch": 0.12936,
      "grad_norm": 0.9350956920809937,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 12936
    },
    {
      "epoch": 0.12937,
      "grad_norm": 0.9610230910043541,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 12937
    },
    {
      "epoch": 0.12938,
      "grad_norm": 1.233023364290806,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 12938
    },
    {
      "epoch": 0.12939,
      "grad_norm": 1.2941221452477212,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 12939
    },
    {
      "epoch": 0.1294,
      "grad_norm": 1.011201607814211,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 12940
    },
    {
      "epoch": 0.12941,
      "grad_norm": 1.151927786045976,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 12941
    },
    {
      "epoch": 0.12942,
      "grad_norm": 1.0484359269577173,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 12942
    },
    {
      "epoch": 0.12943,
      "grad_norm": 1.084312206261362,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 12943
    },
    {
      "epoch": 0.12944,
      "grad_norm": 1.2728509429992616,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 12944
    },
    {
      "epoch": 0.12945,
      "grad_norm": 1.0628719145322947,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12945
    },
    {
      "epoch": 0.12946,
      "grad_norm": 1.1191828860796316,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 12946
    },
    {
      "epoch": 0.12947,
      "grad_norm": 1.104488359673685,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 12947
    },
    {
      "epoch": 0.12948,
      "grad_norm": 1.054862388026599,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 12948
    },
    {
      "epoch": 0.12949,
      "grad_norm": 1.1190613173273303,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 12949
    },
    {
      "epoch": 0.1295,
      "grad_norm": 1.2062057801082409,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 12950
    },
    {
      "epoch": 0.12951,
      "grad_norm": 1.2965595199159816,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 12951
    },
    {
      "epoch": 0.12952,
      "grad_norm": 1.3515064007048385,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 12952
    },
    {
      "epoch": 0.12953,
      "grad_norm": 1.1542883772874049,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 12953
    },
    {
      "epoch": 0.12954,
      "grad_norm": 1.1746570847636975,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 12954
    },
    {
      "epoch": 0.12955,
      "grad_norm": 1.2510953695065115,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 12955
    },
    {
      "epoch": 0.12956,
      "grad_norm": 1.06297432928455,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 12956
    },
    {
      "epoch": 0.12957,
      "grad_norm": 1.1473066198430277,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 12957
    },
    {
      "epoch": 0.12958,
      "grad_norm": 1.0201529204218371,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 12958
    },
    {
      "epoch": 0.12959,
      "grad_norm": 1.0898537437420681,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 12959
    },
    {
      "epoch": 0.1296,
      "grad_norm": 1.1055293846960481,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 12960
    },
    {
      "epoch": 0.12961,
      "grad_norm": 1.2351085707954883,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 12961
    },
    {
      "epoch": 0.12962,
      "grad_norm": 1.1854472228284336,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 12962
    },
    {
      "epoch": 0.12963,
      "grad_norm": 1.227653172728379,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 12963
    },
    {
      "epoch": 0.12964,
      "grad_norm": 1.2789582584621662,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 12964
    },
    {
      "epoch": 0.12965,
      "grad_norm": 1.2087528078557306,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 12965
    },
    {
      "epoch": 0.12966,
      "grad_norm": 1.6112942427002772,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 12966
    },
    {
      "epoch": 0.12967,
      "grad_norm": 0.8976210470667042,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 12967
    },
    {
      "epoch": 0.12968,
      "grad_norm": 0.8608094199061171,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 12968
    },
    {
      "epoch": 0.12969,
      "grad_norm": 1.1370110839187368,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 12969
    },
    {
      "epoch": 0.1297,
      "grad_norm": 1.4481956638153832,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 12970
    },
    {
      "epoch": 0.12971,
      "grad_norm": 1.2543473501113689,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 12971
    },
    {
      "epoch": 0.12972,
      "grad_norm": 1.0145533995976639,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 12972
    },
    {
      "epoch": 0.12973,
      "grad_norm": 1.1360402979394963,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 12973
    },
    {
      "epoch": 0.12974,
      "grad_norm": 1.1046758374769596,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 12974
    },
    {
      "epoch": 0.12975,
      "grad_norm": 1.2369109496401294,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 12975
    },
    {
      "epoch": 0.12976,
      "grad_norm": 0.9704420219201002,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 12976
    },
    {
      "epoch": 0.12977,
      "grad_norm": 1.1921286204207069,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 12977
    },
    {
      "epoch": 0.12978,
      "grad_norm": 0.8809197230731804,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 12978
    },
    {
      "epoch": 0.12979,
      "grad_norm": 1.1979650800023425,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 12979
    },
    {
      "epoch": 0.1298,
      "grad_norm": 1.1362702093641754,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 12980
    },
    {
      "epoch": 0.12981,
      "grad_norm": 1.2225938830516137,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 12981
    },
    {
      "epoch": 0.12982,
      "grad_norm": 1.3885255723407812,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 12982
    },
    {
      "epoch": 0.12983,
      "grad_norm": 1.172983719413079,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 12983
    },
    {
      "epoch": 0.12984,
      "grad_norm": 1.1540178894424364,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 12984
    },
    {
      "epoch": 0.12985,
      "grad_norm": 1.0860372470568143,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 12985
    },
    {
      "epoch": 0.12986,
      "grad_norm": 1.1949009167059308,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 12986
    },
    {
      "epoch": 0.12987,
      "grad_norm": 1.1886249054151257,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 12987
    },
    {
      "epoch": 0.12988,
      "grad_norm": 1.0545386379249677,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 12988
    },
    {
      "epoch": 0.12989,
      "grad_norm": 1.4904358019801487,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 12989
    },
    {
      "epoch": 0.1299,
      "grad_norm": 0.9820795206052854,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 12990
    },
    {
      "epoch": 0.12991,
      "grad_norm": 1.0013014198246373,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 12991
    },
    {
      "epoch": 0.12992,
      "grad_norm": 1.0991845081239433,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 12992
    },
    {
      "epoch": 0.12993,
      "grad_norm": 1.0433208295006966,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 12993
    },
    {
      "epoch": 0.12994,
      "grad_norm": 1.3200638135840004,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 12994
    },
    {
      "epoch": 0.12995,
      "grad_norm": 0.9502599445554497,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 12995
    },
    {
      "epoch": 0.12996,
      "grad_norm": 1.2478926843509446,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 12996
    },
    {
      "epoch": 0.12997,
      "grad_norm": 1.1228555540957137,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 12997
    },
    {
      "epoch": 0.12998,
      "grad_norm": 1.2809069083667275,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 12998
    },
    {
      "epoch": 0.12999,
      "grad_norm": 1.0503689952353794,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 12999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3115533907016008,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 13000
    },
    {
      "epoch": 0.13001,
      "grad_norm": 1.5532057915451387,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 13001
    },
    {
      "epoch": 0.13002,
      "grad_norm": 1.0069991586277387,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 13002
    },
    {
      "epoch": 0.13003,
      "grad_norm": 1.113897741771145,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 13003
    },
    {
      "epoch": 0.13004,
      "grad_norm": 1.3576791887916393,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 13004
    },
    {
      "epoch": 0.13005,
      "grad_norm": 1.1468295753829618,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 13005
    },
    {
      "epoch": 0.13006,
      "grad_norm": 1.1026555314244328,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 13006
    },
    {
      "epoch": 0.13007,
      "grad_norm": 1.2521757772939597,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 13007
    },
    {
      "epoch": 0.13008,
      "grad_norm": 1.1901151612620091,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 13008
    },
    {
      "epoch": 0.13009,
      "grad_norm": 1.075477985606486,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 13009
    },
    {
      "epoch": 0.1301,
      "grad_norm": 1.2593027138009543,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 13010
    },
    {
      "epoch": 0.13011,
      "grad_norm": 1.289946689484295,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 13011
    },
    {
      "epoch": 0.13012,
      "grad_norm": 1.1376906437862906,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 13012
    },
    {
      "epoch": 0.13013,
      "grad_norm": 1.1651300549158004,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 13013
    },
    {
      "epoch": 0.13014,
      "grad_norm": 0.9765529309917151,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 13014
    },
    {
      "epoch": 0.13015,
      "grad_norm": 1.1432089220373196,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 13015
    },
    {
      "epoch": 0.13016,
      "grad_norm": 1.2872956102777362,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 13016
    },
    {
      "epoch": 0.13017,
      "grad_norm": 1.032176218481528,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 13017
    },
    {
      "epoch": 0.13018,
      "grad_norm": 1.3836158601570603,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 13018
    },
    {
      "epoch": 0.13019,
      "grad_norm": 1.2484311132058388,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 13019
    },
    {
      "epoch": 0.1302,
      "grad_norm": 1.133818858935883,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 13020
    },
    {
      "epoch": 0.13021,
      "grad_norm": 1.176538778672168,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 13021
    },
    {
      "epoch": 0.13022,
      "grad_norm": 1.13476736669025,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 13022
    },
    {
      "epoch": 0.13023,
      "grad_norm": 1.2191650827911746,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 13023
    },
    {
      "epoch": 0.13024,
      "grad_norm": 0.9137587500950195,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 13024
    },
    {
      "epoch": 0.13025,
      "grad_norm": 1.056799876749151,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 13025
    },
    {
      "epoch": 0.13026,
      "grad_norm": 1.32997324639056,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 13026
    },
    {
      "epoch": 0.13027,
      "grad_norm": 1.0335813697420737,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 13027
    },
    {
      "epoch": 0.13028,
      "grad_norm": 1.2129057627304443,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 13028
    },
    {
      "epoch": 0.13029,
      "grad_norm": 1.0463769845927975,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 13029
    },
    {
      "epoch": 0.1303,
      "grad_norm": 0.9371220497761807,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 13030
    },
    {
      "epoch": 0.13031,
      "grad_norm": 1.0876623507048622,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 13031
    },
    {
      "epoch": 0.13032,
      "grad_norm": 1.220571969379598,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 13032
    },
    {
      "epoch": 0.13033,
      "grad_norm": 1.283113920569638,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 13033
    },
    {
      "epoch": 0.13034,
      "grad_norm": 1.2716846620882998,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 13034
    },
    {
      "epoch": 0.13035,
      "grad_norm": 0.9581416755949399,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 13035
    },
    {
      "epoch": 0.13036,
      "grad_norm": 1.3266757217606555,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13036
    },
    {
      "epoch": 0.13037,
      "grad_norm": 1.0374416485559421,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 13037
    },
    {
      "epoch": 0.13038,
      "grad_norm": 1.1415059688805005,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 13038
    },
    {
      "epoch": 0.13039,
      "grad_norm": 1.0862234750529718,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 13039
    },
    {
      "epoch": 0.1304,
      "grad_norm": 1.3078031355729784,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 13040
    },
    {
      "epoch": 0.13041,
      "grad_norm": 1.2283871938375608,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 13041
    },
    {
      "epoch": 0.13042,
      "grad_norm": 1.1038751327724663,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 13042
    },
    {
      "epoch": 0.13043,
      "grad_norm": 1.1386760904372004,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 13043
    },
    {
      "epoch": 0.13044,
      "grad_norm": 1.223682986720779,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 13044
    },
    {
      "epoch": 0.13045,
      "grad_norm": 1.1310262796396329,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 13045
    },
    {
      "epoch": 0.13046,
      "grad_norm": 1.1694310039446454,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 13046
    },
    {
      "epoch": 0.13047,
      "grad_norm": 1.2884607512361166,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13047
    },
    {
      "epoch": 0.13048,
      "grad_norm": 1.0162810931708417,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 13048
    },
    {
      "epoch": 0.13049,
      "grad_norm": 1.1075577086696535,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 13049
    },
    {
      "epoch": 0.1305,
      "grad_norm": 0.9801540277850392,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 13050
    },
    {
      "epoch": 0.13051,
      "grad_norm": 1.0137626493739156,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 13051
    },
    {
      "epoch": 0.13052,
      "grad_norm": 1.347414761849935,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 13052
    },
    {
      "epoch": 0.13053,
      "grad_norm": 1.0757138709212104,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 13053
    },
    {
      "epoch": 0.13054,
      "grad_norm": 1.1405350325370918,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 13054
    },
    {
      "epoch": 0.13055,
      "grad_norm": 1.0298726527013953,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 13055
    },
    {
      "epoch": 0.13056,
      "grad_norm": 1.4076665701517697,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 13056
    },
    {
      "epoch": 0.13057,
      "grad_norm": 0.9379627386101226,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 13057
    },
    {
      "epoch": 0.13058,
      "grad_norm": 1.3352121990417045,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 13058
    },
    {
      "epoch": 0.13059,
      "grad_norm": 1.1498020719130333,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 13059
    },
    {
      "epoch": 0.1306,
      "grad_norm": 1.4026732415806318,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 13060
    },
    {
      "epoch": 0.13061,
      "grad_norm": 0.9904325540933382,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 13061
    },
    {
      "epoch": 0.13062,
      "grad_norm": 1.2017031300002665,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 13062
    },
    {
      "epoch": 0.13063,
      "grad_norm": 1.0898718183097498,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 13063
    },
    {
      "epoch": 0.13064,
      "grad_norm": 1.1900076841116998,
      "learning_rate": 0.003,
      "loss": 4.1092,
      "step": 13064
    },
    {
      "epoch": 0.13065,
      "grad_norm": 1.2327062041730883,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13065
    },
    {
      "epoch": 0.13066,
      "grad_norm": 1.1035502075789365,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 13066
    },
    {
      "epoch": 0.13067,
      "grad_norm": 0.9854179039669931,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 13067
    },
    {
      "epoch": 0.13068,
      "grad_norm": 1.1314639070759815,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 13068
    },
    {
      "epoch": 0.13069,
      "grad_norm": 1.0365667525729436,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 13069
    },
    {
      "epoch": 0.1307,
      "grad_norm": 1.1475215298763473,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 13070
    },
    {
      "epoch": 0.13071,
      "grad_norm": 1.0185681100892678,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 13071
    },
    {
      "epoch": 0.13072,
      "grad_norm": 1.220724504459399,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 13072
    },
    {
      "epoch": 0.13073,
      "grad_norm": 1.2797075993239362,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 13073
    },
    {
      "epoch": 0.13074,
      "grad_norm": 1.028420780164218,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 13074
    },
    {
      "epoch": 0.13075,
      "grad_norm": 1.4298434252273828,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 13075
    },
    {
      "epoch": 0.13076,
      "grad_norm": 0.9944607543403813,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 13076
    },
    {
      "epoch": 0.13077,
      "grad_norm": 1.3787263022153806,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 13077
    },
    {
      "epoch": 0.13078,
      "grad_norm": 1.1237099249797804,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 13078
    },
    {
      "epoch": 0.13079,
      "grad_norm": 1.0764073357883925,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 13079
    },
    {
      "epoch": 0.1308,
      "grad_norm": 1.4621915905997236,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 13080
    },
    {
      "epoch": 0.13081,
      "grad_norm": 1.2041644590549643,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 13081
    },
    {
      "epoch": 0.13082,
      "grad_norm": 1.2039357142163236,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 13082
    },
    {
      "epoch": 0.13083,
      "grad_norm": 1.2885711119752967,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 13083
    },
    {
      "epoch": 0.13084,
      "grad_norm": 1.1357064756724613,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 13084
    },
    {
      "epoch": 0.13085,
      "grad_norm": 1.14261875842365,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 13085
    },
    {
      "epoch": 0.13086,
      "grad_norm": 1.1684335033953381,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 13086
    },
    {
      "epoch": 0.13087,
      "grad_norm": 1.1087092618521905,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 13087
    },
    {
      "epoch": 0.13088,
      "grad_norm": 1.2939717696394601,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 13088
    },
    {
      "epoch": 0.13089,
      "grad_norm": 1.0429463650533524,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 13089
    },
    {
      "epoch": 0.1309,
      "grad_norm": 1.3643289914345351,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 13090
    },
    {
      "epoch": 0.13091,
      "grad_norm": 1.1016290315760682,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 13091
    },
    {
      "epoch": 0.13092,
      "grad_norm": 1.2184391063916458,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 13092
    },
    {
      "epoch": 0.13093,
      "grad_norm": 1.1907130843556937,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 13093
    },
    {
      "epoch": 0.13094,
      "grad_norm": 1.0836391576772713,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 13094
    },
    {
      "epoch": 0.13095,
      "grad_norm": 1.3836963149703367,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 13095
    },
    {
      "epoch": 0.13096,
      "grad_norm": 1.1130676307420784,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 13096
    },
    {
      "epoch": 0.13097,
      "grad_norm": 1.1647929026892412,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 13097
    },
    {
      "epoch": 0.13098,
      "grad_norm": 1.2042645357212667,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 13098
    },
    {
      "epoch": 0.13099,
      "grad_norm": 1.2712696162753325,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 13099
    },
    {
      "epoch": 0.131,
      "grad_norm": 0.9122188658437458,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 13100
    },
    {
      "epoch": 0.13101,
      "grad_norm": 1.199374985405497,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 13101
    },
    {
      "epoch": 0.13102,
      "grad_norm": 0.9794571983598035,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 13102
    },
    {
      "epoch": 0.13103,
      "grad_norm": 1.3212059710839936,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 13103
    },
    {
      "epoch": 0.13104,
      "grad_norm": 0.969708064207809,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 13104
    },
    {
      "epoch": 0.13105,
      "grad_norm": 1.4038448146151083,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13105
    },
    {
      "epoch": 0.13106,
      "grad_norm": 1.3397166893368102,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 13106
    },
    {
      "epoch": 0.13107,
      "grad_norm": 1.1263536237964167,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 13107
    },
    {
      "epoch": 0.13108,
      "grad_norm": 1.1421597084589206,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 13108
    },
    {
      "epoch": 0.13109,
      "grad_norm": 1.0402132670011577,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 13109
    },
    {
      "epoch": 0.1311,
      "grad_norm": 1.103323397910269,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 13110
    },
    {
      "epoch": 0.13111,
      "grad_norm": 1.1138973357958888,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 13111
    },
    {
      "epoch": 0.13112,
      "grad_norm": 1.2533795396144758,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 13112
    },
    {
      "epoch": 0.13113,
      "grad_norm": 1.2492591473176269,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 13113
    },
    {
      "epoch": 0.13114,
      "grad_norm": 1.2822561331818683,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 13114
    },
    {
      "epoch": 0.13115,
      "grad_norm": 0.9692542170042184,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 13115
    },
    {
      "epoch": 0.13116,
      "grad_norm": 1.252201337846062,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 13116
    },
    {
      "epoch": 0.13117,
      "grad_norm": 1.124907025503089,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 13117
    },
    {
      "epoch": 0.13118,
      "grad_norm": 1.140182893728732,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 13118
    },
    {
      "epoch": 0.13119,
      "grad_norm": 1.1603913986093715,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 13119
    },
    {
      "epoch": 0.1312,
      "grad_norm": 1.3401520585930122,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 13120
    },
    {
      "epoch": 0.13121,
      "grad_norm": 1.045707895208328,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 13121
    },
    {
      "epoch": 0.13122,
      "grad_norm": 1.3487202343862288,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 13122
    },
    {
      "epoch": 0.13123,
      "grad_norm": 1.2569422638761363,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 13123
    },
    {
      "epoch": 0.13124,
      "grad_norm": 1.1886828284004596,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 13124
    },
    {
      "epoch": 0.13125,
      "grad_norm": 1.0547772494125238,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 13125
    },
    {
      "epoch": 0.13126,
      "grad_norm": 1.0628137619252966,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 13126
    },
    {
      "epoch": 0.13127,
      "grad_norm": 1.3267722281858412,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 13127
    },
    {
      "epoch": 0.13128,
      "grad_norm": 0.9974671959993655,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 13128
    },
    {
      "epoch": 0.13129,
      "grad_norm": 1.2037968613620844,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 13129
    },
    {
      "epoch": 0.1313,
      "grad_norm": 1.0272222870037793,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 13130
    },
    {
      "epoch": 0.13131,
      "grad_norm": 1.4273829392694313,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 13131
    },
    {
      "epoch": 0.13132,
      "grad_norm": 0.9129965729121428,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 13132
    },
    {
      "epoch": 0.13133,
      "grad_norm": 1.4042378210051416,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 13133
    },
    {
      "epoch": 0.13134,
      "grad_norm": 1.2315959320398058,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 13134
    },
    {
      "epoch": 0.13135,
      "grad_norm": 1.145000028319332,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 13135
    },
    {
      "epoch": 0.13136,
      "grad_norm": 1.3626312509951313,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 13136
    },
    {
      "epoch": 0.13137,
      "grad_norm": 0.9208041507990571,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 13137
    },
    {
      "epoch": 0.13138,
      "grad_norm": 0.9949139138439406,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 13138
    },
    {
      "epoch": 0.13139,
      "grad_norm": 1.1330070542618937,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 13139
    },
    {
      "epoch": 0.1314,
      "grad_norm": 1.0277347495649414,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 13140
    },
    {
      "epoch": 0.13141,
      "grad_norm": 1.2157307491872351,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 13141
    },
    {
      "epoch": 0.13142,
      "grad_norm": 1.1540030738824,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 13142
    },
    {
      "epoch": 0.13143,
      "grad_norm": 1.083663671224984,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 13143
    },
    {
      "epoch": 0.13144,
      "grad_norm": 1.1693101820813463,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 13144
    },
    {
      "epoch": 0.13145,
      "grad_norm": 1.0520615237330742,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 13145
    },
    {
      "epoch": 0.13146,
      "grad_norm": 1.3595372029185335,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 13146
    },
    {
      "epoch": 0.13147,
      "grad_norm": 1.2011223687753416,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 13147
    },
    {
      "epoch": 0.13148,
      "grad_norm": 1.3437084532597734,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 13148
    },
    {
      "epoch": 0.13149,
      "grad_norm": 1.0555719973390898,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 13149
    },
    {
      "epoch": 0.1315,
      "grad_norm": 1.10074775022198,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 13150
    },
    {
      "epoch": 0.13151,
      "grad_norm": 1.26607507230037,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 13151
    },
    {
      "epoch": 0.13152,
      "grad_norm": 1.107632811054954,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 13152
    },
    {
      "epoch": 0.13153,
      "grad_norm": 1.1120740461445913,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 13153
    },
    {
      "epoch": 0.13154,
      "grad_norm": 1.0197773346455092,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 13154
    },
    {
      "epoch": 0.13155,
      "grad_norm": 1.3563669767266036,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 13155
    },
    {
      "epoch": 0.13156,
      "grad_norm": 1.2096978560595582,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 13156
    },
    {
      "epoch": 0.13157,
      "grad_norm": 1.201207543586879,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 13157
    },
    {
      "epoch": 0.13158,
      "grad_norm": 1.2986901429844693,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 13158
    },
    {
      "epoch": 0.13159,
      "grad_norm": 0.9337125919864209,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 13159
    },
    {
      "epoch": 0.1316,
      "grad_norm": 1.093091216961512,
      "learning_rate": 0.003,
      "loss": 4.0313,
      "step": 13160
    },
    {
      "epoch": 0.13161,
      "grad_norm": 1.2966463755097752,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 13161
    },
    {
      "epoch": 0.13162,
      "grad_norm": 1.0873521204134697,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 13162
    },
    {
      "epoch": 0.13163,
      "grad_norm": 1.1664387280121102,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 13163
    },
    {
      "epoch": 0.13164,
      "grad_norm": 1.0579034446025735,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 13164
    },
    {
      "epoch": 0.13165,
      "grad_norm": 1.2343600101983327,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 13165
    },
    {
      "epoch": 0.13166,
      "grad_norm": 1.30867519231371,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 13166
    },
    {
      "epoch": 0.13167,
      "grad_norm": 1.4435775289061636,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 13167
    },
    {
      "epoch": 0.13168,
      "grad_norm": 1.2152487189165901,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 13168
    },
    {
      "epoch": 0.13169,
      "grad_norm": 0.9907585100666397,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 13169
    },
    {
      "epoch": 0.1317,
      "grad_norm": 1.166126698203818,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 13170
    },
    {
      "epoch": 0.13171,
      "grad_norm": 1.1213359245000933,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 13171
    },
    {
      "epoch": 0.13172,
      "grad_norm": 1.3021235886651639,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 13172
    },
    {
      "epoch": 0.13173,
      "grad_norm": 1.0134856626832358,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 13173
    },
    {
      "epoch": 0.13174,
      "grad_norm": 1.2731135211937619,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 13174
    },
    {
      "epoch": 0.13175,
      "grad_norm": 0.9865560884011636,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 13175
    },
    {
      "epoch": 0.13176,
      "grad_norm": 1.4327052210028386,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 13176
    },
    {
      "epoch": 0.13177,
      "grad_norm": 1.148008225907461,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 13177
    },
    {
      "epoch": 0.13178,
      "grad_norm": 1.218790609154238,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 13178
    },
    {
      "epoch": 0.13179,
      "grad_norm": 1.1822536749554937,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 13179
    },
    {
      "epoch": 0.1318,
      "grad_norm": 0.9070379499444476,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 13180
    },
    {
      "epoch": 0.13181,
      "grad_norm": 0.9945075791236695,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 13181
    },
    {
      "epoch": 0.13182,
      "grad_norm": 1.240539209282678,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 13182
    },
    {
      "epoch": 0.13183,
      "grad_norm": 1.1850668198222947,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 13183
    },
    {
      "epoch": 0.13184,
      "grad_norm": 1.1527421237373985,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 13184
    },
    {
      "epoch": 0.13185,
      "grad_norm": 1.0952135964750374,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 13185
    },
    {
      "epoch": 0.13186,
      "grad_norm": 1.2506033733384114,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 13186
    },
    {
      "epoch": 0.13187,
      "grad_norm": 0.8635762234024932,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 13187
    },
    {
      "epoch": 0.13188,
      "grad_norm": 0.9892073305520283,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 13188
    },
    {
      "epoch": 0.13189,
      "grad_norm": 1.2420881455372659,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 13189
    },
    {
      "epoch": 0.1319,
      "grad_norm": 1.017536179497929,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 13190
    },
    {
      "epoch": 0.13191,
      "grad_norm": 1.1841956984720243,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 13191
    },
    {
      "epoch": 0.13192,
      "grad_norm": 1.2260648938645817,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 13192
    },
    {
      "epoch": 0.13193,
      "grad_norm": 1.1065175186224798,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 13193
    },
    {
      "epoch": 0.13194,
      "grad_norm": 1.4416607096483425,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 13194
    },
    {
      "epoch": 0.13195,
      "grad_norm": 1.316389643719257,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 13195
    },
    {
      "epoch": 0.13196,
      "grad_norm": 1.1352159192036904,
      "learning_rate": 0.003,
      "loss": 4.0298,
      "step": 13196
    },
    {
      "epoch": 0.13197,
      "grad_norm": 1.1568917969318886,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 13197
    },
    {
      "epoch": 0.13198,
      "grad_norm": 1.1256710643386678,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 13198
    },
    {
      "epoch": 0.13199,
      "grad_norm": 1.2156018543905163,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 13199
    },
    {
      "epoch": 0.132,
      "grad_norm": 1.342164752783955,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 13200
    },
    {
      "epoch": 0.13201,
      "grad_norm": 1.0440248276876525,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 13201
    },
    {
      "epoch": 0.13202,
      "grad_norm": 1.3321336678515836,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 13202
    },
    {
      "epoch": 0.13203,
      "grad_norm": 1.1136943829841504,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 13203
    },
    {
      "epoch": 0.13204,
      "grad_norm": 1.0896162709661188,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 13204
    },
    {
      "epoch": 0.13205,
      "grad_norm": 1.1487113403901221,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 13205
    },
    {
      "epoch": 0.13206,
      "grad_norm": 1.139305369933213,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 13206
    },
    {
      "epoch": 0.13207,
      "grad_norm": 1.1835547312488282,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 13207
    },
    {
      "epoch": 0.13208,
      "grad_norm": 1.33247995754553,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 13208
    },
    {
      "epoch": 0.13209,
      "grad_norm": 1.3513182675506308,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 13209
    },
    {
      "epoch": 0.1321,
      "grad_norm": 1.2551941924992103,
      "learning_rate": 0.003,
      "loss": 4.1134,
      "step": 13210
    },
    {
      "epoch": 0.13211,
      "grad_norm": 0.9604208742486993,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 13211
    },
    {
      "epoch": 0.13212,
      "grad_norm": 1.206886761240646,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 13212
    },
    {
      "epoch": 0.13213,
      "grad_norm": 1.0990701998695058,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 13213
    },
    {
      "epoch": 0.13214,
      "grad_norm": 1.1918705286316655,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 13214
    },
    {
      "epoch": 0.13215,
      "grad_norm": 1.162824123790289,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 13215
    },
    {
      "epoch": 0.13216,
      "grad_norm": 1.2346845464822072,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 13216
    },
    {
      "epoch": 0.13217,
      "grad_norm": 1.1794704948545738,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 13217
    },
    {
      "epoch": 0.13218,
      "grad_norm": 1.1513387641304558,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 13218
    },
    {
      "epoch": 0.13219,
      "grad_norm": 1.2180329095543465,
      "learning_rate": 0.003,
      "loss": 4.1247,
      "step": 13219
    },
    {
      "epoch": 0.1322,
      "grad_norm": 1.484575446629248,
      "learning_rate": 0.003,
      "loss": 4.1142,
      "step": 13220
    },
    {
      "epoch": 0.13221,
      "grad_norm": 0.9424781728975216,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 13221
    },
    {
      "epoch": 0.13222,
      "grad_norm": 1.0684150192432058,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 13222
    },
    {
      "epoch": 0.13223,
      "grad_norm": 1.2191264328565914,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 13223
    },
    {
      "epoch": 0.13224,
      "grad_norm": 0.9350032787837502,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 13224
    },
    {
      "epoch": 0.13225,
      "grad_norm": 1.2614731457692834,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 13225
    },
    {
      "epoch": 0.13226,
      "grad_norm": 1.4290705818533183,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 13226
    },
    {
      "epoch": 0.13227,
      "grad_norm": 1.0077627912110383,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 13227
    },
    {
      "epoch": 0.13228,
      "grad_norm": 1.2211893006066725,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 13228
    },
    {
      "epoch": 0.13229,
      "grad_norm": 1.009855936876147,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 13229
    },
    {
      "epoch": 0.1323,
      "grad_norm": 1.37922530451958,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 13230
    },
    {
      "epoch": 0.13231,
      "grad_norm": 1.0292004867682374,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 13231
    },
    {
      "epoch": 0.13232,
      "grad_norm": 1.1769272669945927,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 13232
    },
    {
      "epoch": 0.13233,
      "grad_norm": 1.1612880435231938,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 13233
    },
    {
      "epoch": 0.13234,
      "grad_norm": 1.2253876757763627,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 13234
    },
    {
      "epoch": 0.13235,
      "grad_norm": 1.3619907271565719,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 13235
    },
    {
      "epoch": 0.13236,
      "grad_norm": 0.9839652554544269,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 13236
    },
    {
      "epoch": 0.13237,
      "grad_norm": 1.1875127600863051,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13237
    },
    {
      "epoch": 0.13238,
      "grad_norm": 1.1055518959048614,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 13238
    },
    {
      "epoch": 0.13239,
      "grad_norm": 1.1295348692789726,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 13239
    },
    {
      "epoch": 0.1324,
      "grad_norm": 1.0981871470208329,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 13240
    },
    {
      "epoch": 0.13241,
      "grad_norm": 1.6840769284460566,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 13241
    },
    {
      "epoch": 0.13242,
      "grad_norm": 0.9366529133323972,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 13242
    },
    {
      "epoch": 0.13243,
      "grad_norm": 1.0424880225179558,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 13243
    },
    {
      "epoch": 0.13244,
      "grad_norm": 1.0848485150228266,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 13244
    },
    {
      "epoch": 0.13245,
      "grad_norm": 1.1482278902934118,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 13245
    },
    {
      "epoch": 0.13246,
      "grad_norm": 1.1663589694334175,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 13246
    },
    {
      "epoch": 0.13247,
      "grad_norm": 1.2100677855701005,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 13247
    },
    {
      "epoch": 0.13248,
      "grad_norm": 1.4384682116904701,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 13248
    },
    {
      "epoch": 0.13249,
      "grad_norm": 1.2706725172309934,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 13249
    },
    {
      "epoch": 0.1325,
      "grad_norm": 1.2777569039571888,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 13250
    },
    {
      "epoch": 0.13251,
      "grad_norm": 1.286598133095499,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13251
    },
    {
      "epoch": 0.13252,
      "grad_norm": 1.1308979039030347,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 13252
    },
    {
      "epoch": 0.13253,
      "grad_norm": 1.1684796593582514,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 13253
    },
    {
      "epoch": 0.13254,
      "grad_norm": 1.2589559908787755,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 13254
    },
    {
      "epoch": 0.13255,
      "grad_norm": 1.113630615783083,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 13255
    },
    {
      "epoch": 0.13256,
      "grad_norm": 1.2670567957920973,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 13256
    },
    {
      "epoch": 0.13257,
      "grad_norm": 1.0472226710354418,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 13257
    },
    {
      "epoch": 0.13258,
      "grad_norm": 1.2427763650400145,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13258
    },
    {
      "epoch": 0.13259,
      "grad_norm": 0.8243964260074956,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 13259
    },
    {
      "epoch": 0.1326,
      "grad_norm": 1.063344790053129,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13260
    },
    {
      "epoch": 0.13261,
      "grad_norm": 1.541367385463494,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 13261
    },
    {
      "epoch": 0.13262,
      "grad_norm": 0.9808939301570836,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 13262
    },
    {
      "epoch": 0.13263,
      "grad_norm": 1.2734804829811215,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 13263
    },
    {
      "epoch": 0.13264,
      "grad_norm": 1.124647905227643,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 13264
    },
    {
      "epoch": 0.13265,
      "grad_norm": 1.319801420352576,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 13265
    },
    {
      "epoch": 0.13266,
      "grad_norm": 1.0439692891652967,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 13266
    },
    {
      "epoch": 0.13267,
      "grad_norm": 1.074415191311497,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 13267
    },
    {
      "epoch": 0.13268,
      "grad_norm": 1.2097445537659586,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 13268
    },
    {
      "epoch": 0.13269,
      "grad_norm": 1.2050238004257336,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 13269
    },
    {
      "epoch": 0.1327,
      "grad_norm": 1.4526043801125104,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 13270
    },
    {
      "epoch": 0.13271,
      "grad_norm": 1.0282734940110956,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 13271
    },
    {
      "epoch": 0.13272,
      "grad_norm": 1.2436480572127762,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 13272
    },
    {
      "epoch": 0.13273,
      "grad_norm": 1.23181245795007,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 13273
    },
    {
      "epoch": 0.13274,
      "grad_norm": 1.1028753704920953,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 13274
    },
    {
      "epoch": 0.13275,
      "grad_norm": 1.1569503153916652,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 13275
    },
    {
      "epoch": 0.13276,
      "grad_norm": 1.066571379296384,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 13276
    },
    {
      "epoch": 0.13277,
      "grad_norm": 1.1392178172837175,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 13277
    },
    {
      "epoch": 0.13278,
      "grad_norm": 1.2438477369686805,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 13278
    },
    {
      "epoch": 0.13279,
      "grad_norm": 1.2866787913315778,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 13279
    },
    {
      "epoch": 0.1328,
      "grad_norm": 1.1591452088705272,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 13280
    },
    {
      "epoch": 0.13281,
      "grad_norm": 1.1830361582764495,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 13281
    },
    {
      "epoch": 0.13282,
      "grad_norm": 1.2505311700849044,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 13282
    },
    {
      "epoch": 0.13283,
      "grad_norm": 1.2096547188191793,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 13283
    },
    {
      "epoch": 0.13284,
      "grad_norm": 1.0036341444276413,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 13284
    },
    {
      "epoch": 0.13285,
      "grad_norm": 1.4806163538861903,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 13285
    },
    {
      "epoch": 0.13286,
      "grad_norm": 1.0813766591338672,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 13286
    },
    {
      "epoch": 0.13287,
      "grad_norm": 1.427730546993291,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 13287
    },
    {
      "epoch": 0.13288,
      "grad_norm": 1.0527188181403344,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 13288
    },
    {
      "epoch": 0.13289,
      "grad_norm": 1.2044844905582028,
      "learning_rate": 0.003,
      "loss": 4.1037,
      "step": 13289
    },
    {
      "epoch": 0.1329,
      "grad_norm": 1.2217949119878093,
      "learning_rate": 0.003,
      "loss": 4.1039,
      "step": 13290
    },
    {
      "epoch": 0.13291,
      "grad_norm": 1.0952651198626726,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 13291
    },
    {
      "epoch": 0.13292,
      "grad_norm": 1.093891287004927,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 13292
    },
    {
      "epoch": 0.13293,
      "grad_norm": 1.2853429222571366,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 13293
    },
    {
      "epoch": 0.13294,
      "grad_norm": 1.0565395357452285,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 13294
    },
    {
      "epoch": 0.13295,
      "grad_norm": 1.129468602840372,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 13295
    },
    {
      "epoch": 0.13296,
      "grad_norm": 1.285359229503194,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 13296
    },
    {
      "epoch": 0.13297,
      "grad_norm": 1.2142336058007384,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 13297
    },
    {
      "epoch": 0.13298,
      "grad_norm": 1.1731333515874773,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 13298
    },
    {
      "epoch": 0.13299,
      "grad_norm": 1.1489028482637664,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 13299
    },
    {
      "epoch": 0.133,
      "grad_norm": 1.445604589879857,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 13300
    },
    {
      "epoch": 0.13301,
      "grad_norm": 1.067028670997227,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 13301
    },
    {
      "epoch": 0.13302,
      "grad_norm": 1.1368262006373977,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 13302
    },
    {
      "epoch": 0.13303,
      "grad_norm": 1.2180401571199988,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 13303
    },
    {
      "epoch": 0.13304,
      "grad_norm": 1.1993665889601892,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 13304
    },
    {
      "epoch": 0.13305,
      "grad_norm": 1.032978558822638,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 13305
    },
    {
      "epoch": 0.13306,
      "grad_norm": 1.2657793854465964,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 13306
    },
    {
      "epoch": 0.13307,
      "grad_norm": 1.1347037754378475,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 13307
    },
    {
      "epoch": 0.13308,
      "grad_norm": 1.0149496856090754,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 13308
    },
    {
      "epoch": 0.13309,
      "grad_norm": 1.4440999617014638,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 13309
    },
    {
      "epoch": 0.1331,
      "grad_norm": 0.9308118300964536,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 13310
    },
    {
      "epoch": 0.13311,
      "grad_norm": 1.2301159217166553,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 13311
    },
    {
      "epoch": 0.13312,
      "grad_norm": 1.1980100279002193,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 13312
    },
    {
      "epoch": 0.13313,
      "grad_norm": 1.1301051915096612,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 13313
    },
    {
      "epoch": 0.13314,
      "grad_norm": 1.1689863478212108,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 13314
    },
    {
      "epoch": 0.13315,
      "grad_norm": 1.062336883714272,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 13315
    },
    {
      "epoch": 0.13316,
      "grad_norm": 1.3501774067650583,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 13316
    },
    {
      "epoch": 0.13317,
      "grad_norm": 1.1449323114599748,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 13317
    },
    {
      "epoch": 0.13318,
      "grad_norm": 1.1220465724242124,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 13318
    },
    {
      "epoch": 0.13319,
      "grad_norm": 1.259102102799351,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 13319
    },
    {
      "epoch": 0.1332,
      "grad_norm": 1.12159952245913,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 13320
    },
    {
      "epoch": 0.13321,
      "grad_norm": 1.2950597137586786,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 13321
    },
    {
      "epoch": 0.13322,
      "grad_norm": 1.0369680795149159,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 13322
    },
    {
      "epoch": 0.13323,
      "grad_norm": 1.2306452875554148,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 13323
    },
    {
      "epoch": 0.13324,
      "grad_norm": 1.2988148242167328,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 13324
    },
    {
      "epoch": 0.13325,
      "grad_norm": 1.0726268592798407,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 13325
    },
    {
      "epoch": 0.13326,
      "grad_norm": 1.201112262648506,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 13326
    },
    {
      "epoch": 0.13327,
      "grad_norm": 1.1806783452463878,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 13327
    },
    {
      "epoch": 0.13328,
      "grad_norm": 1.3125564729665666,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 13328
    },
    {
      "epoch": 0.13329,
      "grad_norm": 1.0701162709873437,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 13329
    },
    {
      "epoch": 0.1333,
      "grad_norm": 1.08267371694619,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 13330
    },
    {
      "epoch": 0.13331,
      "grad_norm": 1.1817818754448577,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 13331
    },
    {
      "epoch": 0.13332,
      "grad_norm": 1.1969977050107905,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 13332
    },
    {
      "epoch": 0.13333,
      "grad_norm": 1.3098841770978693,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 13333
    },
    {
      "epoch": 0.13334,
      "grad_norm": 0.9831900150336861,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 13334
    },
    {
      "epoch": 0.13335,
      "grad_norm": 1.0387924753083626,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 13335
    },
    {
      "epoch": 0.13336,
      "grad_norm": 1.3004967745171618,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 13336
    },
    {
      "epoch": 0.13337,
      "grad_norm": 1.1178283993965488,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 13337
    },
    {
      "epoch": 0.13338,
      "grad_norm": 1.0357008083180437,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 13338
    },
    {
      "epoch": 0.13339,
      "grad_norm": 1.0711535445470308,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 13339
    },
    {
      "epoch": 0.1334,
      "grad_norm": 1.1066555227662631,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 13340
    },
    {
      "epoch": 0.13341,
      "grad_norm": 1.316474283196772,
      "learning_rate": 0.003,
      "loss": 4.1155,
      "step": 13341
    },
    {
      "epoch": 0.13342,
      "grad_norm": 1.245714138650924,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 13342
    },
    {
      "epoch": 0.13343,
      "grad_norm": 1.1062513626932633,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 13343
    },
    {
      "epoch": 0.13344,
      "grad_norm": 1.088863163471069,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 13344
    },
    {
      "epoch": 0.13345,
      "grad_norm": 0.9775251027358349,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 13345
    },
    {
      "epoch": 0.13346,
      "grad_norm": 1.3447630935973398,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 13346
    },
    {
      "epoch": 0.13347,
      "grad_norm": 1.2367279065358332,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 13347
    },
    {
      "epoch": 0.13348,
      "grad_norm": 1.0752274257332177,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 13348
    },
    {
      "epoch": 0.13349,
      "grad_norm": 1.344950271232708,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 13349
    },
    {
      "epoch": 0.1335,
      "grad_norm": 1.2571772033099773,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 13350
    },
    {
      "epoch": 0.13351,
      "grad_norm": 1.235970763029316,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 13351
    },
    {
      "epoch": 0.13352,
      "grad_norm": 1.295337004405207,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 13352
    },
    {
      "epoch": 0.13353,
      "grad_norm": 1.180645390644191,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 13353
    },
    {
      "epoch": 0.13354,
      "grad_norm": 1.1978662749356284,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 13354
    },
    {
      "epoch": 0.13355,
      "grad_norm": 1.1867509477213356,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 13355
    },
    {
      "epoch": 0.13356,
      "grad_norm": 1.245461161844399,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 13356
    },
    {
      "epoch": 0.13357,
      "grad_norm": 1.1018784359679767,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 13357
    },
    {
      "epoch": 0.13358,
      "grad_norm": 1.1035812117049424,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 13358
    },
    {
      "epoch": 0.13359,
      "grad_norm": 1.1254309321450855,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 13359
    },
    {
      "epoch": 0.1336,
      "grad_norm": 1.2553998155691877,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 13360
    },
    {
      "epoch": 0.13361,
      "grad_norm": 1.0222119041098203,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 13361
    },
    {
      "epoch": 0.13362,
      "grad_norm": 1.3022338392042456,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 13362
    },
    {
      "epoch": 0.13363,
      "grad_norm": 0.9521752871091076,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 13363
    },
    {
      "epoch": 0.13364,
      "grad_norm": 1.229941668303482,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 13364
    },
    {
      "epoch": 0.13365,
      "grad_norm": 1.2541762933410678,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 13365
    },
    {
      "epoch": 0.13366,
      "grad_norm": 1.4218542821693612,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 13366
    },
    {
      "epoch": 0.13367,
      "grad_norm": 1.3063230638440648,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 13367
    },
    {
      "epoch": 0.13368,
      "grad_norm": 1.0202774961705887,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 13368
    },
    {
      "epoch": 0.13369,
      "grad_norm": 1.1992099294907612,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 13369
    },
    {
      "epoch": 0.1337,
      "grad_norm": 0.9642267936227421,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 13370
    },
    {
      "epoch": 0.13371,
      "grad_norm": 1.281749927644538,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 13371
    },
    {
      "epoch": 0.13372,
      "grad_norm": 1.1878458622250818,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 13372
    },
    {
      "epoch": 0.13373,
      "grad_norm": 1.5797165771743982,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 13373
    },
    {
      "epoch": 0.13374,
      "grad_norm": 1.0396410036095307,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 13374
    },
    {
      "epoch": 0.13375,
      "grad_norm": 1.2712754912215498,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 13375
    },
    {
      "epoch": 0.13376,
      "grad_norm": 1.2183517382921254,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 13376
    },
    {
      "epoch": 0.13377,
      "grad_norm": 1.0613623993549737,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 13377
    },
    {
      "epoch": 0.13378,
      "grad_norm": 1.3355682372522968,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 13378
    },
    {
      "epoch": 0.13379,
      "grad_norm": 1.0938413286089992,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 13379
    },
    {
      "epoch": 0.1338,
      "grad_norm": 1.1103890552746947,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 13380
    },
    {
      "epoch": 0.13381,
      "grad_norm": 1.2699701548903857,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 13381
    },
    {
      "epoch": 0.13382,
      "grad_norm": 1.1233842540217829,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 13382
    },
    {
      "epoch": 0.13383,
      "grad_norm": 1.2765612797715782,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 13383
    },
    {
      "epoch": 0.13384,
      "grad_norm": 1.2006039862386173,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 13384
    },
    {
      "epoch": 0.13385,
      "grad_norm": 1.1642926025532412,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 13385
    },
    {
      "epoch": 0.13386,
      "grad_norm": 1.2571178812083903,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 13386
    },
    {
      "epoch": 0.13387,
      "grad_norm": 0.9608708361775664,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 13387
    },
    {
      "epoch": 0.13388,
      "grad_norm": 1.1366847605441799,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 13388
    },
    {
      "epoch": 0.13389,
      "grad_norm": 1.2306720467075594,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 13389
    },
    {
      "epoch": 0.1339,
      "grad_norm": 1.1164334604497055,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 13390
    },
    {
      "epoch": 0.13391,
      "grad_norm": 1.0426308201644983,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 13391
    },
    {
      "epoch": 0.13392,
      "grad_norm": 1.210123565635377,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 13392
    },
    {
      "epoch": 0.13393,
      "grad_norm": 1.2414495817178337,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 13393
    },
    {
      "epoch": 0.13394,
      "grad_norm": 1.3451434252302206,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 13394
    },
    {
      "epoch": 0.13395,
      "grad_norm": 1.1878714124216885,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 13395
    },
    {
      "epoch": 0.13396,
      "grad_norm": 1.2257200245169122,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 13396
    },
    {
      "epoch": 0.13397,
      "grad_norm": 1.0356620543924908,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 13397
    },
    {
      "epoch": 0.13398,
      "grad_norm": 1.282872635708081,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 13398
    },
    {
      "epoch": 0.13399,
      "grad_norm": 1.1479540377003152,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 13399
    },
    {
      "epoch": 0.134,
      "grad_norm": 1.1005041773174578,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 13400
    },
    {
      "epoch": 0.13401,
      "grad_norm": 1.268179261420047,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 13401
    },
    {
      "epoch": 0.13402,
      "grad_norm": 1.1738455552542615,
      "learning_rate": 0.003,
      "loss": 4.0296,
      "step": 13402
    },
    {
      "epoch": 0.13403,
      "grad_norm": 1.3879437579648175,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 13403
    },
    {
      "epoch": 0.13404,
      "grad_norm": 1.2541279008654285,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 13404
    },
    {
      "epoch": 0.13405,
      "grad_norm": 1.2605087317660528,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 13405
    },
    {
      "epoch": 0.13406,
      "grad_norm": 1.2681078901785494,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 13406
    },
    {
      "epoch": 0.13407,
      "grad_norm": 1.2605010221084334,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 13407
    },
    {
      "epoch": 0.13408,
      "grad_norm": 0.8239793765808275,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 13408
    },
    {
      "epoch": 0.13409,
      "grad_norm": 0.9311582810932972,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 13409
    },
    {
      "epoch": 0.1341,
      "grad_norm": 1.1252731190236187,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 13410
    },
    {
      "epoch": 0.13411,
      "grad_norm": 1.1060331467637572,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 13411
    },
    {
      "epoch": 0.13412,
      "grad_norm": 0.9573062747071894,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 13412
    },
    {
      "epoch": 0.13413,
      "grad_norm": 1.5146959163459248,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 13413
    },
    {
      "epoch": 0.13414,
      "grad_norm": 1.1917831643750696,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 13414
    },
    {
      "epoch": 0.13415,
      "grad_norm": 1.2904828536059358,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 13415
    },
    {
      "epoch": 0.13416,
      "grad_norm": 0.9840594781612224,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 13416
    },
    {
      "epoch": 0.13417,
      "grad_norm": 1.2528731806628532,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 13417
    },
    {
      "epoch": 0.13418,
      "grad_norm": 1.0942044526081434,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 13418
    },
    {
      "epoch": 0.13419,
      "grad_norm": 1.3026400484830545,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 13419
    },
    {
      "epoch": 0.1342,
      "grad_norm": 1.058350961997098,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 13420
    },
    {
      "epoch": 0.13421,
      "grad_norm": 1.0299704476949245,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 13421
    },
    {
      "epoch": 0.13422,
      "grad_norm": 1.1181419868091191,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 13422
    },
    {
      "epoch": 0.13423,
      "grad_norm": 1.2457760033592935,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 13423
    },
    {
      "epoch": 0.13424,
      "grad_norm": 1.1283267972883926,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 13424
    },
    {
      "epoch": 0.13425,
      "grad_norm": 1.2522641300795905,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 13425
    },
    {
      "epoch": 0.13426,
      "grad_norm": 0.9834937636747162,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 13426
    },
    {
      "epoch": 0.13427,
      "grad_norm": 1.2090305796613712,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 13427
    },
    {
      "epoch": 0.13428,
      "grad_norm": 1.194819039867254,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 13428
    },
    {
      "epoch": 0.13429,
      "grad_norm": 1.3388245141106132,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 13429
    },
    {
      "epoch": 0.1343,
      "grad_norm": 1.1759131578031872,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 13430
    },
    {
      "epoch": 0.13431,
      "grad_norm": 1.4354047016128604,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 13431
    },
    {
      "epoch": 0.13432,
      "grad_norm": 0.9462723573405492,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 13432
    },
    {
      "epoch": 0.13433,
      "grad_norm": 1.1738535078072172,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 13433
    },
    {
      "epoch": 0.13434,
      "grad_norm": 1.1280589093406457,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 13434
    },
    {
      "epoch": 0.13435,
      "grad_norm": 1.1328603065697254,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 13435
    },
    {
      "epoch": 0.13436,
      "grad_norm": 1.3643358412921909,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 13436
    },
    {
      "epoch": 0.13437,
      "grad_norm": 1.1349177609772076,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 13437
    },
    {
      "epoch": 0.13438,
      "grad_norm": 1.3786563909537977,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 13438
    },
    {
      "epoch": 0.13439,
      "grad_norm": 0.9610016782956333,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 13439
    },
    {
      "epoch": 0.1344,
      "grad_norm": 1.2297517952028227,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 13440
    },
    {
      "epoch": 0.13441,
      "grad_norm": 1.1601578060683668,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 13441
    },
    {
      "epoch": 0.13442,
      "grad_norm": 1.3373133683032783,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 13442
    },
    {
      "epoch": 0.13443,
      "grad_norm": 1.0474449161363737,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 13443
    },
    {
      "epoch": 0.13444,
      "grad_norm": 1.4151051682755142,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 13444
    },
    {
      "epoch": 0.13445,
      "grad_norm": 1.0537975131356527,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 13445
    },
    {
      "epoch": 0.13446,
      "grad_norm": 1.186175713874241,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 13446
    },
    {
      "epoch": 0.13447,
      "grad_norm": 1.061759357207979,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 13447
    },
    {
      "epoch": 0.13448,
      "grad_norm": 1.415112960218441,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 13448
    },
    {
      "epoch": 0.13449,
      "grad_norm": 1.1728851918493217,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 13449
    },
    {
      "epoch": 0.1345,
      "grad_norm": 1.4236709724040442,
      "learning_rate": 0.003,
      "loss": 4.0269,
      "step": 13450
    },
    {
      "epoch": 0.13451,
      "grad_norm": 0.9223059572756359,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 13451
    },
    {
      "epoch": 0.13452,
      "grad_norm": 1.2323279228543833,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 13452
    },
    {
      "epoch": 0.13453,
      "grad_norm": 1.291375695669297,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 13453
    },
    {
      "epoch": 0.13454,
      "grad_norm": 1.157160154775136,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 13454
    },
    {
      "epoch": 0.13455,
      "grad_norm": 1.2368519621370178,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 13455
    },
    {
      "epoch": 0.13456,
      "grad_norm": 1.3142726252125636,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 13456
    },
    {
      "epoch": 0.13457,
      "grad_norm": 1.162434038095761,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 13457
    },
    {
      "epoch": 0.13458,
      "grad_norm": 1.1078576104450124,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 13458
    },
    {
      "epoch": 0.13459,
      "grad_norm": 1.2383130680770433,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 13459
    },
    {
      "epoch": 0.1346,
      "grad_norm": 1.1125401488562099,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 13460
    },
    {
      "epoch": 0.13461,
      "grad_norm": 1.2140055487172725,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 13461
    },
    {
      "epoch": 0.13462,
      "grad_norm": 1.1433151082950783,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 13462
    },
    {
      "epoch": 0.13463,
      "grad_norm": 1.1821004125728793,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 13463
    },
    {
      "epoch": 0.13464,
      "grad_norm": 1.2662168014729447,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 13464
    },
    {
      "epoch": 0.13465,
      "grad_norm": 0.9888495946718485,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 13465
    },
    {
      "epoch": 0.13466,
      "grad_norm": 1.1220154671518745,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 13466
    },
    {
      "epoch": 0.13467,
      "grad_norm": 1.0919519821871781,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 13467
    },
    {
      "epoch": 0.13468,
      "grad_norm": 1.340826365379991,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 13468
    },
    {
      "epoch": 0.13469,
      "grad_norm": 1.0121555399016573,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 13469
    },
    {
      "epoch": 0.1347,
      "grad_norm": 1.3067290538201393,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 13470
    },
    {
      "epoch": 0.13471,
      "grad_norm": 1.0880686216576068,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 13471
    },
    {
      "epoch": 0.13472,
      "grad_norm": 1.1831367938514399,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 13472
    },
    {
      "epoch": 0.13473,
      "grad_norm": 1.3612868750769027,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 13473
    },
    {
      "epoch": 0.13474,
      "grad_norm": 0.9508255251277685,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 13474
    },
    {
      "epoch": 0.13475,
      "grad_norm": 1.2856856658658182,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 13475
    },
    {
      "epoch": 0.13476,
      "grad_norm": 1.1606275811173252,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 13476
    },
    {
      "epoch": 0.13477,
      "grad_norm": 1.1524479186013914,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 13477
    },
    {
      "epoch": 0.13478,
      "grad_norm": 1.235184313994686,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 13478
    },
    {
      "epoch": 0.13479,
      "grad_norm": 1.1070378269991312,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 13479
    },
    {
      "epoch": 0.1348,
      "grad_norm": 1.0632334092285003,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 13480
    },
    {
      "epoch": 0.13481,
      "grad_norm": 1.2234028229811578,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 13481
    },
    {
      "epoch": 0.13482,
      "grad_norm": 1.146489255781058,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 13482
    },
    {
      "epoch": 0.13483,
      "grad_norm": 1.1637185523813744,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 13483
    },
    {
      "epoch": 0.13484,
      "grad_norm": 1.1185100360085258,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 13484
    },
    {
      "epoch": 0.13485,
      "grad_norm": 1.193469537143719,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 13485
    },
    {
      "epoch": 0.13486,
      "grad_norm": 1.135558165767479,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 13486
    },
    {
      "epoch": 0.13487,
      "grad_norm": 1.4057601165947837,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 13487
    },
    {
      "epoch": 0.13488,
      "grad_norm": 1.1656899608751365,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 13488
    },
    {
      "epoch": 0.13489,
      "grad_norm": 1.5043333136087835,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 13489
    },
    {
      "epoch": 0.1349,
      "grad_norm": 1.0163820255010707,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 13490
    },
    {
      "epoch": 0.13491,
      "grad_norm": 1.3061494100348068,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 13491
    },
    {
      "epoch": 0.13492,
      "grad_norm": 1.1228916805737252,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 13492
    },
    {
      "epoch": 0.13493,
      "grad_norm": 1.2622587830241703,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 13493
    },
    {
      "epoch": 0.13494,
      "grad_norm": 1.2240498827665462,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 13494
    },
    {
      "epoch": 0.13495,
      "grad_norm": 1.1536114873612044,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 13495
    },
    {
      "epoch": 0.13496,
      "grad_norm": 1.2557495984348974,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 13496
    },
    {
      "epoch": 0.13497,
      "grad_norm": 1.2383652825597389,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 13497
    },
    {
      "epoch": 0.13498,
      "grad_norm": 1.101293070079644,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 13498
    },
    {
      "epoch": 0.13499,
      "grad_norm": 1.2268027447598873,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 13499
    },
    {
      "epoch": 0.135,
      "grad_norm": 1.4412497255523504,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 13500
    },
    {
      "epoch": 0.13501,
      "grad_norm": 1.1916063975445812,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 13501
    },
    {
      "epoch": 0.13502,
      "grad_norm": 1.178800704541382,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 13502
    },
    {
      "epoch": 0.13503,
      "grad_norm": 1.1767774893647651,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 13503
    },
    {
      "epoch": 0.13504,
      "grad_norm": 1.1528621548637201,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 13504
    },
    {
      "epoch": 0.13505,
      "grad_norm": 1.3891953153521712,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 13505
    },
    {
      "epoch": 0.13506,
      "grad_norm": 0.9887205718992813,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 13506
    },
    {
      "epoch": 0.13507,
      "grad_norm": 1.139102638279225,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 13507
    },
    {
      "epoch": 0.13508,
      "grad_norm": 1.0348166624357527,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 13508
    },
    {
      "epoch": 0.13509,
      "grad_norm": 1.266192907540622,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 13509
    },
    {
      "epoch": 0.1351,
      "grad_norm": 1.235255795201123,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 13510
    },
    {
      "epoch": 0.13511,
      "grad_norm": 1.2451094931773148,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 13511
    },
    {
      "epoch": 0.13512,
      "grad_norm": 1.038734480871431,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 13512
    },
    {
      "epoch": 0.13513,
      "grad_norm": 1.3362978558878509,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 13513
    },
    {
      "epoch": 0.13514,
      "grad_norm": 1.0116571880327812,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 13514
    },
    {
      "epoch": 0.13515,
      "grad_norm": 1.5055660158456143,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 13515
    },
    {
      "epoch": 0.13516,
      "grad_norm": 1.0753440149432287,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 13516
    },
    {
      "epoch": 0.13517,
      "grad_norm": 1.475048709045603,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 13517
    },
    {
      "epoch": 0.13518,
      "grad_norm": 1.057908709973592,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 13518
    },
    {
      "epoch": 0.13519,
      "grad_norm": 1.2103503090775813,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 13519
    },
    {
      "epoch": 0.1352,
      "grad_norm": 1.2676298070510073,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 13520
    },
    {
      "epoch": 0.13521,
      "grad_norm": 1.1815116642938954,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 13521
    },
    {
      "epoch": 0.13522,
      "grad_norm": 1.085564867833204,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 13522
    },
    {
      "epoch": 0.13523,
      "grad_norm": 1.0268095579877041,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 13523
    },
    {
      "epoch": 0.13524,
      "grad_norm": 1.3869619517893401,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 13524
    },
    {
      "epoch": 0.13525,
      "grad_norm": 0.9848067653607371,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 13525
    },
    {
      "epoch": 0.13526,
      "grad_norm": 1.4095299409827127,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 13526
    },
    {
      "epoch": 0.13527,
      "grad_norm": 1.1064622843634515,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 13527
    },
    {
      "epoch": 0.13528,
      "grad_norm": 1.2619597308926152,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 13528
    },
    {
      "epoch": 0.13529,
      "grad_norm": 1.3763006823062767,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 13529
    },
    {
      "epoch": 0.1353,
      "grad_norm": 1.1529577761975283,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 13530
    },
    {
      "epoch": 0.13531,
      "grad_norm": 1.4151078239343364,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 13531
    },
    {
      "epoch": 0.13532,
      "grad_norm": 1.0389544412766625,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 13532
    },
    {
      "epoch": 0.13533,
      "grad_norm": 1.1946282506821346,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 13533
    },
    {
      "epoch": 0.13534,
      "grad_norm": 1.2747275025398543,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 13534
    },
    {
      "epoch": 0.13535,
      "grad_norm": 1.1473872652809254,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 13535
    },
    {
      "epoch": 0.13536,
      "grad_norm": 1.1398330387721207,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 13536
    },
    {
      "epoch": 0.13537,
      "grad_norm": 1.437869757657236,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 13537
    },
    {
      "epoch": 0.13538,
      "grad_norm": 1.0497105637254676,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 13538
    },
    {
      "epoch": 0.13539,
      "grad_norm": 1.1803193267606635,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 13539
    },
    {
      "epoch": 0.1354,
      "grad_norm": 1.0150269320384628,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 13540
    },
    {
      "epoch": 0.13541,
      "grad_norm": 1.2965251686076023,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 13541
    },
    {
      "epoch": 0.13542,
      "grad_norm": 1.0030909278074205,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 13542
    },
    {
      "epoch": 0.13543,
      "grad_norm": 1.2468250804520131,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 13543
    },
    {
      "epoch": 0.13544,
      "grad_norm": 1.0310918920010577,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 13544
    },
    {
      "epoch": 0.13545,
      "grad_norm": 1.225367536185542,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 13545
    },
    {
      "epoch": 0.13546,
      "grad_norm": 1.0087671343539084,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 13546
    },
    {
      "epoch": 0.13547,
      "grad_norm": 1.3541882850349083,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 13547
    },
    {
      "epoch": 0.13548,
      "grad_norm": 1.1022860089932935,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 13548
    },
    {
      "epoch": 0.13549,
      "grad_norm": 1.3990431601206805,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 13549
    },
    {
      "epoch": 0.1355,
      "grad_norm": 1.209753798907345,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 13550
    },
    {
      "epoch": 0.13551,
      "grad_norm": 1.2227913643789063,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 13551
    },
    {
      "epoch": 0.13552,
      "grad_norm": 1.1239293994750092,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 13552
    },
    {
      "epoch": 0.13553,
      "grad_norm": 1.1809798944838255,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 13553
    },
    {
      "epoch": 0.13554,
      "grad_norm": 1.1052256243036818,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 13554
    },
    {
      "epoch": 0.13555,
      "grad_norm": 1.2460361801722424,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 13555
    },
    {
      "epoch": 0.13556,
      "grad_norm": 1.0764279421582164,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 13556
    },
    {
      "epoch": 0.13557,
      "grad_norm": 1.1564737937797744,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 13557
    },
    {
      "epoch": 0.13558,
      "grad_norm": 1.1661195493428866,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 13558
    },
    {
      "epoch": 0.13559,
      "grad_norm": 1.3724361794108662,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 13559
    },
    {
      "epoch": 0.1356,
      "grad_norm": 1.137116772236948,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 13560
    },
    {
      "epoch": 0.13561,
      "grad_norm": 1.3021859483441538,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 13561
    },
    {
      "epoch": 0.13562,
      "grad_norm": 0.9913114792500876,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 13562
    },
    {
      "epoch": 0.13563,
      "grad_norm": 1.3501726665731697,
      "learning_rate": 0.003,
      "loss": 4.0243,
      "step": 13563
    },
    {
      "epoch": 0.13564,
      "grad_norm": 1.1109480504606148,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 13564
    },
    {
      "epoch": 0.13565,
      "grad_norm": 1.1932453808361558,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 13565
    },
    {
      "epoch": 0.13566,
      "grad_norm": 1.2968291961796083,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 13566
    },
    {
      "epoch": 0.13567,
      "grad_norm": 0.8928736539916826,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 13567
    },
    {
      "epoch": 0.13568,
      "grad_norm": 0.9474463361605496,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 13568
    },
    {
      "epoch": 0.13569,
      "grad_norm": 1.3066183160874658,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 13569
    },
    {
      "epoch": 0.1357,
      "grad_norm": 1.0298676653080354,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 13570
    },
    {
      "epoch": 0.13571,
      "grad_norm": 1.2340368544245897,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 13571
    },
    {
      "epoch": 0.13572,
      "grad_norm": 1.4403037500764257,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 13572
    },
    {
      "epoch": 0.13573,
      "grad_norm": 1.0939889578066653,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 13573
    },
    {
      "epoch": 0.13574,
      "grad_norm": 1.3088511673177976,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 13574
    },
    {
      "epoch": 0.13575,
      "grad_norm": 1.2768363323140435,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 13575
    },
    {
      "epoch": 0.13576,
      "grad_norm": 1.256661574995121,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 13576
    },
    {
      "epoch": 0.13577,
      "grad_norm": 1.0588483065511913,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 13577
    },
    {
      "epoch": 0.13578,
      "grad_norm": 1.2630132494145923,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 13578
    },
    {
      "epoch": 0.13579,
      "grad_norm": 1.1653312093577335,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13579
    },
    {
      "epoch": 0.1358,
      "grad_norm": 1.223415623412137,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 13580
    },
    {
      "epoch": 0.13581,
      "grad_norm": 1.3238249676184883,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 13581
    },
    {
      "epoch": 0.13582,
      "grad_norm": 1.03416907167867,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 13582
    },
    {
      "epoch": 0.13583,
      "grad_norm": 1.3594672938144423,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 13583
    },
    {
      "epoch": 0.13584,
      "grad_norm": 1.077123244373863,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 13584
    },
    {
      "epoch": 0.13585,
      "grad_norm": 1.1672176890992012,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 13585
    },
    {
      "epoch": 0.13586,
      "grad_norm": 1.2049053246218717,
      "learning_rate": 0.003,
      "loss": 4.1177,
      "step": 13586
    },
    {
      "epoch": 0.13587,
      "grad_norm": 1.3865760587332796,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 13587
    },
    {
      "epoch": 0.13588,
      "grad_norm": 1.1607163231788171,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 13588
    },
    {
      "epoch": 0.13589,
      "grad_norm": 1.174397464749926,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 13589
    },
    {
      "epoch": 0.1359,
      "grad_norm": 1.2792718395574298,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 13590
    },
    {
      "epoch": 0.13591,
      "grad_norm": 0.9721022972830015,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 13591
    },
    {
      "epoch": 0.13592,
      "grad_norm": 1.365827999405511,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 13592
    },
    {
      "epoch": 0.13593,
      "grad_norm": 1.2376559086323935,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 13593
    },
    {
      "epoch": 0.13594,
      "grad_norm": 1.1000453997314037,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 13594
    },
    {
      "epoch": 0.13595,
      "grad_norm": 1.0605799757249121,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 13595
    },
    {
      "epoch": 0.13596,
      "grad_norm": 1.0047523449138953,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 13596
    },
    {
      "epoch": 0.13597,
      "grad_norm": 1.1212769914855838,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 13597
    },
    {
      "epoch": 0.13598,
      "grad_norm": 1.0643512786066576,
      "learning_rate": 0.003,
      "loss": 4.1107,
      "step": 13598
    },
    {
      "epoch": 0.13599,
      "grad_norm": 1.1137772972856694,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 13599
    },
    {
      "epoch": 0.136,
      "grad_norm": 1.238310229737092,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 13600
    },
    {
      "epoch": 0.13601,
      "grad_norm": 1.2288065213057453,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 13601
    },
    {
      "epoch": 0.13602,
      "grad_norm": 1.2127205294525991,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 13602
    },
    {
      "epoch": 0.13603,
      "grad_norm": 1.0688204869182307,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 13603
    },
    {
      "epoch": 0.13604,
      "grad_norm": 1.3238519061580016,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 13604
    },
    {
      "epoch": 0.13605,
      "grad_norm": 1.065978057108014,
      "learning_rate": 0.003,
      "loss": 4.0275,
      "step": 13605
    },
    {
      "epoch": 0.13606,
      "grad_norm": 1.5113214195065328,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 13606
    },
    {
      "epoch": 0.13607,
      "grad_norm": 1.1835044192072945,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 13607
    },
    {
      "epoch": 0.13608,
      "grad_norm": 1.3389232557551585,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 13608
    },
    {
      "epoch": 0.13609,
      "grad_norm": 1.08573694872505,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 13609
    },
    {
      "epoch": 0.1361,
      "grad_norm": 1.3183382408270754,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 13610
    },
    {
      "epoch": 0.13611,
      "grad_norm": 1.088614267022385,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 13611
    },
    {
      "epoch": 0.13612,
      "grad_norm": 1.584591299836842,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 13612
    },
    {
      "epoch": 0.13613,
      "grad_norm": 1.0241167509813345,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 13613
    },
    {
      "epoch": 0.13614,
      "grad_norm": 1.1410279990397971,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 13614
    },
    {
      "epoch": 0.13615,
      "grad_norm": 1.140403522308166,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 13615
    },
    {
      "epoch": 0.13616,
      "grad_norm": 1.1882408037137873,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 13616
    },
    {
      "epoch": 0.13617,
      "grad_norm": 1.2733678401188344,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 13617
    },
    {
      "epoch": 0.13618,
      "grad_norm": 1.182541789342089,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 13618
    },
    {
      "epoch": 0.13619,
      "grad_norm": 1.8158182544769947,
      "learning_rate": 0.003,
      "loss": 4.1042,
      "step": 13619
    },
    {
      "epoch": 0.1362,
      "grad_norm": 0.9262895261876131,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 13620
    },
    {
      "epoch": 0.13621,
      "grad_norm": 1.0048294050605548,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 13621
    },
    {
      "epoch": 0.13622,
      "grad_norm": 1.2932780705564038,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 13622
    },
    {
      "epoch": 0.13623,
      "grad_norm": 1.0931089753864223,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 13623
    },
    {
      "epoch": 0.13624,
      "grad_norm": 1.2219390368450898,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 13624
    },
    {
      "epoch": 0.13625,
      "grad_norm": 1.1703642070006044,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 13625
    },
    {
      "epoch": 0.13626,
      "grad_norm": 1.3588552313638005,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 13626
    },
    {
      "epoch": 0.13627,
      "grad_norm": 0.991279636178976,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 13627
    },
    {
      "epoch": 0.13628,
      "grad_norm": 1.2418972652541356,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 13628
    },
    {
      "epoch": 0.13629,
      "grad_norm": 0.9762054181088988,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 13629
    },
    {
      "epoch": 0.1363,
      "grad_norm": 1.0594583167811744,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 13630
    },
    {
      "epoch": 0.13631,
      "grad_norm": 1.3725521661001745,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 13631
    },
    {
      "epoch": 0.13632,
      "grad_norm": 0.9807824018898389,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 13632
    },
    {
      "epoch": 0.13633,
      "grad_norm": 1.2084168231669876,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 13633
    },
    {
      "epoch": 0.13634,
      "grad_norm": 1.1830272923144818,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 13634
    },
    {
      "epoch": 0.13635,
      "grad_norm": 1.3078598922535416,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 13635
    },
    {
      "epoch": 0.13636,
      "grad_norm": 1.1001066668138082,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 13636
    },
    {
      "epoch": 0.13637,
      "grad_norm": 1.1638935784962923,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 13637
    },
    {
      "epoch": 0.13638,
      "grad_norm": 1.0385485798210434,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 13638
    },
    {
      "epoch": 0.13639,
      "grad_norm": 1.1898560868781058,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 13639
    },
    {
      "epoch": 0.1364,
      "grad_norm": 1.0299533186076069,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 13640
    },
    {
      "epoch": 0.13641,
      "grad_norm": 1.2105958219454982,
      "learning_rate": 0.003,
      "loss": 4.1024,
      "step": 13641
    },
    {
      "epoch": 0.13642,
      "grad_norm": 1.2786757826782007,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 13642
    },
    {
      "epoch": 0.13643,
      "grad_norm": 1.107637562382635,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 13643
    },
    {
      "epoch": 0.13644,
      "grad_norm": 1.4947670952049972,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 13644
    },
    {
      "epoch": 0.13645,
      "grad_norm": 1.043630530838436,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 13645
    },
    {
      "epoch": 0.13646,
      "grad_norm": 1.1074987074965126,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 13646
    },
    {
      "epoch": 0.13647,
      "grad_norm": 1.3183004868184747,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 13647
    },
    {
      "epoch": 0.13648,
      "grad_norm": 1.3334267070048087,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 13648
    },
    {
      "epoch": 0.13649,
      "grad_norm": 0.9769178281211383,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 13649
    },
    {
      "epoch": 0.1365,
      "grad_norm": 0.8940528609794745,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 13650
    },
    {
      "epoch": 0.13651,
      "grad_norm": 1.0203903879437717,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 13651
    },
    {
      "epoch": 0.13652,
      "grad_norm": 1.47641862136386,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 13652
    },
    {
      "epoch": 0.13653,
      "grad_norm": 1.023482671599721,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 13653
    },
    {
      "epoch": 0.13654,
      "grad_norm": 1.4599717145775253,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 13654
    },
    {
      "epoch": 0.13655,
      "grad_norm": 1.1430378029023696,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 13655
    },
    {
      "epoch": 0.13656,
      "grad_norm": 1.2479062092081508,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 13656
    },
    {
      "epoch": 0.13657,
      "grad_norm": 1.1586196485600007,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 13657
    },
    {
      "epoch": 0.13658,
      "grad_norm": 1.2402177149121234,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 13658
    },
    {
      "epoch": 0.13659,
      "grad_norm": 1.1070055244833996,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13659
    },
    {
      "epoch": 0.1366,
      "grad_norm": 1.3387286615081702,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 13660
    },
    {
      "epoch": 0.13661,
      "grad_norm": 0.8802671899533874,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 13661
    },
    {
      "epoch": 0.13662,
      "grad_norm": 1.0961436556444828,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 13662
    },
    {
      "epoch": 0.13663,
      "grad_norm": 1.1659633698089749,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 13663
    },
    {
      "epoch": 0.13664,
      "grad_norm": 1.1625088068087674,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 13664
    },
    {
      "epoch": 0.13665,
      "grad_norm": 1.1952845956658587,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 13665
    },
    {
      "epoch": 0.13666,
      "grad_norm": 1.3022934851628816,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 13666
    },
    {
      "epoch": 0.13667,
      "grad_norm": 1.1024907017076262,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 13667
    },
    {
      "epoch": 0.13668,
      "grad_norm": 1.1053766957068363,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 13668
    },
    {
      "epoch": 0.13669,
      "grad_norm": 1.0384195487713894,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 13669
    },
    {
      "epoch": 0.1367,
      "grad_norm": 1.654512344419585,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 13670
    },
    {
      "epoch": 0.13671,
      "grad_norm": 1.3646809877437882,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 13671
    },
    {
      "epoch": 0.13672,
      "grad_norm": 1.0469603430029577,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 13672
    },
    {
      "epoch": 0.13673,
      "grad_norm": 1.3395441818598068,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 13673
    },
    {
      "epoch": 0.13674,
      "grad_norm": 1.1458122349135758,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 13674
    },
    {
      "epoch": 0.13675,
      "grad_norm": 1.1271907544979884,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 13675
    },
    {
      "epoch": 0.13676,
      "grad_norm": 1.114889662851139,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 13676
    },
    {
      "epoch": 0.13677,
      "grad_norm": 1.182067015827954,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 13677
    },
    {
      "epoch": 0.13678,
      "grad_norm": 1.115747324972659,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 13678
    },
    {
      "epoch": 0.13679,
      "grad_norm": 1.1011917999804097,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 13679
    },
    {
      "epoch": 0.1368,
      "grad_norm": 1.1142952525222676,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 13680
    },
    {
      "epoch": 0.13681,
      "grad_norm": 1.0676043236090158,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 13681
    },
    {
      "epoch": 0.13682,
      "grad_norm": 1.156689715671784,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 13682
    },
    {
      "epoch": 0.13683,
      "grad_norm": 1.1911333513095246,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 13683
    },
    {
      "epoch": 0.13684,
      "grad_norm": 1.2668964686340978,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 13684
    },
    {
      "epoch": 0.13685,
      "grad_norm": 1.2466397953616286,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 13685
    },
    {
      "epoch": 0.13686,
      "grad_norm": 1.064899811062785,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 13686
    },
    {
      "epoch": 0.13687,
      "grad_norm": 1.298345882337796,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 13687
    },
    {
      "epoch": 0.13688,
      "grad_norm": 1.0430799466261411,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 13688
    },
    {
      "epoch": 0.13689,
      "grad_norm": 1.3666586493580102,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 13689
    },
    {
      "epoch": 0.1369,
      "grad_norm": 1.0370636511576221,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 13690
    },
    {
      "epoch": 0.13691,
      "grad_norm": 1.1996813727422466,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 13691
    },
    {
      "epoch": 0.13692,
      "grad_norm": 1.2216185605014462,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 13692
    },
    {
      "epoch": 0.13693,
      "grad_norm": 1.2998804361911467,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 13693
    },
    {
      "epoch": 0.13694,
      "grad_norm": 1.017742494137553,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 13694
    },
    {
      "epoch": 0.13695,
      "grad_norm": 1.3318604210617848,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 13695
    },
    {
      "epoch": 0.13696,
      "grad_norm": 1.2027103780835313,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 13696
    },
    {
      "epoch": 0.13697,
      "grad_norm": 1.2468422630000149,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 13697
    },
    {
      "epoch": 0.13698,
      "grad_norm": 1.1135879945021516,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 13698
    },
    {
      "epoch": 0.13699,
      "grad_norm": 1.33820725858327,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 13699
    },
    {
      "epoch": 0.137,
      "grad_norm": 1.115047651613133,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 13700
    },
    {
      "epoch": 0.13701,
      "grad_norm": 1.0425628725738034,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 13701
    },
    {
      "epoch": 0.13702,
      "grad_norm": 1.3163601188035263,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 13702
    },
    {
      "epoch": 0.13703,
      "grad_norm": 1.078429519381817,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 13703
    },
    {
      "epoch": 0.13704,
      "grad_norm": 1.1876842179490408,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 13704
    },
    {
      "epoch": 0.13705,
      "grad_norm": 1.0233911271089657,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 13705
    },
    {
      "epoch": 0.13706,
      "grad_norm": 1.3133228580224459,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 13706
    },
    {
      "epoch": 0.13707,
      "grad_norm": 0.876190413993581,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 13707
    },
    {
      "epoch": 0.13708,
      "grad_norm": 1.1224763276316478,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 13708
    },
    {
      "epoch": 0.13709,
      "grad_norm": 1.3941391601536905,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 13709
    },
    {
      "epoch": 0.1371,
      "grad_norm": 1.3076888362270127,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 13710
    },
    {
      "epoch": 0.13711,
      "grad_norm": 1.2454693390615892,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 13711
    },
    {
      "epoch": 0.13712,
      "grad_norm": 1.3841143185327094,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 13712
    },
    {
      "epoch": 0.13713,
      "grad_norm": 1.2374480524788867,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 13713
    },
    {
      "epoch": 0.13714,
      "grad_norm": 1.1199426571591335,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 13714
    },
    {
      "epoch": 0.13715,
      "grad_norm": 1.0598577289200253,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 13715
    },
    {
      "epoch": 0.13716,
      "grad_norm": 1.2854522109147988,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 13716
    },
    {
      "epoch": 0.13717,
      "grad_norm": 0.9743690945461065,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 13717
    },
    {
      "epoch": 0.13718,
      "grad_norm": 1.4056823819656654,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 13718
    },
    {
      "epoch": 0.13719,
      "grad_norm": 1.2471282630318425,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 13719
    },
    {
      "epoch": 0.1372,
      "grad_norm": 1.1003533404374917,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 13720
    },
    {
      "epoch": 0.13721,
      "grad_norm": 1.2369300729489772,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 13721
    },
    {
      "epoch": 0.13722,
      "grad_norm": 1.0168342614465826,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 13722
    },
    {
      "epoch": 0.13723,
      "grad_norm": 1.2102545293950289,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 13723
    },
    {
      "epoch": 0.13724,
      "grad_norm": 1.0154559928282219,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 13724
    },
    {
      "epoch": 0.13725,
      "grad_norm": 1.3594160651554588,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 13725
    },
    {
      "epoch": 0.13726,
      "grad_norm": 0.8551637541897258,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 13726
    },
    {
      "epoch": 0.13727,
      "grad_norm": 1.017273693667817,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 13727
    },
    {
      "epoch": 0.13728,
      "grad_norm": 1.4490405897764906,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 13728
    },
    {
      "epoch": 0.13729,
      "grad_norm": 1.2732754804061845,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 13729
    },
    {
      "epoch": 0.1373,
      "grad_norm": 1.3607592856126363,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 13730
    },
    {
      "epoch": 0.13731,
      "grad_norm": 1.4082351054771112,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 13731
    },
    {
      "epoch": 0.13732,
      "grad_norm": 1.416419540367466,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 13732
    },
    {
      "epoch": 0.13733,
      "grad_norm": 1.2685878698524238,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 13733
    },
    {
      "epoch": 0.13734,
      "grad_norm": 1.207884398659683,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 13734
    },
    {
      "epoch": 0.13735,
      "grad_norm": 1.10048154946459,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 13735
    },
    {
      "epoch": 0.13736,
      "grad_norm": 1.1877902199973267,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 13736
    },
    {
      "epoch": 0.13737,
      "grad_norm": 0.9533065968877271,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 13737
    },
    {
      "epoch": 0.13738,
      "grad_norm": 1.1984993751824762,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 13738
    },
    {
      "epoch": 0.13739,
      "grad_norm": 1.1171422689859538,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 13739
    },
    {
      "epoch": 0.1374,
      "grad_norm": 1.1911646618176164,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 13740
    },
    {
      "epoch": 0.13741,
      "grad_norm": 1.6440883465028686,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 13741
    },
    {
      "epoch": 0.13742,
      "grad_norm": 0.8918587280721783,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 13742
    },
    {
      "epoch": 0.13743,
      "grad_norm": 1.2441672699053672,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 13743
    },
    {
      "epoch": 0.13744,
      "grad_norm": 1.2075110129430175,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 13744
    },
    {
      "epoch": 0.13745,
      "grad_norm": 1.3501456347761631,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 13745
    },
    {
      "epoch": 0.13746,
      "grad_norm": 1.0841040758697305,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 13746
    },
    {
      "epoch": 0.13747,
      "grad_norm": 0.9834614633510931,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 13747
    },
    {
      "epoch": 0.13748,
      "grad_norm": 1.276521807710027,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 13748
    },
    {
      "epoch": 0.13749,
      "grad_norm": 1.1909654186897163,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 13749
    },
    {
      "epoch": 0.1375,
      "grad_norm": 1.2622123414540236,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 13750
    },
    {
      "epoch": 0.13751,
      "grad_norm": 1.0297963826688237,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 13751
    },
    {
      "epoch": 0.13752,
      "grad_norm": 1.2260649577899476,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 13752
    },
    {
      "epoch": 0.13753,
      "grad_norm": 1.2058677000367255,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 13753
    },
    {
      "epoch": 0.13754,
      "grad_norm": 1.1248417684321146,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 13754
    },
    {
      "epoch": 0.13755,
      "grad_norm": 1.2510522500262307,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 13755
    },
    {
      "epoch": 0.13756,
      "grad_norm": 1.1205481519257474,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 13756
    },
    {
      "epoch": 0.13757,
      "grad_norm": 1.239229758552996,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 13757
    },
    {
      "epoch": 0.13758,
      "grad_norm": 1.0140626079928357,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 13758
    },
    {
      "epoch": 0.13759,
      "grad_norm": 1.3690637911522001,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 13759
    },
    {
      "epoch": 0.1376,
      "grad_norm": 0.9507319118956362,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 13760
    },
    {
      "epoch": 0.13761,
      "grad_norm": 1.2483510717007327,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 13761
    },
    {
      "epoch": 0.13762,
      "grad_norm": 1.4389340645739723,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 13762
    },
    {
      "epoch": 0.13763,
      "grad_norm": 1.286178570489444,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 13763
    },
    {
      "epoch": 0.13764,
      "grad_norm": 1.1319149093379803,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 13764
    },
    {
      "epoch": 0.13765,
      "grad_norm": 1.2618229318089147,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 13765
    },
    {
      "epoch": 0.13766,
      "grad_norm": 1.1658637765032893,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 13766
    },
    {
      "epoch": 0.13767,
      "grad_norm": 1.18681907560757,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 13767
    },
    {
      "epoch": 0.13768,
      "grad_norm": 1.5055157878611105,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 13768
    },
    {
      "epoch": 0.13769,
      "grad_norm": 1.0051737452007148,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 13769
    },
    {
      "epoch": 0.1377,
      "grad_norm": 1.229234248921126,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 13770
    },
    {
      "epoch": 0.13771,
      "grad_norm": 1.0040775586458914,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 13771
    },
    {
      "epoch": 0.13772,
      "grad_norm": 1.4109407557293536,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 13772
    },
    {
      "epoch": 0.13773,
      "grad_norm": 1.3305022959252513,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 13773
    },
    {
      "epoch": 0.13774,
      "grad_norm": 1.2595580781737046,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 13774
    },
    {
      "epoch": 0.13775,
      "grad_norm": 1.0225571862727028,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 13775
    },
    {
      "epoch": 0.13776,
      "grad_norm": 1.0741113836973304,
      "learning_rate": 0.003,
      "loss": 4.0188,
      "step": 13776
    },
    {
      "epoch": 0.13777,
      "grad_norm": 1.0102454669852285,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 13777
    },
    {
      "epoch": 0.13778,
      "grad_norm": 1.298853127007687,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 13778
    },
    {
      "epoch": 0.13779,
      "grad_norm": 1.193382865103929,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 13779
    },
    {
      "epoch": 0.1378,
      "grad_norm": 1.6791272922067941,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 13780
    },
    {
      "epoch": 0.13781,
      "grad_norm": 1.1043189830034508,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 13781
    },
    {
      "epoch": 0.13782,
      "grad_norm": 1.0874160542628672,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 13782
    },
    {
      "epoch": 0.13783,
      "grad_norm": 1.2642927731824327,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 13783
    },
    {
      "epoch": 0.13784,
      "grad_norm": 1.010104274695837,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 13784
    },
    {
      "epoch": 0.13785,
      "grad_norm": 1.1285935067299098,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 13785
    },
    {
      "epoch": 0.13786,
      "grad_norm": 1.2335300965208245,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 13786
    },
    {
      "epoch": 0.13787,
      "grad_norm": 1.2342508255116131,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 13787
    },
    {
      "epoch": 0.13788,
      "grad_norm": 1.0984236689135678,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 13788
    },
    {
      "epoch": 0.13789,
      "grad_norm": 1.1947137028784856,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 13789
    },
    {
      "epoch": 0.1379,
      "grad_norm": 1.2858819082052524,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 13790
    },
    {
      "epoch": 0.13791,
      "grad_norm": 1.0612760980741958,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 13791
    },
    {
      "epoch": 0.13792,
      "grad_norm": 1.315051291837711,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 13792
    },
    {
      "epoch": 0.13793,
      "grad_norm": 1.2945660930882192,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 13793
    },
    {
      "epoch": 0.13794,
      "grad_norm": 1.1064503623568218,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 13794
    },
    {
      "epoch": 0.13795,
      "grad_norm": 1.1363722862351189,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 13795
    },
    {
      "epoch": 0.13796,
      "grad_norm": 1.2670966849911767,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 13796
    },
    {
      "epoch": 0.13797,
      "grad_norm": 1.2022932708678868,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 13797
    },
    {
      "epoch": 0.13798,
      "grad_norm": 1.3045342419357953,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 13798
    },
    {
      "epoch": 0.13799,
      "grad_norm": 1.1236873174766076,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 13799
    },
    {
      "epoch": 0.138,
      "grad_norm": 1.3005949834912098,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 13800
    },
    {
      "epoch": 0.13801,
      "grad_norm": 1.053257699141777,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 13801
    },
    {
      "epoch": 0.13802,
      "grad_norm": 1.2482038722674353,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 13802
    },
    {
      "epoch": 0.13803,
      "grad_norm": 1.122974392828506,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13803
    },
    {
      "epoch": 0.13804,
      "grad_norm": 1.4588114593504007,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 13804
    },
    {
      "epoch": 0.13805,
      "grad_norm": 1.2518257379913156,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 13805
    },
    {
      "epoch": 0.13806,
      "grad_norm": 1.0687988535135047,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 13806
    },
    {
      "epoch": 0.13807,
      "grad_norm": 1.3961170728719694,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 13807
    },
    {
      "epoch": 0.13808,
      "grad_norm": 1.2103978593187432,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 13808
    },
    {
      "epoch": 0.13809,
      "grad_norm": 1.2205250904177936,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 13809
    },
    {
      "epoch": 0.1381,
      "grad_norm": 1.144928087725286,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 13810
    },
    {
      "epoch": 0.13811,
      "grad_norm": 1.243183918413564,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 13811
    },
    {
      "epoch": 0.13812,
      "grad_norm": 1.1858159425116392,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 13812
    },
    {
      "epoch": 0.13813,
      "grad_norm": 1.2392840620813086,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 13813
    },
    {
      "epoch": 0.13814,
      "grad_norm": 1.1961079829938566,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 13814
    },
    {
      "epoch": 0.13815,
      "grad_norm": 1.2286148445830436,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 13815
    },
    {
      "epoch": 0.13816,
      "grad_norm": 1.059480739896095,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 13816
    },
    {
      "epoch": 0.13817,
      "grad_norm": 1.45459892002146,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 13817
    },
    {
      "epoch": 0.13818,
      "grad_norm": 0.8566130369181839,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 13818
    },
    {
      "epoch": 0.13819,
      "grad_norm": 1.1883964343787299,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 13819
    },
    {
      "epoch": 0.1382,
      "grad_norm": 1.1835827104338559,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 13820
    },
    {
      "epoch": 0.13821,
      "grad_norm": 1.2119907320687489,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 13821
    },
    {
      "epoch": 0.13822,
      "grad_norm": 1.0867692976875394,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 13822
    },
    {
      "epoch": 0.13823,
      "grad_norm": 1.0277948363309009,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 13823
    },
    {
      "epoch": 0.13824,
      "grad_norm": 1.3087394817360376,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 13824
    },
    {
      "epoch": 0.13825,
      "grad_norm": 1.3231505662393632,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 13825
    },
    {
      "epoch": 0.13826,
      "grad_norm": 1.0199844943105496,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 13826
    },
    {
      "epoch": 0.13827,
      "grad_norm": 1.2483024765730963,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 13827
    },
    {
      "epoch": 0.13828,
      "grad_norm": 0.9830237717557309,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 13828
    },
    {
      "epoch": 0.13829,
      "grad_norm": 1.1549798333543255,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 13829
    },
    {
      "epoch": 0.1383,
      "grad_norm": 1.2256141253549033,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 13830
    },
    {
      "epoch": 0.13831,
      "grad_norm": 1.3893703266288122,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 13831
    },
    {
      "epoch": 0.13832,
      "grad_norm": 1.04461634398326,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 13832
    },
    {
      "epoch": 0.13833,
      "grad_norm": 1.2570903342063884,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 13833
    },
    {
      "epoch": 0.13834,
      "grad_norm": 1.2108687512011658,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 13834
    },
    {
      "epoch": 0.13835,
      "grad_norm": 1.1861073808987619,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 13835
    },
    {
      "epoch": 0.13836,
      "grad_norm": 1.2103893164541015,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 13836
    },
    {
      "epoch": 0.13837,
      "grad_norm": 1.1816006360159812,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 13837
    },
    {
      "epoch": 0.13838,
      "grad_norm": 1.0234266452644853,
      "learning_rate": 0.003,
      "loss": 4.0296,
      "step": 13838
    },
    {
      "epoch": 0.13839,
      "grad_norm": 1.1374835719429075,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 13839
    },
    {
      "epoch": 0.1384,
      "grad_norm": 1.113220292818902,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 13840
    },
    {
      "epoch": 0.13841,
      "grad_norm": 1.2108498192604105,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 13841
    },
    {
      "epoch": 0.13842,
      "grad_norm": 1.182047080437874,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 13842
    },
    {
      "epoch": 0.13843,
      "grad_norm": 1.0365640423934581,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 13843
    },
    {
      "epoch": 0.13844,
      "grad_norm": 1.2120314765030384,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 13844
    },
    {
      "epoch": 0.13845,
      "grad_norm": 1.4391785218167055,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 13845
    },
    {
      "epoch": 0.13846,
      "grad_norm": 1.398202203052698,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 13846
    },
    {
      "epoch": 0.13847,
      "grad_norm": 1.2374107107581187,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 13847
    },
    {
      "epoch": 0.13848,
      "grad_norm": 1.142601397450018,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 13848
    },
    {
      "epoch": 0.13849,
      "grad_norm": 1.4394250915044238,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 13849
    },
    {
      "epoch": 0.1385,
      "grad_norm": 0.9071756635676798,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 13850
    },
    {
      "epoch": 0.13851,
      "grad_norm": 1.155898767954042,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 13851
    },
    {
      "epoch": 0.13852,
      "grad_norm": 1.3854446050083606,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 13852
    },
    {
      "epoch": 0.13853,
      "grad_norm": 1.3192226385987496,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 13853
    },
    {
      "epoch": 0.13854,
      "grad_norm": 1.2084015385701228,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 13854
    },
    {
      "epoch": 0.13855,
      "grad_norm": 1.25301652300438,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 13855
    },
    {
      "epoch": 0.13856,
      "grad_norm": 1.0665196097239276,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 13856
    },
    {
      "epoch": 0.13857,
      "grad_norm": 1.1618993258053154,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 13857
    },
    {
      "epoch": 0.13858,
      "grad_norm": 1.3561969206027245,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 13858
    },
    {
      "epoch": 0.13859,
      "grad_norm": 1.091509768863945,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 13859
    },
    {
      "epoch": 0.1386,
      "grad_norm": 1.1411802259550208,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 13860
    },
    {
      "epoch": 0.13861,
      "grad_norm": 1.1427719793225433,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 13861
    },
    {
      "epoch": 0.13862,
      "grad_norm": 1.1738927424268664,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 13862
    },
    {
      "epoch": 0.13863,
      "grad_norm": 1.1997101675962227,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 13863
    },
    {
      "epoch": 0.13864,
      "grad_norm": 1.1630761787979964,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 13864
    },
    {
      "epoch": 0.13865,
      "grad_norm": 1.139098753069408,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 13865
    },
    {
      "epoch": 0.13866,
      "grad_norm": 1.1557888439300514,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 13866
    },
    {
      "epoch": 0.13867,
      "grad_norm": 1.2623839567530815,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 13867
    },
    {
      "epoch": 0.13868,
      "grad_norm": 1.1837947341599429,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 13868
    },
    {
      "epoch": 0.13869,
      "grad_norm": 1.4933916759250025,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 13869
    },
    {
      "epoch": 0.1387,
      "grad_norm": 1.104622387180865,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 13870
    },
    {
      "epoch": 0.13871,
      "grad_norm": 1.2509532855231675,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 13871
    },
    {
      "epoch": 0.13872,
      "grad_norm": 1.3209278407919876,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 13872
    },
    {
      "epoch": 0.13873,
      "grad_norm": 1.1340281379719,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 13873
    },
    {
      "epoch": 0.13874,
      "grad_norm": 1.2491776894445892,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 13874
    },
    {
      "epoch": 0.13875,
      "grad_norm": 0.9903960367573045,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 13875
    },
    {
      "epoch": 0.13876,
      "grad_norm": 1.352282192489039,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 13876
    },
    {
      "epoch": 0.13877,
      "grad_norm": 1.0005486288387502,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 13877
    },
    {
      "epoch": 0.13878,
      "grad_norm": 1.4365204816386525,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 13878
    },
    {
      "epoch": 0.13879,
      "grad_norm": 0.8805469147799052,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 13879
    },
    {
      "epoch": 0.1388,
      "grad_norm": 1.4887720150086994,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 13880
    },
    {
      "epoch": 0.13881,
      "grad_norm": 1.410762056398085,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 13881
    },
    {
      "epoch": 0.13882,
      "grad_norm": 1.3053880278160495,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 13882
    },
    {
      "epoch": 0.13883,
      "grad_norm": 1.3465284752908813,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 13883
    },
    {
      "epoch": 0.13884,
      "grad_norm": 1.2563042257227426,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 13884
    },
    {
      "epoch": 0.13885,
      "grad_norm": 1.2592782036434007,
      "learning_rate": 0.003,
      "loss": 4.1129,
      "step": 13885
    },
    {
      "epoch": 0.13886,
      "grad_norm": 1.0859532416025257,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 13886
    },
    {
      "epoch": 0.13887,
      "grad_norm": 1.491418353096719,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 13887
    },
    {
      "epoch": 0.13888,
      "grad_norm": 0.8654041738780719,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 13888
    },
    {
      "epoch": 0.13889,
      "grad_norm": 0.9210784584786764,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 13889
    },
    {
      "epoch": 0.1389,
      "grad_norm": 1.099693943923993,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 13890
    },
    {
      "epoch": 0.13891,
      "grad_norm": 1.1958793076649281,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 13891
    },
    {
      "epoch": 0.13892,
      "grad_norm": 1.0885220014006605,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 13892
    },
    {
      "epoch": 0.13893,
      "grad_norm": 1.2725398146978182,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 13893
    },
    {
      "epoch": 0.13894,
      "grad_norm": 1.2371715001604757,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 13894
    },
    {
      "epoch": 0.13895,
      "grad_norm": 1.4099389084174885,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 13895
    },
    {
      "epoch": 0.13896,
      "grad_norm": 1.0048792330671146,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 13896
    },
    {
      "epoch": 0.13897,
      "grad_norm": 1.1219108421533248,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 13897
    },
    {
      "epoch": 0.13898,
      "grad_norm": 1.1050201127531942,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 13898
    },
    {
      "epoch": 0.13899,
      "grad_norm": 1.3441208780893406,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 13899
    },
    {
      "epoch": 0.139,
      "grad_norm": 1.0851640605237316,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 13900
    },
    {
      "epoch": 0.13901,
      "grad_norm": 1.3455747589350209,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 13901
    },
    {
      "epoch": 0.13902,
      "grad_norm": 1.189938938341629,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 13902
    },
    {
      "epoch": 0.13903,
      "grad_norm": 1.12316810081831,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 13903
    },
    {
      "epoch": 0.13904,
      "grad_norm": 1.229408118975543,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 13904
    },
    {
      "epoch": 0.13905,
      "grad_norm": 1.0338220238270672,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 13905
    },
    {
      "epoch": 0.13906,
      "grad_norm": 1.320533597961583,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 13906
    },
    {
      "epoch": 0.13907,
      "grad_norm": 1.0975575164129683,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 13907
    },
    {
      "epoch": 0.13908,
      "grad_norm": 1.4485545309639958,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 13908
    },
    {
      "epoch": 0.13909,
      "grad_norm": 1.1849479646868737,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 13909
    },
    {
      "epoch": 0.1391,
      "grad_norm": 1.3591596916616118,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 13910
    },
    {
      "epoch": 0.13911,
      "grad_norm": 1.193285184647018,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 13911
    },
    {
      "epoch": 0.13912,
      "grad_norm": 1.2602725825475507,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 13912
    },
    {
      "epoch": 0.13913,
      "grad_norm": 1.1496508245124075,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 13913
    },
    {
      "epoch": 0.13914,
      "grad_norm": 1.1308946707113137,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 13914
    },
    {
      "epoch": 0.13915,
      "grad_norm": 1.0958359046447819,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 13915
    },
    {
      "epoch": 0.13916,
      "grad_norm": 1.260606596749469,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 13916
    },
    {
      "epoch": 0.13917,
      "grad_norm": 1.0727415215522746,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 13917
    },
    {
      "epoch": 0.13918,
      "grad_norm": 1.1109603984637593,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 13918
    },
    {
      "epoch": 0.13919,
      "grad_norm": 1.2333451431442168,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 13919
    },
    {
      "epoch": 0.1392,
      "grad_norm": 1.4377985912806581,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 13920
    },
    {
      "epoch": 0.13921,
      "grad_norm": 1.0271248694608675,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 13921
    },
    {
      "epoch": 0.13922,
      "grad_norm": 1.4524105427120009,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 13922
    },
    {
      "epoch": 0.13923,
      "grad_norm": 1.0931444063543732,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 13923
    },
    {
      "epoch": 0.13924,
      "grad_norm": 1.377676623237047,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 13924
    },
    {
      "epoch": 0.13925,
      "grad_norm": 1.1584891813511757,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 13925
    },
    {
      "epoch": 0.13926,
      "grad_norm": 1.464207200492954,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 13926
    },
    {
      "epoch": 0.13927,
      "grad_norm": 1.0267295476732703,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 13927
    },
    {
      "epoch": 0.13928,
      "grad_norm": 1.3087279077958458,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 13928
    },
    {
      "epoch": 0.13929,
      "grad_norm": 1.1030413931333924,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 13929
    },
    {
      "epoch": 0.1393,
      "grad_norm": 1.1963510192727531,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 13930
    },
    {
      "epoch": 0.13931,
      "grad_norm": 1.3552900115306397,
      "learning_rate": 0.003,
      "loss": 4.0109,
      "step": 13931
    },
    {
      "epoch": 0.13932,
      "grad_norm": 1.117443482134989,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 13932
    },
    {
      "epoch": 0.13933,
      "grad_norm": 1.4176894352836409,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 13933
    },
    {
      "epoch": 0.13934,
      "grad_norm": 0.9146997303307944,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 13934
    },
    {
      "epoch": 0.13935,
      "grad_norm": 1.1075004455081943,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 13935
    },
    {
      "epoch": 0.13936,
      "grad_norm": 1.3590514377512752,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 13936
    },
    {
      "epoch": 0.13937,
      "grad_norm": 1.2083401040806896,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 13937
    },
    {
      "epoch": 0.13938,
      "grad_norm": 0.9787905320757424,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 13938
    },
    {
      "epoch": 0.13939,
      "grad_norm": 1.251635670336556,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 13939
    },
    {
      "epoch": 0.1394,
      "grad_norm": 1.3134870525772442,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 13940
    },
    {
      "epoch": 0.13941,
      "grad_norm": 1.2293564795195282,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13941
    },
    {
      "epoch": 0.13942,
      "grad_norm": 1.060690054358625,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 13942
    },
    {
      "epoch": 0.13943,
      "grad_norm": 1.090686083165746,
      "learning_rate": 0.003,
      "loss": 4.1143,
      "step": 13943
    },
    {
      "epoch": 0.13944,
      "grad_norm": 1.301586014251149,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 13944
    },
    {
      "epoch": 0.13945,
      "grad_norm": 0.9965137088015933,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 13945
    },
    {
      "epoch": 0.13946,
      "grad_norm": 1.3698183694761605,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 13946
    },
    {
      "epoch": 0.13947,
      "grad_norm": 1.0772867287069972,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 13947
    },
    {
      "epoch": 0.13948,
      "grad_norm": 1.3466737330984049,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 13948
    },
    {
      "epoch": 0.13949,
      "grad_norm": 1.231606661963063,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 13949
    },
    {
      "epoch": 0.1395,
      "grad_norm": 1.1341519641675695,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 13950
    },
    {
      "epoch": 0.13951,
      "grad_norm": 1.1325294230135632,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 13951
    },
    {
      "epoch": 0.13952,
      "grad_norm": 0.995793562007214,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 13952
    },
    {
      "epoch": 0.13953,
      "grad_norm": 1.0873814056283444,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 13953
    },
    {
      "epoch": 0.13954,
      "grad_norm": 1.0919384599880155,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 13954
    },
    {
      "epoch": 0.13955,
      "grad_norm": 1.3122930160032409,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 13955
    },
    {
      "epoch": 0.13956,
      "grad_norm": 1.2936895953678114,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 13956
    },
    {
      "epoch": 0.13957,
      "grad_norm": 1.1638919567860373,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 13957
    },
    {
      "epoch": 0.13958,
      "grad_norm": 1.3219373768753289,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 13958
    },
    {
      "epoch": 0.13959,
      "grad_norm": 1.3180085740090488,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 13959
    },
    {
      "epoch": 0.1396,
      "grad_norm": 1.1477439233655948,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 13960
    },
    {
      "epoch": 0.13961,
      "grad_norm": 1.1681959988248531,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 13961
    },
    {
      "epoch": 0.13962,
      "grad_norm": 1.0433457251588483,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 13962
    },
    {
      "epoch": 0.13963,
      "grad_norm": 1.391823485061002,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 13963
    },
    {
      "epoch": 0.13964,
      "grad_norm": 1.0210707733989797,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 13964
    },
    {
      "epoch": 0.13965,
      "grad_norm": 1.4127300935023928,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 13965
    },
    {
      "epoch": 0.13966,
      "grad_norm": 1.1296663061712948,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 13966
    },
    {
      "epoch": 0.13967,
      "grad_norm": 1.2884345926712524,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 13967
    },
    {
      "epoch": 0.13968,
      "grad_norm": 1.2284051362680708,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 13968
    },
    {
      "epoch": 0.13969,
      "grad_norm": 1.168348777197498,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 13969
    },
    {
      "epoch": 0.1397,
      "grad_norm": 1.244855212775038,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 13970
    },
    {
      "epoch": 0.13971,
      "grad_norm": 1.2413849979162377,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 13971
    },
    {
      "epoch": 0.13972,
      "grad_norm": 1.254429545205275,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 13972
    },
    {
      "epoch": 0.13973,
      "grad_norm": 1.0510032319940372,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 13973
    },
    {
      "epoch": 0.13974,
      "grad_norm": 1.240022161138721,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 13974
    },
    {
      "epoch": 0.13975,
      "grad_norm": 1.1236021708137216,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 13975
    },
    {
      "epoch": 0.13976,
      "grad_norm": 1.282349149962916,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 13976
    },
    {
      "epoch": 0.13977,
      "grad_norm": 1.273656352075562,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 13977
    },
    {
      "epoch": 0.13978,
      "grad_norm": 1.731791547986483,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 13978
    },
    {
      "epoch": 0.13979,
      "grad_norm": 0.9412192924359677,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 13979
    },
    {
      "epoch": 0.1398,
      "grad_norm": 1.1107332178250728,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 13980
    },
    {
      "epoch": 0.13981,
      "grad_norm": 1.616234813264336,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 13981
    },
    {
      "epoch": 0.13982,
      "grad_norm": 0.9711871798503602,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 13982
    },
    {
      "epoch": 0.13983,
      "grad_norm": 1.1601552691916304,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 13983
    },
    {
      "epoch": 0.13984,
      "grad_norm": 1.2597883044670368,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 13984
    },
    {
      "epoch": 0.13985,
      "grad_norm": 1.2256076832195404,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 13985
    },
    {
      "epoch": 0.13986,
      "grad_norm": 1.1600366740074843,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 13986
    },
    {
      "epoch": 0.13987,
      "grad_norm": 1.3271071059940935,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 13987
    },
    {
      "epoch": 0.13988,
      "grad_norm": 1.0963944376064725,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 13988
    },
    {
      "epoch": 0.13989,
      "grad_norm": 1.1583167783910462,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 13989
    },
    {
      "epoch": 0.1399,
      "grad_norm": 1.2763012872823292,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 13990
    },
    {
      "epoch": 0.13991,
      "grad_norm": 1.1761647857400959,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 13991
    },
    {
      "epoch": 0.13992,
      "grad_norm": 1.284745873124637,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 13992
    },
    {
      "epoch": 0.13993,
      "grad_norm": 1.1017391051460177,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 13993
    },
    {
      "epoch": 0.13994,
      "grad_norm": 1.2918818495727538,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 13994
    },
    {
      "epoch": 0.13995,
      "grad_norm": 1.015154283188355,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 13995
    },
    {
      "epoch": 0.13996,
      "grad_norm": 1.1298280746391378,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 13996
    },
    {
      "epoch": 0.13997,
      "grad_norm": 1.0422809770483068,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 13997
    },
    {
      "epoch": 0.13998,
      "grad_norm": 1.3048376609139016,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 13998
    },
    {
      "epoch": 0.13999,
      "grad_norm": 1.158199861113384,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 13999
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4235527130575911,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 14000
    },
    {
      "epoch": 0.14001,
      "grad_norm": 1.1218686754587923,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 14001
    },
    {
      "epoch": 0.14002,
      "grad_norm": 1.073910869987476,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 14002
    },
    {
      "epoch": 0.14003,
      "grad_norm": 1.341400623868328,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 14003
    },
    {
      "epoch": 0.14004,
      "grad_norm": 1.0756948186119029,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 14004
    },
    {
      "epoch": 0.14005,
      "grad_norm": 1.2865374378748764,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 14005
    },
    {
      "epoch": 0.14006,
      "grad_norm": 1.0091892120642572,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 14006
    },
    {
      "epoch": 0.14007,
      "grad_norm": 1.199666970298566,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 14007
    },
    {
      "epoch": 0.14008,
      "grad_norm": 1.1846263521850238,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 14008
    },
    {
      "epoch": 0.14009,
      "grad_norm": 1.1943126712677172,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 14009
    },
    {
      "epoch": 0.1401,
      "grad_norm": 1.1383321883966389,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 14010
    },
    {
      "epoch": 0.14011,
      "grad_norm": 1.2282899338116846,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 14011
    },
    {
      "epoch": 0.14012,
      "grad_norm": 1.0289904331903552,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 14012
    },
    {
      "epoch": 0.14013,
      "grad_norm": 1.238851255206333,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 14013
    },
    {
      "epoch": 0.14014,
      "grad_norm": 1.235452537507704,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 14014
    },
    {
      "epoch": 0.14015,
      "grad_norm": 1.1040128763356043,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 14015
    },
    {
      "epoch": 0.14016,
      "grad_norm": 1.3077179023264822,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 14016
    },
    {
      "epoch": 0.14017,
      "grad_norm": 1.070994763136842,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 14017
    },
    {
      "epoch": 0.14018,
      "grad_norm": 1.5795010445329243,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 14018
    },
    {
      "epoch": 0.14019,
      "grad_norm": 1.110038500243441,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 14019
    },
    {
      "epoch": 0.1402,
      "grad_norm": 1.282640788277247,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 14020
    },
    {
      "epoch": 0.14021,
      "grad_norm": 1.297107847097581,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 14021
    },
    {
      "epoch": 0.14022,
      "grad_norm": 1.1023765488365382,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 14022
    },
    {
      "epoch": 0.14023,
      "grad_norm": 1.460130837344323,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 14023
    },
    {
      "epoch": 0.14024,
      "grad_norm": 1.2242712859475282,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 14024
    },
    {
      "epoch": 0.14025,
      "grad_norm": 1.1691008234239406,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 14025
    },
    {
      "epoch": 0.14026,
      "grad_norm": 1.2835431334905392,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 14026
    },
    {
      "epoch": 0.14027,
      "grad_norm": 1.182980232383101,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 14027
    },
    {
      "epoch": 0.14028,
      "grad_norm": 1.2690266061757915,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 14028
    },
    {
      "epoch": 0.14029,
      "grad_norm": 1.0544734376417177,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 14029
    },
    {
      "epoch": 0.1403,
      "grad_norm": 1.3424873029789186,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 14030
    },
    {
      "epoch": 0.14031,
      "grad_norm": 1.0456050868891826,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 14031
    },
    {
      "epoch": 0.14032,
      "grad_norm": 1.261805056951264,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 14032
    },
    {
      "epoch": 0.14033,
      "grad_norm": 0.980685828701439,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 14033
    },
    {
      "epoch": 0.14034,
      "grad_norm": 1.4253951982871123,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 14034
    },
    {
      "epoch": 0.14035,
      "grad_norm": 1.090918477219447,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 14035
    },
    {
      "epoch": 0.14036,
      "grad_norm": 1.2198947919818297,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 14036
    },
    {
      "epoch": 0.14037,
      "grad_norm": 1.1548789215429767,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 14037
    },
    {
      "epoch": 0.14038,
      "grad_norm": 1.194905070744935,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 14038
    },
    {
      "epoch": 0.14039,
      "grad_norm": 1.335307512644602,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 14039
    },
    {
      "epoch": 0.1404,
      "grad_norm": 1.3541627094142514,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 14040
    },
    {
      "epoch": 0.14041,
      "grad_norm": 1.2668978861497087,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 14041
    },
    {
      "epoch": 0.14042,
      "grad_norm": 1.0961704138846822,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 14042
    },
    {
      "epoch": 0.14043,
      "grad_norm": 1.310633574915397,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 14043
    },
    {
      "epoch": 0.14044,
      "grad_norm": 1.3539709644080737,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 14044
    },
    {
      "epoch": 0.14045,
      "grad_norm": 1.1545400183981318,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 14045
    },
    {
      "epoch": 0.14046,
      "grad_norm": 1.3388500028882995,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 14046
    },
    {
      "epoch": 0.14047,
      "grad_norm": 0.9722385197522674,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 14047
    },
    {
      "epoch": 0.14048,
      "grad_norm": 1.3184871593401073,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 14048
    },
    {
      "epoch": 0.14049,
      "grad_norm": 1.0804800429589527,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 14049
    },
    {
      "epoch": 0.1405,
      "grad_norm": 1.4357860866189762,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 14050
    },
    {
      "epoch": 0.14051,
      "grad_norm": 1.1952804183318617,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 14051
    },
    {
      "epoch": 0.14052,
      "grad_norm": 1.4024191113504245,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 14052
    },
    {
      "epoch": 0.14053,
      "grad_norm": 1.0332715115039801,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 14053
    },
    {
      "epoch": 0.14054,
      "grad_norm": 1.2680229785791972,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 14054
    },
    {
      "epoch": 0.14055,
      "grad_norm": 1.220671894936996,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 14055
    },
    {
      "epoch": 0.14056,
      "grad_norm": 1.2233171041806348,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 14056
    },
    {
      "epoch": 0.14057,
      "grad_norm": 1.3157773723684758,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 14057
    },
    {
      "epoch": 0.14058,
      "grad_norm": 1.124405796378944,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 14058
    },
    {
      "epoch": 0.14059,
      "grad_norm": 1.0499379920133691,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 14059
    },
    {
      "epoch": 0.1406,
      "grad_norm": 1.3067068364078773,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 14060
    },
    {
      "epoch": 0.14061,
      "grad_norm": 1.2410227155176823,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 14061
    },
    {
      "epoch": 0.14062,
      "grad_norm": 1.0952452557827719,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 14062
    },
    {
      "epoch": 0.14063,
      "grad_norm": 1.1135075479715215,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 14063
    },
    {
      "epoch": 0.14064,
      "grad_norm": 1.1132489274054158,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 14064
    },
    {
      "epoch": 0.14065,
      "grad_norm": 1.2691091680210476,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 14065
    },
    {
      "epoch": 0.14066,
      "grad_norm": 1.1945059279642607,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 14066
    },
    {
      "epoch": 0.14067,
      "grad_norm": 1.0849499469417416,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 14067
    },
    {
      "epoch": 0.14068,
      "grad_norm": 1.4788611387245632,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 14068
    },
    {
      "epoch": 0.14069,
      "grad_norm": 1.1060156621939714,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 14069
    },
    {
      "epoch": 0.1407,
      "grad_norm": 1.2579130611915266,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 14070
    },
    {
      "epoch": 0.14071,
      "grad_norm": 1.0615390463572114,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 14071
    },
    {
      "epoch": 0.14072,
      "grad_norm": 1.290325586202983,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 14072
    },
    {
      "epoch": 0.14073,
      "grad_norm": 0.9851816152194139,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 14073
    },
    {
      "epoch": 0.14074,
      "grad_norm": 1.3244838979814133,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 14074
    },
    {
      "epoch": 0.14075,
      "grad_norm": 0.9445330443948831,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 14075
    },
    {
      "epoch": 0.14076,
      "grad_norm": 1.1164429458611043,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 14076
    },
    {
      "epoch": 0.14077,
      "grad_norm": 1.267983868741483,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 14077
    },
    {
      "epoch": 0.14078,
      "grad_norm": 1.1928263559105565,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 14078
    },
    {
      "epoch": 0.14079,
      "grad_norm": 1.122771365310409,
      "learning_rate": 0.003,
      "loss": 4.0333,
      "step": 14079
    },
    {
      "epoch": 0.1408,
      "grad_norm": 1.017592936371798,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 14080
    },
    {
      "epoch": 0.14081,
      "grad_norm": 1.449617758954173,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 14081
    },
    {
      "epoch": 0.14082,
      "grad_norm": 1.1594842458389811,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 14082
    },
    {
      "epoch": 0.14083,
      "grad_norm": 1.257819534262536,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 14083
    },
    {
      "epoch": 0.14084,
      "grad_norm": 1.381279265951169,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 14084
    },
    {
      "epoch": 0.14085,
      "grad_norm": 1.0803274942689212,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 14085
    },
    {
      "epoch": 0.14086,
      "grad_norm": 1.1315871871624874,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 14086
    },
    {
      "epoch": 0.14087,
      "grad_norm": 1.1748763201060617,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 14087
    },
    {
      "epoch": 0.14088,
      "grad_norm": 1.2618259030019516,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 14088
    },
    {
      "epoch": 0.14089,
      "grad_norm": 1.2343867938713182,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 14089
    },
    {
      "epoch": 0.1409,
      "grad_norm": 0.9682727798543243,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 14090
    },
    {
      "epoch": 0.14091,
      "grad_norm": 1.4059866113262258,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 14091
    },
    {
      "epoch": 0.14092,
      "grad_norm": 1.054421179290126,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 14092
    },
    {
      "epoch": 0.14093,
      "grad_norm": 1.351264856046475,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 14093
    },
    {
      "epoch": 0.14094,
      "grad_norm": 1.1926974079925088,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 14094
    },
    {
      "epoch": 0.14095,
      "grad_norm": 1.199123877993281,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 14095
    },
    {
      "epoch": 0.14096,
      "grad_norm": 1.416646770179057,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 14096
    },
    {
      "epoch": 0.14097,
      "grad_norm": 1.0840995665200288,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 14097
    },
    {
      "epoch": 0.14098,
      "grad_norm": 1.4593143604911771,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 14098
    },
    {
      "epoch": 0.14099,
      "grad_norm": 1.372636577899513,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 14099
    },
    {
      "epoch": 0.141,
      "grad_norm": 1.204041439925,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 14100
    },
    {
      "epoch": 0.14101,
      "grad_norm": 1.2853867147082851,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 14101
    },
    {
      "epoch": 0.14102,
      "grad_norm": 1.0729750244090508,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 14102
    },
    {
      "epoch": 0.14103,
      "grad_norm": 1.328516026937791,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 14103
    },
    {
      "epoch": 0.14104,
      "grad_norm": 1.4265493338851565,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 14104
    },
    {
      "epoch": 0.14105,
      "grad_norm": 1.2046846966356402,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 14105
    },
    {
      "epoch": 0.14106,
      "grad_norm": 1.0158974556644869,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 14106
    },
    {
      "epoch": 0.14107,
      "grad_norm": 1.4414729489269356,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 14107
    },
    {
      "epoch": 0.14108,
      "grad_norm": 0.9747188424913261,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 14108
    },
    {
      "epoch": 0.14109,
      "grad_norm": 1.117458651506205,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 14109
    },
    {
      "epoch": 0.1411,
      "grad_norm": 1.1707868125332495,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 14110
    },
    {
      "epoch": 0.14111,
      "grad_norm": 1.1546525831558394,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 14111
    },
    {
      "epoch": 0.14112,
      "grad_norm": 1.1526070159060042,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 14112
    },
    {
      "epoch": 0.14113,
      "grad_norm": 1.1098644078796183,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 14113
    },
    {
      "epoch": 0.14114,
      "grad_norm": 1.0113903919831815,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 14114
    },
    {
      "epoch": 0.14115,
      "grad_norm": 1.2719335196897958,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 14115
    },
    {
      "epoch": 0.14116,
      "grad_norm": 1.0879888158201663,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 14116
    },
    {
      "epoch": 0.14117,
      "grad_norm": 1.4061852816368725,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 14117
    },
    {
      "epoch": 0.14118,
      "grad_norm": 1.155127036615318,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 14118
    },
    {
      "epoch": 0.14119,
      "grad_norm": 1.4390313206171075,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 14119
    },
    {
      "epoch": 0.1412,
      "grad_norm": 1.2243146873664432,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 14120
    },
    {
      "epoch": 0.14121,
      "grad_norm": 1.3477283252085275,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 14121
    },
    {
      "epoch": 0.14122,
      "grad_norm": 1.244334698337371,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 14122
    },
    {
      "epoch": 0.14123,
      "grad_norm": 1.1362973216272718,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 14123
    },
    {
      "epoch": 0.14124,
      "grad_norm": 1.1199342054580403,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 14124
    },
    {
      "epoch": 0.14125,
      "grad_norm": 1.1351963229862383,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 14125
    },
    {
      "epoch": 0.14126,
      "grad_norm": 1.2885471864534528,
      "learning_rate": 0.003,
      "loss": 4.0279,
      "step": 14126
    },
    {
      "epoch": 0.14127,
      "grad_norm": 1.0621581921969891,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14127
    },
    {
      "epoch": 0.14128,
      "grad_norm": 1.5233796698785904,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 14128
    },
    {
      "epoch": 0.14129,
      "grad_norm": 1.0809104836398291,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 14129
    },
    {
      "epoch": 0.1413,
      "grad_norm": 1.2136014062527531,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 14130
    },
    {
      "epoch": 0.14131,
      "grad_norm": 1.3219523759291874,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 14131
    },
    {
      "epoch": 0.14132,
      "grad_norm": 1.0915606916320713,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 14132
    },
    {
      "epoch": 0.14133,
      "grad_norm": 1.1791363407813595,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 14133
    },
    {
      "epoch": 0.14134,
      "grad_norm": 1.1719934454427836,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 14134
    },
    {
      "epoch": 0.14135,
      "grad_norm": 1.2671170437913437,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 14135
    },
    {
      "epoch": 0.14136,
      "grad_norm": 1.2002924475133558,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 14136
    },
    {
      "epoch": 0.14137,
      "grad_norm": 1.2703077591986773,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 14137
    },
    {
      "epoch": 0.14138,
      "grad_norm": 1.2298921842905388,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 14138
    },
    {
      "epoch": 0.14139,
      "grad_norm": 1.423085309232018,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 14139
    },
    {
      "epoch": 0.1414,
      "grad_norm": 1.0237604249414567,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 14140
    },
    {
      "epoch": 0.14141,
      "grad_norm": 1.2418351630645756,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 14141
    },
    {
      "epoch": 0.14142,
      "grad_norm": 1.3108708654007848,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 14142
    },
    {
      "epoch": 0.14143,
      "grad_norm": 1.1746333607980173,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 14143
    },
    {
      "epoch": 0.14144,
      "grad_norm": 1.2919370602961495,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 14144
    },
    {
      "epoch": 0.14145,
      "grad_norm": 1.020497759641275,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 14145
    },
    {
      "epoch": 0.14146,
      "grad_norm": 1.366765879723746,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 14146
    },
    {
      "epoch": 0.14147,
      "grad_norm": 1.068908212949406,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 14147
    },
    {
      "epoch": 0.14148,
      "grad_norm": 1.0305235636939665,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 14148
    },
    {
      "epoch": 0.14149,
      "grad_norm": 1.2279229983109714,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 14149
    },
    {
      "epoch": 0.1415,
      "grad_norm": 1.0316938377792986,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 14150
    },
    {
      "epoch": 0.14151,
      "grad_norm": 1.377595721666909,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 14151
    },
    {
      "epoch": 0.14152,
      "grad_norm": 1.1522508689722177,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 14152
    },
    {
      "epoch": 0.14153,
      "grad_norm": 1.2934979529418558,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 14153
    },
    {
      "epoch": 0.14154,
      "grad_norm": 1.037657039612129,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 14154
    },
    {
      "epoch": 0.14155,
      "grad_norm": 1.4118107667174071,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 14155
    },
    {
      "epoch": 0.14156,
      "grad_norm": 0.9756006396870087,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 14156
    },
    {
      "epoch": 0.14157,
      "grad_norm": 1.19230919409468,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 14157
    },
    {
      "epoch": 0.14158,
      "grad_norm": 1.2705488618993246,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 14158
    },
    {
      "epoch": 0.14159,
      "grad_norm": 1.1206704851826939,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 14159
    },
    {
      "epoch": 0.1416,
      "grad_norm": 1.3111408934130488,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 14160
    },
    {
      "epoch": 0.14161,
      "grad_norm": 0.9504522977031813,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 14161
    },
    {
      "epoch": 0.14162,
      "grad_norm": 1.188242443855758,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 14162
    },
    {
      "epoch": 0.14163,
      "grad_norm": 1.2109152723685594,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 14163
    },
    {
      "epoch": 0.14164,
      "grad_norm": 1.1758764862974531,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 14164
    },
    {
      "epoch": 0.14165,
      "grad_norm": 1.1963761766831424,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 14165
    },
    {
      "epoch": 0.14166,
      "grad_norm": 1.2296011716516042,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 14166
    },
    {
      "epoch": 0.14167,
      "grad_norm": 1.1056886732813094,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 14167
    },
    {
      "epoch": 0.14168,
      "grad_norm": 1.1019334916953336,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 14168
    },
    {
      "epoch": 0.14169,
      "grad_norm": 1.1489295640476946,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 14169
    },
    {
      "epoch": 0.1417,
      "grad_norm": 1.034958735378265,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 14170
    },
    {
      "epoch": 0.14171,
      "grad_norm": 1.164901638667101,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 14171
    },
    {
      "epoch": 0.14172,
      "grad_norm": 1.3051868501269035,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 14172
    },
    {
      "epoch": 0.14173,
      "grad_norm": 1.6416928856078798,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 14173
    },
    {
      "epoch": 0.14174,
      "grad_norm": 0.992803965415215,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 14174
    },
    {
      "epoch": 0.14175,
      "grad_norm": 1.2586048849110307,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 14175
    },
    {
      "epoch": 0.14176,
      "grad_norm": 1.3636801383756292,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 14176
    },
    {
      "epoch": 0.14177,
      "grad_norm": 1.3046002495581674,
      "learning_rate": 0.003,
      "loss": 4.1157,
      "step": 14177
    },
    {
      "epoch": 0.14178,
      "grad_norm": 1.0368564735366113,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 14178
    },
    {
      "epoch": 0.14179,
      "grad_norm": 1.2192823265187933,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 14179
    },
    {
      "epoch": 0.1418,
      "grad_norm": 1.2397683265384485,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 14180
    },
    {
      "epoch": 0.14181,
      "grad_norm": 1.2463906763765813,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 14181
    },
    {
      "epoch": 0.14182,
      "grad_norm": 1.280643109256286,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 14182
    },
    {
      "epoch": 0.14183,
      "grad_norm": 1.1470938139266507,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 14183
    },
    {
      "epoch": 0.14184,
      "grad_norm": 1.2085749166227202,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 14184
    },
    {
      "epoch": 0.14185,
      "grad_norm": 1.2226127200246437,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 14185
    },
    {
      "epoch": 0.14186,
      "grad_norm": 1.1675834090908574,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 14186
    },
    {
      "epoch": 0.14187,
      "grad_norm": 1.273509064648255,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 14187
    },
    {
      "epoch": 0.14188,
      "grad_norm": 1.220538086442717,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 14188
    },
    {
      "epoch": 0.14189,
      "grad_norm": 1.27082425692112,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 14189
    },
    {
      "epoch": 0.1419,
      "grad_norm": 1.2715796466945537,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 14190
    },
    {
      "epoch": 0.14191,
      "grad_norm": 1.0415153909241641,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 14191
    },
    {
      "epoch": 0.14192,
      "grad_norm": 1.2253256958990855,
      "learning_rate": 0.003,
      "loss": 4.1139,
      "step": 14192
    },
    {
      "epoch": 0.14193,
      "grad_norm": 1.03874298384842,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 14193
    },
    {
      "epoch": 0.14194,
      "grad_norm": 1.1986931317570426,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 14194
    },
    {
      "epoch": 0.14195,
      "grad_norm": 1.1557863525220406,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 14195
    },
    {
      "epoch": 0.14196,
      "grad_norm": 1.0770610880955502,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 14196
    },
    {
      "epoch": 0.14197,
      "grad_norm": 1.132517261030939,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 14197
    },
    {
      "epoch": 0.14198,
      "grad_norm": 1.3532234854028709,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 14198
    },
    {
      "epoch": 0.14199,
      "grad_norm": 1.1347193448807606,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 14199
    },
    {
      "epoch": 0.142,
      "grad_norm": 1.3727967137741413,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 14200
    },
    {
      "epoch": 0.14201,
      "grad_norm": 1.0442182792604975,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 14201
    },
    {
      "epoch": 0.14202,
      "grad_norm": 1.2585233889585916,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 14202
    },
    {
      "epoch": 0.14203,
      "grad_norm": 1.3628434262296514,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 14203
    },
    {
      "epoch": 0.14204,
      "grad_norm": 0.9471467248764754,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 14204
    },
    {
      "epoch": 0.14205,
      "grad_norm": 1.2167492529019486,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 14205
    },
    {
      "epoch": 0.14206,
      "grad_norm": 1.0440167721462932,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 14206
    },
    {
      "epoch": 0.14207,
      "grad_norm": 1.360618441047889,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 14207
    },
    {
      "epoch": 0.14208,
      "grad_norm": 1.2537038117019181,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 14208
    },
    {
      "epoch": 0.14209,
      "grad_norm": 1.081791751264513,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 14209
    },
    {
      "epoch": 0.1421,
      "grad_norm": 1.254534712247267,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 14210
    },
    {
      "epoch": 0.14211,
      "grad_norm": 1.5122240518817431,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 14211
    },
    {
      "epoch": 0.14212,
      "grad_norm": 1.0112483003151058,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 14212
    },
    {
      "epoch": 0.14213,
      "grad_norm": 1.5511275971569494,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 14213
    },
    {
      "epoch": 0.14214,
      "grad_norm": 1.1564205582922507,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 14214
    },
    {
      "epoch": 0.14215,
      "grad_norm": 1.4372825515145655,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 14215
    },
    {
      "epoch": 0.14216,
      "grad_norm": 1.1860158195555228,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 14216
    },
    {
      "epoch": 0.14217,
      "grad_norm": 1.3192104639179023,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 14217
    },
    {
      "epoch": 0.14218,
      "grad_norm": 0.9797169344125047,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 14218
    },
    {
      "epoch": 0.14219,
      "grad_norm": 1.455155509542232,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 14219
    },
    {
      "epoch": 0.1422,
      "grad_norm": 1.192258717354992,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 14220
    },
    {
      "epoch": 0.14221,
      "grad_norm": 1.4572850914795752,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 14221
    },
    {
      "epoch": 0.14222,
      "grad_norm": 1.0724039986564513,
      "learning_rate": 0.003,
      "loss": 4.1022,
      "step": 14222
    },
    {
      "epoch": 0.14223,
      "grad_norm": 1.2557558241219466,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 14223
    },
    {
      "epoch": 0.14224,
      "grad_norm": 1.0698137237090464,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 14224
    },
    {
      "epoch": 0.14225,
      "grad_norm": 1.359659837901035,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 14225
    },
    {
      "epoch": 0.14226,
      "grad_norm": 1.121514307256611,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 14226
    },
    {
      "epoch": 0.14227,
      "grad_norm": 1.2797986124941525,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 14227
    },
    {
      "epoch": 0.14228,
      "grad_norm": 1.2148723431841384,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 14228
    },
    {
      "epoch": 0.14229,
      "grad_norm": 1.7105353367182838,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 14229
    },
    {
      "epoch": 0.1423,
      "grad_norm": 0.8998733141850936,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 14230
    },
    {
      "epoch": 0.14231,
      "grad_norm": 1.0518522620748678,
      "learning_rate": 0.003,
      "loss": 4.089,
      "step": 14231
    },
    {
      "epoch": 0.14232,
      "grad_norm": 1.350838942083396,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 14232
    },
    {
      "epoch": 0.14233,
      "grad_norm": 1.0179680902312946,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 14233
    },
    {
      "epoch": 0.14234,
      "grad_norm": 1.2698091867895445,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 14234
    },
    {
      "epoch": 0.14235,
      "grad_norm": 1.2965428655396696,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 14235
    },
    {
      "epoch": 0.14236,
      "grad_norm": 1.0899343077133432,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 14236
    },
    {
      "epoch": 0.14237,
      "grad_norm": 1.3340083275697776,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 14237
    },
    {
      "epoch": 0.14238,
      "grad_norm": 1.246926879826877,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 14238
    },
    {
      "epoch": 0.14239,
      "grad_norm": 1.1718419258346218,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 14239
    },
    {
      "epoch": 0.1424,
      "grad_norm": 1.325326309558883,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 14240
    },
    {
      "epoch": 0.14241,
      "grad_norm": 0.9120505784288971,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 14241
    },
    {
      "epoch": 0.14242,
      "grad_norm": 1.0852945738282669,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 14242
    },
    {
      "epoch": 0.14243,
      "grad_norm": 1.2867726876794063,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 14243
    },
    {
      "epoch": 0.14244,
      "grad_norm": 1.0743183760770276,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 14244
    },
    {
      "epoch": 0.14245,
      "grad_norm": 1.3665750595546649,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 14245
    },
    {
      "epoch": 0.14246,
      "grad_norm": 1.235528909124935,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 14246
    },
    {
      "epoch": 0.14247,
      "grad_norm": 1.1471484614944625,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 14247
    },
    {
      "epoch": 0.14248,
      "grad_norm": 1.1158396969741906,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 14248
    },
    {
      "epoch": 0.14249,
      "grad_norm": 1.2451790763660837,
      "learning_rate": 0.003,
      "loss": 4.0295,
      "step": 14249
    },
    {
      "epoch": 0.1425,
      "grad_norm": 1.1660921278797662,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 14250
    },
    {
      "epoch": 0.14251,
      "grad_norm": 1.337552507610175,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 14251
    },
    {
      "epoch": 0.14252,
      "grad_norm": 1.1237501224202473,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 14252
    },
    {
      "epoch": 0.14253,
      "grad_norm": 1.2401862552002867,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 14253
    },
    {
      "epoch": 0.14254,
      "grad_norm": 1.2076917825407354,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 14254
    },
    {
      "epoch": 0.14255,
      "grad_norm": 1.1075968289184035,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 14255
    },
    {
      "epoch": 0.14256,
      "grad_norm": 1.2753987269843883,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 14256
    },
    {
      "epoch": 0.14257,
      "grad_norm": 1.1585523826878703,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 14257
    },
    {
      "epoch": 0.14258,
      "grad_norm": 1.5092524682885513,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 14258
    },
    {
      "epoch": 0.14259,
      "grad_norm": 0.9875483613399724,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 14259
    },
    {
      "epoch": 0.1426,
      "grad_norm": 1.326626446023979,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 14260
    },
    {
      "epoch": 0.14261,
      "grad_norm": 1.1717803350308014,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 14261
    },
    {
      "epoch": 0.14262,
      "grad_norm": 1.1204453946088764,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 14262
    },
    {
      "epoch": 0.14263,
      "grad_norm": 1.1459022877545624,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 14263
    },
    {
      "epoch": 0.14264,
      "grad_norm": 1.0570187331453484,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 14264
    },
    {
      "epoch": 0.14265,
      "grad_norm": 1.2507717034006678,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 14265
    },
    {
      "epoch": 0.14266,
      "grad_norm": 1.2440790086756464,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 14266
    },
    {
      "epoch": 0.14267,
      "grad_norm": 1.1507090337948278,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 14267
    },
    {
      "epoch": 0.14268,
      "grad_norm": 0.933785452272276,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 14268
    },
    {
      "epoch": 0.14269,
      "grad_norm": 1.33321097143748,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 14269
    },
    {
      "epoch": 0.1427,
      "grad_norm": 1.0385155213078152,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 14270
    },
    {
      "epoch": 0.14271,
      "grad_norm": 1.279339007087537,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 14271
    },
    {
      "epoch": 0.14272,
      "grad_norm": 1.3206856987994857,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 14272
    },
    {
      "epoch": 0.14273,
      "grad_norm": 1.3968104502893839,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 14273
    },
    {
      "epoch": 0.14274,
      "grad_norm": 1.233775105693856,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 14274
    },
    {
      "epoch": 0.14275,
      "grad_norm": 1.0425653767912715,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 14275
    },
    {
      "epoch": 0.14276,
      "grad_norm": 1.4118044096015665,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 14276
    },
    {
      "epoch": 0.14277,
      "grad_norm": 1.471296450615494,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 14277
    },
    {
      "epoch": 0.14278,
      "grad_norm": 1.1820443927859594,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 14278
    },
    {
      "epoch": 0.14279,
      "grad_norm": 1.1725352181243902,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 14279
    },
    {
      "epoch": 0.1428,
      "grad_norm": 1.3310836236444723,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 14280
    },
    {
      "epoch": 0.14281,
      "grad_norm": 1.2274869974280136,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 14281
    },
    {
      "epoch": 0.14282,
      "grad_norm": 1.1488388991529206,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 14282
    },
    {
      "epoch": 0.14283,
      "grad_norm": 1.0685417106519242,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 14283
    },
    {
      "epoch": 0.14284,
      "grad_norm": 1.1333158664486487,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 14284
    },
    {
      "epoch": 0.14285,
      "grad_norm": 1.1609384114256422,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 14285
    },
    {
      "epoch": 0.14286,
      "grad_norm": 1.273127080220649,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 14286
    },
    {
      "epoch": 0.14287,
      "grad_norm": 1.592837902156321,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 14287
    },
    {
      "epoch": 0.14288,
      "grad_norm": 1.1149129312014674,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 14288
    },
    {
      "epoch": 0.14289,
      "grad_norm": 1.3242356847900765,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 14289
    },
    {
      "epoch": 0.1429,
      "grad_norm": 1.3328202982701185,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 14290
    },
    {
      "epoch": 0.14291,
      "grad_norm": 0.990076164424594,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 14291
    },
    {
      "epoch": 0.14292,
      "grad_norm": 1.292375579717724,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 14292
    },
    {
      "epoch": 0.14293,
      "grad_norm": 0.9475587721700227,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 14293
    },
    {
      "epoch": 0.14294,
      "grad_norm": 1.3188992823297991,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 14294
    },
    {
      "epoch": 0.14295,
      "grad_norm": 1.1802765919358251,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 14295
    },
    {
      "epoch": 0.14296,
      "grad_norm": 1.470604664552602,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 14296
    },
    {
      "epoch": 0.14297,
      "grad_norm": 1.2692534422761599,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 14297
    },
    {
      "epoch": 0.14298,
      "grad_norm": 1.190711026045534,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 14298
    },
    {
      "epoch": 0.14299,
      "grad_norm": 1.1527200359274339,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 14299
    },
    {
      "epoch": 0.143,
      "grad_norm": 1.2253939632615483,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 14300
    },
    {
      "epoch": 0.14301,
      "grad_norm": 1.1452804353444426,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 14301
    },
    {
      "epoch": 0.14302,
      "grad_norm": 1.2653095717075904,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 14302
    },
    {
      "epoch": 0.14303,
      "grad_norm": 1.2283386711500326,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 14303
    },
    {
      "epoch": 0.14304,
      "grad_norm": 1.2260239140907923,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 14304
    },
    {
      "epoch": 0.14305,
      "grad_norm": 1.241379595591714,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 14305
    },
    {
      "epoch": 0.14306,
      "grad_norm": 1.2089446349096906,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 14306
    },
    {
      "epoch": 0.14307,
      "grad_norm": 1.317381627080003,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 14307
    },
    {
      "epoch": 0.14308,
      "grad_norm": 1.2697092334474644,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 14308
    },
    {
      "epoch": 0.14309,
      "grad_norm": 1.178546586293183,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 14309
    },
    {
      "epoch": 0.1431,
      "grad_norm": 1.1018919994675551,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 14310
    },
    {
      "epoch": 0.14311,
      "grad_norm": 1.087360520477749,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 14311
    },
    {
      "epoch": 0.14312,
      "grad_norm": 1.1353127920896684,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 14312
    },
    {
      "epoch": 0.14313,
      "grad_norm": 1.3300518405090815,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 14313
    },
    {
      "epoch": 0.14314,
      "grad_norm": 1.1134049920041054,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 14314
    },
    {
      "epoch": 0.14315,
      "grad_norm": 1.4393999140109497,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 14315
    },
    {
      "epoch": 0.14316,
      "grad_norm": 0.9685541012410513,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 14316
    },
    {
      "epoch": 0.14317,
      "grad_norm": 1.2922155314825943,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 14317
    },
    {
      "epoch": 0.14318,
      "grad_norm": 1.0866212897670042,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 14318
    },
    {
      "epoch": 0.14319,
      "grad_norm": 1.1904422597157893,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 14319
    },
    {
      "epoch": 0.1432,
      "grad_norm": 1.0813133187971877,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 14320
    },
    {
      "epoch": 0.14321,
      "grad_norm": 1.2934273949969206,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 14321
    },
    {
      "epoch": 0.14322,
      "grad_norm": 1.3390521441092138,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 14322
    },
    {
      "epoch": 0.14323,
      "grad_norm": 1.052371538996439,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 14323
    },
    {
      "epoch": 0.14324,
      "grad_norm": 1.5072536866034199,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 14324
    },
    {
      "epoch": 0.14325,
      "grad_norm": 1.2621890886460423,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 14325
    },
    {
      "epoch": 0.14326,
      "grad_norm": 1.2722920221123368,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 14326
    },
    {
      "epoch": 0.14327,
      "grad_norm": 1.3018127539015547,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 14327
    },
    {
      "epoch": 0.14328,
      "grad_norm": 1.0657898922750024,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 14328
    },
    {
      "epoch": 0.14329,
      "grad_norm": 1.2160029214303807,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 14329
    },
    {
      "epoch": 0.1433,
      "grad_norm": 1.2569127274770167,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 14330
    },
    {
      "epoch": 0.14331,
      "grad_norm": 1.1996116025902857,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 14331
    },
    {
      "epoch": 0.14332,
      "grad_norm": 1.3788925110863324,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 14332
    },
    {
      "epoch": 0.14333,
      "grad_norm": 1.068258870600438,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 14333
    },
    {
      "epoch": 0.14334,
      "grad_norm": 1.4019931902382263,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 14334
    },
    {
      "epoch": 0.14335,
      "grad_norm": 1.0535736941749134,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 14335
    },
    {
      "epoch": 0.14336,
      "grad_norm": 1.306283302232701,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 14336
    },
    {
      "epoch": 0.14337,
      "grad_norm": 1.111328959070111,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 14337
    },
    {
      "epoch": 0.14338,
      "grad_norm": 1.1975664423135413,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 14338
    },
    {
      "epoch": 0.14339,
      "grad_norm": 1.2191116259359973,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 14339
    },
    {
      "epoch": 0.1434,
      "grad_norm": 1.1840868502775341,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 14340
    },
    {
      "epoch": 0.14341,
      "grad_norm": 1.137478876561676,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 14341
    },
    {
      "epoch": 0.14342,
      "grad_norm": 1.3700259954550276,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 14342
    },
    {
      "epoch": 0.14343,
      "grad_norm": 1.170377210981172,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 14343
    },
    {
      "epoch": 0.14344,
      "grad_norm": 1.2818784705552841,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 14344
    },
    {
      "epoch": 0.14345,
      "grad_norm": 1.4317888747741843,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 14345
    },
    {
      "epoch": 0.14346,
      "grad_norm": 0.9983553444856835,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 14346
    },
    {
      "epoch": 0.14347,
      "grad_norm": 1.356555777043041,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 14347
    },
    {
      "epoch": 0.14348,
      "grad_norm": 1.143216391544617,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 14348
    },
    {
      "epoch": 0.14349,
      "grad_norm": 1.3525650486340823,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 14349
    },
    {
      "epoch": 0.1435,
      "grad_norm": 1.0057561246421187,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 14350
    },
    {
      "epoch": 0.14351,
      "grad_norm": 1.3963353599027775,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 14351
    },
    {
      "epoch": 0.14352,
      "grad_norm": 1.330890632598169,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 14352
    },
    {
      "epoch": 0.14353,
      "grad_norm": 1.1516104014607385,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 14353
    },
    {
      "epoch": 0.14354,
      "grad_norm": 1.1232317363122273,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 14354
    },
    {
      "epoch": 0.14355,
      "grad_norm": 1.1755740122698235,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 14355
    },
    {
      "epoch": 0.14356,
      "grad_norm": 1.3381647039895606,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 14356
    },
    {
      "epoch": 0.14357,
      "grad_norm": 1.0638132963036395,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 14357
    },
    {
      "epoch": 0.14358,
      "grad_norm": 1.1919994563744336,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 14358
    },
    {
      "epoch": 0.14359,
      "grad_norm": 1.1802707096333935,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 14359
    },
    {
      "epoch": 0.1436,
      "grad_norm": 1.196571939548632,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 14360
    },
    {
      "epoch": 0.14361,
      "grad_norm": 1.2288744185635048,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 14361
    },
    {
      "epoch": 0.14362,
      "grad_norm": 1.2168584885220672,
      "learning_rate": 0.003,
      "loss": 4.1105,
      "step": 14362
    },
    {
      "epoch": 0.14363,
      "grad_norm": 1.5046380694248807,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 14363
    },
    {
      "epoch": 0.14364,
      "grad_norm": 1.1395198564200926,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 14364
    },
    {
      "epoch": 0.14365,
      "grad_norm": 1.2341073392599913,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 14365
    },
    {
      "epoch": 0.14366,
      "grad_norm": 1.3233230350475131,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 14366
    },
    {
      "epoch": 0.14367,
      "grad_norm": 1.1290748432048177,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 14367
    },
    {
      "epoch": 0.14368,
      "grad_norm": 1.259093691839906,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 14368
    },
    {
      "epoch": 0.14369,
      "grad_norm": 1.3434206128946158,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 14369
    },
    {
      "epoch": 0.1437,
      "grad_norm": 1.143664186598176,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 14370
    },
    {
      "epoch": 0.14371,
      "grad_norm": 1.2119997531247515,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 14371
    },
    {
      "epoch": 0.14372,
      "grad_norm": 1.068786400245214,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 14372
    },
    {
      "epoch": 0.14373,
      "grad_norm": 1.2530697642056654,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 14373
    },
    {
      "epoch": 0.14374,
      "grad_norm": 1.065215088325605,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 14374
    },
    {
      "epoch": 0.14375,
      "grad_norm": 1.2621075398736497,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 14375
    },
    {
      "epoch": 0.14376,
      "grad_norm": 1.1434925159815088,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 14376
    },
    {
      "epoch": 0.14377,
      "grad_norm": 1.164602629665852,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 14377
    },
    {
      "epoch": 0.14378,
      "grad_norm": 1.2341596077755344,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 14378
    },
    {
      "epoch": 0.14379,
      "grad_norm": 1.4319774842479158,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 14379
    },
    {
      "epoch": 0.1438,
      "grad_norm": 1.181278783262906,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 14380
    },
    {
      "epoch": 0.14381,
      "grad_norm": 1.4463379635472426,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 14381
    },
    {
      "epoch": 0.14382,
      "grad_norm": 1.0250385679941718,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 14382
    },
    {
      "epoch": 0.14383,
      "grad_norm": 1.3341913337356177,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 14383
    },
    {
      "epoch": 0.14384,
      "grad_norm": 1.0578582594846084,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 14384
    },
    {
      "epoch": 0.14385,
      "grad_norm": 1.2006510088537283,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 14385
    },
    {
      "epoch": 0.14386,
      "grad_norm": 1.1735284191020567,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 14386
    },
    {
      "epoch": 0.14387,
      "grad_norm": 1.3144827104637613,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 14387
    },
    {
      "epoch": 0.14388,
      "grad_norm": 1.234010698715963,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 14388
    },
    {
      "epoch": 0.14389,
      "grad_norm": 1.3112248415874672,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 14389
    },
    {
      "epoch": 0.1439,
      "grad_norm": 0.9695468424280841,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 14390
    },
    {
      "epoch": 0.14391,
      "grad_norm": 1.4268333526903854,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 14391
    },
    {
      "epoch": 0.14392,
      "grad_norm": 1.079939896224488,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 14392
    },
    {
      "epoch": 0.14393,
      "grad_norm": 1.2101070030796783,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 14393
    },
    {
      "epoch": 0.14394,
      "grad_norm": 1.2107748764865611,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 14394
    },
    {
      "epoch": 0.14395,
      "grad_norm": 1.1654521766245456,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 14395
    },
    {
      "epoch": 0.14396,
      "grad_norm": 1.2679245581546836,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 14396
    },
    {
      "epoch": 0.14397,
      "grad_norm": 1.2035374322115195,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 14397
    },
    {
      "epoch": 0.14398,
      "grad_norm": 1.35203795746024,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 14398
    },
    {
      "epoch": 0.14399,
      "grad_norm": 1.095843910890968,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 14399
    },
    {
      "epoch": 0.144,
      "grad_norm": 1.1826386393611024,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 14400
    },
    {
      "epoch": 0.14401,
      "grad_norm": 1.3204081501624294,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 14401
    },
    {
      "epoch": 0.14402,
      "grad_norm": 1.1316878090286906,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 14402
    },
    {
      "epoch": 0.14403,
      "grad_norm": 1.27626622488073,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 14403
    },
    {
      "epoch": 0.14404,
      "grad_norm": 1.1663925536898734,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 14404
    },
    {
      "epoch": 0.14405,
      "grad_norm": 1.4300762716857476,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 14405
    },
    {
      "epoch": 0.14406,
      "grad_norm": 0.9779121704087891,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 14406
    },
    {
      "epoch": 0.14407,
      "grad_norm": 1.1832335778071723,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 14407
    },
    {
      "epoch": 0.14408,
      "grad_norm": 1.2603683727876147,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 14408
    },
    {
      "epoch": 0.14409,
      "grad_norm": 1.5610546074596634,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 14409
    },
    {
      "epoch": 0.1441,
      "grad_norm": 1.0511442118032344,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 14410
    },
    {
      "epoch": 0.14411,
      "grad_norm": 1.3488788906343954,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 14411
    },
    {
      "epoch": 0.14412,
      "grad_norm": 1.3284511507449208,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 14412
    },
    {
      "epoch": 0.14413,
      "grad_norm": 0.9772219897331645,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 14413
    },
    {
      "epoch": 0.14414,
      "grad_norm": 1.2426001842398948,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 14414
    },
    {
      "epoch": 0.14415,
      "grad_norm": 1.0887974596263341,
      "learning_rate": 0.003,
      "loss": 4.0278,
      "step": 14415
    },
    {
      "epoch": 0.14416,
      "grad_norm": 1.16577074086037,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 14416
    },
    {
      "epoch": 0.14417,
      "grad_norm": 1.0636381968227409,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 14417
    },
    {
      "epoch": 0.14418,
      "grad_norm": 1.3337924665293321,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 14418
    },
    {
      "epoch": 0.14419,
      "grad_norm": 1.3674897024573358,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14419
    },
    {
      "epoch": 0.1442,
      "grad_norm": 1.0381744655777854,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 14420
    },
    {
      "epoch": 0.14421,
      "grad_norm": 1.229329190596495,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 14421
    },
    {
      "epoch": 0.14422,
      "grad_norm": 1.1932692329223549,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 14422
    },
    {
      "epoch": 0.14423,
      "grad_norm": 1.2891152046270844,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 14423
    },
    {
      "epoch": 0.14424,
      "grad_norm": 1.2381292606460514,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 14424
    },
    {
      "epoch": 0.14425,
      "grad_norm": 1.0439278430058745,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 14425
    },
    {
      "epoch": 0.14426,
      "grad_norm": 1.297336142612756,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 14426
    },
    {
      "epoch": 0.14427,
      "grad_norm": 1.0550985053749837,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 14427
    },
    {
      "epoch": 0.14428,
      "grad_norm": 1.2899119998999717,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 14428
    },
    {
      "epoch": 0.14429,
      "grad_norm": 1.1821243947260043,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 14429
    },
    {
      "epoch": 0.1443,
      "grad_norm": 1.30751709397349,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 14430
    },
    {
      "epoch": 0.14431,
      "grad_norm": 1.093150059274714,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 14431
    },
    {
      "epoch": 0.14432,
      "grad_norm": 1.3784750328904125,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 14432
    },
    {
      "epoch": 0.14433,
      "grad_norm": 1.08182832775297,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 14433
    },
    {
      "epoch": 0.14434,
      "grad_norm": 1.5049206866725384,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 14434
    },
    {
      "epoch": 0.14435,
      "grad_norm": 1.317178243386282,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 14435
    },
    {
      "epoch": 0.14436,
      "grad_norm": 1.2227613692858288,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 14436
    },
    {
      "epoch": 0.14437,
      "grad_norm": 1.304179770019113,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 14437
    },
    {
      "epoch": 0.14438,
      "grad_norm": 1.141831161472869,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 14438
    },
    {
      "epoch": 0.14439,
      "grad_norm": 1.0747145230342643,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 14439
    },
    {
      "epoch": 0.1444,
      "grad_norm": 1.299113550200919,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 14440
    },
    {
      "epoch": 0.14441,
      "grad_norm": 0.9820380080128326,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 14441
    },
    {
      "epoch": 0.14442,
      "grad_norm": 1.3189405568861905,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 14442
    },
    {
      "epoch": 0.14443,
      "grad_norm": 1.1460870888331007,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 14443
    },
    {
      "epoch": 0.14444,
      "grad_norm": 1.3368224171260885,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 14444
    },
    {
      "epoch": 0.14445,
      "grad_norm": 1.291908146511478,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 14445
    },
    {
      "epoch": 0.14446,
      "grad_norm": 1.1620244454533144,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 14446
    },
    {
      "epoch": 0.14447,
      "grad_norm": 1.4377785575175646,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 14447
    },
    {
      "epoch": 0.14448,
      "grad_norm": 1.0584026951747234,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 14448
    },
    {
      "epoch": 0.14449,
      "grad_norm": 1.1819828467010947,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 14449
    },
    {
      "epoch": 0.1445,
      "grad_norm": 1.0794523137185799,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 14450
    },
    {
      "epoch": 0.14451,
      "grad_norm": 1.1818184068695379,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 14451
    },
    {
      "epoch": 0.14452,
      "grad_norm": 1.2710421760018722,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 14452
    },
    {
      "epoch": 0.14453,
      "grad_norm": 1.1622127964996545,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 14453
    },
    {
      "epoch": 0.14454,
      "grad_norm": 1.4394457125821694,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 14454
    },
    {
      "epoch": 0.14455,
      "grad_norm": 1.4153297904619657,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 14455
    },
    {
      "epoch": 0.14456,
      "grad_norm": 1.322307430442675,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 14456
    },
    {
      "epoch": 0.14457,
      "grad_norm": 1.1002488156671828,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 14457
    },
    {
      "epoch": 0.14458,
      "grad_norm": 1.2714201699644943,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 14458
    },
    {
      "epoch": 0.14459,
      "grad_norm": 1.0095749116294088,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 14459
    },
    {
      "epoch": 0.1446,
      "grad_norm": 1.4375780380540866,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 14460
    },
    {
      "epoch": 0.14461,
      "grad_norm": 1.14707387495912,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 14461
    },
    {
      "epoch": 0.14462,
      "grad_norm": 1.0885637677444406,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 14462
    },
    {
      "epoch": 0.14463,
      "grad_norm": 1.1986989329327669,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 14463
    },
    {
      "epoch": 0.14464,
      "grad_norm": 1.0866883339116706,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 14464
    },
    {
      "epoch": 0.14465,
      "grad_norm": 1.1770390100324601,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 14465
    },
    {
      "epoch": 0.14466,
      "grad_norm": 0.9286184501251965,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 14466
    },
    {
      "epoch": 0.14467,
      "grad_norm": 1.3163757201728299,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 14467
    },
    {
      "epoch": 0.14468,
      "grad_norm": 1.17131106150134,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 14468
    },
    {
      "epoch": 0.14469,
      "grad_norm": 1.2925242105191204,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 14469
    },
    {
      "epoch": 0.1447,
      "grad_norm": 1.1042283407728772,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 14470
    },
    {
      "epoch": 0.14471,
      "grad_norm": 1.36741571361758,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 14471
    },
    {
      "epoch": 0.14472,
      "grad_norm": 1.4041101426859173,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 14472
    },
    {
      "epoch": 0.14473,
      "grad_norm": 1.4291062487502415,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 14473
    },
    {
      "epoch": 0.14474,
      "grad_norm": 1.280473028989063,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 14474
    },
    {
      "epoch": 0.14475,
      "grad_norm": 1.0625179105244094,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 14475
    },
    {
      "epoch": 0.14476,
      "grad_norm": 1.328498228319439,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 14476
    },
    {
      "epoch": 0.14477,
      "grad_norm": 1.2486546627827975,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 14477
    },
    {
      "epoch": 0.14478,
      "grad_norm": 1.0105945330782855,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 14478
    },
    {
      "epoch": 0.14479,
      "grad_norm": 1.144118833057518,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 14479
    },
    {
      "epoch": 0.1448,
      "grad_norm": 1.2580688610346515,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 14480
    },
    {
      "epoch": 0.14481,
      "grad_norm": 1.1638733787914244,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 14481
    },
    {
      "epoch": 0.14482,
      "grad_norm": 1.2923456312341166,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 14482
    },
    {
      "epoch": 0.14483,
      "grad_norm": 1.3403808718963135,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 14483
    },
    {
      "epoch": 0.14484,
      "grad_norm": 1.2968967075420859,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 14484
    },
    {
      "epoch": 0.14485,
      "grad_norm": 1.5068711403378459,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 14485
    },
    {
      "epoch": 0.14486,
      "grad_norm": 0.8011818290123951,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 14486
    },
    {
      "epoch": 0.14487,
      "grad_norm": 0.9656806963890344,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 14487
    },
    {
      "epoch": 0.14488,
      "grad_norm": 1.1437377945148044,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 14488
    },
    {
      "epoch": 0.14489,
      "grad_norm": 1.2191665583527924,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 14489
    },
    {
      "epoch": 0.1449,
      "grad_norm": 1.0879306524530827,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 14490
    },
    {
      "epoch": 0.14491,
      "grad_norm": 1.2127406424972869,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 14491
    },
    {
      "epoch": 0.14492,
      "grad_norm": 1.1569332906839882,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 14492
    },
    {
      "epoch": 0.14493,
      "grad_norm": 1.2566899839962806,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 14493
    },
    {
      "epoch": 0.14494,
      "grad_norm": 1.2080703694035553,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 14494
    },
    {
      "epoch": 0.14495,
      "grad_norm": 1.2112001686441412,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 14495
    },
    {
      "epoch": 0.14496,
      "grad_norm": 1.2551719179102072,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 14496
    },
    {
      "epoch": 0.14497,
      "grad_norm": 1.1267006066327367,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14497
    },
    {
      "epoch": 0.14498,
      "grad_norm": 1.173983188727913,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 14498
    },
    {
      "epoch": 0.14499,
      "grad_norm": 1.299307892459785,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 14499
    },
    {
      "epoch": 0.145,
      "grad_norm": 1.2623974500837152,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 14500
    },
    {
      "epoch": 0.14501,
      "grad_norm": 1.2833983713516892,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 14501
    },
    {
      "epoch": 0.14502,
      "grad_norm": 1.1527850471661591,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 14502
    },
    {
      "epoch": 0.14503,
      "grad_norm": 1.3253761895311922,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 14503
    },
    {
      "epoch": 0.14504,
      "grad_norm": 1.302871022894076,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 14504
    },
    {
      "epoch": 0.14505,
      "grad_norm": 1.7242406969598858,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 14505
    },
    {
      "epoch": 0.14506,
      "grad_norm": 1.0678467046870714,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 14506
    },
    {
      "epoch": 0.14507,
      "grad_norm": 1.3537159549441753,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 14507
    },
    {
      "epoch": 0.14508,
      "grad_norm": 1.3472589423143078,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 14508
    },
    {
      "epoch": 0.14509,
      "grad_norm": 1.1619264215472223,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 14509
    },
    {
      "epoch": 0.1451,
      "grad_norm": 1.1637797261869363,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 14510
    },
    {
      "epoch": 0.14511,
      "grad_norm": 1.1056960026610345,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 14511
    },
    {
      "epoch": 0.14512,
      "grad_norm": 1.3473753956609713,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 14512
    },
    {
      "epoch": 0.14513,
      "grad_norm": 1.3543840235294828,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 14513
    },
    {
      "epoch": 0.14514,
      "grad_norm": 1.2032831118156042,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 14514
    },
    {
      "epoch": 0.14515,
      "grad_norm": 1.1997784406249512,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 14515
    },
    {
      "epoch": 0.14516,
      "grad_norm": 1.2434269634164505,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 14516
    },
    {
      "epoch": 0.14517,
      "grad_norm": 1.1062979063611453,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 14517
    },
    {
      "epoch": 0.14518,
      "grad_norm": 1.1578608691856607,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 14518
    },
    {
      "epoch": 0.14519,
      "grad_norm": 1.0622328252719972,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 14519
    },
    {
      "epoch": 0.1452,
      "grad_norm": 1.2651723404892912,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 14520
    },
    {
      "epoch": 0.14521,
      "grad_norm": 1.0940269947093042,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 14521
    },
    {
      "epoch": 0.14522,
      "grad_norm": 1.3365778696794945,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 14522
    },
    {
      "epoch": 0.14523,
      "grad_norm": 1.1536610931751508,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 14523
    },
    {
      "epoch": 0.14524,
      "grad_norm": 1.1571296642834157,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 14524
    },
    {
      "epoch": 0.14525,
      "grad_norm": 1.181679668807532,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 14525
    },
    {
      "epoch": 0.14526,
      "grad_norm": 1.0646818438185597,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 14526
    },
    {
      "epoch": 0.14527,
      "grad_norm": 1.5444754672321084,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 14527
    },
    {
      "epoch": 0.14528,
      "grad_norm": 1.110119502820521,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 14528
    },
    {
      "epoch": 0.14529,
      "grad_norm": 1.3202879624253951,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 14529
    },
    {
      "epoch": 0.1453,
      "grad_norm": 1.0652238096811364,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 14530
    },
    {
      "epoch": 0.14531,
      "grad_norm": 1.3146546875809235,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 14531
    },
    {
      "epoch": 0.14532,
      "grad_norm": 1.316481791494692,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 14532
    },
    {
      "epoch": 0.14533,
      "grad_norm": 1.195736762059849,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 14533
    },
    {
      "epoch": 0.14534,
      "grad_norm": 1.3153453343964692,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 14534
    },
    {
      "epoch": 0.14535,
      "grad_norm": 1.1661697160363889,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 14535
    },
    {
      "epoch": 0.14536,
      "grad_norm": 1.2644956968313958,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 14536
    },
    {
      "epoch": 0.14537,
      "grad_norm": 1.1889914671754607,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 14537
    },
    {
      "epoch": 0.14538,
      "grad_norm": 1.2755526897119813,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 14538
    },
    {
      "epoch": 0.14539,
      "grad_norm": 1.0977307408454553,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 14539
    },
    {
      "epoch": 0.1454,
      "grad_norm": 1.1425193492569852,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 14540
    },
    {
      "epoch": 0.14541,
      "grad_norm": 1.3654741018509804,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 14541
    },
    {
      "epoch": 0.14542,
      "grad_norm": 1.3513178652433542,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 14542
    },
    {
      "epoch": 0.14543,
      "grad_norm": 1.2139546213180623,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 14543
    },
    {
      "epoch": 0.14544,
      "grad_norm": 1.1412586493058874,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 14544
    },
    {
      "epoch": 0.14545,
      "grad_norm": 1.2505032723432452,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 14545
    },
    {
      "epoch": 0.14546,
      "grad_norm": 1.1090289144477474,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 14546
    },
    {
      "epoch": 0.14547,
      "grad_norm": 1.1512395586013198,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 14547
    },
    {
      "epoch": 0.14548,
      "grad_norm": 1.1796163733534573,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 14548
    },
    {
      "epoch": 0.14549,
      "grad_norm": 1.171288245791135,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 14549
    },
    {
      "epoch": 0.1455,
      "grad_norm": 1.0909536639205564,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 14550
    },
    {
      "epoch": 0.14551,
      "grad_norm": 1.2379349235427564,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 14551
    },
    {
      "epoch": 0.14552,
      "grad_norm": 1.1946767662358404,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 14552
    },
    {
      "epoch": 0.14553,
      "grad_norm": 1.3425360041150831,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 14553
    },
    {
      "epoch": 0.14554,
      "grad_norm": 1.2701523855721424,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 14554
    },
    {
      "epoch": 0.14555,
      "grad_norm": 1.4560475136856261,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 14555
    },
    {
      "epoch": 0.14556,
      "grad_norm": 1.0558293006911164,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 14556
    },
    {
      "epoch": 0.14557,
      "grad_norm": 1.277053527905692,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 14557
    },
    {
      "epoch": 0.14558,
      "grad_norm": 1.0996275979016732,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 14558
    },
    {
      "epoch": 0.14559,
      "grad_norm": 1.6489839246077922,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 14559
    },
    {
      "epoch": 0.1456,
      "grad_norm": 0.8985364962976041,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 14560
    },
    {
      "epoch": 0.14561,
      "grad_norm": 1.0939749863890005,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 14561
    },
    {
      "epoch": 0.14562,
      "grad_norm": 1.2240643266714866,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 14562
    },
    {
      "epoch": 0.14563,
      "grad_norm": 1.2048737550612332,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 14563
    },
    {
      "epoch": 0.14564,
      "grad_norm": 1.1849894366036393,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 14564
    },
    {
      "epoch": 0.14565,
      "grad_norm": 1.2868216966661061,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 14565
    },
    {
      "epoch": 0.14566,
      "grad_norm": 1.1401900748690066,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 14566
    },
    {
      "epoch": 0.14567,
      "grad_norm": 1.17255534523372,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 14567
    },
    {
      "epoch": 0.14568,
      "grad_norm": 1.3814064011696163,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 14568
    },
    {
      "epoch": 0.14569,
      "grad_norm": 1.4975264483666402,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 14569
    },
    {
      "epoch": 0.1457,
      "grad_norm": 1.161954516152272,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 14570
    },
    {
      "epoch": 0.14571,
      "grad_norm": 1.3083161772445326,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 14571
    },
    {
      "epoch": 0.14572,
      "grad_norm": 1.0661306176980925,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 14572
    },
    {
      "epoch": 0.14573,
      "grad_norm": 1.168348396939506,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 14573
    },
    {
      "epoch": 0.14574,
      "grad_norm": 1.2995611491474106,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 14574
    },
    {
      "epoch": 0.14575,
      "grad_norm": 1.1051839229605225,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 14575
    },
    {
      "epoch": 0.14576,
      "grad_norm": 1.426897734370533,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 14576
    },
    {
      "epoch": 0.14577,
      "grad_norm": 1.3561950697517535,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 14577
    },
    {
      "epoch": 0.14578,
      "grad_norm": 1.1606752158833817,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 14578
    },
    {
      "epoch": 0.14579,
      "grad_norm": 1.0561133106875344,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 14579
    },
    {
      "epoch": 0.1458,
      "grad_norm": 1.313880559739441,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 14580
    },
    {
      "epoch": 0.14581,
      "grad_norm": 1.175241277341145,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 14581
    },
    {
      "epoch": 0.14582,
      "grad_norm": 1.0354369918458097,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 14582
    },
    {
      "epoch": 0.14583,
      "grad_norm": 1.3564197408932643,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 14583
    },
    {
      "epoch": 0.14584,
      "grad_norm": 1.2592562167383652,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 14584
    },
    {
      "epoch": 0.14585,
      "grad_norm": 1.1655502695806124,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 14585
    },
    {
      "epoch": 0.14586,
      "grad_norm": 1.1270144122520829,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 14586
    },
    {
      "epoch": 0.14587,
      "grad_norm": 1.3758566632654936,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 14587
    },
    {
      "epoch": 0.14588,
      "grad_norm": 1.2272085490990263,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 14588
    },
    {
      "epoch": 0.14589,
      "grad_norm": 1.2677737579782817,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 14589
    },
    {
      "epoch": 0.1459,
      "grad_norm": 1.1360459914972227,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 14590
    },
    {
      "epoch": 0.14591,
      "grad_norm": 1.651258332471172,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 14591
    },
    {
      "epoch": 0.14592,
      "grad_norm": 0.9382332788255714,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 14592
    },
    {
      "epoch": 0.14593,
      "grad_norm": 1.3978775775037138,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 14593
    },
    {
      "epoch": 0.14594,
      "grad_norm": 1.2423191227807584,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 14594
    },
    {
      "epoch": 0.14595,
      "grad_norm": 1.2538985231476125,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 14595
    },
    {
      "epoch": 0.14596,
      "grad_norm": 1.1603969821652953,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 14596
    },
    {
      "epoch": 0.14597,
      "grad_norm": 1.2309084376760162,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 14597
    },
    {
      "epoch": 0.14598,
      "grad_norm": 1.4019712882539683,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 14598
    },
    {
      "epoch": 0.14599,
      "grad_norm": 1.0781586012504811,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 14599
    },
    {
      "epoch": 0.146,
      "grad_norm": 1.2463531132231722,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 14600
    },
    {
      "epoch": 0.14601,
      "grad_norm": 1.095285384160531,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 14601
    },
    {
      "epoch": 0.14602,
      "grad_norm": 1.5438704468250124,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 14602
    },
    {
      "epoch": 0.14603,
      "grad_norm": 1.008786932453542,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 14603
    },
    {
      "epoch": 0.14604,
      "grad_norm": 1.3263740335801186,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 14604
    },
    {
      "epoch": 0.14605,
      "grad_norm": 1.2035455321360693,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 14605
    },
    {
      "epoch": 0.14606,
      "grad_norm": 1.1297030002565212,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 14606
    },
    {
      "epoch": 0.14607,
      "grad_norm": 1.1514523890993336,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 14607
    },
    {
      "epoch": 0.14608,
      "grad_norm": 1.1003428973761358,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 14608
    },
    {
      "epoch": 0.14609,
      "grad_norm": 1.3220980722298383,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 14609
    },
    {
      "epoch": 0.1461,
      "grad_norm": 0.9078562364085505,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 14610
    },
    {
      "epoch": 0.14611,
      "grad_norm": 1.4187892743750314,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 14611
    },
    {
      "epoch": 0.14612,
      "grad_norm": 1.1193931114685458,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 14612
    },
    {
      "epoch": 0.14613,
      "grad_norm": 1.3084042564846994,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 14613
    },
    {
      "epoch": 0.14614,
      "grad_norm": 1.217749523812658,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 14614
    },
    {
      "epoch": 0.14615,
      "grad_norm": 1.1171054826532747,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 14615
    },
    {
      "epoch": 0.14616,
      "grad_norm": 1.1533725317270105,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 14616
    },
    {
      "epoch": 0.14617,
      "grad_norm": 1.2648670840148972,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 14617
    },
    {
      "epoch": 0.14618,
      "grad_norm": 1.207641924938561,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 14618
    },
    {
      "epoch": 0.14619,
      "grad_norm": 1.2927644215866563,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 14619
    },
    {
      "epoch": 0.1462,
      "grad_norm": 1.2135111822809954,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 14620
    },
    {
      "epoch": 0.14621,
      "grad_norm": 1.3374066515976388,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 14621
    },
    {
      "epoch": 0.14622,
      "grad_norm": 1.1798176510548282,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 14622
    },
    {
      "epoch": 0.14623,
      "grad_norm": 1.101241817333466,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 14623
    },
    {
      "epoch": 0.14624,
      "grad_norm": 1.3165157136298984,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 14624
    },
    {
      "epoch": 0.14625,
      "grad_norm": 1.204010145714381,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 14625
    },
    {
      "epoch": 0.14626,
      "grad_norm": 1.1970570695525438,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 14626
    },
    {
      "epoch": 0.14627,
      "grad_norm": 1.3101272713075305,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 14627
    },
    {
      "epoch": 0.14628,
      "grad_norm": 1.2438733569226708,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 14628
    },
    {
      "epoch": 0.14629,
      "grad_norm": 1.1144384917204837,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 14629
    },
    {
      "epoch": 0.1463,
      "grad_norm": 1.3231476742228823,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 14630
    },
    {
      "epoch": 0.14631,
      "grad_norm": 1.2251934200995147,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 14631
    },
    {
      "epoch": 0.14632,
      "grad_norm": 1.4149018197203078,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 14632
    },
    {
      "epoch": 0.14633,
      "grad_norm": 1.111321909938563,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 14633
    },
    {
      "epoch": 0.14634,
      "grad_norm": 1.4009877478969213,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 14634
    },
    {
      "epoch": 0.14635,
      "grad_norm": 0.9757433393240569,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 14635
    },
    {
      "epoch": 0.14636,
      "grad_norm": 1.5448628476612054,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 14636
    },
    {
      "epoch": 0.14637,
      "grad_norm": 1.0681842219058078,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 14637
    },
    {
      "epoch": 0.14638,
      "grad_norm": 1.270278890838018,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 14638
    },
    {
      "epoch": 0.14639,
      "grad_norm": 1.1293518586686984,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 14639
    },
    {
      "epoch": 0.1464,
      "grad_norm": 1.2230323646100287,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 14640
    },
    {
      "epoch": 0.14641,
      "grad_norm": 0.9605823373334033,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 14641
    },
    {
      "epoch": 0.14642,
      "grad_norm": 1.395513703350284,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 14642
    },
    {
      "epoch": 0.14643,
      "grad_norm": 1.111260789455305,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 14643
    },
    {
      "epoch": 0.14644,
      "grad_norm": 1.3088383951677751,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 14644
    },
    {
      "epoch": 0.14645,
      "grad_norm": 1.1924587194471679,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 14645
    },
    {
      "epoch": 0.14646,
      "grad_norm": 1.0230377961179655,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 14646
    },
    {
      "epoch": 0.14647,
      "grad_norm": 1.3166502301604532,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 14647
    },
    {
      "epoch": 0.14648,
      "grad_norm": 1.108098794298806,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 14648
    },
    {
      "epoch": 0.14649,
      "grad_norm": 1.2678976831802262,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 14649
    },
    {
      "epoch": 0.1465,
      "grad_norm": 1.0779314944871512,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 14650
    },
    {
      "epoch": 0.14651,
      "grad_norm": 1.3070108905573998,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 14651
    },
    {
      "epoch": 0.14652,
      "grad_norm": 1.2173180299324342,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 14652
    },
    {
      "epoch": 0.14653,
      "grad_norm": 1.3059875979156272,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 14653
    },
    {
      "epoch": 0.14654,
      "grad_norm": 1.396883465923019,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 14654
    },
    {
      "epoch": 0.14655,
      "grad_norm": 1.0141344618956727,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 14655
    },
    {
      "epoch": 0.14656,
      "grad_norm": 1.4713846091980216,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14656
    },
    {
      "epoch": 0.14657,
      "grad_norm": 1.083585755943934,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 14657
    },
    {
      "epoch": 0.14658,
      "grad_norm": 1.2032480378616042,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 14658
    },
    {
      "epoch": 0.14659,
      "grad_norm": 1.1574205075152437,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 14659
    },
    {
      "epoch": 0.1466,
      "grad_norm": 1.1533865134835117,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 14660
    },
    {
      "epoch": 0.14661,
      "grad_norm": 1.1557425319475578,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 14661
    },
    {
      "epoch": 0.14662,
      "grad_norm": 1.2140824735376958,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 14662
    },
    {
      "epoch": 0.14663,
      "grad_norm": 1.3119847972636143,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 14663
    },
    {
      "epoch": 0.14664,
      "grad_norm": 1.278637541443759,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 14664
    },
    {
      "epoch": 0.14665,
      "grad_norm": 1.5537971245262678,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 14665
    },
    {
      "epoch": 0.14666,
      "grad_norm": 1.0543789191944573,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 14666
    },
    {
      "epoch": 0.14667,
      "grad_norm": 1.321352396499266,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 14667
    },
    {
      "epoch": 0.14668,
      "grad_norm": 1.452374422058879,
      "learning_rate": 0.003,
      "loss": 4.104,
      "step": 14668
    },
    {
      "epoch": 0.14669,
      "grad_norm": 0.9483661975026233,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 14669
    },
    {
      "epoch": 0.1467,
      "grad_norm": 1.2457290239459582,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 14670
    },
    {
      "epoch": 0.14671,
      "grad_norm": 1.166673424790351,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 14671
    },
    {
      "epoch": 0.14672,
      "grad_norm": 1.134676582970826,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 14672
    },
    {
      "epoch": 0.14673,
      "grad_norm": 1.4049530615079515,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 14673
    },
    {
      "epoch": 0.14674,
      "grad_norm": 1.0791538339719724,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 14674
    },
    {
      "epoch": 0.14675,
      "grad_norm": 1.3387749229849049,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 14675
    },
    {
      "epoch": 0.14676,
      "grad_norm": 1.0816776734773716,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 14676
    },
    {
      "epoch": 0.14677,
      "grad_norm": 1.3010340266823335,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 14677
    },
    {
      "epoch": 0.14678,
      "grad_norm": 1.273904467089526,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 14678
    },
    {
      "epoch": 0.14679,
      "grad_norm": 1.2050737400060467,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 14679
    },
    {
      "epoch": 0.1468,
      "grad_norm": 1.1782396903671484,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 14680
    },
    {
      "epoch": 0.14681,
      "grad_norm": 1.2641672769658576,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 14681
    },
    {
      "epoch": 0.14682,
      "grad_norm": 1.3683531524304238,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 14682
    },
    {
      "epoch": 0.14683,
      "grad_norm": 1.1681281783864517,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 14683
    },
    {
      "epoch": 0.14684,
      "grad_norm": 1.2147665860320387,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 14684
    },
    {
      "epoch": 0.14685,
      "grad_norm": 1.1383565292447009,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 14685
    },
    {
      "epoch": 0.14686,
      "grad_norm": 1.2606728631128676,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 14686
    },
    {
      "epoch": 0.14687,
      "grad_norm": 1.2339376261541728,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 14687
    },
    {
      "epoch": 0.14688,
      "grad_norm": 1.1587680031981418,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 14688
    },
    {
      "epoch": 0.14689,
      "grad_norm": 1.1302148406483674,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 14689
    },
    {
      "epoch": 0.1469,
      "grad_norm": 1.160721415065776,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 14690
    },
    {
      "epoch": 0.14691,
      "grad_norm": 1.2096541604524635,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 14691
    },
    {
      "epoch": 0.14692,
      "grad_norm": 1.098711790790942,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 14692
    },
    {
      "epoch": 0.14693,
      "grad_norm": 1.5608335345805666,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 14693
    },
    {
      "epoch": 0.14694,
      "grad_norm": 1.3805529775587744,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 14694
    },
    {
      "epoch": 0.14695,
      "grad_norm": 1.1496335459014944,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 14695
    },
    {
      "epoch": 0.14696,
      "grad_norm": 1.2934564888383862,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 14696
    },
    {
      "epoch": 0.14697,
      "grad_norm": 1.0298392730293933,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 14697
    },
    {
      "epoch": 0.14698,
      "grad_norm": 1.2493236769696854,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 14698
    },
    {
      "epoch": 0.14699,
      "grad_norm": 1.1966206064149618,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 14699
    },
    {
      "epoch": 0.147,
      "grad_norm": 1.2008163126396347,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 14700
    },
    {
      "epoch": 0.14701,
      "grad_norm": 1.307359185613977,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 14701
    },
    {
      "epoch": 0.14702,
      "grad_norm": 1.17130778803398,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 14702
    },
    {
      "epoch": 0.14703,
      "grad_norm": 0.9278549741557135,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 14703
    },
    {
      "epoch": 0.14704,
      "grad_norm": 1.2416731233037417,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 14704
    },
    {
      "epoch": 0.14705,
      "grad_norm": 1.443478884981792,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 14705
    },
    {
      "epoch": 0.14706,
      "grad_norm": 1.169988258938255,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 14706
    },
    {
      "epoch": 0.14707,
      "grad_norm": 1.3884268894595029,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 14707
    },
    {
      "epoch": 0.14708,
      "grad_norm": 1.2376664296974007,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 14708
    },
    {
      "epoch": 0.14709,
      "grad_norm": 1.3585251644666752,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 14709
    },
    {
      "epoch": 0.1471,
      "grad_norm": 1.1070441241571778,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 14710
    },
    {
      "epoch": 0.14711,
      "grad_norm": 1.209342609620908,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 14711
    },
    {
      "epoch": 0.14712,
      "grad_norm": 1.1481596216904917,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 14712
    },
    {
      "epoch": 0.14713,
      "grad_norm": 1.362689261051619,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 14713
    },
    {
      "epoch": 0.14714,
      "grad_norm": 1.1744847375395815,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 14714
    },
    {
      "epoch": 0.14715,
      "grad_norm": 1.2132539753611855,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 14715
    },
    {
      "epoch": 0.14716,
      "grad_norm": 1.3284941942614448,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 14716
    },
    {
      "epoch": 0.14717,
      "grad_norm": 0.9466611944947958,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 14717
    },
    {
      "epoch": 0.14718,
      "grad_norm": 1.2938546324781965,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 14718
    },
    {
      "epoch": 0.14719,
      "grad_norm": 1.1099514972231117,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 14719
    },
    {
      "epoch": 0.1472,
      "grad_norm": 1.3147842488951789,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 14720
    },
    {
      "epoch": 0.14721,
      "grad_norm": 1.3027693784383552,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 14721
    },
    {
      "epoch": 0.14722,
      "grad_norm": 1.1463487338501712,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 14722
    },
    {
      "epoch": 0.14723,
      "grad_norm": 1.244333943250865,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 14723
    },
    {
      "epoch": 0.14724,
      "grad_norm": 1.1755040875323115,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 14724
    },
    {
      "epoch": 0.14725,
      "grad_norm": 1.3023666248547623,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 14725
    },
    {
      "epoch": 0.14726,
      "grad_norm": 1.1758954249843652,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 14726
    },
    {
      "epoch": 0.14727,
      "grad_norm": 1.4709314206982071,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 14727
    },
    {
      "epoch": 0.14728,
      "grad_norm": 0.9819314270248706,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 14728
    },
    {
      "epoch": 0.14729,
      "grad_norm": 1.2375661512872944,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 14729
    },
    {
      "epoch": 0.1473,
      "grad_norm": 1.2885113061978055,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 14730
    },
    {
      "epoch": 0.14731,
      "grad_norm": 1.0505392875289823,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 14731
    },
    {
      "epoch": 0.14732,
      "grad_norm": 1.5160304880082762,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 14732
    },
    {
      "epoch": 0.14733,
      "grad_norm": 1.0131695855765106,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 14733
    },
    {
      "epoch": 0.14734,
      "grad_norm": 1.2967729921181999,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 14734
    },
    {
      "epoch": 0.14735,
      "grad_norm": 1.154259022839686,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 14735
    },
    {
      "epoch": 0.14736,
      "grad_norm": 1.1909045056718182,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 14736
    },
    {
      "epoch": 0.14737,
      "grad_norm": 1.2765410017020464,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 14737
    },
    {
      "epoch": 0.14738,
      "grad_norm": 1.1805332411128984,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 14738
    },
    {
      "epoch": 0.14739,
      "grad_norm": 1.2642126914844365,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 14739
    },
    {
      "epoch": 0.1474,
      "grad_norm": 1.0699407445012783,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 14740
    },
    {
      "epoch": 0.14741,
      "grad_norm": 1.1499692761506615,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 14741
    },
    {
      "epoch": 0.14742,
      "grad_norm": 1.0295183618628216,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 14742
    },
    {
      "epoch": 0.14743,
      "grad_norm": 1.3084781291711867,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 14743
    },
    {
      "epoch": 0.14744,
      "grad_norm": 1.0996251198527003,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 14744
    },
    {
      "epoch": 0.14745,
      "grad_norm": 1.134401093824059,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 14745
    },
    {
      "epoch": 0.14746,
      "grad_norm": 1.2803354774841316,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 14746
    },
    {
      "epoch": 0.14747,
      "grad_norm": 1.3988377679385435,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 14747
    },
    {
      "epoch": 0.14748,
      "grad_norm": 1.0568472895986643,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 14748
    },
    {
      "epoch": 0.14749,
      "grad_norm": 1.4024469804051825,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 14749
    },
    {
      "epoch": 0.1475,
      "grad_norm": 1.1838146800837628,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 14750
    },
    {
      "epoch": 0.14751,
      "grad_norm": 1.1286320877224711,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 14751
    },
    {
      "epoch": 0.14752,
      "grad_norm": 1.3952019890043283,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 14752
    },
    {
      "epoch": 0.14753,
      "grad_norm": 1.3268177274947321,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 14753
    },
    {
      "epoch": 0.14754,
      "grad_norm": 1.3229095606049164,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 14754
    },
    {
      "epoch": 0.14755,
      "grad_norm": 1.1673253364225489,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 14755
    },
    {
      "epoch": 0.14756,
      "grad_norm": 1.382402758451,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 14756
    },
    {
      "epoch": 0.14757,
      "grad_norm": 1.1441186256744826,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 14757
    },
    {
      "epoch": 0.14758,
      "grad_norm": 1.1621501001512715,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 14758
    },
    {
      "epoch": 0.14759,
      "grad_norm": 1.3877917651303693,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 14759
    },
    {
      "epoch": 0.1476,
      "grad_norm": 1.3096270746637773,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 14760
    },
    {
      "epoch": 0.14761,
      "grad_norm": 1.107020187704195,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 14761
    },
    {
      "epoch": 0.14762,
      "grad_norm": 1.1961993881582595,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 14762
    },
    {
      "epoch": 0.14763,
      "grad_norm": 1.303135898447819,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 14763
    },
    {
      "epoch": 0.14764,
      "grad_norm": 1.2515419072382497,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 14764
    },
    {
      "epoch": 0.14765,
      "grad_norm": 1.2859986321936037,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 14765
    },
    {
      "epoch": 0.14766,
      "grad_norm": 1.021156011483574,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 14766
    },
    {
      "epoch": 0.14767,
      "grad_norm": 1.3546518718683798,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 14767
    },
    {
      "epoch": 0.14768,
      "grad_norm": 1.1048674849286966,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 14768
    },
    {
      "epoch": 0.14769,
      "grad_norm": 1.213721024463415,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 14769
    },
    {
      "epoch": 0.1477,
      "grad_norm": 1.181611806859025,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 14770
    },
    {
      "epoch": 0.14771,
      "grad_norm": 1.4015824040772773,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 14771
    },
    {
      "epoch": 0.14772,
      "grad_norm": 1.197184787661011,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 14772
    },
    {
      "epoch": 0.14773,
      "grad_norm": 1.1226164282132731,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 14773
    },
    {
      "epoch": 0.14774,
      "grad_norm": 1.3344600923991445,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 14774
    },
    {
      "epoch": 0.14775,
      "grad_norm": 1.1506419423756051,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 14775
    },
    {
      "epoch": 0.14776,
      "grad_norm": 1.293786746697097,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 14776
    },
    {
      "epoch": 0.14777,
      "grad_norm": 1.1234974672375049,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 14777
    },
    {
      "epoch": 0.14778,
      "grad_norm": 1.3024728459130335,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 14778
    },
    {
      "epoch": 0.14779,
      "grad_norm": 0.9917384616067323,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 14779
    },
    {
      "epoch": 0.1478,
      "grad_norm": 1.4676339518358084,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 14780
    },
    {
      "epoch": 0.14781,
      "grad_norm": 1.034039632520674,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 14781
    },
    {
      "epoch": 0.14782,
      "grad_norm": 1.407785614525778,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 14782
    },
    {
      "epoch": 0.14783,
      "grad_norm": 1.13730215414004,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 14783
    },
    {
      "epoch": 0.14784,
      "grad_norm": 1.2776348796678363,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 14784
    },
    {
      "epoch": 0.14785,
      "grad_norm": 1.2582277140130618,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 14785
    },
    {
      "epoch": 0.14786,
      "grad_norm": 1.1181549682198384,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 14786
    },
    {
      "epoch": 0.14787,
      "grad_norm": 1.3305483061294012,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 14787
    },
    {
      "epoch": 0.14788,
      "grad_norm": 1.3643327116171096,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 14788
    },
    {
      "epoch": 0.14789,
      "grad_norm": 1.2123896710327138,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 14789
    },
    {
      "epoch": 0.1479,
      "grad_norm": 1.257666517907204,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 14790
    },
    {
      "epoch": 0.14791,
      "grad_norm": 1.2042170863016024,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 14791
    },
    {
      "epoch": 0.14792,
      "grad_norm": 1.2417665699180211,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 14792
    },
    {
      "epoch": 0.14793,
      "grad_norm": 1.2227523296617067,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 14793
    },
    {
      "epoch": 0.14794,
      "grad_norm": 1.224635598347942,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 14794
    },
    {
      "epoch": 0.14795,
      "grad_norm": 1.068073493344534,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 14795
    },
    {
      "epoch": 0.14796,
      "grad_norm": 1.2125563865221247,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 14796
    },
    {
      "epoch": 0.14797,
      "grad_norm": 1.1118542142201944,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 14797
    },
    {
      "epoch": 0.14798,
      "grad_norm": 1.3720803310657834,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 14798
    },
    {
      "epoch": 0.14799,
      "grad_norm": 1.0967274438274146,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 14799
    },
    {
      "epoch": 0.148,
      "grad_norm": 1.4347201117761799,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 14800
    },
    {
      "epoch": 0.14801,
      "grad_norm": 1.0571807698765756,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 14801
    },
    {
      "epoch": 0.14802,
      "grad_norm": 1.33974434466865,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 14802
    },
    {
      "epoch": 0.14803,
      "grad_norm": 1.0461335273789005,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 14803
    },
    {
      "epoch": 0.14804,
      "grad_norm": 1.5678993159546522,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 14804
    },
    {
      "epoch": 0.14805,
      "grad_norm": 1.5017874509817701,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 14805
    },
    {
      "epoch": 0.14806,
      "grad_norm": 1.3135443467584271,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 14806
    },
    {
      "epoch": 0.14807,
      "grad_norm": 1.2573607343279003,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 14807
    },
    {
      "epoch": 0.14808,
      "grad_norm": 1.340007504898721,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 14808
    },
    {
      "epoch": 0.14809,
      "grad_norm": 1.2737817685965107,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 14809
    },
    {
      "epoch": 0.1481,
      "grad_norm": 1.231657153429162,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 14810
    },
    {
      "epoch": 0.14811,
      "grad_norm": 0.9652700786951047,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 14811
    },
    {
      "epoch": 0.14812,
      "grad_norm": 1.134367013325509,
      "learning_rate": 0.003,
      "loss": 4.0192,
      "step": 14812
    },
    {
      "epoch": 0.14813,
      "grad_norm": 1.2739003479310576,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 14813
    },
    {
      "epoch": 0.14814,
      "grad_norm": 1.2158291141016366,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 14814
    },
    {
      "epoch": 0.14815,
      "grad_norm": 1.2046555948864488,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 14815
    },
    {
      "epoch": 0.14816,
      "grad_norm": 0.9447172429128873,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 14816
    },
    {
      "epoch": 0.14817,
      "grad_norm": 1.2092539474962263,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 14817
    },
    {
      "epoch": 0.14818,
      "grad_norm": 1.2998028016770053,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 14818
    },
    {
      "epoch": 0.14819,
      "grad_norm": 1.5738605231395526,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 14819
    },
    {
      "epoch": 0.1482,
      "grad_norm": 1.1981548986992587,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 14820
    },
    {
      "epoch": 0.14821,
      "grad_norm": 1.2884537538247403,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 14821
    },
    {
      "epoch": 0.14822,
      "grad_norm": 1.2873606531452093,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 14822
    },
    {
      "epoch": 0.14823,
      "grad_norm": 1.2819850551690075,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 14823
    },
    {
      "epoch": 0.14824,
      "grad_norm": 1.1941581998763113,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 14824
    },
    {
      "epoch": 0.14825,
      "grad_norm": 1.3312177938775116,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 14825
    },
    {
      "epoch": 0.14826,
      "grad_norm": 1.133401460840803,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 14826
    },
    {
      "epoch": 0.14827,
      "grad_norm": 1.2103649040830275,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 14827
    },
    {
      "epoch": 0.14828,
      "grad_norm": 1.2024206746365804,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 14828
    },
    {
      "epoch": 0.14829,
      "grad_norm": 1.049474499056886,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 14829
    },
    {
      "epoch": 0.1483,
      "grad_norm": 1.1404466141485063,
      "learning_rate": 0.003,
      "loss": 4.0237,
      "step": 14830
    },
    {
      "epoch": 0.14831,
      "grad_norm": 1.2028619755384187,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 14831
    },
    {
      "epoch": 0.14832,
      "grad_norm": 1.274719239737427,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 14832
    },
    {
      "epoch": 0.14833,
      "grad_norm": 1.168398802387037,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 14833
    },
    {
      "epoch": 0.14834,
      "grad_norm": 1.2480203646129735,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 14834
    },
    {
      "epoch": 0.14835,
      "grad_norm": 1.3441893456511147,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 14835
    },
    {
      "epoch": 0.14836,
      "grad_norm": 1.263215019256534,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14836
    },
    {
      "epoch": 0.14837,
      "grad_norm": 1.0878443963119684,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 14837
    },
    {
      "epoch": 0.14838,
      "grad_norm": 1.4808670361142087,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 14838
    },
    {
      "epoch": 0.14839,
      "grad_norm": 1.026590985882426,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 14839
    },
    {
      "epoch": 0.1484,
      "grad_norm": 1.4501261403621308,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 14840
    },
    {
      "epoch": 0.14841,
      "grad_norm": 1.0557531101030002,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 14841
    },
    {
      "epoch": 0.14842,
      "grad_norm": 1.3984652412164706,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 14842
    },
    {
      "epoch": 0.14843,
      "grad_norm": 1.2635315636832918,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 14843
    },
    {
      "epoch": 0.14844,
      "grad_norm": 1.3736676538535175,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 14844
    },
    {
      "epoch": 0.14845,
      "grad_norm": 1.351947964711682,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 14845
    },
    {
      "epoch": 0.14846,
      "grad_norm": 1.2076037307068557,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 14846
    },
    {
      "epoch": 0.14847,
      "grad_norm": 1.2853195481414905,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 14847
    },
    {
      "epoch": 0.14848,
      "grad_norm": 1.189469804334418,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 14848
    },
    {
      "epoch": 0.14849,
      "grad_norm": 1.3956045687534402,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 14849
    },
    {
      "epoch": 0.1485,
      "grad_norm": 0.9904765968057987,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 14850
    },
    {
      "epoch": 0.14851,
      "grad_norm": 1.4490773260474896,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 14851
    },
    {
      "epoch": 0.14852,
      "grad_norm": 1.179826635925976,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 14852
    },
    {
      "epoch": 0.14853,
      "grad_norm": 1.3174642687265041,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 14853
    },
    {
      "epoch": 0.14854,
      "grad_norm": 1.2344900327272792,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 14854
    },
    {
      "epoch": 0.14855,
      "grad_norm": 1.2097266505343784,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 14855
    },
    {
      "epoch": 0.14856,
      "grad_norm": 1.3055087196965998,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 14856
    },
    {
      "epoch": 0.14857,
      "grad_norm": 1.107104085111295,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 14857
    },
    {
      "epoch": 0.14858,
      "grad_norm": 1.2705730843296212,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 14858
    },
    {
      "epoch": 0.14859,
      "grad_norm": 0.9637858708447968,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 14859
    },
    {
      "epoch": 0.1486,
      "grad_norm": 1.1497883211281064,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 14860
    },
    {
      "epoch": 0.14861,
      "grad_norm": 1.2953516340935691,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 14861
    },
    {
      "epoch": 0.14862,
      "grad_norm": 1.054124052608267,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 14862
    },
    {
      "epoch": 0.14863,
      "grad_norm": 1.3789660702493187,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 14863
    },
    {
      "epoch": 0.14864,
      "grad_norm": 1.042024050437487,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 14864
    },
    {
      "epoch": 0.14865,
      "grad_norm": 1.4555256458496995,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 14865
    },
    {
      "epoch": 0.14866,
      "grad_norm": 1.1897332584183165,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 14866
    },
    {
      "epoch": 0.14867,
      "grad_norm": 1.490532399523928,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 14867
    },
    {
      "epoch": 0.14868,
      "grad_norm": 1.441816676541135,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 14868
    },
    {
      "epoch": 0.14869,
      "grad_norm": 1.0043574852900115,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 14869
    },
    {
      "epoch": 0.1487,
      "grad_norm": 1.4047636039319338,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 14870
    },
    {
      "epoch": 0.14871,
      "grad_norm": 1.3884438161536075,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 14871
    },
    {
      "epoch": 0.14872,
      "grad_norm": 1.1132691376640853,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 14872
    },
    {
      "epoch": 0.14873,
      "grad_norm": 1.1977345192860807,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 14873
    },
    {
      "epoch": 0.14874,
      "grad_norm": 1.245883895286039,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 14874
    },
    {
      "epoch": 0.14875,
      "grad_norm": 1.207678386998773,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 14875
    },
    {
      "epoch": 0.14876,
      "grad_norm": 1.2774366088212148,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 14876
    },
    {
      "epoch": 0.14877,
      "grad_norm": 1.6502687525664734,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 14877
    },
    {
      "epoch": 0.14878,
      "grad_norm": 0.9857659560796367,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 14878
    },
    {
      "epoch": 0.14879,
      "grad_norm": 1.4082595417112858,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 14879
    },
    {
      "epoch": 0.1488,
      "grad_norm": 1.0771591045335969,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 14880
    },
    {
      "epoch": 0.14881,
      "grad_norm": 1.3166686548948419,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 14881
    },
    {
      "epoch": 0.14882,
      "grad_norm": 1.233792941159962,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 14882
    },
    {
      "epoch": 0.14883,
      "grad_norm": 1.2791038993936084,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14883
    },
    {
      "epoch": 0.14884,
      "grad_norm": 1.2305968988624012,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 14884
    },
    {
      "epoch": 0.14885,
      "grad_norm": 1.0241945735482276,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 14885
    },
    {
      "epoch": 0.14886,
      "grad_norm": 1.3974200331093105,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 14886
    },
    {
      "epoch": 0.14887,
      "grad_norm": 1.0833386368940932,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 14887
    },
    {
      "epoch": 0.14888,
      "grad_norm": 1.1964013561459441,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 14888
    },
    {
      "epoch": 0.14889,
      "grad_norm": 1.1799582435469052,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 14889
    },
    {
      "epoch": 0.1489,
      "grad_norm": 1.390973260807786,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 14890
    },
    {
      "epoch": 0.14891,
      "grad_norm": 0.9525587988948696,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 14891
    },
    {
      "epoch": 0.14892,
      "grad_norm": 1.204310502728258,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 14892
    },
    {
      "epoch": 0.14893,
      "grad_norm": 1.0966924371763758,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 14893
    },
    {
      "epoch": 0.14894,
      "grad_norm": 1.2787229752399252,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 14894
    },
    {
      "epoch": 0.14895,
      "grad_norm": 1.2750937206101463,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 14895
    },
    {
      "epoch": 0.14896,
      "grad_norm": 1.2756110885462355,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 14896
    },
    {
      "epoch": 0.14897,
      "grad_norm": 1.0988749320591142,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 14897
    },
    {
      "epoch": 0.14898,
      "grad_norm": 1.1670682622351207,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 14898
    },
    {
      "epoch": 0.14899,
      "grad_norm": 1.1043180262888312,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 14899
    },
    {
      "epoch": 0.149,
      "grad_norm": 1.4613050465833892,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 14900
    },
    {
      "epoch": 0.14901,
      "grad_norm": 1.4317226682036521,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 14901
    },
    {
      "epoch": 0.14902,
      "grad_norm": 1.3030218748593272,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 14902
    },
    {
      "epoch": 0.14903,
      "grad_norm": 1.2349257456957607,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 14903
    },
    {
      "epoch": 0.14904,
      "grad_norm": 1.222522424979833,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 14904
    },
    {
      "epoch": 0.14905,
      "grad_norm": 1.3452550190203154,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 14905
    },
    {
      "epoch": 0.14906,
      "grad_norm": 0.9832696376167528,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 14906
    },
    {
      "epoch": 0.14907,
      "grad_norm": 1.3907971745147707,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 14907
    },
    {
      "epoch": 0.14908,
      "grad_norm": 1.1372164335048494,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 14908
    },
    {
      "epoch": 0.14909,
      "grad_norm": 1.1219052450795157,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 14909
    },
    {
      "epoch": 0.1491,
      "grad_norm": 1.1825502432967192,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 14910
    },
    {
      "epoch": 0.14911,
      "grad_norm": 1.2010941132677941,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 14911
    },
    {
      "epoch": 0.14912,
      "grad_norm": 1.3070054434027416,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 14912
    },
    {
      "epoch": 0.14913,
      "grad_norm": 1.2596303439594332,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 14913
    },
    {
      "epoch": 0.14914,
      "grad_norm": 1.1702506535315904,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 14914
    },
    {
      "epoch": 0.14915,
      "grad_norm": 1.3948642013157526,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 14915
    },
    {
      "epoch": 0.14916,
      "grad_norm": 1.0853748743773275,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 14916
    },
    {
      "epoch": 0.14917,
      "grad_norm": 1.4939579311245224,
      "learning_rate": 0.003,
      "loss": 4.1058,
      "step": 14917
    },
    {
      "epoch": 0.14918,
      "grad_norm": 1.1587444653275318,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 14918
    },
    {
      "epoch": 0.14919,
      "grad_norm": 1.2655904622340168,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 14919
    },
    {
      "epoch": 0.1492,
      "grad_norm": 1.0995955087144285,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 14920
    },
    {
      "epoch": 0.14921,
      "grad_norm": 1.3928205441425652,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 14921
    },
    {
      "epoch": 0.14922,
      "grad_norm": 1.0865550751685151,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 14922
    },
    {
      "epoch": 0.14923,
      "grad_norm": 1.2093099353750922,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 14923
    },
    {
      "epoch": 0.14924,
      "grad_norm": 1.2166173806791813,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 14924
    },
    {
      "epoch": 0.14925,
      "grad_norm": 1.1776264155167722,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 14925
    },
    {
      "epoch": 0.14926,
      "grad_norm": 1.3314021975619488,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 14926
    },
    {
      "epoch": 0.14927,
      "grad_norm": 1.060505965970766,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 14927
    },
    {
      "epoch": 0.14928,
      "grad_norm": 1.4406791970002304,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 14928
    },
    {
      "epoch": 0.14929,
      "grad_norm": 1.082162795184589,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 14929
    },
    {
      "epoch": 0.1493,
      "grad_norm": 1.2086048080899723,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 14930
    },
    {
      "epoch": 0.14931,
      "grad_norm": 1.2695538610859594,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 14931
    },
    {
      "epoch": 0.14932,
      "grad_norm": 1.1318361431289865,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 14932
    },
    {
      "epoch": 0.14933,
      "grad_norm": 1.446875788888267,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 14933
    },
    {
      "epoch": 0.14934,
      "grad_norm": 1.1462224150787974,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 14934
    },
    {
      "epoch": 0.14935,
      "grad_norm": 1.4912216225843362,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 14935
    },
    {
      "epoch": 0.14936,
      "grad_norm": 1.0533209518682838,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 14936
    },
    {
      "epoch": 0.14937,
      "grad_norm": 1.1819293779876494,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 14937
    },
    {
      "epoch": 0.14938,
      "grad_norm": 1.5128539093937539,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 14938
    },
    {
      "epoch": 0.14939,
      "grad_norm": 0.9915716999665365,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 14939
    },
    {
      "epoch": 0.1494,
      "grad_norm": 1.187533299776154,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 14940
    },
    {
      "epoch": 0.14941,
      "grad_norm": 1.2326825324147908,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 14941
    },
    {
      "epoch": 0.14942,
      "grad_norm": 1.1264556500091154,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 14942
    },
    {
      "epoch": 0.14943,
      "grad_norm": 1.411943555864094,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 14943
    },
    {
      "epoch": 0.14944,
      "grad_norm": 0.998547571587907,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 14944
    },
    {
      "epoch": 0.14945,
      "grad_norm": 1.3654248880724962,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 14945
    },
    {
      "epoch": 0.14946,
      "grad_norm": 1.1308853175079046,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 14946
    },
    {
      "epoch": 0.14947,
      "grad_norm": 1.1934755301548037,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 14947
    },
    {
      "epoch": 0.14948,
      "grad_norm": 1.2485781004994043,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 14948
    },
    {
      "epoch": 0.14949,
      "grad_norm": 1.2913617741593948,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 14949
    },
    {
      "epoch": 0.1495,
      "grad_norm": 1.487539410018331,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 14950
    },
    {
      "epoch": 0.14951,
      "grad_norm": 1.1026109116379552,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 14951
    },
    {
      "epoch": 0.14952,
      "grad_norm": 1.2953541884433633,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 14952
    },
    {
      "epoch": 0.14953,
      "grad_norm": 1.3740274670281512,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 14953
    },
    {
      "epoch": 0.14954,
      "grad_norm": 1.0925317694287173,
      "learning_rate": 0.003,
      "loss": 4.1083,
      "step": 14954
    },
    {
      "epoch": 0.14955,
      "grad_norm": 1.44005155343115,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 14955
    },
    {
      "epoch": 0.14956,
      "grad_norm": 1.1721710223911144,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 14956
    },
    {
      "epoch": 0.14957,
      "grad_norm": 1.2069504172841292,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 14957
    },
    {
      "epoch": 0.14958,
      "grad_norm": 1.2642042169712921,
      "learning_rate": 0.003,
      "loss": 4.1,
      "step": 14958
    },
    {
      "epoch": 0.14959,
      "grad_norm": 1.3063621719604683,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 14959
    },
    {
      "epoch": 0.1496,
      "grad_norm": 1.3659620018831007,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 14960
    },
    {
      "epoch": 0.14961,
      "grad_norm": 1.1731188763545322,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 14961
    },
    {
      "epoch": 0.14962,
      "grad_norm": 1.0893103154843458,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 14962
    },
    {
      "epoch": 0.14963,
      "grad_norm": 1.1524601757780435,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 14963
    },
    {
      "epoch": 0.14964,
      "grad_norm": 1.208167351977869,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 14964
    },
    {
      "epoch": 0.14965,
      "grad_norm": 1.23294526359167,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 14965
    },
    {
      "epoch": 0.14966,
      "grad_norm": 1.1538788312035761,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 14966
    },
    {
      "epoch": 0.14967,
      "grad_norm": 1.1682462372892446,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 14967
    },
    {
      "epoch": 0.14968,
      "grad_norm": 1.182504116597818,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 14968
    },
    {
      "epoch": 0.14969,
      "grad_norm": 1.223622558305158,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 14969
    },
    {
      "epoch": 0.1497,
      "grad_norm": 1.077729223328631,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 14970
    },
    {
      "epoch": 0.14971,
      "grad_norm": 1.146948410877107,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 14971
    },
    {
      "epoch": 0.14972,
      "grad_norm": 1.2407325286138748,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 14972
    },
    {
      "epoch": 0.14973,
      "grad_norm": 1.2644632188687903,
      "learning_rate": 0.003,
      "loss": 4.0289,
      "step": 14973
    },
    {
      "epoch": 0.14974,
      "grad_norm": 1.15196408084641,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 14974
    },
    {
      "epoch": 0.14975,
      "grad_norm": 1.3690869462360291,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 14975
    },
    {
      "epoch": 0.14976,
      "grad_norm": 1.146252532722954,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 14976
    },
    {
      "epoch": 0.14977,
      "grad_norm": 1.2893203491323464,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 14977
    },
    {
      "epoch": 0.14978,
      "grad_norm": 1.5339944589904086,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 14978
    },
    {
      "epoch": 0.14979,
      "grad_norm": 0.9614418891020783,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 14979
    },
    {
      "epoch": 0.1498,
      "grad_norm": 1.000608635648666,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 14980
    },
    {
      "epoch": 0.14981,
      "grad_norm": 1.3009567765546934,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 14981
    },
    {
      "epoch": 0.14982,
      "grad_norm": 1.3233483659166978,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 14982
    },
    {
      "epoch": 0.14983,
      "grad_norm": 1.2615111957089797,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 14983
    },
    {
      "epoch": 0.14984,
      "grad_norm": 1.3554690619827399,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 14984
    },
    {
      "epoch": 0.14985,
      "grad_norm": 1.1924340871785533,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 14985
    },
    {
      "epoch": 0.14986,
      "grad_norm": 1.2318918468737168,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 14986
    },
    {
      "epoch": 0.14987,
      "grad_norm": 1.0930518175296557,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 14987
    },
    {
      "epoch": 0.14988,
      "grad_norm": 1.1663896180400055,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 14988
    },
    {
      "epoch": 0.14989,
      "grad_norm": 1.2063698055393761,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 14989
    },
    {
      "epoch": 0.1499,
      "grad_norm": 1.2874535734867687,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 14990
    },
    {
      "epoch": 0.14991,
      "grad_norm": 1.1759325809215206,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 14991
    },
    {
      "epoch": 0.14992,
      "grad_norm": 1.4122874229515472,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 14992
    },
    {
      "epoch": 0.14993,
      "grad_norm": 1.2843662582190924,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 14993
    },
    {
      "epoch": 0.14994,
      "grad_norm": 1.5176369330876351,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 14994
    },
    {
      "epoch": 0.14995,
      "grad_norm": 1.1588560544407271,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 14995
    },
    {
      "epoch": 0.14996,
      "grad_norm": 1.1271968264084253,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 14996
    },
    {
      "epoch": 0.14997,
      "grad_norm": 1.2102659856566516,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 14997
    },
    {
      "epoch": 0.14998,
      "grad_norm": 1.275108957785287,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 14998
    },
    {
      "epoch": 0.14999,
      "grad_norm": 1.265708011373228,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 14999
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1439138688198385,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 15000
    },
    {
      "epoch": 0.15001,
      "grad_norm": 1.2421886806834463,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 15001
    },
    {
      "epoch": 0.15002,
      "grad_norm": 1.1004519397192334,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 15002
    },
    {
      "epoch": 0.15003,
      "grad_norm": 1.158455354790534,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 15003
    },
    {
      "epoch": 0.15004,
      "grad_norm": 1.2229740132210145,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 15004
    },
    {
      "epoch": 0.15005,
      "grad_norm": 1.2085418202639537,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 15005
    },
    {
      "epoch": 0.15006,
      "grad_norm": 1.3294954845260496,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 15006
    },
    {
      "epoch": 0.15007,
      "grad_norm": 1.3285655789422341,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 15007
    },
    {
      "epoch": 0.15008,
      "grad_norm": 1.2991360798789722,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 15008
    },
    {
      "epoch": 0.15009,
      "grad_norm": 1.3008766761442365,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 15009
    },
    {
      "epoch": 0.1501,
      "grad_norm": 1.157389568496755,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 15010
    },
    {
      "epoch": 0.15011,
      "grad_norm": 1.309388746363442,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 15011
    },
    {
      "epoch": 0.15012,
      "grad_norm": 1.1518291599840276,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 15012
    },
    {
      "epoch": 0.15013,
      "grad_norm": 1.130969825922695,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 15013
    },
    {
      "epoch": 0.15014,
      "grad_norm": 1.2675786794096975,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 15014
    },
    {
      "epoch": 0.15015,
      "grad_norm": 1.254882157454107,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 15015
    },
    {
      "epoch": 0.15016,
      "grad_norm": 1.2056073937925937,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 15016
    },
    {
      "epoch": 0.15017,
      "grad_norm": 1.2294431401855699,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 15017
    },
    {
      "epoch": 0.15018,
      "grad_norm": 1.1514030329240634,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 15018
    },
    {
      "epoch": 0.15019,
      "grad_norm": 1.0910474852609486,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 15019
    },
    {
      "epoch": 0.1502,
      "grad_norm": 1.3194572190844807,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 15020
    },
    {
      "epoch": 0.15021,
      "grad_norm": 1.0878947290892265,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 15021
    },
    {
      "epoch": 0.15022,
      "grad_norm": 1.359752959173593,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15022
    },
    {
      "epoch": 0.15023,
      "grad_norm": 1.3272200351660286,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 15023
    },
    {
      "epoch": 0.15024,
      "grad_norm": 1.3650082466937745,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 15024
    },
    {
      "epoch": 0.15025,
      "grad_norm": 1.0567603693339083,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 15025
    },
    {
      "epoch": 0.15026,
      "grad_norm": 1.2385010885446035,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 15026
    },
    {
      "epoch": 0.15027,
      "grad_norm": 0.9679446810635529,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 15027
    },
    {
      "epoch": 0.15028,
      "grad_norm": 1.503314405318618,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 15028
    },
    {
      "epoch": 0.15029,
      "grad_norm": 1.0213436660067043,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 15029
    },
    {
      "epoch": 0.1503,
      "grad_norm": 1.6028553178422178,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 15030
    },
    {
      "epoch": 0.15031,
      "grad_norm": 0.9355461869193742,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 15031
    },
    {
      "epoch": 0.15032,
      "grad_norm": 1.3076981026785015,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 15032
    },
    {
      "epoch": 0.15033,
      "grad_norm": 1.3188057675135088,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 15033
    },
    {
      "epoch": 0.15034,
      "grad_norm": 1.4290179136677195,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 15034
    },
    {
      "epoch": 0.15035,
      "grad_norm": 1.1285669561399247,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 15035
    },
    {
      "epoch": 0.15036,
      "grad_norm": 1.3493413030420565,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 15036
    },
    {
      "epoch": 0.15037,
      "grad_norm": 1.1927330301156278,
      "learning_rate": 0.003,
      "loss": 4.1009,
      "step": 15037
    },
    {
      "epoch": 0.15038,
      "grad_norm": 1.4071175189669094,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 15038
    },
    {
      "epoch": 0.15039,
      "grad_norm": 1.1432060701143985,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 15039
    },
    {
      "epoch": 0.1504,
      "grad_norm": 1.3626859879510167,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 15040
    },
    {
      "epoch": 0.15041,
      "grad_norm": 1.1375881639804108,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 15041
    },
    {
      "epoch": 0.15042,
      "grad_norm": 1.5718864410411195,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 15042
    },
    {
      "epoch": 0.15043,
      "grad_norm": 1.0101162506936627,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 15043
    },
    {
      "epoch": 0.15044,
      "grad_norm": 1.4177728319729006,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 15044
    },
    {
      "epoch": 0.15045,
      "grad_norm": 1.2584862320579329,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 15045
    },
    {
      "epoch": 0.15046,
      "grad_norm": 1.0371645732582666,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 15046
    },
    {
      "epoch": 0.15047,
      "grad_norm": 1.4680775466139233,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 15047
    },
    {
      "epoch": 0.15048,
      "grad_norm": 1.240976206826836,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 15048
    },
    {
      "epoch": 0.15049,
      "grad_norm": 1.4725510420309051,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 15049
    },
    {
      "epoch": 0.1505,
      "grad_norm": 1.2503013556073201,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 15050
    },
    {
      "epoch": 0.15051,
      "grad_norm": 1.3187053136228997,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 15051
    },
    {
      "epoch": 0.15052,
      "grad_norm": 1.1261744199142463,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 15052
    },
    {
      "epoch": 0.15053,
      "grad_norm": 1.1922341562009968,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 15053
    },
    {
      "epoch": 0.15054,
      "grad_norm": 1.280375040618315,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 15054
    },
    {
      "epoch": 0.15055,
      "grad_norm": 1.2626409514997128,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 15055
    },
    {
      "epoch": 0.15056,
      "grad_norm": 1.3031502941906068,
      "learning_rate": 0.003,
      "loss": 4.0268,
      "step": 15056
    },
    {
      "epoch": 0.15057,
      "grad_norm": 1.146443655378379,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 15057
    },
    {
      "epoch": 0.15058,
      "grad_norm": 1.234747124111013,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 15058
    },
    {
      "epoch": 0.15059,
      "grad_norm": 1.1111204119286546,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 15059
    },
    {
      "epoch": 0.1506,
      "grad_norm": 1.2782788932590696,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 15060
    },
    {
      "epoch": 0.15061,
      "grad_norm": 1.0913728828104843,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15061
    },
    {
      "epoch": 0.15062,
      "grad_norm": 1.273029391481596,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 15062
    },
    {
      "epoch": 0.15063,
      "grad_norm": 1.136396228675531,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 15063
    },
    {
      "epoch": 0.15064,
      "grad_norm": 1.1737411571075842,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 15064
    },
    {
      "epoch": 0.15065,
      "grad_norm": 1.2904500760837567,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 15065
    },
    {
      "epoch": 0.15066,
      "grad_norm": 1.4351156073359232,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 15066
    },
    {
      "epoch": 0.15067,
      "grad_norm": 1.1296351993384457,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 15067
    },
    {
      "epoch": 0.15068,
      "grad_norm": 1.4385586669901966,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 15068
    },
    {
      "epoch": 0.15069,
      "grad_norm": 1.104913445664609,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 15069
    },
    {
      "epoch": 0.1507,
      "grad_norm": 1.0053274139043227,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 15070
    },
    {
      "epoch": 0.15071,
      "grad_norm": 1.2780511731178645,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 15071
    },
    {
      "epoch": 0.15072,
      "grad_norm": 1.4261993717400032,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 15072
    },
    {
      "epoch": 0.15073,
      "grad_norm": 1.2855849210568024,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 15073
    },
    {
      "epoch": 0.15074,
      "grad_norm": 1.5074041648077694,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 15074
    },
    {
      "epoch": 0.15075,
      "grad_norm": 0.8953110771587716,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 15075
    },
    {
      "epoch": 0.15076,
      "grad_norm": 1.1679691805137673,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 15076
    },
    {
      "epoch": 0.15077,
      "grad_norm": 1.2715640089224092,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 15077
    },
    {
      "epoch": 0.15078,
      "grad_norm": 1.3102155377085043,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 15078
    },
    {
      "epoch": 0.15079,
      "grad_norm": 1.197144269451487,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 15079
    },
    {
      "epoch": 0.1508,
      "grad_norm": 1.2880148382239833,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 15080
    },
    {
      "epoch": 0.15081,
      "grad_norm": 1.2900635608171667,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 15081
    },
    {
      "epoch": 0.15082,
      "grad_norm": 1.2647862818670748,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 15082
    },
    {
      "epoch": 0.15083,
      "grad_norm": 1.2757016919839912,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 15083
    },
    {
      "epoch": 0.15084,
      "grad_norm": 1.4092964104632957,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 15084
    },
    {
      "epoch": 0.15085,
      "grad_norm": 1.2947858384427136,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 15085
    },
    {
      "epoch": 0.15086,
      "grad_norm": 1.3530504623742912,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 15086
    },
    {
      "epoch": 0.15087,
      "grad_norm": 1.2112907542851699,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 15087
    },
    {
      "epoch": 0.15088,
      "grad_norm": 1.3015483423935452,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 15088
    },
    {
      "epoch": 0.15089,
      "grad_norm": 1.1048537637635403,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 15089
    },
    {
      "epoch": 0.1509,
      "grad_norm": 1.437913899117062,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 15090
    },
    {
      "epoch": 0.15091,
      "grad_norm": 0.9485030774003165,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 15091
    },
    {
      "epoch": 0.15092,
      "grad_norm": 1.2492208854554452,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 15092
    },
    {
      "epoch": 0.15093,
      "grad_norm": 1.233147518537233,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 15093
    },
    {
      "epoch": 0.15094,
      "grad_norm": 1.254102187917705,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 15094
    },
    {
      "epoch": 0.15095,
      "grad_norm": 1.0789350378474252,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 15095
    },
    {
      "epoch": 0.15096,
      "grad_norm": 1.2446487998554652,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 15096
    },
    {
      "epoch": 0.15097,
      "grad_norm": 1.3909024001347274,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 15097
    },
    {
      "epoch": 0.15098,
      "grad_norm": 1.6132558578797918,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 15098
    },
    {
      "epoch": 0.15099,
      "grad_norm": 1.3620799585695513,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 15099
    },
    {
      "epoch": 0.151,
      "grad_norm": 1.177533687643014,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 15100
    },
    {
      "epoch": 0.15101,
      "grad_norm": 1.2328881779449214,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 15101
    },
    {
      "epoch": 0.15102,
      "grad_norm": 1.128148852596184,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15102
    },
    {
      "epoch": 0.15103,
      "grad_norm": 1.3633640245779437,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 15103
    },
    {
      "epoch": 0.15104,
      "grad_norm": 1.0922438358343403,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 15104
    },
    {
      "epoch": 0.15105,
      "grad_norm": 1.2945037789067804,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 15105
    },
    {
      "epoch": 0.15106,
      "grad_norm": 1.3206419658757098,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 15106
    },
    {
      "epoch": 0.15107,
      "grad_norm": 1.1755714323357802,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 15107
    },
    {
      "epoch": 0.15108,
      "grad_norm": 1.2204786349814887,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 15108
    },
    {
      "epoch": 0.15109,
      "grad_norm": 1.1234280468452336,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 15109
    },
    {
      "epoch": 0.1511,
      "grad_norm": 1.128668822196183,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 15110
    },
    {
      "epoch": 0.15111,
      "grad_norm": 0.9421908290843221,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 15111
    },
    {
      "epoch": 0.15112,
      "grad_norm": 1.2068180002027014,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 15112
    },
    {
      "epoch": 0.15113,
      "grad_norm": 1.2186237898118937,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 15113
    },
    {
      "epoch": 0.15114,
      "grad_norm": 1.4641406766827232,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 15114
    },
    {
      "epoch": 0.15115,
      "grad_norm": 1.1919232290832495,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 15115
    },
    {
      "epoch": 0.15116,
      "grad_norm": 1.4728097731686103,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 15116
    },
    {
      "epoch": 0.15117,
      "grad_norm": 1.3330637176888094,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 15117
    },
    {
      "epoch": 0.15118,
      "grad_norm": 1.2385592706364272,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 15118
    },
    {
      "epoch": 0.15119,
      "grad_norm": 1.4067574442613973,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 15119
    },
    {
      "epoch": 0.1512,
      "grad_norm": 1.0519760618581901,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 15120
    },
    {
      "epoch": 0.15121,
      "grad_norm": 1.2466957115531099,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 15121
    },
    {
      "epoch": 0.15122,
      "grad_norm": 0.9444434919056375,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 15122
    },
    {
      "epoch": 0.15123,
      "grad_norm": 1.1491075553362577,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 15123
    },
    {
      "epoch": 0.15124,
      "grad_norm": 1.1463798435149848,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 15124
    },
    {
      "epoch": 0.15125,
      "grad_norm": 1.2427319169671964,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 15125
    },
    {
      "epoch": 0.15126,
      "grad_norm": 1.2947749681988245,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 15126
    },
    {
      "epoch": 0.15127,
      "grad_norm": 1.2967915386781237,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 15127
    },
    {
      "epoch": 0.15128,
      "grad_norm": 1.4545692904296599,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 15128
    },
    {
      "epoch": 0.15129,
      "grad_norm": 1.2941914586636714,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 15129
    },
    {
      "epoch": 0.1513,
      "grad_norm": 1.472383577636909,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 15130
    },
    {
      "epoch": 0.15131,
      "grad_norm": 1.0286634832854549,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 15131
    },
    {
      "epoch": 0.15132,
      "grad_norm": 1.2498407259586195,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 15132
    },
    {
      "epoch": 0.15133,
      "grad_norm": 1.0802144509705898,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 15133
    },
    {
      "epoch": 0.15134,
      "grad_norm": 1.246959344241732,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 15134
    },
    {
      "epoch": 0.15135,
      "grad_norm": 1.0771065009220033,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 15135
    },
    {
      "epoch": 0.15136,
      "grad_norm": 1.1340540660269705,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 15136
    },
    {
      "epoch": 0.15137,
      "grad_norm": 1.1862769462900704,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 15137
    },
    {
      "epoch": 0.15138,
      "grad_norm": 1.6039770521958727,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 15138
    },
    {
      "epoch": 0.15139,
      "grad_norm": 1.550446799994394,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 15139
    },
    {
      "epoch": 0.1514,
      "grad_norm": 1.087988808876127,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 15140
    },
    {
      "epoch": 0.15141,
      "grad_norm": 1.2878979634976375,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 15141
    },
    {
      "epoch": 0.15142,
      "grad_norm": 1.4691629254672802,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 15142
    },
    {
      "epoch": 0.15143,
      "grad_norm": 1.249094550079718,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 15143
    },
    {
      "epoch": 0.15144,
      "grad_norm": 1.2074184009542037,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 15144
    },
    {
      "epoch": 0.15145,
      "grad_norm": 1.1486188631745466,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 15145
    },
    {
      "epoch": 0.15146,
      "grad_norm": 1.3538820591910932,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 15146
    },
    {
      "epoch": 0.15147,
      "grad_norm": 1.2287767445354425,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 15147
    },
    {
      "epoch": 0.15148,
      "grad_norm": 1.0631822777129871,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 15148
    },
    {
      "epoch": 0.15149,
      "grad_norm": 1.3665721519318366,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 15149
    },
    {
      "epoch": 0.1515,
      "grad_norm": 1.0531931029076098,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 15150
    },
    {
      "epoch": 0.15151,
      "grad_norm": 1.3861834427558128,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 15151
    },
    {
      "epoch": 0.15152,
      "grad_norm": 1.110165551093399,
      "learning_rate": 0.003,
      "loss": 4.1115,
      "step": 15152
    },
    {
      "epoch": 0.15153,
      "grad_norm": 1.373577010685897,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 15153
    },
    {
      "epoch": 0.15154,
      "grad_norm": 1.0768957629492795,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 15154
    },
    {
      "epoch": 0.15155,
      "grad_norm": 1.3036681944200013,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 15155
    },
    {
      "epoch": 0.15156,
      "grad_norm": 1.0134078913275204,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 15156
    },
    {
      "epoch": 0.15157,
      "grad_norm": 1.3149188647412473,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 15157
    },
    {
      "epoch": 0.15158,
      "grad_norm": 1.0714720525448034,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 15158
    },
    {
      "epoch": 0.15159,
      "grad_norm": 1.4667820080180005,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 15159
    },
    {
      "epoch": 0.1516,
      "grad_norm": 1.0332003953815936,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 15160
    },
    {
      "epoch": 0.15161,
      "grad_norm": 1.5389260212058098,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 15161
    },
    {
      "epoch": 0.15162,
      "grad_norm": 1.3362982356832491,
      "learning_rate": 0.003,
      "loss": 4.1102,
      "step": 15162
    },
    {
      "epoch": 0.15163,
      "grad_norm": 1.3142600675432459,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 15163
    },
    {
      "epoch": 0.15164,
      "grad_norm": 1.0328530820609112,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 15164
    },
    {
      "epoch": 0.15165,
      "grad_norm": 1.3797164887047848,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 15165
    },
    {
      "epoch": 0.15166,
      "grad_norm": 1.4007279063856606,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 15166
    },
    {
      "epoch": 0.15167,
      "grad_norm": 1.0361904823528456,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 15167
    },
    {
      "epoch": 0.15168,
      "grad_norm": 1.3243661623460543,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 15168
    },
    {
      "epoch": 0.15169,
      "grad_norm": 1.3822565238845799,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 15169
    },
    {
      "epoch": 0.1517,
      "grad_norm": 1.2175661961955546,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 15170
    },
    {
      "epoch": 0.15171,
      "grad_norm": 1.1780681534154256,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 15171
    },
    {
      "epoch": 0.15172,
      "grad_norm": 1.2003824494889688,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 15172
    },
    {
      "epoch": 0.15173,
      "grad_norm": 1.1151207640916274,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 15173
    },
    {
      "epoch": 0.15174,
      "grad_norm": 1.1997119339423539,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 15174
    },
    {
      "epoch": 0.15175,
      "grad_norm": 1.0815699925379336,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 15175
    },
    {
      "epoch": 0.15176,
      "grad_norm": 1.6680875842354892,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 15176
    },
    {
      "epoch": 0.15177,
      "grad_norm": 1.177468208006591,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 15177
    },
    {
      "epoch": 0.15178,
      "grad_norm": 1.4624994320532954,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 15178
    },
    {
      "epoch": 0.15179,
      "grad_norm": 1.246454152068424,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 15179
    },
    {
      "epoch": 0.1518,
      "grad_norm": 1.1371684814178287,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 15180
    },
    {
      "epoch": 0.15181,
      "grad_norm": 1.4943192277886241,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 15181
    },
    {
      "epoch": 0.15182,
      "grad_norm": 1.2107772520176932,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 15182
    },
    {
      "epoch": 0.15183,
      "grad_norm": 1.1054710008402246,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 15183
    },
    {
      "epoch": 0.15184,
      "grad_norm": 1.1351215569960855,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 15184
    },
    {
      "epoch": 0.15185,
      "grad_norm": 1.4329950415901205,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 15185
    },
    {
      "epoch": 0.15186,
      "grad_norm": 1.149140906166895,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 15186
    },
    {
      "epoch": 0.15187,
      "grad_norm": 1.2625000256864467,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 15187
    },
    {
      "epoch": 0.15188,
      "grad_norm": 0.9386358846680741,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 15188
    },
    {
      "epoch": 0.15189,
      "grad_norm": 1.1714196410569182,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 15189
    },
    {
      "epoch": 0.1519,
      "grad_norm": 1.2510415235166548,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 15190
    },
    {
      "epoch": 0.15191,
      "grad_norm": 1.2872169549150057,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 15191
    },
    {
      "epoch": 0.15192,
      "grad_norm": 1.3119991182040003,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 15192
    },
    {
      "epoch": 0.15193,
      "grad_norm": 1.0603377530914684,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 15193
    },
    {
      "epoch": 0.15194,
      "grad_norm": 1.2822519431801203,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 15194
    },
    {
      "epoch": 0.15195,
      "grad_norm": 1.004259428252377,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 15195
    },
    {
      "epoch": 0.15196,
      "grad_norm": 1.4231490250930479,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 15196
    },
    {
      "epoch": 0.15197,
      "grad_norm": 1.420169991810222,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 15197
    },
    {
      "epoch": 0.15198,
      "grad_norm": 1.1784357099586809,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 15198
    },
    {
      "epoch": 0.15199,
      "grad_norm": 1.392187510395877,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 15199
    },
    {
      "epoch": 0.152,
      "grad_norm": 1.116211059199764,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 15200
    },
    {
      "epoch": 0.15201,
      "grad_norm": 1.4263333656125654,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 15201
    },
    {
      "epoch": 0.15202,
      "grad_norm": 1.2286807081200959,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 15202
    },
    {
      "epoch": 0.15203,
      "grad_norm": 1.0920509862065366,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 15203
    },
    {
      "epoch": 0.15204,
      "grad_norm": 1.2614998089297877,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 15204
    },
    {
      "epoch": 0.15205,
      "grad_norm": 1.1692633950773663,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 15205
    },
    {
      "epoch": 0.15206,
      "grad_norm": 1.409550414533334,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 15206
    },
    {
      "epoch": 0.15207,
      "grad_norm": 1.3113605146737568,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 15207
    },
    {
      "epoch": 0.15208,
      "grad_norm": 1.309460276567801,
      "learning_rate": 0.003,
      "loss": 4.0322,
      "step": 15208
    },
    {
      "epoch": 0.15209,
      "grad_norm": 1.1894917430330245,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 15209
    },
    {
      "epoch": 0.1521,
      "grad_norm": 1.1703660954370134,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 15210
    },
    {
      "epoch": 0.15211,
      "grad_norm": 1.2989528506275134,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 15211
    },
    {
      "epoch": 0.15212,
      "grad_norm": 1.0765810928628827,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 15212
    },
    {
      "epoch": 0.15213,
      "grad_norm": 1.4756608926855814,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 15213
    },
    {
      "epoch": 0.15214,
      "grad_norm": 1.1912439313479937,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 15214
    },
    {
      "epoch": 0.15215,
      "grad_norm": 1.006524948376339,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 15215
    },
    {
      "epoch": 0.15216,
      "grad_norm": 1.261103228412681,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 15216
    },
    {
      "epoch": 0.15217,
      "grad_norm": 1.1249186122517645,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 15217
    },
    {
      "epoch": 0.15218,
      "grad_norm": 1.278349145361421,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 15218
    },
    {
      "epoch": 0.15219,
      "grad_norm": 1.0713858095086553,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 15219
    },
    {
      "epoch": 0.1522,
      "grad_norm": 1.6026457850936104,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 15220
    },
    {
      "epoch": 0.15221,
      "grad_norm": 1.0833468242383575,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 15221
    },
    {
      "epoch": 0.15222,
      "grad_norm": 1.3253246748494185,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 15222
    },
    {
      "epoch": 0.15223,
      "grad_norm": 1.3226742938326956,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 15223
    },
    {
      "epoch": 0.15224,
      "grad_norm": 1.0668314677110395,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 15224
    },
    {
      "epoch": 0.15225,
      "grad_norm": 1.339893976702488,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 15225
    },
    {
      "epoch": 0.15226,
      "grad_norm": 0.9909076328643073,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 15226
    },
    {
      "epoch": 0.15227,
      "grad_norm": 1.3457475132936831,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 15227
    },
    {
      "epoch": 0.15228,
      "grad_norm": 1.223750730023511,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 15228
    },
    {
      "epoch": 0.15229,
      "grad_norm": 1.256632194523041,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 15229
    },
    {
      "epoch": 0.1523,
      "grad_norm": 1.059870090530304,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 15230
    },
    {
      "epoch": 0.15231,
      "grad_norm": 1.689885438587469,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 15231
    },
    {
      "epoch": 0.15232,
      "grad_norm": 1.181258046402351,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 15232
    },
    {
      "epoch": 0.15233,
      "grad_norm": 1.5769078838020736,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 15233
    },
    {
      "epoch": 0.15234,
      "grad_norm": 0.9274230056976925,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 15234
    },
    {
      "epoch": 0.15235,
      "grad_norm": 1.2926060469909377,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 15235
    },
    {
      "epoch": 0.15236,
      "grad_norm": 1.240377557258737,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 15236
    },
    {
      "epoch": 0.15237,
      "grad_norm": 1.298389514130426,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 15237
    },
    {
      "epoch": 0.15238,
      "grad_norm": 1.2163521689730095,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 15238
    },
    {
      "epoch": 0.15239,
      "grad_norm": 1.0937131665994184,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 15239
    },
    {
      "epoch": 0.1524,
      "grad_norm": 1.3494526264556403,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 15240
    },
    {
      "epoch": 0.15241,
      "grad_norm": 1.2086076171424276,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 15241
    },
    {
      "epoch": 0.15242,
      "grad_norm": 1.2275639454465173,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 15242
    },
    {
      "epoch": 0.15243,
      "grad_norm": 1.1768041134744753,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 15243
    },
    {
      "epoch": 0.15244,
      "grad_norm": 1.1460609617592146,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 15244
    },
    {
      "epoch": 0.15245,
      "grad_norm": 1.0362631941590839,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 15245
    },
    {
      "epoch": 0.15246,
      "grad_norm": 1.2221248760518095,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 15246
    },
    {
      "epoch": 0.15247,
      "grad_norm": 1.095691988113859,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 15247
    },
    {
      "epoch": 0.15248,
      "grad_norm": 1.4252171113365828,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 15248
    },
    {
      "epoch": 0.15249,
      "grad_norm": 0.9907212320633807,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 15249
    },
    {
      "epoch": 0.1525,
      "grad_norm": 1.1648163444936928,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 15250
    },
    {
      "epoch": 0.15251,
      "grad_norm": 1.3792381729375842,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 15251
    },
    {
      "epoch": 0.15252,
      "grad_norm": 1.0966643893758867,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 15252
    },
    {
      "epoch": 0.15253,
      "grad_norm": 1.352568157703781,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 15253
    },
    {
      "epoch": 0.15254,
      "grad_norm": 1.162965904860133,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 15254
    },
    {
      "epoch": 0.15255,
      "grad_norm": 1.230108589317307,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 15255
    },
    {
      "epoch": 0.15256,
      "grad_norm": 1.306415003864345,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 15256
    },
    {
      "epoch": 0.15257,
      "grad_norm": 1.2698699716294912,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 15257
    },
    {
      "epoch": 0.15258,
      "grad_norm": 1.2202436427791954,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 15258
    },
    {
      "epoch": 0.15259,
      "grad_norm": 1.3699491692866101,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15259
    },
    {
      "epoch": 0.1526,
      "grad_norm": 1.2865829303939478,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 15260
    },
    {
      "epoch": 0.15261,
      "grad_norm": 1.1907326806337453,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 15261
    },
    {
      "epoch": 0.15262,
      "grad_norm": 1.3741574656043807,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 15262
    },
    {
      "epoch": 0.15263,
      "grad_norm": 1.2461198602893133,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 15263
    },
    {
      "epoch": 0.15264,
      "grad_norm": 1.3037038166580812,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 15264
    },
    {
      "epoch": 0.15265,
      "grad_norm": 1.0973837487129527,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 15265
    },
    {
      "epoch": 0.15266,
      "grad_norm": 1.17550255547033,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 15266
    },
    {
      "epoch": 0.15267,
      "grad_norm": 1.350055319387001,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 15267
    },
    {
      "epoch": 0.15268,
      "grad_norm": 1.2135196925321778,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 15268
    },
    {
      "epoch": 0.15269,
      "grad_norm": 1.052807123240445,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 15269
    },
    {
      "epoch": 0.1527,
      "grad_norm": 1.5162917289975957,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 15270
    },
    {
      "epoch": 0.15271,
      "grad_norm": 0.9087099666900801,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 15271
    },
    {
      "epoch": 0.15272,
      "grad_norm": 1.2080481623654342,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 15272
    },
    {
      "epoch": 0.15273,
      "grad_norm": 1.205366680073686,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 15273
    },
    {
      "epoch": 0.15274,
      "grad_norm": 1.2646904527024778,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 15274
    },
    {
      "epoch": 0.15275,
      "grad_norm": 1.5286956874796136,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 15275
    },
    {
      "epoch": 0.15276,
      "grad_norm": 1.0009018904534552,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 15276
    },
    {
      "epoch": 0.15277,
      "grad_norm": 1.2494785309856136,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 15277
    },
    {
      "epoch": 0.15278,
      "grad_norm": 1.2861304222588785,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 15278
    },
    {
      "epoch": 0.15279,
      "grad_norm": 1.1892301729765877,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 15279
    },
    {
      "epoch": 0.1528,
      "grad_norm": 1.1204842553347314,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 15280
    },
    {
      "epoch": 0.15281,
      "grad_norm": 1.379822288030074,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 15281
    },
    {
      "epoch": 0.15282,
      "grad_norm": 1.183577530960742,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 15282
    },
    {
      "epoch": 0.15283,
      "grad_norm": 1.2435364515876217,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 15283
    },
    {
      "epoch": 0.15284,
      "grad_norm": 1.1296979253872497,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 15284
    },
    {
      "epoch": 0.15285,
      "grad_norm": 1.3123458983768324,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 15285
    },
    {
      "epoch": 0.15286,
      "grad_norm": 1.3325798748619977,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 15286
    },
    {
      "epoch": 0.15287,
      "grad_norm": 1.250202260518972,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 15287
    },
    {
      "epoch": 0.15288,
      "grad_norm": 1.322467529630218,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 15288
    },
    {
      "epoch": 0.15289,
      "grad_norm": 1.294869647610876,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 15289
    },
    {
      "epoch": 0.1529,
      "grad_norm": 1.0919572359477165,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 15290
    },
    {
      "epoch": 0.15291,
      "grad_norm": 1.2314711363464752,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 15291
    },
    {
      "epoch": 0.15292,
      "grad_norm": 1.1864497583011213,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15292
    },
    {
      "epoch": 0.15293,
      "grad_norm": 1.273989288109886,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 15293
    },
    {
      "epoch": 0.15294,
      "grad_norm": 1.1428627955755606,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 15294
    },
    {
      "epoch": 0.15295,
      "grad_norm": 1.3318276639584221,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 15295
    },
    {
      "epoch": 0.15296,
      "grad_norm": 0.9996627595976725,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 15296
    },
    {
      "epoch": 0.15297,
      "grad_norm": 1.3567545643645589,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 15297
    },
    {
      "epoch": 0.15298,
      "grad_norm": 1.0888841094184967,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15298
    },
    {
      "epoch": 0.15299,
      "grad_norm": 1.3143604999336704,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 15299
    },
    {
      "epoch": 0.153,
      "grad_norm": 1.10478291680258,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 15300
    },
    {
      "epoch": 0.15301,
      "grad_norm": 1.332542631690166,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 15301
    },
    {
      "epoch": 0.15302,
      "grad_norm": 1.3610143355801558,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 15302
    },
    {
      "epoch": 0.15303,
      "grad_norm": 1.6499931279519562,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 15303
    },
    {
      "epoch": 0.15304,
      "grad_norm": 0.9729268294633743,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 15304
    },
    {
      "epoch": 0.15305,
      "grad_norm": 1.2812268645301028,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 15305
    },
    {
      "epoch": 0.15306,
      "grad_norm": 1.3219200685292423,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 15306
    },
    {
      "epoch": 0.15307,
      "grad_norm": 1.161016281668427,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 15307
    },
    {
      "epoch": 0.15308,
      "grad_norm": 1.117450920302711,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 15308
    },
    {
      "epoch": 0.15309,
      "grad_norm": 1.197542675446433,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 15309
    },
    {
      "epoch": 0.1531,
      "grad_norm": 1.3599553040681978,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 15310
    },
    {
      "epoch": 0.15311,
      "grad_norm": 1.3874569351603045,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 15311
    },
    {
      "epoch": 0.15312,
      "grad_norm": 1.2480329082893813,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 15312
    },
    {
      "epoch": 0.15313,
      "grad_norm": 1.1836604804725042,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 15313
    },
    {
      "epoch": 0.15314,
      "grad_norm": 1.3363089275512905,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 15314
    },
    {
      "epoch": 0.15315,
      "grad_norm": 1.2005744273055012,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 15315
    },
    {
      "epoch": 0.15316,
      "grad_norm": 1.272784295793868,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 15316
    },
    {
      "epoch": 0.15317,
      "grad_norm": 1.028788498800255,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 15317
    },
    {
      "epoch": 0.15318,
      "grad_norm": 1.5934021220240417,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 15318
    },
    {
      "epoch": 0.15319,
      "grad_norm": 1.0956734498965002,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 15319
    },
    {
      "epoch": 0.1532,
      "grad_norm": 1.6551031206932727,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 15320
    },
    {
      "epoch": 0.15321,
      "grad_norm": 1.2102759249965225,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 15321
    },
    {
      "epoch": 0.15322,
      "grad_norm": 1.1370647599775683,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 15322
    },
    {
      "epoch": 0.15323,
      "grad_norm": 1.3220558450236222,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 15323
    },
    {
      "epoch": 0.15324,
      "grad_norm": 1.2840062337198395,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 15324
    },
    {
      "epoch": 0.15325,
      "grad_norm": 1.3013782425747533,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 15325
    },
    {
      "epoch": 0.15326,
      "grad_norm": 1.1975442037482786,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 15326
    },
    {
      "epoch": 0.15327,
      "grad_norm": 1.1269489485313762,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 15327
    },
    {
      "epoch": 0.15328,
      "grad_norm": 1.1410612864581509,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 15328
    },
    {
      "epoch": 0.15329,
      "grad_norm": 1.3460362322188562,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 15329
    },
    {
      "epoch": 0.1533,
      "grad_norm": 1.3637334924767632,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 15330
    },
    {
      "epoch": 0.15331,
      "grad_norm": 1.0996693942208156,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 15331
    },
    {
      "epoch": 0.15332,
      "grad_norm": 1.3134945624775438,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 15332
    },
    {
      "epoch": 0.15333,
      "grad_norm": 0.9603032493816029,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 15333
    },
    {
      "epoch": 0.15334,
      "grad_norm": 1.4163721083411738,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 15334
    },
    {
      "epoch": 0.15335,
      "grad_norm": 1.0705217711953943,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 15335
    },
    {
      "epoch": 0.15336,
      "grad_norm": 1.2004877087986423,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 15336
    },
    {
      "epoch": 0.15337,
      "grad_norm": 1.2784195202324355,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 15337
    },
    {
      "epoch": 0.15338,
      "grad_norm": 0.9428270674974255,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 15338
    },
    {
      "epoch": 0.15339,
      "grad_norm": 1.232716499678335,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 15339
    },
    {
      "epoch": 0.1534,
      "grad_norm": 1.3422285090284969,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 15340
    },
    {
      "epoch": 0.15341,
      "grad_norm": 1.0887037346989066,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 15341
    },
    {
      "epoch": 0.15342,
      "grad_norm": 1.40533166454307,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 15342
    },
    {
      "epoch": 0.15343,
      "grad_norm": 1.2808704916445435,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 15343
    },
    {
      "epoch": 0.15344,
      "grad_norm": 1.7381559003301226,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 15344
    },
    {
      "epoch": 0.15345,
      "grad_norm": 0.8981993766105962,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 15345
    },
    {
      "epoch": 0.15346,
      "grad_norm": 1.021591731175303,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 15346
    },
    {
      "epoch": 0.15347,
      "grad_norm": 1.342833780617317,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 15347
    },
    {
      "epoch": 0.15348,
      "grad_norm": 1.055504084451775,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 15348
    },
    {
      "epoch": 0.15349,
      "grad_norm": 1.4292975403438253,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 15349
    },
    {
      "epoch": 0.1535,
      "grad_norm": 1.242457260105871,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 15350
    },
    {
      "epoch": 0.15351,
      "grad_norm": 1.607477705777395,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 15351
    },
    {
      "epoch": 0.15352,
      "grad_norm": 1.0042952049385223,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 15352
    },
    {
      "epoch": 0.15353,
      "grad_norm": 1.5215221222668274,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15353
    },
    {
      "epoch": 0.15354,
      "grad_norm": 1.1167885093181256,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 15354
    },
    {
      "epoch": 0.15355,
      "grad_norm": 1.337051204486555,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 15355
    },
    {
      "epoch": 0.15356,
      "grad_norm": 1.2014364989112793,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 15356
    },
    {
      "epoch": 0.15357,
      "grad_norm": 1.0611147472963682,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15357
    },
    {
      "epoch": 0.15358,
      "grad_norm": 1.4014457304898078,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 15358
    },
    {
      "epoch": 0.15359,
      "grad_norm": 1.0441717054483164,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 15359
    },
    {
      "epoch": 0.1536,
      "grad_norm": 1.2532453030452138,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 15360
    },
    {
      "epoch": 0.15361,
      "grad_norm": 1.0668416216593732,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 15361
    },
    {
      "epoch": 0.15362,
      "grad_norm": 1.44348968864017,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 15362
    },
    {
      "epoch": 0.15363,
      "grad_norm": 1.0324397895486972,
      "learning_rate": 0.003,
      "loss": 4.1054,
      "step": 15363
    },
    {
      "epoch": 0.15364,
      "grad_norm": 1.4302096146858072,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 15364
    },
    {
      "epoch": 0.15365,
      "grad_norm": 1.445770752562098,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 15365
    },
    {
      "epoch": 0.15366,
      "grad_norm": 1.4021681092834855,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 15366
    },
    {
      "epoch": 0.15367,
      "grad_norm": 1.1844554825553057,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 15367
    },
    {
      "epoch": 0.15368,
      "grad_norm": 1.3157634932169993,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 15368
    },
    {
      "epoch": 0.15369,
      "grad_norm": 1.5005684371665555,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 15369
    },
    {
      "epoch": 0.1537,
      "grad_norm": 1.1268061902873954,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 15370
    },
    {
      "epoch": 0.15371,
      "grad_norm": 1.2839413286799333,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 15371
    },
    {
      "epoch": 0.15372,
      "grad_norm": 1.401179204053029,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 15372
    },
    {
      "epoch": 0.15373,
      "grad_norm": 1.1167028590635872,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 15373
    },
    {
      "epoch": 0.15374,
      "grad_norm": 1.4268777343963048,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 15374
    },
    {
      "epoch": 0.15375,
      "grad_norm": 1.2064984743607405,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 15375
    },
    {
      "epoch": 0.15376,
      "grad_norm": 1.0546248966109226,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 15376
    },
    {
      "epoch": 0.15377,
      "grad_norm": 1.157082907867689,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 15377
    },
    {
      "epoch": 0.15378,
      "grad_norm": 1.0936352419546245,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 15378
    },
    {
      "epoch": 0.15379,
      "grad_norm": 1.2548739883846005,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 15379
    },
    {
      "epoch": 0.1538,
      "grad_norm": 1.4031950487192069,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 15380
    },
    {
      "epoch": 0.15381,
      "grad_norm": 1.2224787943842523,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 15381
    },
    {
      "epoch": 0.15382,
      "grad_norm": 1.2236705526083234,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 15382
    },
    {
      "epoch": 0.15383,
      "grad_norm": 1.0648812817113953,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 15383
    },
    {
      "epoch": 0.15384,
      "grad_norm": 1.2879123629622429,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 15384
    },
    {
      "epoch": 0.15385,
      "grad_norm": 1.012268736905604,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 15385
    },
    {
      "epoch": 0.15386,
      "grad_norm": 1.2187582005774895,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 15386
    },
    {
      "epoch": 0.15387,
      "grad_norm": 1.1884376651243376,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 15387
    },
    {
      "epoch": 0.15388,
      "grad_norm": 1.3226916097039858,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15388
    },
    {
      "epoch": 0.15389,
      "grad_norm": 1.3303186594915881,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 15389
    },
    {
      "epoch": 0.1539,
      "grad_norm": 1.3866925840894608,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 15390
    },
    {
      "epoch": 0.15391,
      "grad_norm": 1.3728111518143102,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 15391
    },
    {
      "epoch": 0.15392,
      "grad_norm": 1.3402808808497404,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 15392
    },
    {
      "epoch": 0.15393,
      "grad_norm": 1.1579581855029002,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 15393
    },
    {
      "epoch": 0.15394,
      "grad_norm": 1.2034269596203029,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 15394
    },
    {
      "epoch": 0.15395,
      "grad_norm": 1.1618115125409054,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 15395
    },
    {
      "epoch": 0.15396,
      "grad_norm": 1.2863291450699967,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 15396
    },
    {
      "epoch": 0.15397,
      "grad_norm": 1.2100596132475223,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 15397
    },
    {
      "epoch": 0.15398,
      "grad_norm": 1.3418803257466523,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 15398
    },
    {
      "epoch": 0.15399,
      "grad_norm": 0.8759960471617386,
      "learning_rate": 0.003,
      "loss": 4.0244,
      "step": 15399
    },
    {
      "epoch": 0.154,
      "grad_norm": 0.9909269155497589,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 15400
    },
    {
      "epoch": 0.15401,
      "grad_norm": 1.280760201565616,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 15401
    },
    {
      "epoch": 0.15402,
      "grad_norm": 1.3039485153919603,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 15402
    },
    {
      "epoch": 0.15403,
      "grad_norm": 1.1319622563485687,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 15403
    },
    {
      "epoch": 0.15404,
      "grad_norm": 1.5872253219858796,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 15404
    },
    {
      "epoch": 0.15405,
      "grad_norm": 1.1610910153502412,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 15405
    },
    {
      "epoch": 0.15406,
      "grad_norm": 1.3946380911331175,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 15406
    },
    {
      "epoch": 0.15407,
      "grad_norm": 1.1320665509097738,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 15407
    },
    {
      "epoch": 0.15408,
      "grad_norm": 1.3176567139986404,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 15408
    },
    {
      "epoch": 0.15409,
      "grad_norm": 1.3445347155612486,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 15409
    },
    {
      "epoch": 0.1541,
      "grad_norm": 1.2194685040205706,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 15410
    },
    {
      "epoch": 0.15411,
      "grad_norm": 1.217023193541309,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 15411
    },
    {
      "epoch": 0.15412,
      "grad_norm": 1.1999213085696572,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 15412
    },
    {
      "epoch": 0.15413,
      "grad_norm": 1.2812853286852148,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 15413
    },
    {
      "epoch": 0.15414,
      "grad_norm": 1.3143368525425625,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 15414
    },
    {
      "epoch": 0.15415,
      "grad_norm": 1.3786879212302876,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 15415
    },
    {
      "epoch": 0.15416,
      "grad_norm": 1.1378575541792832,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 15416
    },
    {
      "epoch": 0.15417,
      "grad_norm": 1.4468712668422241,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 15417
    },
    {
      "epoch": 0.15418,
      "grad_norm": 1.1350194620282688,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 15418
    },
    {
      "epoch": 0.15419,
      "grad_norm": 1.3058088501430103,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 15419
    },
    {
      "epoch": 0.1542,
      "grad_norm": 1.2290528501643871,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 15420
    },
    {
      "epoch": 0.15421,
      "grad_norm": 1.34099276740831,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 15421
    },
    {
      "epoch": 0.15422,
      "grad_norm": 1.2689686644532474,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 15422
    },
    {
      "epoch": 0.15423,
      "grad_norm": 1.1798623382145859,
      "learning_rate": 0.003,
      "loss": 4.0193,
      "step": 15423
    },
    {
      "epoch": 0.15424,
      "grad_norm": 1.1414205811208402,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 15424
    },
    {
      "epoch": 0.15425,
      "grad_norm": 1.382624074165894,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 15425
    },
    {
      "epoch": 0.15426,
      "grad_norm": 1.0098622350888455,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 15426
    },
    {
      "epoch": 0.15427,
      "grad_norm": 1.2158447740165523,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 15427
    },
    {
      "epoch": 0.15428,
      "grad_norm": 1.0800505764651744,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 15428
    },
    {
      "epoch": 0.15429,
      "grad_norm": 1.4389797603354793,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 15429
    },
    {
      "epoch": 0.1543,
      "grad_norm": 0.9131784264756577,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 15430
    },
    {
      "epoch": 0.15431,
      "grad_norm": 1.1406774347349673,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 15431
    },
    {
      "epoch": 0.15432,
      "grad_norm": 1.4160712255500911,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 15432
    },
    {
      "epoch": 0.15433,
      "grad_norm": 1.0321017123231477,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 15433
    },
    {
      "epoch": 0.15434,
      "grad_norm": 1.3129401837773598,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 15434
    },
    {
      "epoch": 0.15435,
      "grad_norm": 1.295097177277385,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 15435
    },
    {
      "epoch": 0.15436,
      "grad_norm": 1.3445014318375115,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 15436
    },
    {
      "epoch": 0.15437,
      "grad_norm": 1.2037512242094492,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 15437
    },
    {
      "epoch": 0.15438,
      "grad_norm": 1.2204961008983182,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 15438
    },
    {
      "epoch": 0.15439,
      "grad_norm": 1.1936893451355557,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 15439
    },
    {
      "epoch": 0.1544,
      "grad_norm": 1.110082884413141,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 15440
    },
    {
      "epoch": 0.15441,
      "grad_norm": 1.17060566072975,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 15441
    },
    {
      "epoch": 0.15442,
      "grad_norm": 1.2001477195120134,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 15442
    },
    {
      "epoch": 0.15443,
      "grad_norm": 1.2014003725220923,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 15443
    },
    {
      "epoch": 0.15444,
      "grad_norm": 1.2729262505707326,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 15444
    },
    {
      "epoch": 0.15445,
      "grad_norm": 1.3667928177643467,
      "learning_rate": 0.003,
      "loss": 4.0304,
      "step": 15445
    },
    {
      "epoch": 0.15446,
      "grad_norm": 1.1717222335822768,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 15446
    },
    {
      "epoch": 0.15447,
      "grad_norm": 1.4097833628228227,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 15447
    },
    {
      "epoch": 0.15448,
      "grad_norm": 1.4843293064081833,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 15448
    },
    {
      "epoch": 0.15449,
      "grad_norm": 1.073061414519865,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 15449
    },
    {
      "epoch": 0.1545,
      "grad_norm": 1.4751810338681888,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 15450
    },
    {
      "epoch": 0.15451,
      "grad_norm": 1.1619538323488983,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 15451
    },
    {
      "epoch": 0.15452,
      "grad_norm": 1.4427250183680014,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 15452
    },
    {
      "epoch": 0.15453,
      "grad_norm": 1.0283716662610662,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 15453
    },
    {
      "epoch": 0.15454,
      "grad_norm": 1.2388449133718995,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 15454
    },
    {
      "epoch": 0.15455,
      "grad_norm": 1.2397170781988538,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 15455
    },
    {
      "epoch": 0.15456,
      "grad_norm": 1.335769548973706,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 15456
    },
    {
      "epoch": 0.15457,
      "grad_norm": 1.2550274367262313,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 15457
    },
    {
      "epoch": 0.15458,
      "grad_norm": 1.6013706385545896,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 15458
    },
    {
      "epoch": 0.15459,
      "grad_norm": 1.4113908926528906,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 15459
    },
    {
      "epoch": 0.1546,
      "grad_norm": 1.1872880676167363,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 15460
    },
    {
      "epoch": 0.15461,
      "grad_norm": 1.1602721882141036,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 15461
    },
    {
      "epoch": 0.15462,
      "grad_norm": 1.2657351307427533,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 15462
    },
    {
      "epoch": 0.15463,
      "grad_norm": 1.017266066341593,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 15463
    },
    {
      "epoch": 0.15464,
      "grad_norm": 1.22408385113059,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 15464
    },
    {
      "epoch": 0.15465,
      "grad_norm": 1.2388371752459086,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 15465
    },
    {
      "epoch": 0.15466,
      "grad_norm": 1.3378166995203025,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 15466
    },
    {
      "epoch": 0.15467,
      "grad_norm": 1.2673971792020895,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 15467
    },
    {
      "epoch": 0.15468,
      "grad_norm": 1.2670871657029013,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 15468
    },
    {
      "epoch": 0.15469,
      "grad_norm": 1.4855845526650822,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 15469
    },
    {
      "epoch": 0.1547,
      "grad_norm": 1.1664566423581224,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 15470
    },
    {
      "epoch": 0.15471,
      "grad_norm": 1.43333578071656,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 15471
    },
    {
      "epoch": 0.15472,
      "grad_norm": 0.8916827173933783,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 15472
    },
    {
      "epoch": 0.15473,
      "grad_norm": 1.2228661035293775,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 15473
    },
    {
      "epoch": 0.15474,
      "grad_norm": 1.3045348002003534,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 15474
    },
    {
      "epoch": 0.15475,
      "grad_norm": 0.9752922594250674,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15475
    },
    {
      "epoch": 0.15476,
      "grad_norm": 1.399714944150617,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 15476
    },
    {
      "epoch": 0.15477,
      "grad_norm": 0.967828553650773,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 15477
    },
    {
      "epoch": 0.15478,
      "grad_norm": 1.54376698998061,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 15478
    },
    {
      "epoch": 0.15479,
      "grad_norm": 1.111321915290337,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 15479
    },
    {
      "epoch": 0.1548,
      "grad_norm": 1.3432186028132298,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 15480
    },
    {
      "epoch": 0.15481,
      "grad_norm": 1.2045266952666287,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 15481
    },
    {
      "epoch": 0.15482,
      "grad_norm": 1.5509962795872299,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 15482
    },
    {
      "epoch": 0.15483,
      "grad_norm": 1.038507871352347,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 15483
    },
    {
      "epoch": 0.15484,
      "grad_norm": 1.3999752602376994,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 15484
    },
    {
      "epoch": 0.15485,
      "grad_norm": 1.2488920313017315,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 15485
    },
    {
      "epoch": 0.15486,
      "grad_norm": 1.2928076344043185,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 15486
    },
    {
      "epoch": 0.15487,
      "grad_norm": 1.1813696870402486,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 15487
    },
    {
      "epoch": 0.15488,
      "grad_norm": 1.4851073401211132,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 15488
    },
    {
      "epoch": 0.15489,
      "grad_norm": 1.005290261773753,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 15489
    },
    {
      "epoch": 0.1549,
      "grad_norm": 1.4472636495984204,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 15490
    },
    {
      "epoch": 0.15491,
      "grad_norm": 1.1113667748653209,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 15491
    },
    {
      "epoch": 0.15492,
      "grad_norm": 1.1923215483799454,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 15492
    },
    {
      "epoch": 0.15493,
      "grad_norm": 1.3557008672418909,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 15493
    },
    {
      "epoch": 0.15494,
      "grad_norm": 1.2625023638728134,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 15494
    },
    {
      "epoch": 0.15495,
      "grad_norm": 1.486884034569414,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 15495
    },
    {
      "epoch": 0.15496,
      "grad_norm": 1.0947414987682853,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 15496
    },
    {
      "epoch": 0.15497,
      "grad_norm": 1.3444649580568884,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 15497
    },
    {
      "epoch": 0.15498,
      "grad_norm": 1.2517682754123742,
      "learning_rate": 0.003,
      "loss": 4.0229,
      "step": 15498
    },
    {
      "epoch": 0.15499,
      "grad_norm": 1.2939405259023302,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 15499
    },
    {
      "epoch": 0.155,
      "grad_norm": 1.2833960560786328,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 15500
    },
    {
      "epoch": 0.15501,
      "grad_norm": 1.1128859882325806,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 15501
    },
    {
      "epoch": 0.15502,
      "grad_norm": 1.1979063438901953,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 15502
    },
    {
      "epoch": 0.15503,
      "grad_norm": 1.3011241842981904,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 15503
    },
    {
      "epoch": 0.15504,
      "grad_norm": 1.327429675523034,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 15504
    },
    {
      "epoch": 0.15505,
      "grad_norm": 1.037004976848662,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 15505
    },
    {
      "epoch": 0.15506,
      "grad_norm": 1.358924356732899,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 15506
    },
    {
      "epoch": 0.15507,
      "grad_norm": 1.0333510095833958,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 15507
    },
    {
      "epoch": 0.15508,
      "grad_norm": 1.6523371746178406,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 15508
    },
    {
      "epoch": 0.15509,
      "grad_norm": 1.151636731367369,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 15509
    },
    {
      "epoch": 0.1551,
      "grad_norm": 1.5218639023836638,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 15510
    },
    {
      "epoch": 0.15511,
      "grad_norm": 0.9612598194560834,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 15511
    },
    {
      "epoch": 0.15512,
      "grad_norm": 1.1547449258340936,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 15512
    },
    {
      "epoch": 0.15513,
      "grad_norm": 0.9827899496468655,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 15513
    },
    {
      "epoch": 0.15514,
      "grad_norm": 1.3755150660300457,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 15514
    },
    {
      "epoch": 0.15515,
      "grad_norm": 1.1498310420367837,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 15515
    },
    {
      "epoch": 0.15516,
      "grad_norm": 1.26169842384535,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 15516
    },
    {
      "epoch": 0.15517,
      "grad_norm": 1.2913450985149533,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 15517
    },
    {
      "epoch": 0.15518,
      "grad_norm": 1.3689027325340566,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 15518
    },
    {
      "epoch": 0.15519,
      "grad_norm": 1.11909032130159,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 15519
    },
    {
      "epoch": 0.1552,
      "grad_norm": 1.3565836554568418,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 15520
    },
    {
      "epoch": 0.15521,
      "grad_norm": 1.1839474669898502,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 15521
    },
    {
      "epoch": 0.15522,
      "grad_norm": 1.2984200170694058,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 15522
    },
    {
      "epoch": 0.15523,
      "grad_norm": 1.0926685665559404,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 15523
    },
    {
      "epoch": 0.15524,
      "grad_norm": 1.2953543711923454,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 15524
    },
    {
      "epoch": 0.15525,
      "grad_norm": 1.1528653541075322,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 15525
    },
    {
      "epoch": 0.15526,
      "grad_norm": 1.5481099803544527,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 15526
    },
    {
      "epoch": 0.15527,
      "grad_norm": 1.1766676107693896,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 15527
    },
    {
      "epoch": 0.15528,
      "grad_norm": 1.323063424820846,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 15528
    },
    {
      "epoch": 0.15529,
      "grad_norm": 1.2147770426421365,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 15529
    },
    {
      "epoch": 0.1553,
      "grad_norm": 1.4566094408042551,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 15530
    },
    {
      "epoch": 0.15531,
      "grad_norm": 1.0736352163009661,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 15531
    },
    {
      "epoch": 0.15532,
      "grad_norm": 1.2415821433548437,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 15532
    },
    {
      "epoch": 0.15533,
      "grad_norm": 1.2764666863250977,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 15533
    },
    {
      "epoch": 0.15534,
      "grad_norm": 1.2784009664834688,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 15534
    },
    {
      "epoch": 0.15535,
      "grad_norm": 1.2456733455217581,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 15535
    },
    {
      "epoch": 0.15536,
      "grad_norm": 1.4517271950516681,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 15536
    },
    {
      "epoch": 0.15537,
      "grad_norm": 1.054271128214894,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 15537
    },
    {
      "epoch": 0.15538,
      "grad_norm": 1.3506048182368122,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 15538
    },
    {
      "epoch": 0.15539,
      "grad_norm": 1.1058801360274975,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 15539
    },
    {
      "epoch": 0.1554,
      "grad_norm": 1.1772540177736688,
      "learning_rate": 0.003,
      "loss": 4.1028,
      "step": 15540
    },
    {
      "epoch": 0.15541,
      "grad_norm": 1.148712931209117,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 15541
    },
    {
      "epoch": 0.15542,
      "grad_norm": 1.2879276566085887,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 15542
    },
    {
      "epoch": 0.15543,
      "grad_norm": 1.276845100107273,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 15543
    },
    {
      "epoch": 0.15544,
      "grad_norm": 1.3452703390926941,
      "learning_rate": 0.003,
      "loss": 4.1167,
      "step": 15544
    },
    {
      "epoch": 0.15545,
      "grad_norm": 1.472933607340616,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 15545
    },
    {
      "epoch": 0.15546,
      "grad_norm": 1.119633066296943,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 15546
    },
    {
      "epoch": 0.15547,
      "grad_norm": 1.370763514248208,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 15547
    },
    {
      "epoch": 0.15548,
      "grad_norm": 1.112782868303845,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 15548
    },
    {
      "epoch": 0.15549,
      "grad_norm": 1.3146426369934066,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 15549
    },
    {
      "epoch": 0.1555,
      "grad_norm": 1.1262815670981912,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 15550
    },
    {
      "epoch": 0.15551,
      "grad_norm": 1.4412305467903548,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 15551
    },
    {
      "epoch": 0.15552,
      "grad_norm": 1.0376583191100974,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 15552
    },
    {
      "epoch": 0.15553,
      "grad_norm": 1.4330994659926226,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 15553
    },
    {
      "epoch": 0.15554,
      "grad_norm": 1.0885961883362105,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 15554
    },
    {
      "epoch": 0.15555,
      "grad_norm": 1.6627063289448987,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 15555
    },
    {
      "epoch": 0.15556,
      "grad_norm": 1.1627993597262296,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 15556
    },
    {
      "epoch": 0.15557,
      "grad_norm": 1.181147045209965,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 15557
    },
    {
      "epoch": 0.15558,
      "grad_norm": 1.2773965711076947,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 15558
    },
    {
      "epoch": 0.15559,
      "grad_norm": 1.3169844120615914,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 15559
    },
    {
      "epoch": 0.1556,
      "grad_norm": 1.141942025857773,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 15560
    },
    {
      "epoch": 0.15561,
      "grad_norm": 1.1580066271965455,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 15561
    },
    {
      "epoch": 0.15562,
      "grad_norm": 1.2786956245622974,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 15562
    },
    {
      "epoch": 0.15563,
      "grad_norm": 1.1445034130736649,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 15563
    },
    {
      "epoch": 0.15564,
      "grad_norm": 1.2313003477981035,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 15564
    },
    {
      "epoch": 0.15565,
      "grad_norm": 1.0935705973884606,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 15565
    },
    {
      "epoch": 0.15566,
      "grad_norm": 1.324605016905411,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 15566
    },
    {
      "epoch": 0.15567,
      "grad_norm": 1.115542163926082,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 15567
    },
    {
      "epoch": 0.15568,
      "grad_norm": 1.2483062696111795,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 15568
    },
    {
      "epoch": 0.15569,
      "grad_norm": 1.5644969184712574,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 15569
    },
    {
      "epoch": 0.1557,
      "grad_norm": 1.2953495515696785,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 15570
    },
    {
      "epoch": 0.15571,
      "grad_norm": 1.4506731944335949,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 15571
    },
    {
      "epoch": 0.15572,
      "grad_norm": 1.1401530574177883,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 15572
    },
    {
      "epoch": 0.15573,
      "grad_norm": 1.4880401906606766,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 15573
    },
    {
      "epoch": 0.15574,
      "grad_norm": 0.9606893059570549,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 15574
    },
    {
      "epoch": 0.15575,
      "grad_norm": 1.364810777937475,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 15575
    },
    {
      "epoch": 0.15576,
      "grad_norm": 1.2727795282851142,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 15576
    },
    {
      "epoch": 0.15577,
      "grad_norm": 1.1629652085594882,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 15577
    },
    {
      "epoch": 0.15578,
      "grad_norm": 1.512206308929831,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 15578
    },
    {
      "epoch": 0.15579,
      "grad_norm": 1.2138603565111312,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 15579
    },
    {
      "epoch": 0.1558,
      "grad_norm": 1.1349088167246613,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 15580
    },
    {
      "epoch": 0.15581,
      "grad_norm": 1.3414979120902917,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 15581
    },
    {
      "epoch": 0.15582,
      "grad_norm": 1.201843636397302,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 15582
    },
    {
      "epoch": 0.15583,
      "grad_norm": 1.1345880097815193,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 15583
    },
    {
      "epoch": 0.15584,
      "grad_norm": 1.0766859641248807,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 15584
    },
    {
      "epoch": 0.15585,
      "grad_norm": 1.5509132059441428,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 15585
    },
    {
      "epoch": 0.15586,
      "grad_norm": 1.2169131996422387,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 15586
    },
    {
      "epoch": 0.15587,
      "grad_norm": 1.097994332980497,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 15587
    },
    {
      "epoch": 0.15588,
      "grad_norm": 1.2743281404250246,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 15588
    },
    {
      "epoch": 0.15589,
      "grad_norm": 1.1074285724019182,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 15589
    },
    {
      "epoch": 0.1559,
      "grad_norm": 1.1906649516913077,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 15590
    },
    {
      "epoch": 0.15591,
      "grad_norm": 1.560290123206607,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 15591
    },
    {
      "epoch": 0.15592,
      "grad_norm": 1.2565321454769511,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 15592
    },
    {
      "epoch": 0.15593,
      "grad_norm": 1.4275719733904104,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 15593
    },
    {
      "epoch": 0.15594,
      "grad_norm": 0.9907865538415417,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 15594
    },
    {
      "epoch": 0.15595,
      "grad_norm": 1.4361657276878441,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 15595
    },
    {
      "epoch": 0.15596,
      "grad_norm": 0.9457249881724901,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 15596
    },
    {
      "epoch": 0.15597,
      "grad_norm": 1.4679972367121594,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 15597
    },
    {
      "epoch": 0.15598,
      "grad_norm": 1.261988434296396,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 15598
    },
    {
      "epoch": 0.15599,
      "grad_norm": 1.1806554796870041,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 15599
    },
    {
      "epoch": 0.156,
      "grad_norm": 1.3879861431774005,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 15600
    },
    {
      "epoch": 0.15601,
      "grad_norm": 0.9985869460398353,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 15601
    },
    {
      "epoch": 0.15602,
      "grad_norm": 1.2678765858709244,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 15602
    },
    {
      "epoch": 0.15603,
      "grad_norm": 0.9736360743365328,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 15603
    },
    {
      "epoch": 0.15604,
      "grad_norm": 1.2809955932621875,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 15604
    },
    {
      "epoch": 0.15605,
      "grad_norm": 1.148291316111474,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 15605
    },
    {
      "epoch": 0.15606,
      "grad_norm": 1.3178519897612266,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 15606
    },
    {
      "epoch": 0.15607,
      "grad_norm": 1.2750891947071228,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 15607
    },
    {
      "epoch": 0.15608,
      "grad_norm": 1.4865355424381292,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 15608
    },
    {
      "epoch": 0.15609,
      "grad_norm": 1.2259715257260813,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 15609
    },
    {
      "epoch": 0.1561,
      "grad_norm": 1.2060706359677797,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 15610
    },
    {
      "epoch": 0.15611,
      "grad_norm": 1.2223137806997504,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 15611
    },
    {
      "epoch": 0.15612,
      "grad_norm": 1.2042145879676174,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 15612
    },
    {
      "epoch": 0.15613,
      "grad_norm": 1.3304156227239157,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 15613
    },
    {
      "epoch": 0.15614,
      "grad_norm": 1.1535381824815405,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 15614
    },
    {
      "epoch": 0.15615,
      "grad_norm": 1.2497216772094886,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 15615
    },
    {
      "epoch": 0.15616,
      "grad_norm": 1.1589575226176436,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 15616
    },
    {
      "epoch": 0.15617,
      "grad_norm": 1.214026917629878,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 15617
    },
    {
      "epoch": 0.15618,
      "grad_norm": 1.3279395508300051,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 15618
    },
    {
      "epoch": 0.15619,
      "grad_norm": 1.1274343927832446,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 15619
    },
    {
      "epoch": 0.1562,
      "grad_norm": 1.4311028687528993,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 15620
    },
    {
      "epoch": 0.15621,
      "grad_norm": 1.0211279971828127,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 15621
    },
    {
      "epoch": 0.15622,
      "grad_norm": 1.6800417751483885,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 15622
    },
    {
      "epoch": 0.15623,
      "grad_norm": 1.1727981745287182,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 15623
    },
    {
      "epoch": 0.15624,
      "grad_norm": 1.117591637939842,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 15624
    },
    {
      "epoch": 0.15625,
      "grad_norm": 1.2896942605248825,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 15625
    },
    {
      "epoch": 0.15626,
      "grad_norm": 1.3008859506847203,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 15626
    },
    {
      "epoch": 0.15627,
      "grad_norm": 1.3904328415175207,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 15627
    },
    {
      "epoch": 0.15628,
      "grad_norm": 1.234645460229858,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 15628
    },
    {
      "epoch": 0.15629,
      "grad_norm": 1.1809067210150166,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 15629
    },
    {
      "epoch": 0.1563,
      "grad_norm": 1.3195640443236507,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 15630
    },
    {
      "epoch": 0.15631,
      "grad_norm": 1.1508156778152754,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 15631
    },
    {
      "epoch": 0.15632,
      "grad_norm": 1.4927637092891726,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 15632
    },
    {
      "epoch": 0.15633,
      "grad_norm": 1.105076574968614,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 15633
    },
    {
      "epoch": 0.15634,
      "grad_norm": 1.4510587816784541,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 15634
    },
    {
      "epoch": 0.15635,
      "grad_norm": 1.183188151744009,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 15635
    },
    {
      "epoch": 0.15636,
      "grad_norm": 1.2823032120205844,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 15636
    },
    {
      "epoch": 0.15637,
      "grad_norm": 1.3473647772899666,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 15637
    },
    {
      "epoch": 0.15638,
      "grad_norm": 1.1079315472829223,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 15638
    },
    {
      "epoch": 0.15639,
      "grad_norm": 1.2366179096289918,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 15639
    },
    {
      "epoch": 0.1564,
      "grad_norm": 1.1307362871644169,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 15640
    },
    {
      "epoch": 0.15641,
      "grad_norm": 1.1955411780508391,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 15641
    },
    {
      "epoch": 0.15642,
      "grad_norm": 1.1580441924185543,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 15642
    },
    {
      "epoch": 0.15643,
      "grad_norm": 1.2320659842797703,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 15643
    },
    {
      "epoch": 0.15644,
      "grad_norm": 1.3220035146985878,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 15644
    },
    {
      "epoch": 0.15645,
      "grad_norm": 1.2045901852908378,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 15645
    },
    {
      "epoch": 0.15646,
      "grad_norm": 1.1124091615092142,
      "learning_rate": 0.003,
      "loss": 4.0264,
      "step": 15646
    },
    {
      "epoch": 0.15647,
      "grad_norm": 1.271170063053241,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 15647
    },
    {
      "epoch": 0.15648,
      "grad_norm": 1.3214408727366123,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 15648
    },
    {
      "epoch": 0.15649,
      "grad_norm": 1.187372164818155,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 15649
    },
    {
      "epoch": 0.1565,
      "grad_norm": 1.6176252258318626,
      "learning_rate": 0.003,
      "loss": 4.0289,
      "step": 15650
    },
    {
      "epoch": 0.15651,
      "grad_norm": 1.1386870708317176,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 15651
    },
    {
      "epoch": 0.15652,
      "grad_norm": 1.3083055135317863,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 15652
    },
    {
      "epoch": 0.15653,
      "grad_norm": 1.40915147872283,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 15653
    },
    {
      "epoch": 0.15654,
      "grad_norm": 0.8538626665931397,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 15654
    },
    {
      "epoch": 0.15655,
      "grad_norm": 0.9678335918623943,
      "learning_rate": 0.003,
      "loss": 4.1045,
      "step": 15655
    },
    {
      "epoch": 0.15656,
      "grad_norm": 1.5401136030257625,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 15656
    },
    {
      "epoch": 0.15657,
      "grad_norm": 0.9994550280432748,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 15657
    },
    {
      "epoch": 0.15658,
      "grad_norm": 1.4934189160213556,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 15658
    },
    {
      "epoch": 0.15659,
      "grad_norm": 1.0734913281470804,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 15659
    },
    {
      "epoch": 0.1566,
      "grad_norm": 1.3201131927167937,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 15660
    },
    {
      "epoch": 0.15661,
      "grad_norm": 1.1840754960357602,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15661
    },
    {
      "epoch": 0.15662,
      "grad_norm": 1.3540471955572964,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 15662
    },
    {
      "epoch": 0.15663,
      "grad_norm": 1.1679447196377668,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 15663
    },
    {
      "epoch": 0.15664,
      "grad_norm": 1.2586333202804696,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 15664
    },
    {
      "epoch": 0.15665,
      "grad_norm": 1.3458507787994516,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 15665
    },
    {
      "epoch": 0.15666,
      "grad_norm": 1.2384653285613236,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 15666
    },
    {
      "epoch": 0.15667,
      "grad_norm": 1.2330148927447746,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 15667
    },
    {
      "epoch": 0.15668,
      "grad_norm": 1.3681828569052863,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 15668
    },
    {
      "epoch": 0.15669,
      "grad_norm": 1.2420398174536744,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 15669
    },
    {
      "epoch": 0.1567,
      "grad_norm": 1.0635111815096139,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 15670
    },
    {
      "epoch": 0.15671,
      "grad_norm": 1.4294078233540295,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 15671
    },
    {
      "epoch": 0.15672,
      "grad_norm": 1.235341350191663,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 15672
    },
    {
      "epoch": 0.15673,
      "grad_norm": 1.2249451316293496,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 15673
    },
    {
      "epoch": 0.15674,
      "grad_norm": 1.3152866164886607,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 15674
    },
    {
      "epoch": 0.15675,
      "grad_norm": 1.2188115890433888,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 15675
    },
    {
      "epoch": 0.15676,
      "grad_norm": 1.2487670894851186,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 15676
    },
    {
      "epoch": 0.15677,
      "grad_norm": 1.1509127594545203,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 15677
    },
    {
      "epoch": 0.15678,
      "grad_norm": 1.2295049671565317,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 15678
    },
    {
      "epoch": 0.15679,
      "grad_norm": 1.2564199180091304,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 15679
    },
    {
      "epoch": 0.1568,
      "grad_norm": 1.2011586651713093,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 15680
    },
    {
      "epoch": 0.15681,
      "grad_norm": 1.4087063015759582,
      "learning_rate": 0.003,
      "loss": 4.1048,
      "step": 15681
    },
    {
      "epoch": 0.15682,
      "grad_norm": 1.191937364455209,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 15682
    },
    {
      "epoch": 0.15683,
      "grad_norm": 1.4745662297874194,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 15683
    },
    {
      "epoch": 0.15684,
      "grad_norm": 1.001726375746545,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 15684
    },
    {
      "epoch": 0.15685,
      "grad_norm": 1.29555761953909,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 15685
    },
    {
      "epoch": 0.15686,
      "grad_norm": 1.2168139453733735,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 15686
    },
    {
      "epoch": 0.15687,
      "grad_norm": 1.1113718267499884,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 15687
    },
    {
      "epoch": 0.15688,
      "grad_norm": 1.2747869571151393,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 15688
    },
    {
      "epoch": 0.15689,
      "grad_norm": 1.0391721449384115,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 15689
    },
    {
      "epoch": 0.1569,
      "grad_norm": 1.329338261019493,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 15690
    },
    {
      "epoch": 0.15691,
      "grad_norm": 1.038286920678713,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 15691
    },
    {
      "epoch": 0.15692,
      "grad_norm": 1.5112599084156186,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 15692
    },
    {
      "epoch": 0.15693,
      "grad_norm": 1.0740745559288278,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 15693
    },
    {
      "epoch": 0.15694,
      "grad_norm": 1.6952246256536485,
      "learning_rate": 0.003,
      "loss": 4.1034,
      "step": 15694
    },
    {
      "epoch": 0.15695,
      "grad_norm": 1.1356207728741108,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 15695
    },
    {
      "epoch": 0.15696,
      "grad_norm": 1.600737678873874,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 15696
    },
    {
      "epoch": 0.15697,
      "grad_norm": 1.3798245399881897,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 15697
    },
    {
      "epoch": 0.15698,
      "grad_norm": 1.2319880529302756,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 15698
    },
    {
      "epoch": 0.15699,
      "grad_norm": 1.162085383918692,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 15699
    },
    {
      "epoch": 0.157,
      "grad_norm": 1.1861369428968611,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 15700
    },
    {
      "epoch": 0.15701,
      "grad_norm": 1.1022028086837685,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 15701
    },
    {
      "epoch": 0.15702,
      "grad_norm": 1.3519041828221627,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 15702
    },
    {
      "epoch": 0.15703,
      "grad_norm": 1.2546178713888105,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 15703
    },
    {
      "epoch": 0.15704,
      "grad_norm": 1.3885754744550656,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 15704
    },
    {
      "epoch": 0.15705,
      "grad_norm": 1.0136676281040595,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 15705
    },
    {
      "epoch": 0.15706,
      "grad_norm": 1.479423464466427,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 15706
    },
    {
      "epoch": 0.15707,
      "grad_norm": 1.0692242216649765,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 15707
    },
    {
      "epoch": 0.15708,
      "grad_norm": 1.6514457947652423,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 15708
    },
    {
      "epoch": 0.15709,
      "grad_norm": 1.0980061962964487,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 15709
    },
    {
      "epoch": 0.1571,
      "grad_norm": 1.1629910875995582,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 15710
    },
    {
      "epoch": 0.15711,
      "grad_norm": 1.3835072420169645,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 15711
    },
    {
      "epoch": 0.15712,
      "grad_norm": 0.9814395529013048,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 15712
    },
    {
      "epoch": 0.15713,
      "grad_norm": 1.3044701713468254,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 15713
    },
    {
      "epoch": 0.15714,
      "grad_norm": 1.0128719798895127,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 15714
    },
    {
      "epoch": 0.15715,
      "grad_norm": 1.4611873649306613,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 15715
    },
    {
      "epoch": 0.15716,
      "grad_norm": 1.1673127264656047,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 15716
    },
    {
      "epoch": 0.15717,
      "grad_norm": 1.6321591972869431,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 15717
    },
    {
      "epoch": 0.15718,
      "grad_norm": 1.194441152695339,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 15718
    },
    {
      "epoch": 0.15719,
      "grad_norm": 1.4729419167542548,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 15719
    },
    {
      "epoch": 0.1572,
      "grad_norm": 1.153977332613549,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 15720
    },
    {
      "epoch": 0.15721,
      "grad_norm": 1.3469995913475945,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 15721
    },
    {
      "epoch": 0.15722,
      "grad_norm": 1.1257746335495376,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 15722
    },
    {
      "epoch": 0.15723,
      "grad_norm": 1.273311840186899,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 15723
    },
    {
      "epoch": 0.15724,
      "grad_norm": 1.2425020612828945,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 15724
    },
    {
      "epoch": 0.15725,
      "grad_norm": 1.0998561464008263,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 15725
    },
    {
      "epoch": 0.15726,
      "grad_norm": 1.2074450589504484,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 15726
    },
    {
      "epoch": 0.15727,
      "grad_norm": 1.2465016405631146,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 15727
    },
    {
      "epoch": 0.15728,
      "grad_norm": 1.2308298672569458,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 15728
    },
    {
      "epoch": 0.15729,
      "grad_norm": 1.2353477693601442,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 15729
    },
    {
      "epoch": 0.1573,
      "grad_norm": 1.1997809145637395,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 15730
    },
    {
      "epoch": 0.15731,
      "grad_norm": 1.055664502187246,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 15731
    },
    {
      "epoch": 0.15732,
      "grad_norm": 1.1251817771876742,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 15732
    },
    {
      "epoch": 0.15733,
      "grad_norm": 1.2684826751675962,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 15733
    },
    {
      "epoch": 0.15734,
      "grad_norm": 1.176261122634088,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 15734
    },
    {
      "epoch": 0.15735,
      "grad_norm": 1.3011228658316907,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 15735
    },
    {
      "epoch": 0.15736,
      "grad_norm": 1.4022393363425198,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 15736
    },
    {
      "epoch": 0.15737,
      "grad_norm": 1.4943081663486804,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 15737
    },
    {
      "epoch": 0.15738,
      "grad_norm": 1.2095662083063563,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 15738
    },
    {
      "epoch": 0.15739,
      "grad_norm": 1.1900215263482639,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 15739
    },
    {
      "epoch": 0.1574,
      "grad_norm": 1.498445894217828,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 15740
    },
    {
      "epoch": 0.15741,
      "grad_norm": 1.247342175522623,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 15741
    },
    {
      "epoch": 0.15742,
      "grad_norm": 1.281881136558292,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 15742
    },
    {
      "epoch": 0.15743,
      "grad_norm": 1.1637487580323629,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 15743
    },
    {
      "epoch": 0.15744,
      "grad_norm": 1.6250204266507926,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 15744
    },
    {
      "epoch": 0.15745,
      "grad_norm": 1.1120053816868876,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 15745
    },
    {
      "epoch": 0.15746,
      "grad_norm": 1.1806132697403393,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 15746
    },
    {
      "epoch": 0.15747,
      "grad_norm": 1.3753350852825859,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 15747
    },
    {
      "epoch": 0.15748,
      "grad_norm": 1.3617899171508971,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 15748
    },
    {
      "epoch": 0.15749,
      "grad_norm": 1.224551816489951,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 15749
    },
    {
      "epoch": 0.1575,
      "grad_norm": 1.2329503063424165,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 15750
    },
    {
      "epoch": 0.15751,
      "grad_norm": 1.3487282436055932,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 15751
    },
    {
      "epoch": 0.15752,
      "grad_norm": 0.8983113736476528,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 15752
    },
    {
      "epoch": 0.15753,
      "grad_norm": 1.1205134642362236,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 15753
    },
    {
      "epoch": 0.15754,
      "grad_norm": 1.1108146780065227,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 15754
    },
    {
      "epoch": 0.15755,
      "grad_norm": 1.3179332530517076,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 15755
    },
    {
      "epoch": 0.15756,
      "grad_norm": 1.2340293618335434,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 15756
    },
    {
      "epoch": 0.15757,
      "grad_norm": 1.2067630444776245,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 15757
    },
    {
      "epoch": 0.15758,
      "grad_norm": 1.358633075141287,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 15758
    },
    {
      "epoch": 0.15759,
      "grad_norm": 1.2713680632283646,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 15759
    },
    {
      "epoch": 0.1576,
      "grad_norm": 1.429606807726027,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 15760
    },
    {
      "epoch": 0.15761,
      "grad_norm": 1.1698066698894496,
      "learning_rate": 0.003,
      "loss": 4.0225,
      "step": 15761
    },
    {
      "epoch": 0.15762,
      "grad_norm": 1.3301859759951484,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 15762
    },
    {
      "epoch": 0.15763,
      "grad_norm": 1.0043360109228892,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 15763
    },
    {
      "epoch": 0.15764,
      "grad_norm": 1.2805820162353942,
      "learning_rate": 0.003,
      "loss": 4.0285,
      "step": 15764
    },
    {
      "epoch": 0.15765,
      "grad_norm": 1.084130383097619,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 15765
    },
    {
      "epoch": 0.15766,
      "grad_norm": 1.4455323926932548,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 15766
    },
    {
      "epoch": 0.15767,
      "grad_norm": 1.2458212253441476,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 15767
    },
    {
      "epoch": 0.15768,
      "grad_norm": 1.4079166675160757,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 15768
    },
    {
      "epoch": 0.15769,
      "grad_norm": 1.0886689940232694,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 15769
    },
    {
      "epoch": 0.1577,
      "grad_norm": 1.33669784595432,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 15770
    },
    {
      "epoch": 0.15771,
      "grad_norm": 1.2659884771872834,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 15771
    },
    {
      "epoch": 0.15772,
      "grad_norm": 1.2676175178618903,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 15772
    },
    {
      "epoch": 0.15773,
      "grad_norm": 1.3247024497021374,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 15773
    },
    {
      "epoch": 0.15774,
      "grad_norm": 1.1798670009014236,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 15774
    },
    {
      "epoch": 0.15775,
      "grad_norm": 1.2745844735563765,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 15775
    },
    {
      "epoch": 0.15776,
      "grad_norm": 1.2586926574914314,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 15776
    },
    {
      "epoch": 0.15777,
      "grad_norm": 1.13855349226103,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 15777
    },
    {
      "epoch": 0.15778,
      "grad_norm": 1.1802061114019076,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 15778
    },
    {
      "epoch": 0.15779,
      "grad_norm": 1.2664395752802722,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 15779
    },
    {
      "epoch": 0.1578,
      "grad_norm": 1.3060934404648121,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 15780
    },
    {
      "epoch": 0.15781,
      "grad_norm": 1.380042239870967,
      "learning_rate": 0.003,
      "loss": 4.0086,
      "step": 15781
    },
    {
      "epoch": 0.15782,
      "grad_norm": 1.087924226172932,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 15782
    },
    {
      "epoch": 0.15783,
      "grad_norm": 1.3430527522168907,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 15783
    },
    {
      "epoch": 0.15784,
      "grad_norm": 1.0104217717628523,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 15784
    },
    {
      "epoch": 0.15785,
      "grad_norm": 1.4267809040482224,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 15785
    },
    {
      "epoch": 0.15786,
      "grad_norm": 1.216574584344044,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 15786
    },
    {
      "epoch": 0.15787,
      "grad_norm": 1.212189380409753,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 15787
    },
    {
      "epoch": 0.15788,
      "grad_norm": 1.2650434917429774,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 15788
    },
    {
      "epoch": 0.15789,
      "grad_norm": 1.3134562170382684,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 15789
    },
    {
      "epoch": 0.1579,
      "grad_norm": 1.0200074018220817,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 15790
    },
    {
      "epoch": 0.15791,
      "grad_norm": 1.508952445135445,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 15791
    },
    {
      "epoch": 0.15792,
      "grad_norm": 1.0099881102959938,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 15792
    },
    {
      "epoch": 0.15793,
      "grad_norm": 1.5182111830652727,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 15793
    },
    {
      "epoch": 0.15794,
      "grad_norm": 1.0140123650871258,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 15794
    },
    {
      "epoch": 0.15795,
      "grad_norm": 1.630216472140459,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 15795
    },
    {
      "epoch": 0.15796,
      "grad_norm": 1.2356276021322978,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 15796
    },
    {
      "epoch": 0.15797,
      "grad_norm": 1.1195009082517002,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 15797
    },
    {
      "epoch": 0.15798,
      "grad_norm": 1.2554496879615797,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 15798
    },
    {
      "epoch": 0.15799,
      "grad_norm": 1.1882650169054183,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 15799
    },
    {
      "epoch": 0.158,
      "grad_norm": 1.025180895775637,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 15800
    },
    {
      "epoch": 0.15801,
      "grad_norm": 1.3323650505869045,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 15801
    },
    {
      "epoch": 0.15802,
      "grad_norm": 1.1648391893077823,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 15802
    },
    {
      "epoch": 0.15803,
      "grad_norm": 1.3263065106366811,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 15803
    },
    {
      "epoch": 0.15804,
      "grad_norm": 1.019953277288765,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 15804
    },
    {
      "epoch": 0.15805,
      "grad_norm": 1.3199518646915922,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 15805
    },
    {
      "epoch": 0.15806,
      "grad_norm": 1.1567590802816259,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 15806
    },
    {
      "epoch": 0.15807,
      "grad_norm": 1.2112833131281036,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 15807
    },
    {
      "epoch": 0.15808,
      "grad_norm": 1.543700800987061,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 15808
    },
    {
      "epoch": 0.15809,
      "grad_norm": 1.2148372812557238,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 15809
    },
    {
      "epoch": 0.1581,
      "grad_norm": 1.64595741304971,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 15810
    },
    {
      "epoch": 0.15811,
      "grad_norm": 0.9473241818368024,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 15811
    },
    {
      "epoch": 0.15812,
      "grad_norm": 1.2074555335242219,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 15812
    },
    {
      "epoch": 0.15813,
      "grad_norm": 1.7567532028410888,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 15813
    },
    {
      "epoch": 0.15814,
      "grad_norm": 1.1609304991514573,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 15814
    },
    {
      "epoch": 0.15815,
      "grad_norm": 1.315570986779992,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 15815
    },
    {
      "epoch": 0.15816,
      "grad_norm": 1.0790403099689905,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 15816
    },
    {
      "epoch": 0.15817,
      "grad_norm": 1.4012912339396493,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 15817
    },
    {
      "epoch": 0.15818,
      "grad_norm": 0.9990085088007167,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 15818
    },
    {
      "epoch": 0.15819,
      "grad_norm": 1.3108872436591201,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 15819
    },
    {
      "epoch": 0.1582,
      "grad_norm": 1.2860627087013567,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 15820
    },
    {
      "epoch": 0.15821,
      "grad_norm": 1.313384160400915,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 15821
    },
    {
      "epoch": 0.15822,
      "grad_norm": 1.2846586054114109,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 15822
    },
    {
      "epoch": 0.15823,
      "grad_norm": 1.1358164618290276,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 15823
    },
    {
      "epoch": 0.15824,
      "grad_norm": 1.4458733395957897,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 15824
    },
    {
      "epoch": 0.15825,
      "grad_norm": 1.1737861594039298,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 15825
    },
    {
      "epoch": 0.15826,
      "grad_norm": 1.2185578330836349,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 15826
    },
    {
      "epoch": 0.15827,
      "grad_norm": 1.1982989834195201,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 15827
    },
    {
      "epoch": 0.15828,
      "grad_norm": 1.3110406043806941,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 15828
    },
    {
      "epoch": 0.15829,
      "grad_norm": 1.466595140772327,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 15829
    },
    {
      "epoch": 0.1583,
      "grad_norm": 1.1028802305304912,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 15830
    },
    {
      "epoch": 0.15831,
      "grad_norm": 1.4401414313447025,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 15831
    },
    {
      "epoch": 0.15832,
      "grad_norm": 1.0121391822966723,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 15832
    },
    {
      "epoch": 0.15833,
      "grad_norm": 1.3701569406549603,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 15833
    },
    {
      "epoch": 0.15834,
      "grad_norm": 1.0467743466126278,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 15834
    },
    {
      "epoch": 0.15835,
      "grad_norm": 1.4350110033054864,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15835
    },
    {
      "epoch": 0.15836,
      "grad_norm": 1.1655596804028503,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 15836
    },
    {
      "epoch": 0.15837,
      "grad_norm": 1.366809734456757,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 15837
    },
    {
      "epoch": 0.15838,
      "grad_norm": 1.1015343654202836,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 15838
    },
    {
      "epoch": 0.15839,
      "grad_norm": 1.497404289489368,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 15839
    },
    {
      "epoch": 0.1584,
      "grad_norm": 1.162686134233669,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 15840
    },
    {
      "epoch": 0.15841,
      "grad_norm": 1.4122265105717202,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 15841
    },
    {
      "epoch": 0.15842,
      "grad_norm": 1.0646479436981007,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 15842
    },
    {
      "epoch": 0.15843,
      "grad_norm": 1.4378996909096384,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 15843
    },
    {
      "epoch": 0.15844,
      "grad_norm": 1.1335472518084901,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 15844
    },
    {
      "epoch": 0.15845,
      "grad_norm": 1.202254551336859,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 15845
    },
    {
      "epoch": 0.15846,
      "grad_norm": 1.1023046241915795,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 15846
    },
    {
      "epoch": 0.15847,
      "grad_norm": 1.2241215062872985,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 15847
    },
    {
      "epoch": 0.15848,
      "grad_norm": 1.2360989485035663,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 15848
    },
    {
      "epoch": 0.15849,
      "grad_norm": 1.2851274263498322,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 15849
    },
    {
      "epoch": 0.1585,
      "grad_norm": 1.5063698866099746,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 15850
    },
    {
      "epoch": 0.15851,
      "grad_norm": 1.3370699560578674,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15851
    },
    {
      "epoch": 0.15852,
      "grad_norm": 1.0649414968565,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 15852
    },
    {
      "epoch": 0.15853,
      "grad_norm": 1.2923127203605576,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 15853
    },
    {
      "epoch": 0.15854,
      "grad_norm": 1.17886527455772,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 15854
    },
    {
      "epoch": 0.15855,
      "grad_norm": 1.110334224682506,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 15855
    },
    {
      "epoch": 0.15856,
      "grad_norm": 1.014714199014075,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 15856
    },
    {
      "epoch": 0.15857,
      "grad_norm": 1.3186310147508855,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 15857
    },
    {
      "epoch": 0.15858,
      "grad_norm": 1.0082773124492816,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 15858
    },
    {
      "epoch": 0.15859,
      "grad_norm": 1.341166900129246,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 15859
    },
    {
      "epoch": 0.1586,
      "grad_norm": 0.9757690221429678,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 15860
    },
    {
      "epoch": 0.15861,
      "grad_norm": 1.5181678524071738,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 15861
    },
    {
      "epoch": 0.15862,
      "grad_norm": 1.4198118300073572,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 15862
    },
    {
      "epoch": 0.15863,
      "grad_norm": 1.3250680079462993,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 15863
    },
    {
      "epoch": 0.15864,
      "grad_norm": 1.1219890722185781,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 15864
    },
    {
      "epoch": 0.15865,
      "grad_norm": 1.3773526564893919,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 15865
    },
    {
      "epoch": 0.15866,
      "grad_norm": 1.1187512363140182,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 15866
    },
    {
      "epoch": 0.15867,
      "grad_norm": 1.3873696864503684,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 15867
    },
    {
      "epoch": 0.15868,
      "grad_norm": 1.05050287732908,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 15868
    },
    {
      "epoch": 0.15869,
      "grad_norm": 1.3226400567507672,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 15869
    },
    {
      "epoch": 0.1587,
      "grad_norm": 1.112817535635143,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 15870
    },
    {
      "epoch": 0.15871,
      "grad_norm": 1.1031885680213505,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 15871
    },
    {
      "epoch": 0.15872,
      "grad_norm": 1.3063976521351819,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 15872
    },
    {
      "epoch": 0.15873,
      "grad_norm": 1.3075266758431832,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 15873
    },
    {
      "epoch": 0.15874,
      "grad_norm": 1.4796353974658847,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 15874
    },
    {
      "epoch": 0.15875,
      "grad_norm": 1.1544673497341547,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 15875
    },
    {
      "epoch": 0.15876,
      "grad_norm": 1.2104504771981357,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 15876
    },
    {
      "epoch": 0.15877,
      "grad_norm": 1.3874317918786958,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 15877
    },
    {
      "epoch": 0.15878,
      "grad_norm": 0.8694977231904792,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 15878
    },
    {
      "epoch": 0.15879,
      "grad_norm": 0.9167007165846688,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 15879
    },
    {
      "epoch": 0.1588,
      "grad_norm": 1.3094180982937673,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 15880
    },
    {
      "epoch": 0.15881,
      "grad_norm": 1.0940458833334197,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 15881
    },
    {
      "epoch": 0.15882,
      "grad_norm": 1.6272294752909982,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 15882
    },
    {
      "epoch": 0.15883,
      "grad_norm": 0.953058667318075,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 15883
    },
    {
      "epoch": 0.15884,
      "grad_norm": 1.3800984250047426,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 15884
    },
    {
      "epoch": 0.15885,
      "grad_norm": 1.0469522999452336,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 15885
    },
    {
      "epoch": 0.15886,
      "grad_norm": 1.3807583853414365,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 15886
    },
    {
      "epoch": 0.15887,
      "grad_norm": 1.1689417177816546,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 15887
    },
    {
      "epoch": 0.15888,
      "grad_norm": 1.166352545373565,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 15888
    },
    {
      "epoch": 0.15889,
      "grad_norm": 1.1497410035203872,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 15889
    },
    {
      "epoch": 0.1589,
      "grad_norm": 1.1474521902301942,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 15890
    },
    {
      "epoch": 0.15891,
      "grad_norm": 1.378472287993341,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 15891
    },
    {
      "epoch": 0.15892,
      "grad_norm": 1.0524190751041174,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 15892
    },
    {
      "epoch": 0.15893,
      "grad_norm": 1.4838436770972758,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 15893
    },
    {
      "epoch": 0.15894,
      "grad_norm": 1.4774807045801406,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 15894
    },
    {
      "epoch": 0.15895,
      "grad_norm": 1.6066140641730953,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 15895
    },
    {
      "epoch": 0.15896,
      "grad_norm": 1.1622148299667838,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 15896
    },
    {
      "epoch": 0.15897,
      "grad_norm": 1.1931281726448333,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 15897
    },
    {
      "epoch": 0.15898,
      "grad_norm": 1.302918768938021,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 15898
    },
    {
      "epoch": 0.15899,
      "grad_norm": 1.2655608973386931,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 15899
    },
    {
      "epoch": 0.159,
      "grad_norm": 1.3007264409876362,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 15900
    },
    {
      "epoch": 0.15901,
      "grad_norm": 1.4017728789526172,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 15901
    },
    {
      "epoch": 0.15902,
      "grad_norm": 1.096907458183354,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 15902
    },
    {
      "epoch": 0.15903,
      "grad_norm": 1.148682715983762,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 15903
    },
    {
      "epoch": 0.15904,
      "grad_norm": 1.2946860163634188,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 15904
    },
    {
      "epoch": 0.15905,
      "grad_norm": 1.122481448668236,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 15905
    },
    {
      "epoch": 0.15906,
      "grad_norm": 1.5089008973010962,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 15906
    },
    {
      "epoch": 0.15907,
      "grad_norm": 1.2836271686701424,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 15907
    },
    {
      "epoch": 0.15908,
      "grad_norm": 1.05237679745181,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 15908
    },
    {
      "epoch": 0.15909,
      "grad_norm": 1.2775159608553672,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 15909
    },
    {
      "epoch": 0.1591,
      "grad_norm": 1.2042252711867507,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 15910
    },
    {
      "epoch": 0.15911,
      "grad_norm": 1.2424910856410036,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 15911
    },
    {
      "epoch": 0.15912,
      "grad_norm": 1.004066165602063,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 15912
    },
    {
      "epoch": 0.15913,
      "grad_norm": 1.2899718102539786,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 15913
    },
    {
      "epoch": 0.15914,
      "grad_norm": 1.265411655883751,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 15914
    },
    {
      "epoch": 0.15915,
      "grad_norm": 1.3841302073176251,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 15915
    },
    {
      "epoch": 0.15916,
      "grad_norm": 1.0780930650512748,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 15916
    },
    {
      "epoch": 0.15917,
      "grad_norm": 1.3632698488776467,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 15917
    },
    {
      "epoch": 0.15918,
      "grad_norm": 1.3193444004494173,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 15918
    },
    {
      "epoch": 0.15919,
      "grad_norm": 1.2523241950513657,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 15919
    },
    {
      "epoch": 0.1592,
      "grad_norm": 1.1129624685015624,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 15920
    },
    {
      "epoch": 0.15921,
      "grad_norm": 1.2390879797094592,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 15921
    },
    {
      "epoch": 0.15922,
      "grad_norm": 1.018065075102559,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 15922
    },
    {
      "epoch": 0.15923,
      "grad_norm": 1.3992497510637925,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 15923
    },
    {
      "epoch": 0.15924,
      "grad_norm": 0.9807181832191064,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 15924
    },
    {
      "epoch": 0.15925,
      "grad_norm": 1.443845094545865,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 15925
    },
    {
      "epoch": 0.15926,
      "grad_norm": 1.3191998073770685,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 15926
    },
    {
      "epoch": 0.15927,
      "grad_norm": 1.4318050152423651,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 15927
    },
    {
      "epoch": 0.15928,
      "grad_norm": 1.5445127896410225,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 15928
    },
    {
      "epoch": 0.15929,
      "grad_norm": 1.646914315146204,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 15929
    },
    {
      "epoch": 0.1593,
      "grad_norm": 0.9881917367102422,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 15930
    },
    {
      "epoch": 0.15931,
      "grad_norm": 1.4051895948329052,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 15931
    },
    {
      "epoch": 0.15932,
      "grad_norm": 1.3037727964692034,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 15932
    },
    {
      "epoch": 0.15933,
      "grad_norm": 1.4378668261881007,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 15933
    },
    {
      "epoch": 0.15934,
      "grad_norm": 1.1431656766108897,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 15934
    },
    {
      "epoch": 0.15935,
      "grad_norm": 1.124566753972068,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 15935
    },
    {
      "epoch": 0.15936,
      "grad_norm": 1.16275651830534,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 15936
    },
    {
      "epoch": 0.15937,
      "grad_norm": 1.2342222553688793,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 15937
    },
    {
      "epoch": 0.15938,
      "grad_norm": 1.19713324059584,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 15938
    },
    {
      "epoch": 0.15939,
      "grad_norm": 1.1509074055799497,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 15939
    },
    {
      "epoch": 0.1594,
      "grad_norm": 1.3990238341878423,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 15940
    },
    {
      "epoch": 0.15941,
      "grad_norm": 1.12517011632491,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 15941
    },
    {
      "epoch": 0.15942,
      "grad_norm": 1.4062696973627182,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 15942
    },
    {
      "epoch": 0.15943,
      "grad_norm": 1.2298148700416418,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 15943
    },
    {
      "epoch": 0.15944,
      "grad_norm": 1.0423298803055772,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 15944
    },
    {
      "epoch": 0.15945,
      "grad_norm": 1.5119355212233925,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 15945
    },
    {
      "epoch": 0.15946,
      "grad_norm": 1.1043321719037542,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 15946
    },
    {
      "epoch": 0.15947,
      "grad_norm": 1.5754318100897307,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 15947
    },
    {
      "epoch": 0.15948,
      "grad_norm": 1.2247612008785362,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 15948
    },
    {
      "epoch": 0.15949,
      "grad_norm": 1.2780057860668521,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 15949
    },
    {
      "epoch": 0.1595,
      "grad_norm": 1.0935819702783598,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 15950
    },
    {
      "epoch": 0.15951,
      "grad_norm": 1.4264660530503124,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 15951
    },
    {
      "epoch": 0.15952,
      "grad_norm": 1.4432776864714585,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 15952
    },
    {
      "epoch": 0.15953,
      "grad_norm": 1.0936306035911116,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 15953
    },
    {
      "epoch": 0.15954,
      "grad_norm": 1.3093780200539074,
      "learning_rate": 0.003,
      "loss": 4.1032,
      "step": 15954
    },
    {
      "epoch": 0.15955,
      "grad_norm": 1.1337523344829086,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 15955
    },
    {
      "epoch": 0.15956,
      "grad_norm": 1.2976729310465538,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 15956
    },
    {
      "epoch": 0.15957,
      "grad_norm": 1.1486881406964997,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 15957
    },
    {
      "epoch": 0.15958,
      "grad_norm": 1.1649483196714951,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 15958
    },
    {
      "epoch": 0.15959,
      "grad_norm": 1.0751428111362795,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 15959
    },
    {
      "epoch": 0.1596,
      "grad_norm": 1.525072047246104,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 15960
    },
    {
      "epoch": 0.15961,
      "grad_norm": 1.2430663142139295,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 15961
    },
    {
      "epoch": 0.15962,
      "grad_norm": 1.3229566512908846,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 15962
    },
    {
      "epoch": 0.15963,
      "grad_norm": 1.2326710037686879,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 15963
    },
    {
      "epoch": 0.15964,
      "grad_norm": 1.178974318726838,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 15964
    },
    {
      "epoch": 0.15965,
      "grad_norm": 1.4016731036877232,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 15965
    },
    {
      "epoch": 0.15966,
      "grad_norm": 1.346665660360529,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 15966
    },
    {
      "epoch": 0.15967,
      "grad_norm": 1.5714256770776949,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 15967
    },
    {
      "epoch": 0.15968,
      "grad_norm": 1.174841832382644,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 15968
    },
    {
      "epoch": 0.15969,
      "grad_norm": 1.1664948385353457,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 15969
    },
    {
      "epoch": 0.1597,
      "grad_norm": 1.1363541719355623,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 15970
    },
    {
      "epoch": 0.15971,
      "grad_norm": 1.2671228496799418,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 15971
    },
    {
      "epoch": 0.15972,
      "grad_norm": 1.3127969455545418,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 15972
    },
    {
      "epoch": 0.15973,
      "grad_norm": 1.353167208597876,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 15973
    },
    {
      "epoch": 0.15974,
      "grad_norm": 0.9988635512343115,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 15974
    },
    {
      "epoch": 0.15975,
      "grad_norm": 1.4102942267563905,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 15975
    },
    {
      "epoch": 0.15976,
      "grad_norm": 1.0329889844965574,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 15976
    },
    {
      "epoch": 0.15977,
      "grad_norm": 1.4709554695572877,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 15977
    },
    {
      "epoch": 0.15978,
      "grad_norm": 1.0835729764776088,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 15978
    },
    {
      "epoch": 0.15979,
      "grad_norm": 1.3346053161633171,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 15979
    },
    {
      "epoch": 0.1598,
      "grad_norm": 1.4367864476983094,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 15980
    },
    {
      "epoch": 0.15981,
      "grad_norm": 1.211044054434938,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 15981
    },
    {
      "epoch": 0.15982,
      "grad_norm": 1.130377839272506,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 15982
    },
    {
      "epoch": 0.15983,
      "grad_norm": 1.199590883702763,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 15983
    },
    {
      "epoch": 0.15984,
      "grad_norm": 1.2234798884021005,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 15984
    },
    {
      "epoch": 0.15985,
      "grad_norm": 1.0621335951197872,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 15985
    },
    {
      "epoch": 0.15986,
      "grad_norm": 1.4802870810804452,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 15986
    },
    {
      "epoch": 0.15987,
      "grad_norm": 1.3745377945452901,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 15987
    },
    {
      "epoch": 0.15988,
      "grad_norm": 1.2800826333131476,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 15988
    },
    {
      "epoch": 0.15989,
      "grad_norm": 1.249017924347872,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 15989
    },
    {
      "epoch": 0.1599,
      "grad_norm": 1.087436131985846,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 15990
    },
    {
      "epoch": 0.15991,
      "grad_norm": 1.6208015102893236,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 15991
    },
    {
      "epoch": 0.15992,
      "grad_norm": 0.9553147533503118,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 15992
    },
    {
      "epoch": 0.15993,
      "grad_norm": 1.2399004682165844,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 15993
    },
    {
      "epoch": 0.15994,
      "grad_norm": 1.151216794905892,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 15994
    },
    {
      "epoch": 0.15995,
      "grad_norm": 1.209520667594038,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 15995
    },
    {
      "epoch": 0.15996,
      "grad_norm": 1.2654515123798205,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 15996
    },
    {
      "epoch": 0.15997,
      "grad_norm": 1.090738242669251,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 15997
    },
    {
      "epoch": 0.15998,
      "grad_norm": 1.6241870427660114,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 15998
    },
    {
      "epoch": 0.15999,
      "grad_norm": 1.037839226991951,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 15999
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.468254338539844,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 16000
    },
    {
      "epoch": 0.16001,
      "grad_norm": 1.0569871997225846,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 16001
    },
    {
      "epoch": 0.16002,
      "grad_norm": 1.3543454173351195,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 16002
    },
    {
      "epoch": 0.16003,
      "grad_norm": 1.219238499758112,
      "learning_rate": 0.003,
      "loss": 4.1014,
      "step": 16003
    },
    {
      "epoch": 0.16004,
      "grad_norm": 1.2148919686004898,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 16004
    },
    {
      "epoch": 0.16005,
      "grad_norm": 1.3208705578435487,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 16005
    },
    {
      "epoch": 0.16006,
      "grad_norm": 1.3542892074931192,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 16006
    },
    {
      "epoch": 0.16007,
      "grad_norm": 1.3671824625496356,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 16007
    },
    {
      "epoch": 0.16008,
      "grad_norm": 1.0668311187906165,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 16008
    },
    {
      "epoch": 0.16009,
      "grad_norm": 1.3470746547863566,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 16009
    },
    {
      "epoch": 0.1601,
      "grad_norm": 1.0772154800713234,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 16010
    },
    {
      "epoch": 0.16011,
      "grad_norm": 1.3510199835884709,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 16011
    },
    {
      "epoch": 0.16012,
      "grad_norm": 1.1924190650854338,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 16012
    },
    {
      "epoch": 0.16013,
      "grad_norm": 1.1614570082005813,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 16013
    },
    {
      "epoch": 0.16014,
      "grad_norm": 1.2943798068788903,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 16014
    },
    {
      "epoch": 0.16015,
      "grad_norm": 1.3157424343117976,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 16015
    },
    {
      "epoch": 0.16016,
      "grad_norm": 1.2977011987523548,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 16016
    },
    {
      "epoch": 0.16017,
      "grad_norm": 1.2413929690186232,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 16017
    },
    {
      "epoch": 0.16018,
      "grad_norm": 1.6940441644090516,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 16018
    },
    {
      "epoch": 0.16019,
      "grad_norm": 1.3383232864298142,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 16019
    },
    {
      "epoch": 0.1602,
      "grad_norm": 1.214208237044485,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 16020
    },
    {
      "epoch": 0.16021,
      "grad_norm": 1.385184359152161,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 16021
    },
    {
      "epoch": 0.16022,
      "grad_norm": 1.1521194968074087,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 16022
    },
    {
      "epoch": 0.16023,
      "grad_norm": 1.2034260544674427,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 16023
    },
    {
      "epoch": 0.16024,
      "grad_norm": 1.024534410119776,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 16024
    },
    {
      "epoch": 0.16025,
      "grad_norm": 1.535277051756849,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 16025
    },
    {
      "epoch": 0.16026,
      "grad_norm": 0.9833674829901832,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 16026
    },
    {
      "epoch": 0.16027,
      "grad_norm": 1.4822598816434278,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 16027
    },
    {
      "epoch": 0.16028,
      "grad_norm": 1.2765333691830032,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 16028
    },
    {
      "epoch": 0.16029,
      "grad_norm": 1.1821927698993437,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 16029
    },
    {
      "epoch": 0.1603,
      "grad_norm": 1.3831232505540298,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 16030
    },
    {
      "epoch": 0.16031,
      "grad_norm": 1.0981332122382825,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 16031
    },
    {
      "epoch": 0.16032,
      "grad_norm": 1.184711305254577,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 16032
    },
    {
      "epoch": 0.16033,
      "grad_norm": 1.1372011397446296,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 16033
    },
    {
      "epoch": 0.16034,
      "grad_norm": 1.5536688790305972,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16034
    },
    {
      "epoch": 0.16035,
      "grad_norm": 1.2204597945609714,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 16035
    },
    {
      "epoch": 0.16036,
      "grad_norm": 1.2896458181398158,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 16036
    },
    {
      "epoch": 0.16037,
      "grad_norm": 1.169264447255346,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 16037
    },
    {
      "epoch": 0.16038,
      "grad_norm": 1.2634712517724833,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 16038
    },
    {
      "epoch": 0.16039,
      "grad_norm": 1.0855605432845257,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 16039
    },
    {
      "epoch": 0.1604,
      "grad_norm": 1.224483811509826,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16040
    },
    {
      "epoch": 0.16041,
      "grad_norm": 1.3331507875882669,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 16041
    },
    {
      "epoch": 0.16042,
      "grad_norm": 1.3891453967678526,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 16042
    },
    {
      "epoch": 0.16043,
      "grad_norm": 1.0795881114598527,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 16043
    },
    {
      "epoch": 0.16044,
      "grad_norm": 1.3023423697862442,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 16044
    },
    {
      "epoch": 0.16045,
      "grad_norm": 1.3246120268063921,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 16045
    },
    {
      "epoch": 0.16046,
      "grad_norm": 1.1689040553881331,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 16046
    },
    {
      "epoch": 0.16047,
      "grad_norm": 1.3412929735739219,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 16047
    },
    {
      "epoch": 0.16048,
      "grad_norm": 1.079196558883636,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 16048
    },
    {
      "epoch": 0.16049,
      "grad_norm": 1.203615237844193,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 16049
    },
    {
      "epoch": 0.1605,
      "grad_norm": 1.2286539119716313,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 16050
    },
    {
      "epoch": 0.16051,
      "grad_norm": 1.1185597063613093,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 16051
    },
    {
      "epoch": 0.16052,
      "grad_norm": 1.163864268386705,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 16052
    },
    {
      "epoch": 0.16053,
      "grad_norm": 1.1613314754989137,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 16053
    },
    {
      "epoch": 0.16054,
      "grad_norm": 1.4948819073330097,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 16054
    },
    {
      "epoch": 0.16055,
      "grad_norm": 1.3371682943048482,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16055
    },
    {
      "epoch": 0.16056,
      "grad_norm": 1.3145085674969466,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 16056
    },
    {
      "epoch": 0.16057,
      "grad_norm": 1.2269871285738352,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 16057
    },
    {
      "epoch": 0.16058,
      "grad_norm": 1.0244775848847234,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 16058
    },
    {
      "epoch": 0.16059,
      "grad_norm": 1.4063519288689355,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 16059
    },
    {
      "epoch": 0.1606,
      "grad_norm": 0.9709192756974219,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16060
    },
    {
      "epoch": 0.16061,
      "grad_norm": 1.411425188042891,
      "learning_rate": 0.003,
      "loss": 4.0934,
      "step": 16061
    },
    {
      "epoch": 0.16062,
      "grad_norm": 1.1177862472664484,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 16062
    },
    {
      "epoch": 0.16063,
      "grad_norm": 1.536173225451104,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 16063
    },
    {
      "epoch": 0.16064,
      "grad_norm": 0.9707829048308666,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 16064
    },
    {
      "epoch": 0.16065,
      "grad_norm": 1.2900190071381814,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 16065
    },
    {
      "epoch": 0.16066,
      "grad_norm": 1.1189790366172518,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 16066
    },
    {
      "epoch": 0.16067,
      "grad_norm": 1.3640985307959768,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 16067
    },
    {
      "epoch": 0.16068,
      "grad_norm": 1.296619773757928,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 16068
    },
    {
      "epoch": 0.16069,
      "grad_norm": 1.2669442982556671,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 16069
    },
    {
      "epoch": 0.1607,
      "grad_norm": 1.4834545045440388,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 16070
    },
    {
      "epoch": 0.16071,
      "grad_norm": 1.02338157717909,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 16071
    },
    {
      "epoch": 0.16072,
      "grad_norm": 1.1439734656409877,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 16072
    },
    {
      "epoch": 0.16073,
      "grad_norm": 1.5239911455146988,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 16073
    },
    {
      "epoch": 0.16074,
      "grad_norm": 1.0346397079451986,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 16074
    },
    {
      "epoch": 0.16075,
      "grad_norm": 1.497834128932681,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 16075
    },
    {
      "epoch": 0.16076,
      "grad_norm": 0.9777741720449256,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 16076
    },
    {
      "epoch": 0.16077,
      "grad_norm": 1.301822090815537,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 16077
    },
    {
      "epoch": 0.16078,
      "grad_norm": 1.1638833095483208,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 16078
    },
    {
      "epoch": 0.16079,
      "grad_norm": 1.421476786130954,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 16079
    },
    {
      "epoch": 0.1608,
      "grad_norm": 1.42572680340593,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 16080
    },
    {
      "epoch": 0.16081,
      "grad_norm": 1.6065065791865625,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 16081
    },
    {
      "epoch": 0.16082,
      "grad_norm": 1.0614834918617888,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 16082
    },
    {
      "epoch": 0.16083,
      "grad_norm": 1.0720843631531674,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 16083
    },
    {
      "epoch": 0.16084,
      "grad_norm": 1.2596062426915247,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 16084
    },
    {
      "epoch": 0.16085,
      "grad_norm": 1.3197603521255723,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 16085
    },
    {
      "epoch": 0.16086,
      "grad_norm": 1.209075789456963,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 16086
    },
    {
      "epoch": 0.16087,
      "grad_norm": 1.3986140934480322,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 16087
    },
    {
      "epoch": 0.16088,
      "grad_norm": 1.29174831939878,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 16088
    },
    {
      "epoch": 0.16089,
      "grad_norm": 1.1118705050907147,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 16089
    },
    {
      "epoch": 0.1609,
      "grad_norm": 1.3356641655821764,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 16090
    },
    {
      "epoch": 0.16091,
      "grad_norm": 1.199188973900642,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 16091
    },
    {
      "epoch": 0.16092,
      "grad_norm": 1.793560222834461,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 16092
    },
    {
      "epoch": 0.16093,
      "grad_norm": 1.1309312644698293,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 16093
    },
    {
      "epoch": 0.16094,
      "grad_norm": 1.3730381439115702,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 16094
    },
    {
      "epoch": 0.16095,
      "grad_norm": 1.4378123039792088,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 16095
    },
    {
      "epoch": 0.16096,
      "grad_norm": 1.3389759471446794,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 16096
    },
    {
      "epoch": 0.16097,
      "grad_norm": 1.3158504027456166,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 16097
    },
    {
      "epoch": 0.16098,
      "grad_norm": 0.9877951756956128,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 16098
    },
    {
      "epoch": 0.16099,
      "grad_norm": 1.4479428336464673,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16099
    },
    {
      "epoch": 0.161,
      "grad_norm": 1.0690018663453107,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 16100
    },
    {
      "epoch": 0.16101,
      "grad_norm": 1.1087175494936976,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 16101
    },
    {
      "epoch": 0.16102,
      "grad_norm": 1.2190483165806931,
      "learning_rate": 0.003,
      "loss": 4.0273,
      "step": 16102
    },
    {
      "epoch": 0.16103,
      "grad_norm": 1.2464640512391514,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 16103
    },
    {
      "epoch": 0.16104,
      "grad_norm": 1.2605449243517843,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 16104
    },
    {
      "epoch": 0.16105,
      "grad_norm": 1.3609265610053145,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 16105
    },
    {
      "epoch": 0.16106,
      "grad_norm": 1.1561289517191853,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 16106
    },
    {
      "epoch": 0.16107,
      "grad_norm": 1.2108900446658217,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 16107
    },
    {
      "epoch": 0.16108,
      "grad_norm": 1.3871226865725534,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 16108
    },
    {
      "epoch": 0.16109,
      "grad_norm": 1.2416300741278905,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 16109
    },
    {
      "epoch": 0.1611,
      "grad_norm": 1.3549279352566184,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 16110
    },
    {
      "epoch": 0.16111,
      "grad_norm": 1.3554516222304485,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 16111
    },
    {
      "epoch": 0.16112,
      "grad_norm": 1.1836975651978479,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 16112
    },
    {
      "epoch": 0.16113,
      "grad_norm": 1.215756601682861,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 16113
    },
    {
      "epoch": 0.16114,
      "grad_norm": 1.077720507183036,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 16114
    },
    {
      "epoch": 0.16115,
      "grad_norm": 1.2836790769497726,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 16115
    },
    {
      "epoch": 0.16116,
      "grad_norm": 1.2067943231225828,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 16116
    },
    {
      "epoch": 0.16117,
      "grad_norm": 1.2529076106920993,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 16117
    },
    {
      "epoch": 0.16118,
      "grad_norm": 1.261711886375417,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 16118
    },
    {
      "epoch": 0.16119,
      "grad_norm": 1.1898514059835839,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 16119
    },
    {
      "epoch": 0.1612,
      "grad_norm": 1.042256094147418,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 16120
    },
    {
      "epoch": 0.16121,
      "grad_norm": 1.2608642545166504,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 16121
    },
    {
      "epoch": 0.16122,
      "grad_norm": 1.2194499849594356,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 16122
    },
    {
      "epoch": 0.16123,
      "grad_norm": 1.3722797412364909,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 16123
    },
    {
      "epoch": 0.16124,
      "grad_norm": 1.1532481352638193,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 16124
    },
    {
      "epoch": 0.16125,
      "grad_norm": 1.3444693723979186,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 16125
    },
    {
      "epoch": 0.16126,
      "grad_norm": 1.3587021060251283,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 16126
    },
    {
      "epoch": 0.16127,
      "grad_norm": 1.4680093616394054,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 16127
    },
    {
      "epoch": 0.16128,
      "grad_norm": 1.3708040186891726,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 16128
    },
    {
      "epoch": 0.16129,
      "grad_norm": 1.265944768899528,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16129
    },
    {
      "epoch": 0.1613,
      "grad_norm": 1.2147761587916155,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 16130
    },
    {
      "epoch": 0.16131,
      "grad_norm": 1.2783421175922953,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 16131
    },
    {
      "epoch": 0.16132,
      "grad_norm": 1.2751345573542021,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 16132
    },
    {
      "epoch": 0.16133,
      "grad_norm": 1.2163991871109832,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 16133
    },
    {
      "epoch": 0.16134,
      "grad_norm": 1.2495786526427402,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 16134
    },
    {
      "epoch": 0.16135,
      "grad_norm": 1.2552880181808581,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 16135
    },
    {
      "epoch": 0.16136,
      "grad_norm": 1.4879017034123712,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 16136
    },
    {
      "epoch": 0.16137,
      "grad_norm": 1.1368841189315837,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 16137
    },
    {
      "epoch": 0.16138,
      "grad_norm": 1.4063956846527264,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 16138
    },
    {
      "epoch": 0.16139,
      "grad_norm": 1.2446175451255377,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 16139
    },
    {
      "epoch": 0.1614,
      "grad_norm": 1.0876939463342667,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 16140
    },
    {
      "epoch": 0.16141,
      "grad_norm": 1.3324680737003314,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 16141
    },
    {
      "epoch": 0.16142,
      "grad_norm": 1.341488522037896,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 16142
    },
    {
      "epoch": 0.16143,
      "grad_norm": 1.0284822379365979,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 16143
    },
    {
      "epoch": 0.16144,
      "grad_norm": 1.446545691671991,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 16144
    },
    {
      "epoch": 0.16145,
      "grad_norm": 1.198007631522013,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 16145
    },
    {
      "epoch": 0.16146,
      "grad_norm": 1.068927102975408,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 16146
    },
    {
      "epoch": 0.16147,
      "grad_norm": 1.5003008190770473,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 16147
    },
    {
      "epoch": 0.16148,
      "grad_norm": 1.0764359861587243,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 16148
    },
    {
      "epoch": 0.16149,
      "grad_norm": 1.2309420090166217,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 16149
    },
    {
      "epoch": 0.1615,
      "grad_norm": 1.3078507897419687,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 16150
    },
    {
      "epoch": 0.16151,
      "grad_norm": 1.462687140006635,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 16151
    },
    {
      "epoch": 0.16152,
      "grad_norm": 1.1113703241677952,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 16152
    },
    {
      "epoch": 0.16153,
      "grad_norm": 1.2475908893404852,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 16153
    },
    {
      "epoch": 0.16154,
      "grad_norm": 1.6105637724327737,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 16154
    },
    {
      "epoch": 0.16155,
      "grad_norm": 1.0413256463913592,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 16155
    },
    {
      "epoch": 0.16156,
      "grad_norm": 1.3574353999341298,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 16156
    },
    {
      "epoch": 0.16157,
      "grad_norm": 1.2846752995761646,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 16157
    },
    {
      "epoch": 0.16158,
      "grad_norm": 1.3225066680499187,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 16158
    },
    {
      "epoch": 0.16159,
      "grad_norm": 1.3116346575986928,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 16159
    },
    {
      "epoch": 0.1616,
      "grad_norm": 1.0986634848005652,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 16160
    },
    {
      "epoch": 0.16161,
      "grad_norm": 1.489009004397367,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 16161
    },
    {
      "epoch": 0.16162,
      "grad_norm": 1.0671443540009917,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 16162
    },
    {
      "epoch": 0.16163,
      "grad_norm": 1.3852087044343444,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 16163
    },
    {
      "epoch": 0.16164,
      "grad_norm": 1.305599729039529,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 16164
    },
    {
      "epoch": 0.16165,
      "grad_norm": 1.151567243082139,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 16165
    },
    {
      "epoch": 0.16166,
      "grad_norm": 1.501787296654248,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 16166
    },
    {
      "epoch": 0.16167,
      "grad_norm": 1.1041427692019146,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 16167
    },
    {
      "epoch": 0.16168,
      "grad_norm": 1.4258051081646463,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 16168
    },
    {
      "epoch": 0.16169,
      "grad_norm": 1.1599263469979852,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 16169
    },
    {
      "epoch": 0.1617,
      "grad_norm": 1.1963195979131709,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 16170
    },
    {
      "epoch": 0.16171,
      "grad_norm": 1.2205864585786506,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 16171
    },
    {
      "epoch": 0.16172,
      "grad_norm": 1.0770953868565933,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 16172
    },
    {
      "epoch": 0.16173,
      "grad_norm": 1.3909570250524284,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 16173
    },
    {
      "epoch": 0.16174,
      "grad_norm": 0.9878212489967368,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 16174
    },
    {
      "epoch": 0.16175,
      "grad_norm": 1.3457320433704199,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 16175
    },
    {
      "epoch": 0.16176,
      "grad_norm": 0.9850536798331347,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 16176
    },
    {
      "epoch": 0.16177,
      "grad_norm": 1.4579496653506565,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 16177
    },
    {
      "epoch": 0.16178,
      "grad_norm": 1.3846670682448412,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 16178
    },
    {
      "epoch": 0.16179,
      "grad_norm": 1.559565431975786,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 16179
    },
    {
      "epoch": 0.1618,
      "grad_norm": 0.9908894518485826,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 16180
    },
    {
      "epoch": 0.16181,
      "grad_norm": 1.348064109065105,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 16181
    },
    {
      "epoch": 0.16182,
      "grad_norm": 1.399950817707883,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 16182
    },
    {
      "epoch": 0.16183,
      "grad_norm": 1.118259926151866,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16183
    },
    {
      "epoch": 0.16184,
      "grad_norm": 1.394829148863611,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 16184
    },
    {
      "epoch": 0.16185,
      "grad_norm": 1.1745608278679487,
      "learning_rate": 0.003,
      "loss": 4.1002,
      "step": 16185
    },
    {
      "epoch": 0.16186,
      "grad_norm": 1.251648889195004,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 16186
    },
    {
      "epoch": 0.16187,
      "grad_norm": 1.2524277254239389,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 16187
    },
    {
      "epoch": 0.16188,
      "grad_norm": 1.2350728005269838,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 16188
    },
    {
      "epoch": 0.16189,
      "grad_norm": 1.291575283921151,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 16189
    },
    {
      "epoch": 0.1619,
      "grad_norm": 1.36391824636541,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 16190
    },
    {
      "epoch": 0.16191,
      "grad_norm": 1.5770002178564555,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 16191
    },
    {
      "epoch": 0.16192,
      "grad_norm": 1.2634525353872517,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16192
    },
    {
      "epoch": 0.16193,
      "grad_norm": 1.28488165744424,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 16193
    },
    {
      "epoch": 0.16194,
      "grad_norm": 1.3465642431600706,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 16194
    },
    {
      "epoch": 0.16195,
      "grad_norm": 1.2280001595506032,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 16195
    },
    {
      "epoch": 0.16196,
      "grad_norm": 1.3282442151996745,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 16196
    },
    {
      "epoch": 0.16197,
      "grad_norm": 0.9150787690474047,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 16197
    },
    {
      "epoch": 0.16198,
      "grad_norm": 1.1928188777396278,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 16198
    },
    {
      "epoch": 0.16199,
      "grad_norm": 1.20056096196458,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 16199
    },
    {
      "epoch": 0.162,
      "grad_norm": 1.2854712720823054,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 16200
    },
    {
      "epoch": 0.16201,
      "grad_norm": 1.0471238400168474,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 16201
    },
    {
      "epoch": 0.16202,
      "grad_norm": 1.2771481878611228,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 16202
    },
    {
      "epoch": 0.16203,
      "grad_norm": 1.2991642908445467,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 16203
    },
    {
      "epoch": 0.16204,
      "grad_norm": 1.391096078521732,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 16204
    },
    {
      "epoch": 0.16205,
      "grad_norm": 1.2138677594564928,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16205
    },
    {
      "epoch": 0.16206,
      "grad_norm": 1.5280425359595857,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 16206
    },
    {
      "epoch": 0.16207,
      "grad_norm": 1.2393184803608204,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 16207
    },
    {
      "epoch": 0.16208,
      "grad_norm": 1.1685867078437393,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 16208
    },
    {
      "epoch": 0.16209,
      "grad_norm": 1.313631968426678,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 16209
    },
    {
      "epoch": 0.1621,
      "grad_norm": 1.183468025996727,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 16210
    },
    {
      "epoch": 0.16211,
      "grad_norm": 1.3508023550431656,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 16211
    },
    {
      "epoch": 0.16212,
      "grad_norm": 1.2543446443891058,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 16212
    },
    {
      "epoch": 0.16213,
      "grad_norm": 1.07806009180168,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 16213
    },
    {
      "epoch": 0.16214,
      "grad_norm": 1.2482642071439212,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 16214
    },
    {
      "epoch": 0.16215,
      "grad_norm": 1.12070374858544,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 16215
    },
    {
      "epoch": 0.16216,
      "grad_norm": 1.2417339257100313,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 16216
    },
    {
      "epoch": 0.16217,
      "grad_norm": 1.021932077670497,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 16217
    },
    {
      "epoch": 0.16218,
      "grad_norm": 1.4476588358430686,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 16218
    },
    {
      "epoch": 0.16219,
      "grad_norm": 1.2801147532941688,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 16219
    },
    {
      "epoch": 0.1622,
      "grad_norm": 1.3666638771024235,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 16220
    },
    {
      "epoch": 0.16221,
      "grad_norm": 1.6919655916293102,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 16221
    },
    {
      "epoch": 0.16222,
      "grad_norm": 1.133020693571414,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 16222
    },
    {
      "epoch": 0.16223,
      "grad_norm": 1.2765201950142349,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 16223
    },
    {
      "epoch": 0.16224,
      "grad_norm": 1.292954656723287,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 16224
    },
    {
      "epoch": 0.16225,
      "grad_norm": 1.4715150749213328,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 16225
    },
    {
      "epoch": 0.16226,
      "grad_norm": 0.9149378601180623,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 16226
    },
    {
      "epoch": 0.16227,
      "grad_norm": 1.230883163804639,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 16227
    },
    {
      "epoch": 0.16228,
      "grad_norm": 1.3347318552909315,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 16228
    },
    {
      "epoch": 0.16229,
      "grad_norm": 1.0674427620211304,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 16229
    },
    {
      "epoch": 0.1623,
      "grad_norm": 1.338595938504359,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 16230
    },
    {
      "epoch": 0.16231,
      "grad_norm": 1.101809798055724,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16231
    },
    {
      "epoch": 0.16232,
      "grad_norm": 1.2083469494542567,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 16232
    },
    {
      "epoch": 0.16233,
      "grad_norm": 1.1788584236976805,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 16233
    },
    {
      "epoch": 0.16234,
      "grad_norm": 1.438190409187352,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 16234
    },
    {
      "epoch": 0.16235,
      "grad_norm": 1.3217102563848495,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 16235
    },
    {
      "epoch": 0.16236,
      "grad_norm": 1.1574464022417847,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 16236
    },
    {
      "epoch": 0.16237,
      "grad_norm": 1.6586220373587746,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 16237
    },
    {
      "epoch": 0.16238,
      "grad_norm": 1.1904629378808893,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 16238
    },
    {
      "epoch": 0.16239,
      "grad_norm": 1.4029682136773813,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 16239
    },
    {
      "epoch": 0.1624,
      "grad_norm": 1.309501521771856,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 16240
    },
    {
      "epoch": 0.16241,
      "grad_norm": 1.1912634851965251,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 16241
    },
    {
      "epoch": 0.16242,
      "grad_norm": 1.1957762603851831,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 16242
    },
    {
      "epoch": 0.16243,
      "grad_norm": 0.9978203017900059,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 16243
    },
    {
      "epoch": 0.16244,
      "grad_norm": 1.3631017302455204,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 16244
    },
    {
      "epoch": 0.16245,
      "grad_norm": 1.1172482592038133,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 16245
    },
    {
      "epoch": 0.16246,
      "grad_norm": 1.291457136090636,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 16246
    },
    {
      "epoch": 0.16247,
      "grad_norm": 1.241648838090917,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 16247
    },
    {
      "epoch": 0.16248,
      "grad_norm": 1.1912627487580116,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 16248
    },
    {
      "epoch": 0.16249,
      "grad_norm": 1.4376949678364168,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 16249
    },
    {
      "epoch": 0.1625,
      "grad_norm": 1.332877486817461,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 16250
    },
    {
      "epoch": 0.16251,
      "grad_norm": 1.450341644372587,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 16251
    },
    {
      "epoch": 0.16252,
      "grad_norm": 1.2216004688391267,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 16252
    },
    {
      "epoch": 0.16253,
      "grad_norm": 1.38458126587778,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 16253
    },
    {
      "epoch": 0.16254,
      "grad_norm": 1.1578657329660573,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 16254
    },
    {
      "epoch": 0.16255,
      "grad_norm": 1.3584737020831286,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 16255
    },
    {
      "epoch": 0.16256,
      "grad_norm": 1.0626796569996118,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 16256
    },
    {
      "epoch": 0.16257,
      "grad_norm": 1.1854307824297667,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 16257
    },
    {
      "epoch": 0.16258,
      "grad_norm": 1.0881261326227933,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 16258
    },
    {
      "epoch": 0.16259,
      "grad_norm": 1.1159855556877203,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 16259
    },
    {
      "epoch": 0.1626,
      "grad_norm": 1.1850051078218073,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 16260
    },
    {
      "epoch": 0.16261,
      "grad_norm": 1.2954838458818088,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 16261
    },
    {
      "epoch": 0.16262,
      "grad_norm": 1.3844454619665678,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 16262
    },
    {
      "epoch": 0.16263,
      "grad_norm": 1.4326657049181495,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 16263
    },
    {
      "epoch": 0.16264,
      "grad_norm": 1.0972597975116023,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 16264
    },
    {
      "epoch": 0.16265,
      "grad_norm": 1.456728688815861,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 16265
    },
    {
      "epoch": 0.16266,
      "grad_norm": 1.0893231721970618,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 16266
    },
    {
      "epoch": 0.16267,
      "grad_norm": 1.4265836466647381,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 16267
    },
    {
      "epoch": 0.16268,
      "grad_norm": 1.026908516804711,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 16268
    },
    {
      "epoch": 0.16269,
      "grad_norm": 1.2867505437326605,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 16269
    },
    {
      "epoch": 0.1627,
      "grad_norm": 1.3769738983162028,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 16270
    },
    {
      "epoch": 0.16271,
      "grad_norm": 1.1512571132733376,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 16271
    },
    {
      "epoch": 0.16272,
      "grad_norm": 1.2946312032229343,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 16272
    },
    {
      "epoch": 0.16273,
      "grad_norm": 1.160598306973601,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 16273
    },
    {
      "epoch": 0.16274,
      "grad_norm": 1.190913322895234,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 16274
    },
    {
      "epoch": 0.16275,
      "grad_norm": 1.3565131412588454,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 16275
    },
    {
      "epoch": 0.16276,
      "grad_norm": 1.2228374178624037,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 16276
    },
    {
      "epoch": 0.16277,
      "grad_norm": 1.157740305349302,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 16277
    },
    {
      "epoch": 0.16278,
      "grad_norm": 1.1589390738190832,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 16278
    },
    {
      "epoch": 0.16279,
      "grad_norm": 1.1427874625875416,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 16279
    },
    {
      "epoch": 0.1628,
      "grad_norm": 1.4394281861045208,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 16280
    },
    {
      "epoch": 0.16281,
      "grad_norm": 1.2479386614109227,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 16281
    },
    {
      "epoch": 0.16282,
      "grad_norm": 1.4435228993586355,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 16282
    },
    {
      "epoch": 0.16283,
      "grad_norm": 1.2367412237166255,
      "learning_rate": 0.003,
      "loss": 4.025,
      "step": 16283
    },
    {
      "epoch": 0.16284,
      "grad_norm": 1.630935112607969,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 16284
    },
    {
      "epoch": 0.16285,
      "grad_norm": 1.2151600151267754,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 16285
    },
    {
      "epoch": 0.16286,
      "grad_norm": 0.99844286924819,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 16286
    },
    {
      "epoch": 0.16287,
      "grad_norm": 1.3311858498500226,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 16287
    },
    {
      "epoch": 0.16288,
      "grad_norm": 1.1277454830696854,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 16288
    },
    {
      "epoch": 0.16289,
      "grad_norm": 1.3751420588987922,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 16289
    },
    {
      "epoch": 0.1629,
      "grad_norm": 1.2574333033294496,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 16290
    },
    {
      "epoch": 0.16291,
      "grad_norm": 1.2709011614362218,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 16291
    },
    {
      "epoch": 0.16292,
      "grad_norm": 1.1731475105596485,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 16292
    },
    {
      "epoch": 0.16293,
      "grad_norm": 1.362351505374211,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 16293
    },
    {
      "epoch": 0.16294,
      "grad_norm": 1.2153049142800547,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 16294
    },
    {
      "epoch": 0.16295,
      "grad_norm": 1.2855140261647056,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 16295
    },
    {
      "epoch": 0.16296,
      "grad_norm": 1.2451835918537577,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 16296
    },
    {
      "epoch": 0.16297,
      "grad_norm": 1.0899963185254051,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 16297
    },
    {
      "epoch": 0.16298,
      "grad_norm": 1.5368176411159515,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 16298
    },
    {
      "epoch": 0.16299,
      "grad_norm": 1.1951570026687528,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 16299
    },
    {
      "epoch": 0.163,
      "grad_norm": 1.677062653199514,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 16300
    },
    {
      "epoch": 0.16301,
      "grad_norm": 1.3448873086832585,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 16301
    },
    {
      "epoch": 0.16302,
      "grad_norm": 1.1602759708115664,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 16302
    },
    {
      "epoch": 0.16303,
      "grad_norm": 1.3811138715383926,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 16303
    },
    {
      "epoch": 0.16304,
      "grad_norm": 1.1069545503836433,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 16304
    },
    {
      "epoch": 0.16305,
      "grad_norm": 1.570100339629662,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 16305
    },
    {
      "epoch": 0.16306,
      "grad_norm": 1.232831755226583,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 16306
    },
    {
      "epoch": 0.16307,
      "grad_norm": 1.2644840112025777,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 16307
    },
    {
      "epoch": 0.16308,
      "grad_norm": 1.1619975932512598,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 16308
    },
    {
      "epoch": 0.16309,
      "grad_norm": 1.3731086711484763,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16309
    },
    {
      "epoch": 0.1631,
      "grad_norm": 1.4049245801410037,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 16310
    },
    {
      "epoch": 0.16311,
      "grad_norm": 1.064243387322423,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 16311
    },
    {
      "epoch": 0.16312,
      "grad_norm": 1.3639565242761726,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 16312
    },
    {
      "epoch": 0.16313,
      "grad_norm": 1.1172881919093074,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 16313
    },
    {
      "epoch": 0.16314,
      "grad_norm": 1.165383167634807,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 16314
    },
    {
      "epoch": 0.16315,
      "grad_norm": 1.205942669437338,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 16315
    },
    {
      "epoch": 0.16316,
      "grad_norm": 1.1958451581976022,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 16316
    },
    {
      "epoch": 0.16317,
      "grad_norm": 1.1744914363089543,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 16317
    },
    {
      "epoch": 0.16318,
      "grad_norm": 1.466582108406998,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 16318
    },
    {
      "epoch": 0.16319,
      "grad_norm": 1.0385254669559079,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 16319
    },
    {
      "epoch": 0.1632,
      "grad_norm": 1.619029579687159,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 16320
    },
    {
      "epoch": 0.16321,
      "grad_norm": 1.1827478809755751,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 16321
    },
    {
      "epoch": 0.16322,
      "grad_norm": 1.4517056777140702,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 16322
    },
    {
      "epoch": 0.16323,
      "grad_norm": 1.3651946112406492,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 16323
    },
    {
      "epoch": 0.16324,
      "grad_norm": 1.661911846035659,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 16324
    },
    {
      "epoch": 0.16325,
      "grad_norm": 1.0679565552916765,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 16325
    },
    {
      "epoch": 0.16326,
      "grad_norm": 1.3346690818098346,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 16326
    },
    {
      "epoch": 0.16327,
      "grad_norm": 1.389890495562059,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 16327
    },
    {
      "epoch": 0.16328,
      "grad_norm": 1.0516321084279026,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 16328
    },
    {
      "epoch": 0.16329,
      "grad_norm": 1.3894814680474321,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 16329
    },
    {
      "epoch": 0.1633,
      "grad_norm": 1.1799369245058706,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 16330
    },
    {
      "epoch": 0.16331,
      "grad_norm": 1.2172999397129118,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 16331
    },
    {
      "epoch": 0.16332,
      "grad_norm": 1.4402792548750516,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 16332
    },
    {
      "epoch": 0.16333,
      "grad_norm": 1.0962335816678261,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 16333
    },
    {
      "epoch": 0.16334,
      "grad_norm": 1.0535394881053008,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 16334
    },
    {
      "epoch": 0.16335,
      "grad_norm": 1.3723287192785087,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 16335
    },
    {
      "epoch": 0.16336,
      "grad_norm": 1.2712505974707535,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 16336
    },
    {
      "epoch": 0.16337,
      "grad_norm": 1.4053550627250662,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 16337
    },
    {
      "epoch": 0.16338,
      "grad_norm": 1.3102522704393156,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 16338
    },
    {
      "epoch": 0.16339,
      "grad_norm": 1.0884281320779658,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 16339
    },
    {
      "epoch": 0.1634,
      "grad_norm": 1.369979709062412,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 16340
    },
    {
      "epoch": 0.16341,
      "grad_norm": 1.2454029918205807,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 16341
    },
    {
      "epoch": 0.16342,
      "grad_norm": 1.1386512829299478,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 16342
    },
    {
      "epoch": 0.16343,
      "grad_norm": 1.4316064097597532,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 16343
    },
    {
      "epoch": 0.16344,
      "grad_norm": 1.3025839922559512,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 16344
    },
    {
      "epoch": 0.16345,
      "grad_norm": 1.4062081379943796,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 16345
    },
    {
      "epoch": 0.16346,
      "grad_norm": 1.380426926743026,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 16346
    },
    {
      "epoch": 0.16347,
      "grad_norm": 1.048439154887281,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 16347
    },
    {
      "epoch": 0.16348,
      "grad_norm": 1.4187306660020738,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 16348
    },
    {
      "epoch": 0.16349,
      "grad_norm": 1.133000049884522,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 16349
    },
    {
      "epoch": 0.1635,
      "grad_norm": 1.152077902344025,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 16350
    },
    {
      "epoch": 0.16351,
      "grad_norm": 1.2543029760980908,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 16351
    },
    {
      "epoch": 0.16352,
      "grad_norm": 1.1991339006489674,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 16352
    },
    {
      "epoch": 0.16353,
      "grad_norm": 1.3279973071452171,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 16353
    },
    {
      "epoch": 0.16354,
      "grad_norm": 1.3209555143437766,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 16354
    },
    {
      "epoch": 0.16355,
      "grad_norm": 1.6069005203635782,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 16355
    },
    {
      "epoch": 0.16356,
      "grad_norm": 1.3874060586460157,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 16356
    },
    {
      "epoch": 0.16357,
      "grad_norm": 1.2573377261700582,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 16357
    },
    {
      "epoch": 0.16358,
      "grad_norm": 1.5693088348261484,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 16358
    },
    {
      "epoch": 0.16359,
      "grad_norm": 1.2199974621945564,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 16359
    },
    {
      "epoch": 0.1636,
      "grad_norm": 1.198373588602752,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 16360
    },
    {
      "epoch": 0.16361,
      "grad_norm": 1.2415033730433311,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 16361
    },
    {
      "epoch": 0.16362,
      "grad_norm": 1.300340553941177,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 16362
    },
    {
      "epoch": 0.16363,
      "grad_norm": 1.3048339173505643,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 16363
    },
    {
      "epoch": 0.16364,
      "grad_norm": 1.2023696939439263,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 16364
    },
    {
      "epoch": 0.16365,
      "grad_norm": 1.4217090836305526,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 16365
    },
    {
      "epoch": 0.16366,
      "grad_norm": 1.1233430770093011,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 16366
    },
    {
      "epoch": 0.16367,
      "grad_norm": 1.4427845967168846,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 16367
    },
    {
      "epoch": 0.16368,
      "grad_norm": 1.1155905332168412,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 16368
    },
    {
      "epoch": 0.16369,
      "grad_norm": 1.3464489557382935,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 16369
    },
    {
      "epoch": 0.1637,
      "grad_norm": 1.0578550159183047,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 16370
    },
    {
      "epoch": 0.16371,
      "grad_norm": 1.511918444515664,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 16371
    },
    {
      "epoch": 0.16372,
      "grad_norm": 1.129299153317865,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 16372
    },
    {
      "epoch": 0.16373,
      "grad_norm": 1.587989632774138,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 16373
    },
    {
      "epoch": 0.16374,
      "grad_norm": 1.1700883074011486,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 16374
    },
    {
      "epoch": 0.16375,
      "grad_norm": 1.3040537631563571,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 16375
    },
    {
      "epoch": 0.16376,
      "grad_norm": 1.0856008882922459,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 16376
    },
    {
      "epoch": 0.16377,
      "grad_norm": 1.3168505447124415,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 16377
    },
    {
      "epoch": 0.16378,
      "grad_norm": 1.409911310129991,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 16378
    },
    {
      "epoch": 0.16379,
      "grad_norm": 1.1578313905653308,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 16379
    },
    {
      "epoch": 0.1638,
      "grad_norm": 1.2243553009162318,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 16380
    },
    {
      "epoch": 0.16381,
      "grad_norm": 1.3211151537680381,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 16381
    },
    {
      "epoch": 0.16382,
      "grad_norm": 1.0003765638365243,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 16382
    },
    {
      "epoch": 0.16383,
      "grad_norm": 1.4431212791886239,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 16383
    },
    {
      "epoch": 0.16384,
      "grad_norm": 1.1418299209144636,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 16384
    },
    {
      "epoch": 0.16385,
      "grad_norm": 1.2792884718943385,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 16385
    },
    {
      "epoch": 0.16386,
      "grad_norm": 1.1404948764661138,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 16386
    },
    {
      "epoch": 0.16387,
      "grad_norm": 1.284319680515499,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 16387
    },
    {
      "epoch": 0.16388,
      "grad_norm": 1.4158203236465858,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 16388
    },
    {
      "epoch": 0.16389,
      "grad_norm": 1.4336931757387177,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 16389
    },
    {
      "epoch": 0.1639,
      "grad_norm": 1.2536201072137658,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 16390
    },
    {
      "epoch": 0.16391,
      "grad_norm": 1.2836875225183824,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 16391
    },
    {
      "epoch": 0.16392,
      "grad_norm": 1.2176262982047852,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 16392
    },
    {
      "epoch": 0.16393,
      "grad_norm": 1.4483108282285946,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 16393
    },
    {
      "epoch": 0.16394,
      "grad_norm": 1.246150339402114,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 16394
    },
    {
      "epoch": 0.16395,
      "grad_norm": 1.381316017329402,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 16395
    },
    {
      "epoch": 0.16396,
      "grad_norm": 1.3577173637161588,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 16396
    },
    {
      "epoch": 0.16397,
      "grad_norm": 1.1911270331753483,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 16397
    },
    {
      "epoch": 0.16398,
      "grad_norm": 1.420612509964672,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 16398
    },
    {
      "epoch": 0.16399,
      "grad_norm": 1.1716767922518303,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 16399
    },
    {
      "epoch": 0.164,
      "grad_norm": 1.380990633593178,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 16400
    },
    {
      "epoch": 0.16401,
      "grad_norm": 1.2059872705480625,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 16401
    },
    {
      "epoch": 0.16402,
      "grad_norm": 1.2082775563370716,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 16402
    },
    {
      "epoch": 0.16403,
      "grad_norm": 1.3569918868573316,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 16403
    },
    {
      "epoch": 0.16404,
      "grad_norm": 1.2132090912442626,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 16404
    },
    {
      "epoch": 0.16405,
      "grad_norm": 1.3219062034164897,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 16405
    },
    {
      "epoch": 0.16406,
      "grad_norm": 1.2571787124909675,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 16406
    },
    {
      "epoch": 0.16407,
      "grad_norm": 1.090618551506124,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 16407
    },
    {
      "epoch": 0.16408,
      "grad_norm": 1.4297594624104537,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 16408
    },
    {
      "epoch": 0.16409,
      "grad_norm": 1.1132480973468442,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 16409
    },
    {
      "epoch": 0.1641,
      "grad_norm": 1.5933083288070427,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 16410
    },
    {
      "epoch": 0.16411,
      "grad_norm": 1.1376891610757405,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 16411
    },
    {
      "epoch": 0.16412,
      "grad_norm": 1.114888268375818,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 16412
    },
    {
      "epoch": 0.16413,
      "grad_norm": 1.2345716566950218,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 16413
    },
    {
      "epoch": 0.16414,
      "grad_norm": 1.246216303691841,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 16414
    },
    {
      "epoch": 0.16415,
      "grad_norm": 1.1663422313702552,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 16415
    },
    {
      "epoch": 0.16416,
      "grad_norm": 1.3673709590515428,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 16416
    },
    {
      "epoch": 0.16417,
      "grad_norm": 1.4693437831528886,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 16417
    },
    {
      "epoch": 0.16418,
      "grad_norm": 1.1785751760532168,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 16418
    },
    {
      "epoch": 0.16419,
      "grad_norm": 1.2877570452345621,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 16419
    },
    {
      "epoch": 0.1642,
      "grad_norm": 1.3446446582739822,
      "learning_rate": 0.003,
      "loss": 4.1047,
      "step": 16420
    },
    {
      "epoch": 0.16421,
      "grad_norm": 1.0676062088698817,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 16421
    },
    {
      "epoch": 0.16422,
      "grad_norm": 1.1319626846229354,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 16422
    },
    {
      "epoch": 0.16423,
      "grad_norm": 1.5613091015565685,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 16423
    },
    {
      "epoch": 0.16424,
      "grad_norm": 1.4368794722217562,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 16424
    },
    {
      "epoch": 0.16425,
      "grad_norm": 1.139300263428175,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 16425
    },
    {
      "epoch": 0.16426,
      "grad_norm": 1.2165822346927826,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 16426
    },
    {
      "epoch": 0.16427,
      "grad_norm": 1.32474169950318,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 16427
    },
    {
      "epoch": 0.16428,
      "grad_norm": 1.2124515065784107,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 16428
    },
    {
      "epoch": 0.16429,
      "grad_norm": 1.2099272814720674,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 16429
    },
    {
      "epoch": 0.1643,
      "grad_norm": 1.3165838134959365,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 16430
    },
    {
      "epoch": 0.16431,
      "grad_norm": 1.3124340867269597,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 16431
    },
    {
      "epoch": 0.16432,
      "grad_norm": 1.3083488161900734,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 16432
    },
    {
      "epoch": 0.16433,
      "grad_norm": 1.2212582053693837,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 16433
    },
    {
      "epoch": 0.16434,
      "grad_norm": 1.248712178980779,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 16434
    },
    {
      "epoch": 0.16435,
      "grad_norm": 1.10247019771398,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 16435
    },
    {
      "epoch": 0.16436,
      "grad_norm": 1.3138991041673305,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 16436
    },
    {
      "epoch": 0.16437,
      "grad_norm": 1.1426017630552887,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 16437
    },
    {
      "epoch": 0.16438,
      "grad_norm": 1.3516739937256734,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 16438
    },
    {
      "epoch": 0.16439,
      "grad_norm": 1.1325632245553459,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 16439
    },
    {
      "epoch": 0.1644,
      "grad_norm": 1.5530152726734836,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 16440
    },
    {
      "epoch": 0.16441,
      "grad_norm": 1.0258403136890335,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 16441
    },
    {
      "epoch": 0.16442,
      "grad_norm": 1.6038201349055385,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 16442
    },
    {
      "epoch": 0.16443,
      "grad_norm": 1.1902209962381898,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 16443
    },
    {
      "epoch": 0.16444,
      "grad_norm": 1.2619373346371396,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 16444
    },
    {
      "epoch": 0.16445,
      "grad_norm": 1.0269781849552861,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16445
    },
    {
      "epoch": 0.16446,
      "grad_norm": 1.3911572338821512,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16446
    },
    {
      "epoch": 0.16447,
      "grad_norm": 1.0130236852413237,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 16447
    },
    {
      "epoch": 0.16448,
      "grad_norm": 1.3752196679624056,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 16448
    },
    {
      "epoch": 0.16449,
      "grad_norm": 1.2228405074153301,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 16449
    },
    {
      "epoch": 0.1645,
      "grad_norm": 1.5814816887535803,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 16450
    },
    {
      "epoch": 0.16451,
      "grad_norm": 1.315260931427507,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 16451
    },
    {
      "epoch": 0.16452,
      "grad_norm": 1.4668306986859874,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 16452
    },
    {
      "epoch": 0.16453,
      "grad_norm": 1.4764637017528175,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 16453
    },
    {
      "epoch": 0.16454,
      "grad_norm": 1.1302037993586078,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 16454
    },
    {
      "epoch": 0.16455,
      "grad_norm": 1.2718503840267625,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 16455
    },
    {
      "epoch": 0.16456,
      "grad_norm": 1.1737062799099678,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 16456
    },
    {
      "epoch": 0.16457,
      "grad_norm": 1.1129495146380497,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 16457
    },
    {
      "epoch": 0.16458,
      "grad_norm": 1.0785969337090933,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 16458
    },
    {
      "epoch": 0.16459,
      "grad_norm": 1.271296796031856,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 16459
    },
    {
      "epoch": 0.1646,
      "grad_norm": 1.0523375357333073,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 16460
    },
    {
      "epoch": 0.16461,
      "grad_norm": 1.368714109205373,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 16461
    },
    {
      "epoch": 0.16462,
      "grad_norm": 1.0702950855721094,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 16462
    },
    {
      "epoch": 0.16463,
      "grad_norm": 1.57216920795091,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 16463
    },
    {
      "epoch": 0.16464,
      "grad_norm": 1.137099088097417,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 16464
    },
    {
      "epoch": 0.16465,
      "grad_norm": 1.1996377255304156,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 16465
    },
    {
      "epoch": 0.16466,
      "grad_norm": 1.6724182629270892,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 16466
    },
    {
      "epoch": 0.16467,
      "grad_norm": 1.5553433654665862,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 16467
    },
    {
      "epoch": 0.16468,
      "grad_norm": 1.1064706461046685,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 16468
    },
    {
      "epoch": 0.16469,
      "grad_norm": 1.3002040341038863,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 16469
    },
    {
      "epoch": 0.1647,
      "grad_norm": 1.2451762501791963,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 16470
    },
    {
      "epoch": 0.16471,
      "grad_norm": 1.4559204829357344,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 16471
    },
    {
      "epoch": 0.16472,
      "grad_norm": 1.2892392909664752,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 16472
    },
    {
      "epoch": 0.16473,
      "grad_norm": 1.4956841084217132,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 16473
    },
    {
      "epoch": 0.16474,
      "grad_norm": 1.0187601518228537,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 16474
    },
    {
      "epoch": 0.16475,
      "grad_norm": 1.2428533920598925,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 16475
    },
    {
      "epoch": 0.16476,
      "grad_norm": 1.1668821201355952,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 16476
    },
    {
      "epoch": 0.16477,
      "grad_norm": 1.1982463835413693,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 16477
    },
    {
      "epoch": 0.16478,
      "grad_norm": 1.2485534588013527,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 16478
    },
    {
      "epoch": 0.16479,
      "grad_norm": 1.3181502788303614,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16479
    },
    {
      "epoch": 0.1648,
      "grad_norm": 1.1974651532537637,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 16480
    },
    {
      "epoch": 0.16481,
      "grad_norm": 1.5443286790141675,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 16481
    },
    {
      "epoch": 0.16482,
      "grad_norm": 1.0580173964338795,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 16482
    },
    {
      "epoch": 0.16483,
      "grad_norm": 1.4011464567334797,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 16483
    },
    {
      "epoch": 0.16484,
      "grad_norm": 1.2466234410214732,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 16484
    },
    {
      "epoch": 0.16485,
      "grad_norm": 1.4197020019380282,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 16485
    },
    {
      "epoch": 0.16486,
      "grad_norm": 1.3467791156955868,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 16486
    },
    {
      "epoch": 0.16487,
      "grad_norm": 1.1286023331651984,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 16487
    },
    {
      "epoch": 0.16488,
      "grad_norm": 1.3406858944819247,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 16488
    },
    {
      "epoch": 0.16489,
      "grad_norm": 1.3346421533134032,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 16489
    },
    {
      "epoch": 0.1649,
      "grad_norm": 1.299839265832347,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 16490
    },
    {
      "epoch": 0.16491,
      "grad_norm": 1.3447492313470124,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 16491
    },
    {
      "epoch": 0.16492,
      "grad_norm": 1.480927716158016,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 16492
    },
    {
      "epoch": 0.16493,
      "grad_norm": 1.15363763860873,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 16493
    },
    {
      "epoch": 0.16494,
      "grad_norm": 1.088843248716626,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 16494
    },
    {
      "epoch": 0.16495,
      "grad_norm": 1.3271018110522426,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 16495
    },
    {
      "epoch": 0.16496,
      "grad_norm": 1.3579986770082573,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 16496
    },
    {
      "epoch": 0.16497,
      "grad_norm": 1.246765843625915,
      "learning_rate": 0.003,
      "loss": 4.0226,
      "step": 16497
    },
    {
      "epoch": 0.16498,
      "grad_norm": 1.3336299537332825,
      "learning_rate": 0.003,
      "loss": 4.0176,
      "step": 16498
    },
    {
      "epoch": 0.16499,
      "grad_norm": 1.1407031711918372,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 16499
    },
    {
      "epoch": 0.165,
      "grad_norm": 1.3208808146452224,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 16500
    },
    {
      "epoch": 0.16501,
      "grad_norm": 1.4543504750256484,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 16501
    },
    {
      "epoch": 0.16502,
      "grad_norm": 1.3363482030749658,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 16502
    },
    {
      "epoch": 0.16503,
      "grad_norm": 1.329011551322635,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 16503
    },
    {
      "epoch": 0.16504,
      "grad_norm": 1.3054054631462377,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 16504
    },
    {
      "epoch": 0.16505,
      "grad_norm": 1.1677461912364604,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 16505
    },
    {
      "epoch": 0.16506,
      "grad_norm": 1.1645870622599563,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 16506
    },
    {
      "epoch": 0.16507,
      "grad_norm": 1.2277219991165256,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 16507
    },
    {
      "epoch": 0.16508,
      "grad_norm": 1.1194755996128067,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 16508
    },
    {
      "epoch": 0.16509,
      "grad_norm": 1.5244986955115396,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 16509
    },
    {
      "epoch": 0.1651,
      "grad_norm": 0.8323002691807789,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 16510
    },
    {
      "epoch": 0.16511,
      "grad_norm": 1.1334584499206646,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 16511
    },
    {
      "epoch": 0.16512,
      "grad_norm": 1.2896129322877243,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 16512
    },
    {
      "epoch": 0.16513,
      "grad_norm": 1.0856970973611928,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 16513
    },
    {
      "epoch": 0.16514,
      "grad_norm": 1.1817625275124146,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 16514
    },
    {
      "epoch": 0.16515,
      "grad_norm": 1.3875030521708789,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 16515
    },
    {
      "epoch": 0.16516,
      "grad_norm": 1.5322349487881628,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 16516
    },
    {
      "epoch": 0.16517,
      "grad_norm": 1.214093787219609,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 16517
    },
    {
      "epoch": 0.16518,
      "grad_norm": 1.3264218281675586,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 16518
    },
    {
      "epoch": 0.16519,
      "grad_norm": 1.3037814719409515,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 16519
    },
    {
      "epoch": 0.1652,
      "grad_norm": 1.347256831948668,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 16520
    },
    {
      "epoch": 0.16521,
      "grad_norm": 1.135833220945594,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 16521
    },
    {
      "epoch": 0.16522,
      "grad_norm": 1.1635330169048987,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 16522
    },
    {
      "epoch": 0.16523,
      "grad_norm": 1.0707791474315451,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 16523
    },
    {
      "epoch": 0.16524,
      "grad_norm": 1.25384814518774,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 16524
    },
    {
      "epoch": 0.16525,
      "grad_norm": 1.2132133705864718,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 16525
    },
    {
      "epoch": 0.16526,
      "grad_norm": 1.7649845481238309,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 16526
    },
    {
      "epoch": 0.16527,
      "grad_norm": 1.1197744287796618,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 16527
    },
    {
      "epoch": 0.16528,
      "grad_norm": 1.5118003809505407,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 16528
    },
    {
      "epoch": 0.16529,
      "grad_norm": 1.4730620724374555,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 16529
    },
    {
      "epoch": 0.1653,
      "grad_norm": 0.90921825476028,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 16530
    },
    {
      "epoch": 0.16531,
      "grad_norm": 1.2822372958361714,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 16531
    },
    {
      "epoch": 0.16532,
      "grad_norm": 1.2598019825381823,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 16532
    },
    {
      "epoch": 0.16533,
      "grad_norm": 1.3640189473188689,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 16533
    },
    {
      "epoch": 0.16534,
      "grad_norm": 1.250697213719519,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 16534
    },
    {
      "epoch": 0.16535,
      "grad_norm": 1.1863462394733493,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 16535
    },
    {
      "epoch": 0.16536,
      "grad_norm": 1.193772061769314,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 16536
    },
    {
      "epoch": 0.16537,
      "grad_norm": 1.3875022671160369,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 16537
    },
    {
      "epoch": 0.16538,
      "grad_norm": 1.1032716696200444,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 16538
    },
    {
      "epoch": 0.16539,
      "grad_norm": 1.2085962067759086,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 16539
    },
    {
      "epoch": 0.1654,
      "grad_norm": 1.4437133351830067,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 16540
    },
    {
      "epoch": 0.16541,
      "grad_norm": 1.1113670923330192,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 16541
    },
    {
      "epoch": 0.16542,
      "grad_norm": 1.2923329685370313,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 16542
    },
    {
      "epoch": 0.16543,
      "grad_norm": 1.2265620690901557,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 16543
    },
    {
      "epoch": 0.16544,
      "grad_norm": 1.2889039235514912,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 16544
    },
    {
      "epoch": 0.16545,
      "grad_norm": 1.0361539017314416,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 16545
    },
    {
      "epoch": 0.16546,
      "grad_norm": 1.5710515904146527,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16546
    },
    {
      "epoch": 0.16547,
      "grad_norm": 0.9675541228764893,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 16547
    },
    {
      "epoch": 0.16548,
      "grad_norm": 1.3690807668603275,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 16548
    },
    {
      "epoch": 0.16549,
      "grad_norm": 1.2238178932528583,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 16549
    },
    {
      "epoch": 0.1655,
      "grad_norm": 1.3624593844958657,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 16550
    },
    {
      "epoch": 0.16551,
      "grad_norm": 1.1864513195867061,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 16551
    },
    {
      "epoch": 0.16552,
      "grad_norm": 1.516787527114139,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 16552
    },
    {
      "epoch": 0.16553,
      "grad_norm": 0.9501015686654448,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 16553
    },
    {
      "epoch": 0.16554,
      "grad_norm": 1.3711422680821514,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 16554
    },
    {
      "epoch": 0.16555,
      "grad_norm": 1.3117752994242997,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 16555
    },
    {
      "epoch": 0.16556,
      "grad_norm": 1.6255604613126764,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 16556
    },
    {
      "epoch": 0.16557,
      "grad_norm": 1.0632114951015792,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 16557
    },
    {
      "epoch": 0.16558,
      "grad_norm": 1.59166839672321,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 16558
    },
    {
      "epoch": 0.16559,
      "grad_norm": 1.0469939320043071,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 16559
    },
    {
      "epoch": 0.1656,
      "grad_norm": 1.4773552474017304,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 16560
    },
    {
      "epoch": 0.16561,
      "grad_norm": 1.130145898065509,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 16561
    },
    {
      "epoch": 0.16562,
      "grad_norm": 1.205195347039316,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 16562
    },
    {
      "epoch": 0.16563,
      "grad_norm": 1.276745680601048,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 16563
    },
    {
      "epoch": 0.16564,
      "grad_norm": 1.340516278687605,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 16564
    },
    {
      "epoch": 0.16565,
      "grad_norm": 1.2179714282624683,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 16565
    },
    {
      "epoch": 0.16566,
      "grad_norm": 1.5093543786751196,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 16566
    },
    {
      "epoch": 0.16567,
      "grad_norm": 1.4059023012697849,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 16567
    },
    {
      "epoch": 0.16568,
      "grad_norm": 1.1544566308338915,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 16568
    },
    {
      "epoch": 0.16569,
      "grad_norm": 1.4033121208928,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 16569
    },
    {
      "epoch": 0.1657,
      "grad_norm": 1.2238377475927165,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 16570
    },
    {
      "epoch": 0.16571,
      "grad_norm": 1.219524154636383,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 16571
    },
    {
      "epoch": 0.16572,
      "grad_norm": 1.308759384697221,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 16572
    },
    {
      "epoch": 0.16573,
      "grad_norm": 1.0418892214396567,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 16573
    },
    {
      "epoch": 0.16574,
      "grad_norm": 1.2344157645941867,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 16574
    },
    {
      "epoch": 0.16575,
      "grad_norm": 1.186935435165905,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 16575
    },
    {
      "epoch": 0.16576,
      "grad_norm": 1.3678784952055867,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 16576
    },
    {
      "epoch": 0.16577,
      "grad_norm": 1.0878995328997643,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 16577
    },
    {
      "epoch": 0.16578,
      "grad_norm": 1.3512824724156975,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 16578
    },
    {
      "epoch": 0.16579,
      "grad_norm": 1.1906761604147342,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 16579
    },
    {
      "epoch": 0.1658,
      "grad_norm": 1.1456813687747813,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 16580
    },
    {
      "epoch": 0.16581,
      "grad_norm": 1.2775782086054193,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 16581
    },
    {
      "epoch": 0.16582,
      "grad_norm": 1.6744773766700716,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 16582
    },
    {
      "epoch": 0.16583,
      "grad_norm": 1.2677766208582708,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 16583
    },
    {
      "epoch": 0.16584,
      "grad_norm": 1.4270773934465257,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 16584
    },
    {
      "epoch": 0.16585,
      "grad_norm": 1.3975056354370472,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 16585
    },
    {
      "epoch": 0.16586,
      "grad_norm": 1.2458642694686481,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 16586
    },
    {
      "epoch": 0.16587,
      "grad_norm": 1.2736822759593291,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 16587
    },
    {
      "epoch": 0.16588,
      "grad_norm": 1.236275197540774,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 16588
    },
    {
      "epoch": 0.16589,
      "grad_norm": 1.1066772994727698,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 16589
    },
    {
      "epoch": 0.1659,
      "grad_norm": 1.1232249599314457,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 16590
    },
    {
      "epoch": 0.16591,
      "grad_norm": 1.4093779362858834,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 16591
    },
    {
      "epoch": 0.16592,
      "grad_norm": 1.429195971694986,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 16592
    },
    {
      "epoch": 0.16593,
      "grad_norm": 1.1240960247071905,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 16593
    },
    {
      "epoch": 0.16594,
      "grad_norm": 1.2425919778673997,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 16594
    },
    {
      "epoch": 0.16595,
      "grad_norm": 1.397375625252414,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 16595
    },
    {
      "epoch": 0.16596,
      "grad_norm": 1.2009407512656327,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 16596
    },
    {
      "epoch": 0.16597,
      "grad_norm": 1.276527633707103,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 16597
    },
    {
      "epoch": 0.16598,
      "grad_norm": 1.533636505639626,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 16598
    },
    {
      "epoch": 0.16599,
      "grad_norm": 0.9518186678655945,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 16599
    },
    {
      "epoch": 0.166,
      "grad_norm": 1.2160159835882107,
      "learning_rate": 0.003,
      "loss": 4.0259,
      "step": 16600
    },
    {
      "epoch": 0.16601,
      "grad_norm": 1.2152796106285066,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 16601
    },
    {
      "epoch": 0.16602,
      "grad_norm": 1.2746660739876183,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 16602
    },
    {
      "epoch": 0.16603,
      "grad_norm": 1.4252702373047241,
      "learning_rate": 0.003,
      "loss": 4.0985,
      "step": 16603
    },
    {
      "epoch": 0.16604,
      "grad_norm": 1.0635959392211614,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 16604
    },
    {
      "epoch": 0.16605,
      "grad_norm": 1.552258145640101,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 16605
    },
    {
      "epoch": 0.16606,
      "grad_norm": 1.204222339131148,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 16606
    },
    {
      "epoch": 0.16607,
      "grad_norm": 1.5978386275991108,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 16607
    },
    {
      "epoch": 0.16608,
      "grad_norm": 1.1061872445758147,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 16608
    },
    {
      "epoch": 0.16609,
      "grad_norm": 1.2701346668331372,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 16609
    },
    {
      "epoch": 0.1661,
      "grad_norm": 1.3289461078989357,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 16610
    },
    {
      "epoch": 0.16611,
      "grad_norm": 1.1168024268611252,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 16611
    },
    {
      "epoch": 0.16612,
      "grad_norm": 1.3048247232069652,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 16612
    },
    {
      "epoch": 0.16613,
      "grad_norm": 1.018674315159564,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 16613
    },
    {
      "epoch": 0.16614,
      "grad_norm": 1.2739071182800958,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 16614
    },
    {
      "epoch": 0.16615,
      "grad_norm": 1.448527645704851,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 16615
    },
    {
      "epoch": 0.16616,
      "grad_norm": 1.2618724216619026,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 16616
    },
    {
      "epoch": 0.16617,
      "grad_norm": 1.4716514933801228,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 16617
    },
    {
      "epoch": 0.16618,
      "grad_norm": 1.000641246318044,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 16618
    },
    {
      "epoch": 0.16619,
      "grad_norm": 1.5442007930216206,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 16619
    },
    {
      "epoch": 0.1662,
      "grad_norm": 1.0895470650253285,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 16620
    },
    {
      "epoch": 0.16621,
      "grad_norm": 1.4581178360506257,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 16621
    },
    {
      "epoch": 0.16622,
      "grad_norm": 1.2356933981899747,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 16622
    },
    {
      "epoch": 0.16623,
      "grad_norm": 1.176423111853083,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 16623
    },
    {
      "epoch": 0.16624,
      "grad_norm": 1.3403710983985258,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 16624
    },
    {
      "epoch": 0.16625,
      "grad_norm": 1.167415870709617,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 16625
    },
    {
      "epoch": 0.16626,
      "grad_norm": 1.3870341030724083,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 16626
    },
    {
      "epoch": 0.16627,
      "grad_norm": 1.207894431592241,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 16627
    },
    {
      "epoch": 0.16628,
      "grad_norm": 1.4728929292785442,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 16628
    },
    {
      "epoch": 0.16629,
      "grad_norm": 1.1112205171923735,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 16629
    },
    {
      "epoch": 0.1663,
      "grad_norm": 1.1590305896874513,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 16630
    },
    {
      "epoch": 0.16631,
      "grad_norm": 1.482380357341026,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 16631
    },
    {
      "epoch": 0.16632,
      "grad_norm": 1.0773970821438095,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 16632
    },
    {
      "epoch": 0.16633,
      "grad_norm": 1.3817570315473837,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 16633
    },
    {
      "epoch": 0.16634,
      "grad_norm": 1.1575282789470405,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 16634
    },
    {
      "epoch": 0.16635,
      "grad_norm": 1.415087075334309,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 16635
    },
    {
      "epoch": 0.16636,
      "grad_norm": 0.9255282579734383,
      "learning_rate": 0.003,
      "loss": 4.0112,
      "step": 16636
    },
    {
      "epoch": 0.16637,
      "grad_norm": 1.2282673114747942,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 16637
    },
    {
      "epoch": 0.16638,
      "grad_norm": 1.6337941574914259,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 16638
    },
    {
      "epoch": 0.16639,
      "grad_norm": 1.1293362893753414,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 16639
    },
    {
      "epoch": 0.1664,
      "grad_norm": 1.2601235958197192,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 16640
    },
    {
      "epoch": 0.16641,
      "grad_norm": 1.2557510037069466,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 16641
    },
    {
      "epoch": 0.16642,
      "grad_norm": 1.1529014840095058,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 16642
    },
    {
      "epoch": 0.16643,
      "grad_norm": 1.4280815118907935,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 16643
    },
    {
      "epoch": 0.16644,
      "grad_norm": 1.1052315863725835,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 16644
    },
    {
      "epoch": 0.16645,
      "grad_norm": 1.2503694285321836,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 16645
    },
    {
      "epoch": 0.16646,
      "grad_norm": 1.2495900248899456,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 16646
    },
    {
      "epoch": 0.16647,
      "grad_norm": 1.375768919260447,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 16647
    },
    {
      "epoch": 0.16648,
      "grad_norm": 1.2305083673039703,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 16648
    },
    {
      "epoch": 0.16649,
      "grad_norm": 1.564214022867252,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 16649
    },
    {
      "epoch": 0.1665,
      "grad_norm": 1.2337311610277657,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 16650
    },
    {
      "epoch": 0.16651,
      "grad_norm": 1.3018102841398165,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 16651
    },
    {
      "epoch": 0.16652,
      "grad_norm": 1.4591675818303105,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 16652
    },
    {
      "epoch": 0.16653,
      "grad_norm": 1.3195737330554282,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 16653
    },
    {
      "epoch": 0.16654,
      "grad_norm": 1.3569877794562422,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 16654
    },
    {
      "epoch": 0.16655,
      "grad_norm": 1.1633207592710884,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 16655
    },
    {
      "epoch": 0.16656,
      "grad_norm": 1.2950788549807473,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 16656
    },
    {
      "epoch": 0.16657,
      "grad_norm": 1.2092542748592239,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 16657
    },
    {
      "epoch": 0.16658,
      "grad_norm": 1.0939329273704779,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 16658
    },
    {
      "epoch": 0.16659,
      "grad_norm": 1.123124855614399,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 16659
    },
    {
      "epoch": 0.1666,
      "grad_norm": 1.25646383028317,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 16660
    },
    {
      "epoch": 0.16661,
      "grad_norm": 1.2927711530505899,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 16661
    },
    {
      "epoch": 0.16662,
      "grad_norm": 1.5065359649910308,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 16662
    },
    {
      "epoch": 0.16663,
      "grad_norm": 1.2217810993901903,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 16663
    },
    {
      "epoch": 0.16664,
      "grad_norm": 1.2506747027717897,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 16664
    },
    {
      "epoch": 0.16665,
      "grad_norm": 1.3679567019358465,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 16665
    },
    {
      "epoch": 0.16666,
      "grad_norm": 1.0847656585217225,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 16666
    },
    {
      "epoch": 0.16667,
      "grad_norm": 1.2440115678312678,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 16667
    },
    {
      "epoch": 0.16668,
      "grad_norm": 1.2986418728887241,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 16668
    },
    {
      "epoch": 0.16669,
      "grad_norm": 1.3937775860243589,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 16669
    },
    {
      "epoch": 0.1667,
      "grad_norm": 1.1444490580546132,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 16670
    },
    {
      "epoch": 0.16671,
      "grad_norm": 1.2547154137614827,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 16671
    },
    {
      "epoch": 0.16672,
      "grad_norm": 1.4686589795806708,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 16672
    },
    {
      "epoch": 0.16673,
      "grad_norm": 1.2966678797714233,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 16673
    },
    {
      "epoch": 0.16674,
      "grad_norm": 1.4008422920220023,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 16674
    },
    {
      "epoch": 0.16675,
      "grad_norm": 1.4374214591850536,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 16675
    },
    {
      "epoch": 0.16676,
      "grad_norm": 1.333975506995556,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 16676
    },
    {
      "epoch": 0.16677,
      "grad_norm": 1.3659889263727105,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 16677
    },
    {
      "epoch": 0.16678,
      "grad_norm": 1.3232398475591127,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 16678
    },
    {
      "epoch": 0.16679,
      "grad_norm": 1.372056133834555,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 16679
    },
    {
      "epoch": 0.1668,
      "grad_norm": 1.115400796578888,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 16680
    },
    {
      "epoch": 0.16681,
      "grad_norm": 1.1462923783523356,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 16681
    },
    {
      "epoch": 0.16682,
      "grad_norm": 1.2822705786345197,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 16682
    },
    {
      "epoch": 0.16683,
      "grad_norm": 1.033317131360321,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 16683
    },
    {
      "epoch": 0.16684,
      "grad_norm": 1.5289135933913132,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 16684
    },
    {
      "epoch": 0.16685,
      "grad_norm": 1.3185212238437796,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 16685
    },
    {
      "epoch": 0.16686,
      "grad_norm": 1.2775429286296354,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 16686
    },
    {
      "epoch": 0.16687,
      "grad_norm": 1.310704859490125,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 16687
    },
    {
      "epoch": 0.16688,
      "grad_norm": 1.1067398773051462,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 16688
    },
    {
      "epoch": 0.16689,
      "grad_norm": 1.4483876947604948,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 16689
    },
    {
      "epoch": 0.1669,
      "grad_norm": 1.1567384752757102,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 16690
    },
    {
      "epoch": 0.16691,
      "grad_norm": 1.278726343385809,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 16691
    },
    {
      "epoch": 0.16692,
      "grad_norm": 1.0648070520419866,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 16692
    },
    {
      "epoch": 0.16693,
      "grad_norm": 1.4081478664976732,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 16693
    },
    {
      "epoch": 0.16694,
      "grad_norm": 1.1594991371369678,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16694
    },
    {
      "epoch": 0.16695,
      "grad_norm": 1.4636847314266446,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 16695
    },
    {
      "epoch": 0.16696,
      "grad_norm": 1.3951427322084577,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 16696
    },
    {
      "epoch": 0.16697,
      "grad_norm": 1.0347629487687562,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 16697
    },
    {
      "epoch": 0.16698,
      "grad_norm": 1.684232743457682,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 16698
    },
    {
      "epoch": 0.16699,
      "grad_norm": 0.8511245416361911,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 16699
    },
    {
      "epoch": 0.167,
      "grad_norm": 1.1386843043359771,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 16700
    },
    {
      "epoch": 0.16701,
      "grad_norm": 1.7512591591733795,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 16701
    },
    {
      "epoch": 0.16702,
      "grad_norm": 0.9962913652943098,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 16702
    },
    {
      "epoch": 0.16703,
      "grad_norm": 1.2565831934183758,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 16703
    },
    {
      "epoch": 0.16704,
      "grad_norm": 1.3433961964467531,
      "learning_rate": 0.003,
      "loss": 4.1071,
      "step": 16704
    },
    {
      "epoch": 0.16705,
      "grad_norm": 1.3910816757951838,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 16705
    },
    {
      "epoch": 0.16706,
      "grad_norm": 1.3316360416361006,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 16706
    },
    {
      "epoch": 0.16707,
      "grad_norm": 0.991523558879406,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 16707
    },
    {
      "epoch": 0.16708,
      "grad_norm": 1.4048042006132784,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 16708
    },
    {
      "epoch": 0.16709,
      "grad_norm": 1.3886487992408054,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 16709
    },
    {
      "epoch": 0.1671,
      "grad_norm": 1.3016910098704964,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 16710
    },
    {
      "epoch": 0.16711,
      "grad_norm": 1.3477782241628462,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 16711
    },
    {
      "epoch": 0.16712,
      "grad_norm": 1.134131580682221,
      "learning_rate": 0.003,
      "loss": 4.0227,
      "step": 16712
    },
    {
      "epoch": 0.16713,
      "grad_norm": 1.3070182485559603,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 16713
    },
    {
      "epoch": 0.16714,
      "grad_norm": 1.1352574486120421,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 16714
    },
    {
      "epoch": 0.16715,
      "grad_norm": 1.3861551298051138,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 16715
    },
    {
      "epoch": 0.16716,
      "grad_norm": 1.221713531815258,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 16716
    },
    {
      "epoch": 0.16717,
      "grad_norm": 1.126946267158167,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 16717
    },
    {
      "epoch": 0.16718,
      "grad_norm": 1.391457071408226,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 16718
    },
    {
      "epoch": 0.16719,
      "grad_norm": 1.259439961310894,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 16719
    },
    {
      "epoch": 0.1672,
      "grad_norm": 1.3993421115373907,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 16720
    },
    {
      "epoch": 0.16721,
      "grad_norm": 1.080223645130753,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 16721
    },
    {
      "epoch": 0.16722,
      "grad_norm": 1.5730549980475002,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 16722
    },
    {
      "epoch": 0.16723,
      "grad_norm": 1.1297822430243398,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 16723
    },
    {
      "epoch": 0.16724,
      "grad_norm": 1.186892184655547,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 16724
    },
    {
      "epoch": 0.16725,
      "grad_norm": 1.2311396580283709,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 16725
    },
    {
      "epoch": 0.16726,
      "grad_norm": 1.140182561961307,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 16726
    },
    {
      "epoch": 0.16727,
      "grad_norm": 1.32465115746718,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 16727
    },
    {
      "epoch": 0.16728,
      "grad_norm": 1.2796869537548115,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 16728
    },
    {
      "epoch": 0.16729,
      "grad_norm": 1.181847879728677,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 16729
    },
    {
      "epoch": 0.1673,
      "grad_norm": 1.3776713468609083,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 16730
    },
    {
      "epoch": 0.16731,
      "grad_norm": 1.2565759096060845,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 16731
    },
    {
      "epoch": 0.16732,
      "grad_norm": 1.2550901339996967,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 16732
    },
    {
      "epoch": 0.16733,
      "grad_norm": 1.1175831388571573,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 16733
    },
    {
      "epoch": 0.16734,
      "grad_norm": 1.3450730195508063,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 16734
    },
    {
      "epoch": 0.16735,
      "grad_norm": 1.116004009871605,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 16735
    },
    {
      "epoch": 0.16736,
      "grad_norm": 1.621623953487288,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 16736
    },
    {
      "epoch": 0.16737,
      "grad_norm": 1.2525261016436242,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 16737
    },
    {
      "epoch": 0.16738,
      "grad_norm": 1.4305063793483448,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 16738
    },
    {
      "epoch": 0.16739,
      "grad_norm": 1.662838940432477,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 16739
    },
    {
      "epoch": 0.1674,
      "grad_norm": 0.9891646601073584,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 16740
    },
    {
      "epoch": 0.16741,
      "grad_norm": 1.1384552528164928,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 16741
    },
    {
      "epoch": 0.16742,
      "grad_norm": 1.3547129028711147,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 16742
    },
    {
      "epoch": 0.16743,
      "grad_norm": 1.6243793382905138,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 16743
    },
    {
      "epoch": 0.16744,
      "grad_norm": 1.0975972967796066,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 16744
    },
    {
      "epoch": 0.16745,
      "grad_norm": 1.3074619617397136,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 16745
    },
    {
      "epoch": 0.16746,
      "grad_norm": 1.230182685581038,
      "learning_rate": 0.003,
      "loss": 4.1025,
      "step": 16746
    },
    {
      "epoch": 0.16747,
      "grad_norm": 1.1312420920964736,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 16747
    },
    {
      "epoch": 0.16748,
      "grad_norm": 1.2507249543126753,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 16748
    },
    {
      "epoch": 0.16749,
      "grad_norm": 1.1980369151237098,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 16749
    },
    {
      "epoch": 0.1675,
      "grad_norm": 1.4655643665371967,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 16750
    },
    {
      "epoch": 0.16751,
      "grad_norm": 1.286853091051945,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 16751
    },
    {
      "epoch": 0.16752,
      "grad_norm": 1.415363481460103,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 16752
    },
    {
      "epoch": 0.16753,
      "grad_norm": 1.377393120294197,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 16753
    },
    {
      "epoch": 0.16754,
      "grad_norm": 1.148828273784319,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 16754
    },
    {
      "epoch": 0.16755,
      "grad_norm": 1.2696627055678036,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 16755
    },
    {
      "epoch": 0.16756,
      "grad_norm": 1.3370050291485667,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 16756
    },
    {
      "epoch": 0.16757,
      "grad_norm": 1.1439509199832336,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 16757
    },
    {
      "epoch": 0.16758,
      "grad_norm": 1.2863755871018208,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 16758
    },
    {
      "epoch": 0.16759,
      "grad_norm": 1.0127227549338795,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 16759
    },
    {
      "epoch": 0.1676,
      "grad_norm": 1.9709684350556145,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 16760
    },
    {
      "epoch": 0.16761,
      "grad_norm": 0.8838581402290888,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 16761
    },
    {
      "epoch": 0.16762,
      "grad_norm": 1.327185229227163,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 16762
    },
    {
      "epoch": 0.16763,
      "grad_norm": 1.0901753046638558,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 16763
    },
    {
      "epoch": 0.16764,
      "grad_norm": 1.1942403829502464,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 16764
    },
    {
      "epoch": 0.16765,
      "grad_norm": 1.4232219300432742,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 16765
    },
    {
      "epoch": 0.16766,
      "grad_norm": 1.1262783920801236,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 16766
    },
    {
      "epoch": 0.16767,
      "grad_norm": 1.4223258265073562,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 16767
    },
    {
      "epoch": 0.16768,
      "grad_norm": 1.4295881431617428,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 16768
    },
    {
      "epoch": 0.16769,
      "grad_norm": 1.0963666305613717,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16769
    },
    {
      "epoch": 0.1677,
      "grad_norm": 1.1414442762819592,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16770
    },
    {
      "epoch": 0.16771,
      "grad_norm": 1.2238226630858162,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 16771
    },
    {
      "epoch": 0.16772,
      "grad_norm": 1.6783808507277331,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 16772
    },
    {
      "epoch": 0.16773,
      "grad_norm": 1.095242159359064,
      "learning_rate": 0.003,
      "loss": 4.0291,
      "step": 16773
    },
    {
      "epoch": 0.16774,
      "grad_norm": 1.332248501252304,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 16774
    },
    {
      "epoch": 0.16775,
      "grad_norm": 1.155182945765907,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 16775
    },
    {
      "epoch": 0.16776,
      "grad_norm": 1.3650135785233826,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 16776
    },
    {
      "epoch": 0.16777,
      "grad_norm": 1.4514998796554472,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 16777
    },
    {
      "epoch": 0.16778,
      "grad_norm": 1.287162740762738,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 16778
    },
    {
      "epoch": 0.16779,
      "grad_norm": 1.3125614663209144,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 16779
    },
    {
      "epoch": 0.1678,
      "grad_norm": 1.0247349687445304,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 16780
    },
    {
      "epoch": 0.16781,
      "grad_norm": 1.454803588060753,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 16781
    },
    {
      "epoch": 0.16782,
      "grad_norm": 1.1087921509488263,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 16782
    },
    {
      "epoch": 0.16783,
      "grad_norm": 1.3142906154006264,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 16783
    },
    {
      "epoch": 0.16784,
      "grad_norm": 1.1099279060836573,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 16784
    },
    {
      "epoch": 0.16785,
      "grad_norm": 1.220567366028567,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 16785
    },
    {
      "epoch": 0.16786,
      "grad_norm": 1.2919404227923874,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 16786
    },
    {
      "epoch": 0.16787,
      "grad_norm": 1.4363324714839323,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 16787
    },
    {
      "epoch": 0.16788,
      "grad_norm": 1.170231924201678,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 16788
    },
    {
      "epoch": 0.16789,
      "grad_norm": 1.393351981675191,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 16789
    },
    {
      "epoch": 0.1679,
      "grad_norm": 1.1336535552565001,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 16790
    },
    {
      "epoch": 0.16791,
      "grad_norm": 1.1474977162170739,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 16791
    },
    {
      "epoch": 0.16792,
      "grad_norm": 1.3920508668947844,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 16792
    },
    {
      "epoch": 0.16793,
      "grad_norm": 1.4227528877948599,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 16793
    },
    {
      "epoch": 0.16794,
      "grad_norm": 1.2060257722629177,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 16794
    },
    {
      "epoch": 0.16795,
      "grad_norm": 1.3600827230748966,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 16795
    },
    {
      "epoch": 0.16796,
      "grad_norm": 1.1770420210337125,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 16796
    },
    {
      "epoch": 0.16797,
      "grad_norm": 1.4393574934891733,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 16797
    },
    {
      "epoch": 0.16798,
      "grad_norm": 1.0901274871323479,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 16798
    },
    {
      "epoch": 0.16799,
      "grad_norm": 1.5252149919240858,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 16799
    },
    {
      "epoch": 0.168,
      "grad_norm": 1.14032647596702,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 16800
    },
    {
      "epoch": 0.16801,
      "grad_norm": 1.327648166260483,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 16801
    },
    {
      "epoch": 0.16802,
      "grad_norm": 1.1002849680966285,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 16802
    },
    {
      "epoch": 0.16803,
      "grad_norm": 1.5068799667060553,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 16803
    },
    {
      "epoch": 0.16804,
      "grad_norm": 1.3470880046221414,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 16804
    },
    {
      "epoch": 0.16805,
      "grad_norm": 1.2200143859080324,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 16805
    },
    {
      "epoch": 0.16806,
      "grad_norm": 1.307463612754022,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 16806
    },
    {
      "epoch": 0.16807,
      "grad_norm": 1.266542756627587,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 16807
    },
    {
      "epoch": 0.16808,
      "grad_norm": 1.123860411350716,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 16808
    },
    {
      "epoch": 0.16809,
      "grad_norm": 1.2527402200254805,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 16809
    },
    {
      "epoch": 0.1681,
      "grad_norm": 1.2286560649219227,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 16810
    },
    {
      "epoch": 0.16811,
      "grad_norm": 1.2794688476481573,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 16811
    },
    {
      "epoch": 0.16812,
      "grad_norm": 1.0624837677134609,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 16812
    },
    {
      "epoch": 0.16813,
      "grad_norm": 1.3450544802064075,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 16813
    },
    {
      "epoch": 0.16814,
      "grad_norm": 1.1588218630193072,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 16814
    },
    {
      "epoch": 0.16815,
      "grad_norm": 1.276754195656693,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 16815
    },
    {
      "epoch": 0.16816,
      "grad_norm": 1.2745892758250235,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 16816
    },
    {
      "epoch": 0.16817,
      "grad_norm": 1.3571051184806497,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 16817
    },
    {
      "epoch": 0.16818,
      "grad_norm": 1.267336203960523,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 16818
    },
    {
      "epoch": 0.16819,
      "grad_norm": 1.1924544666810764,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 16819
    },
    {
      "epoch": 0.1682,
      "grad_norm": 1.207256301555306,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 16820
    },
    {
      "epoch": 0.16821,
      "grad_norm": 1.370854375794414,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 16821
    },
    {
      "epoch": 0.16822,
      "grad_norm": 1.3663088529821639,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 16822
    },
    {
      "epoch": 0.16823,
      "grad_norm": 1.1774873108786654,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 16823
    },
    {
      "epoch": 0.16824,
      "grad_norm": 1.4194242487443058,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 16824
    },
    {
      "epoch": 0.16825,
      "grad_norm": 1.3822584662925912,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 16825
    },
    {
      "epoch": 0.16826,
      "grad_norm": 1.1102142883788981,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 16826
    },
    {
      "epoch": 0.16827,
      "grad_norm": 1.38003821159165,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 16827
    },
    {
      "epoch": 0.16828,
      "grad_norm": 1.1851290243500439,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 16828
    },
    {
      "epoch": 0.16829,
      "grad_norm": 1.6430352870086788,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 16829
    },
    {
      "epoch": 0.1683,
      "grad_norm": 1.0075497669286884,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 16830
    },
    {
      "epoch": 0.16831,
      "grad_norm": 1.5054189554987507,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 16831
    },
    {
      "epoch": 0.16832,
      "grad_norm": 1.3496529029391273,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 16832
    },
    {
      "epoch": 0.16833,
      "grad_norm": 1.4899614656492906,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 16833
    },
    {
      "epoch": 0.16834,
      "grad_norm": 1.2239859880131512,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 16834
    },
    {
      "epoch": 0.16835,
      "grad_norm": 1.0447703047011303,
      "learning_rate": 0.003,
      "loss": 4.0135,
      "step": 16835
    },
    {
      "epoch": 0.16836,
      "grad_norm": 1.2672507917954132,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 16836
    },
    {
      "epoch": 0.16837,
      "grad_norm": 1.2027880989996271,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 16837
    },
    {
      "epoch": 0.16838,
      "grad_norm": 1.4245659086392621,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 16838
    },
    {
      "epoch": 0.16839,
      "grad_norm": 1.2037598344828246,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 16839
    },
    {
      "epoch": 0.1684,
      "grad_norm": 1.0951149498838402,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 16840
    },
    {
      "epoch": 0.16841,
      "grad_norm": 1.4364129329916702,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 16841
    },
    {
      "epoch": 0.16842,
      "grad_norm": 1.068806632324706,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 16842
    },
    {
      "epoch": 0.16843,
      "grad_norm": 1.4819403695542883,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 16843
    },
    {
      "epoch": 0.16844,
      "grad_norm": 1.1490258055091571,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 16844
    },
    {
      "epoch": 0.16845,
      "grad_norm": 1.3541141006956319,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 16845
    },
    {
      "epoch": 0.16846,
      "grad_norm": 1.1206996339264277,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 16846
    },
    {
      "epoch": 0.16847,
      "grad_norm": 1.5244419982059054,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 16847
    },
    {
      "epoch": 0.16848,
      "grad_norm": 1.09708306061788,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 16848
    },
    {
      "epoch": 0.16849,
      "grad_norm": 1.470418708707691,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 16849
    },
    {
      "epoch": 0.1685,
      "grad_norm": 1.2319714683828253,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 16850
    },
    {
      "epoch": 0.16851,
      "grad_norm": 1.4095576657043976,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 16851
    },
    {
      "epoch": 0.16852,
      "grad_norm": 1.1033045827696277,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 16852
    },
    {
      "epoch": 0.16853,
      "grad_norm": 1.3847096334809648,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 16853
    },
    {
      "epoch": 0.16854,
      "grad_norm": 1.1113328080159366,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 16854
    },
    {
      "epoch": 0.16855,
      "grad_norm": 1.224386372223287,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 16855
    },
    {
      "epoch": 0.16856,
      "grad_norm": 1.1916273075115396,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 16856
    },
    {
      "epoch": 0.16857,
      "grad_norm": 1.1481876709715975,
      "learning_rate": 0.003,
      "loss": 4.0221,
      "step": 16857
    },
    {
      "epoch": 0.16858,
      "grad_norm": 1.3593802424987913,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 16858
    },
    {
      "epoch": 0.16859,
      "grad_norm": 1.273283610859603,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 16859
    },
    {
      "epoch": 0.1686,
      "grad_norm": 1.496853202283384,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 16860
    },
    {
      "epoch": 0.16861,
      "grad_norm": 1.40736923510857,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 16861
    },
    {
      "epoch": 0.16862,
      "grad_norm": 1.490879432399349,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 16862
    },
    {
      "epoch": 0.16863,
      "grad_norm": 1.31948245652066,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 16863
    },
    {
      "epoch": 0.16864,
      "grad_norm": 1.301382904637517,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 16864
    },
    {
      "epoch": 0.16865,
      "grad_norm": 1.5323866546447675,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 16865
    },
    {
      "epoch": 0.16866,
      "grad_norm": 1.1631458714171514,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 16866
    },
    {
      "epoch": 0.16867,
      "grad_norm": 1.158729013732875,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 16867
    },
    {
      "epoch": 0.16868,
      "grad_norm": 1.1228514993443777,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 16868
    },
    {
      "epoch": 0.16869,
      "grad_norm": 1.393778205896287,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 16869
    },
    {
      "epoch": 0.1687,
      "grad_norm": 1.3243045374960487,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 16870
    },
    {
      "epoch": 0.16871,
      "grad_norm": 1.2010567934501997,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 16871
    },
    {
      "epoch": 0.16872,
      "grad_norm": 1.440514056954753,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 16872
    },
    {
      "epoch": 0.16873,
      "grad_norm": 1.2139601121876216,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 16873
    },
    {
      "epoch": 0.16874,
      "grad_norm": 1.4099829826682746,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 16874
    },
    {
      "epoch": 0.16875,
      "grad_norm": 1.146165363835504,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 16875
    },
    {
      "epoch": 0.16876,
      "grad_norm": 1.241179998010361,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 16876
    },
    {
      "epoch": 0.16877,
      "grad_norm": 1.1944182913192236,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 16877
    },
    {
      "epoch": 0.16878,
      "grad_norm": 1.2242141822087804,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 16878
    },
    {
      "epoch": 0.16879,
      "grad_norm": 1.115057333475243,
      "learning_rate": 0.003,
      "loss": 4.0268,
      "step": 16879
    },
    {
      "epoch": 0.1688,
      "grad_norm": 1.270200199752863,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 16880
    },
    {
      "epoch": 0.16881,
      "grad_norm": 1.2019256836950207,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 16881
    },
    {
      "epoch": 0.16882,
      "grad_norm": 1.4479286266212343,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 16882
    },
    {
      "epoch": 0.16883,
      "grad_norm": 1.062123034185907,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 16883
    },
    {
      "epoch": 0.16884,
      "grad_norm": 1.3882209746942173,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 16884
    },
    {
      "epoch": 0.16885,
      "grad_norm": 1.109180520492913,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 16885
    },
    {
      "epoch": 0.16886,
      "grad_norm": 1.2836788033972242,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 16886
    },
    {
      "epoch": 0.16887,
      "grad_norm": 1.2521604536286681,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 16887
    },
    {
      "epoch": 0.16888,
      "grad_norm": 1.4136543025600252,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 16888
    },
    {
      "epoch": 0.16889,
      "grad_norm": 1.5782665241111962,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 16889
    },
    {
      "epoch": 0.1689,
      "grad_norm": 1.1959929754457943,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 16890
    },
    {
      "epoch": 0.16891,
      "grad_norm": 1.2352924174891171,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 16891
    },
    {
      "epoch": 0.16892,
      "grad_norm": 1.58818430273992,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 16892
    },
    {
      "epoch": 0.16893,
      "grad_norm": 1.1601970694296755,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 16893
    },
    {
      "epoch": 0.16894,
      "grad_norm": 1.4143245558241038,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 16894
    },
    {
      "epoch": 0.16895,
      "grad_norm": 1.1094466555637557,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 16895
    },
    {
      "epoch": 0.16896,
      "grad_norm": 1.2896671978581782,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 16896
    },
    {
      "epoch": 0.16897,
      "grad_norm": 1.2833778034226433,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 16897
    },
    {
      "epoch": 0.16898,
      "grad_norm": 1.2460864940528413,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 16898
    },
    {
      "epoch": 0.16899,
      "grad_norm": 1.2819890523227007,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 16899
    },
    {
      "epoch": 0.169,
      "grad_norm": 1.2147009440965972,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 16900
    },
    {
      "epoch": 0.16901,
      "grad_norm": 1.544973459892542,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 16901
    },
    {
      "epoch": 0.16902,
      "grad_norm": 1.3220660850852575,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 16902
    },
    {
      "epoch": 0.16903,
      "grad_norm": 1.399095635772759,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 16903
    },
    {
      "epoch": 0.16904,
      "grad_norm": 1.1102399477345783,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 16904
    },
    {
      "epoch": 0.16905,
      "grad_norm": 1.4593558716593782,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 16905
    },
    {
      "epoch": 0.16906,
      "grad_norm": 1.3101786260146533,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 16906
    },
    {
      "epoch": 0.16907,
      "grad_norm": 1.0702024985287735,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 16907
    },
    {
      "epoch": 0.16908,
      "grad_norm": 1.3887646224909347,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 16908
    },
    {
      "epoch": 0.16909,
      "grad_norm": 1.1828192580708548,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 16909
    },
    {
      "epoch": 0.1691,
      "grad_norm": 1.3176191915351163,
      "learning_rate": 0.003,
      "loss": 4.0192,
      "step": 16910
    },
    {
      "epoch": 0.16911,
      "grad_norm": 1.4570497416853039,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 16911
    },
    {
      "epoch": 0.16912,
      "grad_norm": 1.347768060910466,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 16912
    },
    {
      "epoch": 0.16913,
      "grad_norm": 1.4406348710462704,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 16913
    },
    {
      "epoch": 0.16914,
      "grad_norm": 1.1302934221198233,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 16914
    },
    {
      "epoch": 0.16915,
      "grad_norm": 1.2401024203800077,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 16915
    },
    {
      "epoch": 0.16916,
      "grad_norm": 1.2956588105084923,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 16916
    },
    {
      "epoch": 0.16917,
      "grad_norm": 1.24168297084552,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 16917
    },
    {
      "epoch": 0.16918,
      "grad_norm": 1.2961998472371414,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 16918
    },
    {
      "epoch": 0.16919,
      "grad_norm": 1.2452220446431184,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 16919
    },
    {
      "epoch": 0.1692,
      "grad_norm": 1.2714969308512136,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 16920
    },
    {
      "epoch": 0.16921,
      "grad_norm": 1.0320651012042785,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 16921
    },
    {
      "epoch": 0.16922,
      "grad_norm": 1.3198445938865049,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 16922
    },
    {
      "epoch": 0.16923,
      "grad_norm": 1.1314290964153324,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 16923
    },
    {
      "epoch": 0.16924,
      "grad_norm": 1.4219952718125404,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 16924
    },
    {
      "epoch": 0.16925,
      "grad_norm": 1.1234743637945024,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 16925
    },
    {
      "epoch": 0.16926,
      "grad_norm": 1.4123328622018225,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 16926
    },
    {
      "epoch": 0.16927,
      "grad_norm": 1.4473500817037783,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 16927
    },
    {
      "epoch": 0.16928,
      "grad_norm": 1.2799292837639202,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 16928
    },
    {
      "epoch": 0.16929,
      "grad_norm": 1.4288548030602994,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 16929
    },
    {
      "epoch": 0.1693,
      "grad_norm": 1.3195056052785827,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 16930
    },
    {
      "epoch": 0.16931,
      "grad_norm": 1.1625917507435435,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 16931
    },
    {
      "epoch": 0.16932,
      "grad_norm": 1.4378633883226684,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 16932
    },
    {
      "epoch": 0.16933,
      "grad_norm": 1.1256401510921825,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 16933
    },
    {
      "epoch": 0.16934,
      "grad_norm": 1.3449069483606546,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 16934
    },
    {
      "epoch": 0.16935,
      "grad_norm": 1.1393050219429626,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 16935
    },
    {
      "epoch": 0.16936,
      "grad_norm": 1.2662484153953546,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 16936
    },
    {
      "epoch": 0.16937,
      "grad_norm": 1.2963480614111844,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 16937
    },
    {
      "epoch": 0.16938,
      "grad_norm": 1.1074316734336083,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 16938
    },
    {
      "epoch": 0.16939,
      "grad_norm": 1.3065192526603162,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 16939
    },
    {
      "epoch": 0.1694,
      "grad_norm": 1.225136446838548,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 16940
    },
    {
      "epoch": 0.16941,
      "grad_norm": 1.5104008620908072,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 16941
    },
    {
      "epoch": 0.16942,
      "grad_norm": 1.1440377001531545,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 16942
    },
    {
      "epoch": 0.16943,
      "grad_norm": 1.3343754663051575,
      "learning_rate": 0.003,
      "loss": 4.0253,
      "step": 16943
    },
    {
      "epoch": 0.16944,
      "grad_norm": 1.278040988184129,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 16944
    },
    {
      "epoch": 0.16945,
      "grad_norm": 1.3936716350946827,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 16945
    },
    {
      "epoch": 0.16946,
      "grad_norm": 0.9878081793633148,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 16946
    },
    {
      "epoch": 0.16947,
      "grad_norm": 1.4320519568494658,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 16947
    },
    {
      "epoch": 0.16948,
      "grad_norm": 1.244286346836353,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 16948
    },
    {
      "epoch": 0.16949,
      "grad_norm": 1.2299209004693132,
      "learning_rate": 0.003,
      "loss": 4.0232,
      "step": 16949
    },
    {
      "epoch": 0.1695,
      "grad_norm": 1.0409565052429766,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 16950
    },
    {
      "epoch": 0.16951,
      "grad_norm": 1.2508836847989877,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 16951
    },
    {
      "epoch": 0.16952,
      "grad_norm": 1.2967029164784285,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 16952
    },
    {
      "epoch": 0.16953,
      "grad_norm": 1.4077178182440349,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 16953
    },
    {
      "epoch": 0.16954,
      "grad_norm": 1.0826038727213632,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 16954
    },
    {
      "epoch": 0.16955,
      "grad_norm": 1.2115900177002683,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 16955
    },
    {
      "epoch": 0.16956,
      "grad_norm": 1.4481540542492375,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 16956
    },
    {
      "epoch": 0.16957,
      "grad_norm": 1.54688189499958,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 16957
    },
    {
      "epoch": 0.16958,
      "grad_norm": 1.3514547824099727,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 16958
    },
    {
      "epoch": 0.16959,
      "grad_norm": 1.3902847878239193,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 16959
    },
    {
      "epoch": 0.1696,
      "grad_norm": 1.0485211572345803,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 16960
    },
    {
      "epoch": 0.16961,
      "grad_norm": 1.4504868829279804,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 16961
    },
    {
      "epoch": 0.16962,
      "grad_norm": 1.2840105861582294,
      "learning_rate": 0.003,
      "loss": 4.0936,
      "step": 16962
    },
    {
      "epoch": 0.16963,
      "grad_norm": 1.2296728562735413,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 16963
    },
    {
      "epoch": 0.16964,
      "grad_norm": 1.1315041822083285,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 16964
    },
    {
      "epoch": 0.16965,
      "grad_norm": 1.3289468427844238,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 16965
    },
    {
      "epoch": 0.16966,
      "grad_norm": 1.0464856718548334,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 16966
    },
    {
      "epoch": 0.16967,
      "grad_norm": 1.3937570112756352,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 16967
    },
    {
      "epoch": 0.16968,
      "grad_norm": 1.3874378247783272,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 16968
    },
    {
      "epoch": 0.16969,
      "grad_norm": 1.1011033569991182,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 16969
    },
    {
      "epoch": 0.1697,
      "grad_norm": 1.5819831429822668,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 16970
    },
    {
      "epoch": 0.16971,
      "grad_norm": 0.9604786741124874,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 16971
    },
    {
      "epoch": 0.16972,
      "grad_norm": 1.4722719123412784,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 16972
    },
    {
      "epoch": 0.16973,
      "grad_norm": 1.0411160539916922,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 16973
    },
    {
      "epoch": 0.16974,
      "grad_norm": 1.3526490148604315,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 16974
    },
    {
      "epoch": 0.16975,
      "grad_norm": 1.2802469104047067,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 16975
    },
    {
      "epoch": 0.16976,
      "grad_norm": 1.2589720717564747,
      "learning_rate": 0.003,
      "loss": 4.0238,
      "step": 16976
    },
    {
      "epoch": 0.16977,
      "grad_norm": 1.2634228865755068,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 16977
    },
    {
      "epoch": 0.16978,
      "grad_norm": 1.1191953796902516,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 16978
    },
    {
      "epoch": 0.16979,
      "grad_norm": 1.2790069978179737,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 16979
    },
    {
      "epoch": 0.1698,
      "grad_norm": 1.5249885874241196,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 16980
    },
    {
      "epoch": 0.16981,
      "grad_norm": 1.1204813867109078,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 16981
    },
    {
      "epoch": 0.16982,
      "grad_norm": 1.6120602744889545,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 16982
    },
    {
      "epoch": 0.16983,
      "grad_norm": 1.2142625856634415,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 16983
    },
    {
      "epoch": 0.16984,
      "grad_norm": 1.5650189553425617,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 16984
    },
    {
      "epoch": 0.16985,
      "grad_norm": 1.2220873866641324,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 16985
    },
    {
      "epoch": 0.16986,
      "grad_norm": 1.251526545562485,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 16986
    },
    {
      "epoch": 0.16987,
      "grad_norm": 1.2099160891817955,
      "learning_rate": 0.003,
      "loss": 4.1026,
      "step": 16987
    },
    {
      "epoch": 0.16988,
      "grad_norm": 1.1561308162598547,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 16988
    },
    {
      "epoch": 0.16989,
      "grad_norm": 1.3716886639666412,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 16989
    },
    {
      "epoch": 0.1699,
      "grad_norm": 1.2276839460561324,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 16990
    },
    {
      "epoch": 0.16991,
      "grad_norm": 1.4220253366428155,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 16991
    },
    {
      "epoch": 0.16992,
      "grad_norm": 1.4182198336528233,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 16992
    },
    {
      "epoch": 0.16993,
      "grad_norm": 1.3889111493859785,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 16993
    },
    {
      "epoch": 0.16994,
      "grad_norm": 1.2130238797491997,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 16994
    },
    {
      "epoch": 0.16995,
      "grad_norm": 1.1780230433886487,
      "learning_rate": 0.003,
      "loss": 4.0296,
      "step": 16995
    },
    {
      "epoch": 0.16996,
      "grad_norm": 1.3603919110346885,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 16996
    },
    {
      "epoch": 0.16997,
      "grad_norm": 1.2760748429890882,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 16997
    },
    {
      "epoch": 0.16998,
      "grad_norm": 1.2502815178778632,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 16998
    },
    {
      "epoch": 0.16999,
      "grad_norm": 1.6165519910026174,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 16999
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0053911199760457,
      "learning_rate": 0.003,
      "loss": 4.0222,
      "step": 17000
    },
    {
      "epoch": 0.17001,
      "grad_norm": 1.2888523071738722,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 17001
    },
    {
      "epoch": 0.17002,
      "grad_norm": 1.0247128604707927,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 17002
    },
    {
      "epoch": 0.17003,
      "grad_norm": 1.4999931056125426,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 17003
    },
    {
      "epoch": 0.17004,
      "grad_norm": 1.3596282609549202,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 17004
    },
    {
      "epoch": 0.17005,
      "grad_norm": 1.1514114776069913,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 17005
    },
    {
      "epoch": 0.17006,
      "grad_norm": 1.3312429685018117,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 17006
    },
    {
      "epoch": 0.17007,
      "grad_norm": 1.218792080089117,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 17007
    },
    {
      "epoch": 0.17008,
      "grad_norm": 1.197802950001522,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 17008
    },
    {
      "epoch": 0.17009,
      "grad_norm": 1.3763630217530811,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 17009
    },
    {
      "epoch": 0.1701,
      "grad_norm": 1.4212308500084787,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 17010
    },
    {
      "epoch": 0.17011,
      "grad_norm": 0.953128092845532,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 17011
    },
    {
      "epoch": 0.17012,
      "grad_norm": 1.324057361967305,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 17012
    },
    {
      "epoch": 0.17013,
      "grad_norm": 1.4539562806509976,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 17013
    },
    {
      "epoch": 0.17014,
      "grad_norm": 1.3728094854799713,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 17014
    },
    {
      "epoch": 0.17015,
      "grad_norm": 1.7138404683265989,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 17015
    },
    {
      "epoch": 0.17016,
      "grad_norm": 1.0001556175972248,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 17016
    },
    {
      "epoch": 0.17017,
      "grad_norm": 1.3144709201023723,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 17017
    },
    {
      "epoch": 0.17018,
      "grad_norm": 1.183298451331619,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 17018
    },
    {
      "epoch": 0.17019,
      "grad_norm": 1.2976131957707802,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17019
    },
    {
      "epoch": 0.1702,
      "grad_norm": 1.4441694521919246,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 17020
    },
    {
      "epoch": 0.17021,
      "grad_norm": 1.0231368351229009,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 17021
    },
    {
      "epoch": 0.17022,
      "grad_norm": 1.4522769663622839,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 17022
    },
    {
      "epoch": 0.17023,
      "grad_norm": 1.2115552248199124,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 17023
    },
    {
      "epoch": 0.17024,
      "grad_norm": 1.3955549248970327,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 17024
    },
    {
      "epoch": 0.17025,
      "grad_norm": 1.767413058772527,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 17025
    },
    {
      "epoch": 0.17026,
      "grad_norm": 1.1257427274255005,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 17026
    },
    {
      "epoch": 0.17027,
      "grad_norm": 1.3977854342088645,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 17027
    },
    {
      "epoch": 0.17028,
      "grad_norm": 1.4133438305646633,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 17028
    },
    {
      "epoch": 0.17029,
      "grad_norm": 0.9857828699963934,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 17029
    },
    {
      "epoch": 0.1703,
      "grad_norm": 1.2909867094838485,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 17030
    },
    {
      "epoch": 0.17031,
      "grad_norm": 1.2136917950238058,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 17031
    },
    {
      "epoch": 0.17032,
      "grad_norm": 1.3308725833491453,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 17032
    },
    {
      "epoch": 0.17033,
      "grad_norm": 1.1916228887328304,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 17033
    },
    {
      "epoch": 0.17034,
      "grad_norm": 1.3075246141432557,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 17034
    },
    {
      "epoch": 0.17035,
      "grad_norm": 1.2721657649715863,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 17035
    },
    {
      "epoch": 0.17036,
      "grad_norm": 1.2104509544352788,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 17036
    },
    {
      "epoch": 0.17037,
      "grad_norm": 1.1525415633438325,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 17037
    },
    {
      "epoch": 0.17038,
      "grad_norm": 1.1475974463818974,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 17038
    },
    {
      "epoch": 0.17039,
      "grad_norm": 1.5018976469381597,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 17039
    },
    {
      "epoch": 0.1704,
      "grad_norm": 1.0270191604314975,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 17040
    },
    {
      "epoch": 0.17041,
      "grad_norm": 1.3971266040112438,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 17041
    },
    {
      "epoch": 0.17042,
      "grad_norm": 1.3538705145785344,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 17042
    },
    {
      "epoch": 0.17043,
      "grad_norm": 1.2683846167389807,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 17043
    },
    {
      "epoch": 0.17044,
      "grad_norm": 1.4541193508004426,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 17044
    },
    {
      "epoch": 0.17045,
      "grad_norm": 1.0988360012839928,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 17045
    },
    {
      "epoch": 0.17046,
      "grad_norm": 1.6000719523297462,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 17046
    },
    {
      "epoch": 0.17047,
      "grad_norm": 1.2817610928722536,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 17047
    },
    {
      "epoch": 0.17048,
      "grad_norm": 1.2292969932309563,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 17048
    },
    {
      "epoch": 0.17049,
      "grad_norm": 1.2630107557417314,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 17049
    },
    {
      "epoch": 0.1705,
      "grad_norm": 1.276507448279547,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 17050
    },
    {
      "epoch": 0.17051,
      "grad_norm": 1.3492516633248473,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 17051
    },
    {
      "epoch": 0.17052,
      "grad_norm": 1.1940749532981079,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 17052
    },
    {
      "epoch": 0.17053,
      "grad_norm": 1.573189456458545,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 17053
    },
    {
      "epoch": 0.17054,
      "grad_norm": 1.210068121530533,
      "learning_rate": 0.003,
      "loss": 4.1033,
      "step": 17054
    },
    {
      "epoch": 0.17055,
      "grad_norm": 1.169180023801874,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 17055
    },
    {
      "epoch": 0.17056,
      "grad_norm": 1.2195067811040499,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 17056
    },
    {
      "epoch": 0.17057,
      "grad_norm": 1.360570643958307,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 17057
    },
    {
      "epoch": 0.17058,
      "grad_norm": 1.0253700488945972,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 17058
    },
    {
      "epoch": 0.17059,
      "grad_norm": 1.5336780724536332,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 17059
    },
    {
      "epoch": 0.1706,
      "grad_norm": 1.178011295450224,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 17060
    },
    {
      "epoch": 0.17061,
      "grad_norm": 1.3498779313088982,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 17061
    },
    {
      "epoch": 0.17062,
      "grad_norm": 1.1502540681934281,
      "learning_rate": 0.003,
      "loss": 4.0262,
      "step": 17062
    },
    {
      "epoch": 0.17063,
      "grad_norm": 1.384156634041985,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 17063
    },
    {
      "epoch": 0.17064,
      "grad_norm": 1.2550908972261812,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 17064
    },
    {
      "epoch": 0.17065,
      "grad_norm": 1.3201841544989974,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 17065
    },
    {
      "epoch": 0.17066,
      "grad_norm": 1.3370917019906838,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 17066
    },
    {
      "epoch": 0.17067,
      "grad_norm": 1.2869269678962787,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 17067
    },
    {
      "epoch": 0.17068,
      "grad_norm": 1.219233000989476,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 17068
    },
    {
      "epoch": 0.17069,
      "grad_norm": 1.4603560265924007,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 17069
    },
    {
      "epoch": 0.1707,
      "grad_norm": 1.313142605798217,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 17070
    },
    {
      "epoch": 0.17071,
      "grad_norm": 1.354739867548382,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 17071
    },
    {
      "epoch": 0.17072,
      "grad_norm": 1.322882693187208,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 17072
    },
    {
      "epoch": 0.17073,
      "grad_norm": 1.324393461801636,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 17073
    },
    {
      "epoch": 0.17074,
      "grad_norm": 1.1466469394594943,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 17074
    },
    {
      "epoch": 0.17075,
      "grad_norm": 1.184709216749383,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 17075
    },
    {
      "epoch": 0.17076,
      "grad_norm": 1.380052308242218,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 17076
    },
    {
      "epoch": 0.17077,
      "grad_norm": 1.273079251798578,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 17077
    },
    {
      "epoch": 0.17078,
      "grad_norm": 1.3272735724648692,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 17078
    },
    {
      "epoch": 0.17079,
      "grad_norm": 1.0803079847919785,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17079
    },
    {
      "epoch": 0.1708,
      "grad_norm": 1.1239999521431894,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 17080
    },
    {
      "epoch": 0.17081,
      "grad_norm": 1.430104041450772,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 17081
    },
    {
      "epoch": 0.17082,
      "grad_norm": 1.4949733715992284,
      "learning_rate": 0.003,
      "loss": 4.1044,
      "step": 17082
    },
    {
      "epoch": 0.17083,
      "grad_norm": 1.2917046206070553,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 17083
    },
    {
      "epoch": 0.17084,
      "grad_norm": 1.2323312232063808,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 17084
    },
    {
      "epoch": 0.17085,
      "grad_norm": 1.4111662613496083,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 17085
    },
    {
      "epoch": 0.17086,
      "grad_norm": 1.3477746976894227,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17086
    },
    {
      "epoch": 0.17087,
      "grad_norm": 1.233966993941767,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 17087
    },
    {
      "epoch": 0.17088,
      "grad_norm": 1.3116288329107746,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 17088
    },
    {
      "epoch": 0.17089,
      "grad_norm": 1.273728819976221,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 17089
    },
    {
      "epoch": 0.1709,
      "grad_norm": 1.505682932680468,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 17090
    },
    {
      "epoch": 0.17091,
      "grad_norm": 1.337474022173787,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 17091
    },
    {
      "epoch": 0.17092,
      "grad_norm": 1.4887830578497672,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 17092
    },
    {
      "epoch": 0.17093,
      "grad_norm": 1.315462724256315,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 17093
    },
    {
      "epoch": 0.17094,
      "grad_norm": 1.2402138789365362,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 17094
    },
    {
      "epoch": 0.17095,
      "grad_norm": 1.2766321856539715,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 17095
    },
    {
      "epoch": 0.17096,
      "grad_norm": 1.3477141194221316,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 17096
    },
    {
      "epoch": 0.17097,
      "grad_norm": 1.0294500312951556,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 17097
    },
    {
      "epoch": 0.17098,
      "grad_norm": 1.2372975873508332,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 17098
    },
    {
      "epoch": 0.17099,
      "grad_norm": 1.2999064965554796,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 17099
    },
    {
      "epoch": 0.171,
      "grad_norm": 1.2421551693192254,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 17100
    },
    {
      "epoch": 0.17101,
      "grad_norm": 1.3029925757656373,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 17101
    },
    {
      "epoch": 0.17102,
      "grad_norm": 1.413628790406156,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17102
    },
    {
      "epoch": 0.17103,
      "grad_norm": 1.270813481606152,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17103
    },
    {
      "epoch": 0.17104,
      "grad_norm": 1.4491769084696768,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 17104
    },
    {
      "epoch": 0.17105,
      "grad_norm": 1.2999105264354298,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 17105
    },
    {
      "epoch": 0.17106,
      "grad_norm": 1.493524806336209,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 17106
    },
    {
      "epoch": 0.17107,
      "grad_norm": 1.1253355514126384,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 17107
    },
    {
      "epoch": 0.17108,
      "grad_norm": 1.211621058987755,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 17108
    },
    {
      "epoch": 0.17109,
      "grad_norm": 1.2953679514880798,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 17109
    },
    {
      "epoch": 0.1711,
      "grad_norm": 1.2760926805103348,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 17110
    },
    {
      "epoch": 0.17111,
      "grad_norm": 1.2362842209560756,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 17111
    },
    {
      "epoch": 0.17112,
      "grad_norm": 1.2360788813061327,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 17112
    },
    {
      "epoch": 0.17113,
      "grad_norm": 1.339630283544096,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 17113
    },
    {
      "epoch": 0.17114,
      "grad_norm": 1.2114235919349148,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 17114
    },
    {
      "epoch": 0.17115,
      "grad_norm": 1.4198339218631812,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 17115
    },
    {
      "epoch": 0.17116,
      "grad_norm": 1.1717499098535376,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 17116
    },
    {
      "epoch": 0.17117,
      "grad_norm": 1.2909379254052165,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 17117
    },
    {
      "epoch": 0.17118,
      "grad_norm": 1.283910040750877,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 17118
    },
    {
      "epoch": 0.17119,
      "grad_norm": 1.1893986812955086,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 17119
    },
    {
      "epoch": 0.1712,
      "grad_norm": 1.265651088184533,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 17120
    },
    {
      "epoch": 0.17121,
      "grad_norm": 1.071795774668601,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 17121
    },
    {
      "epoch": 0.17122,
      "grad_norm": 1.648050976357266,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 17122
    },
    {
      "epoch": 0.17123,
      "grad_norm": 0.9528963605296452,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 17123
    },
    {
      "epoch": 0.17124,
      "grad_norm": 1.5673856382501787,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 17124
    },
    {
      "epoch": 0.17125,
      "grad_norm": 0.9182606986835793,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 17125
    },
    {
      "epoch": 0.17126,
      "grad_norm": 1.5321951332412338,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 17126
    },
    {
      "epoch": 0.17127,
      "grad_norm": 1.1578440872666076,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 17127
    },
    {
      "epoch": 0.17128,
      "grad_norm": 1.221124572752152,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 17128
    },
    {
      "epoch": 0.17129,
      "grad_norm": 1.3130787941979272,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 17129
    },
    {
      "epoch": 0.1713,
      "grad_norm": 1.1637149225438805,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 17130
    },
    {
      "epoch": 0.17131,
      "grad_norm": 1.2916207885201532,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 17131
    },
    {
      "epoch": 0.17132,
      "grad_norm": 1.4271216764913037,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 17132
    },
    {
      "epoch": 0.17133,
      "grad_norm": 1.3067500766517877,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 17133
    },
    {
      "epoch": 0.17134,
      "grad_norm": 1.32243194743986,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 17134
    },
    {
      "epoch": 0.17135,
      "grad_norm": 1.484578398902238,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 17135
    },
    {
      "epoch": 0.17136,
      "grad_norm": 1.2726072690525767,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 17136
    },
    {
      "epoch": 0.17137,
      "grad_norm": 1.393464284738465,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 17137
    },
    {
      "epoch": 0.17138,
      "grad_norm": 1.282219676916128,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 17138
    },
    {
      "epoch": 0.17139,
      "grad_norm": 1.4666292925377877,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 17139
    },
    {
      "epoch": 0.1714,
      "grad_norm": 1.0967948384984816,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 17140
    },
    {
      "epoch": 0.17141,
      "grad_norm": 1.4127740532449033,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 17141
    },
    {
      "epoch": 0.17142,
      "grad_norm": 1.1499694824347622,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 17142
    },
    {
      "epoch": 0.17143,
      "grad_norm": 1.250961346843233,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 17143
    },
    {
      "epoch": 0.17144,
      "grad_norm": 1.4443254164579862,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 17144
    },
    {
      "epoch": 0.17145,
      "grad_norm": 1.133039401868114,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 17145
    },
    {
      "epoch": 0.17146,
      "grad_norm": 1.5210457604440224,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 17146
    },
    {
      "epoch": 0.17147,
      "grad_norm": 1.3175377150941554,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 17147
    },
    {
      "epoch": 0.17148,
      "grad_norm": 1.2857199420481968,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 17148
    },
    {
      "epoch": 0.17149,
      "grad_norm": 1.2421283281105395,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 17149
    },
    {
      "epoch": 0.1715,
      "grad_norm": 1.0859540141491995,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 17150
    },
    {
      "epoch": 0.17151,
      "grad_norm": 1.3327157766348299,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 17151
    },
    {
      "epoch": 0.17152,
      "grad_norm": 1.2225879267594058,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 17152
    },
    {
      "epoch": 0.17153,
      "grad_norm": 1.4078252435542349,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 17153
    },
    {
      "epoch": 0.17154,
      "grad_norm": 1.2346220383369924,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 17154
    },
    {
      "epoch": 0.17155,
      "grad_norm": 1.3211567280413448,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 17155
    },
    {
      "epoch": 0.17156,
      "grad_norm": 1.6188785958146443,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 17156
    },
    {
      "epoch": 0.17157,
      "grad_norm": 1.1252843117347187,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 17157
    },
    {
      "epoch": 0.17158,
      "grad_norm": 1.3454771111737065,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 17158
    },
    {
      "epoch": 0.17159,
      "grad_norm": 1.3145208555465913,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 17159
    },
    {
      "epoch": 0.1716,
      "grad_norm": 1.3265742013322044,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 17160
    },
    {
      "epoch": 0.17161,
      "grad_norm": 1.0264878314735832,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 17161
    },
    {
      "epoch": 0.17162,
      "grad_norm": 1.5524565318758603,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 17162
    },
    {
      "epoch": 0.17163,
      "grad_norm": 1.1832378641723125,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 17163
    },
    {
      "epoch": 0.17164,
      "grad_norm": 1.3867140127479662,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 17164
    },
    {
      "epoch": 0.17165,
      "grad_norm": 1.269827437724221,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 17165
    },
    {
      "epoch": 0.17166,
      "grad_norm": 1.3508449911267215,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 17166
    },
    {
      "epoch": 0.17167,
      "grad_norm": 1.0951431787498167,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 17167
    },
    {
      "epoch": 0.17168,
      "grad_norm": 1.34342161127684,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 17168
    },
    {
      "epoch": 0.17169,
      "grad_norm": 1.2916502046230676,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 17169
    },
    {
      "epoch": 0.1717,
      "grad_norm": 1.1609694039144327,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 17170
    },
    {
      "epoch": 0.17171,
      "grad_norm": 1.320479587536383,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 17171
    },
    {
      "epoch": 0.17172,
      "grad_norm": 1.4114748920774176,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 17172
    },
    {
      "epoch": 0.17173,
      "grad_norm": 1.1061921435867776,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 17173
    },
    {
      "epoch": 0.17174,
      "grad_norm": 1.63429457533596,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 17174
    },
    {
      "epoch": 0.17175,
      "grad_norm": 1.1211868271676955,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 17175
    },
    {
      "epoch": 0.17176,
      "grad_norm": 1.347355707028804,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 17176
    },
    {
      "epoch": 0.17177,
      "grad_norm": 1.3774178524334308,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 17177
    },
    {
      "epoch": 0.17178,
      "grad_norm": 1.293544172783026,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 17178
    },
    {
      "epoch": 0.17179,
      "grad_norm": 1.4128510891275874,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 17179
    },
    {
      "epoch": 0.1718,
      "grad_norm": 1.1957458135373926,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 17180
    },
    {
      "epoch": 0.17181,
      "grad_norm": 1.3046029433494455,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 17181
    },
    {
      "epoch": 0.17182,
      "grad_norm": 1.1799759959603753,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 17182
    },
    {
      "epoch": 0.17183,
      "grad_norm": 1.1758628162170435,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 17183
    },
    {
      "epoch": 0.17184,
      "grad_norm": 1.341975268725816,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 17184
    },
    {
      "epoch": 0.17185,
      "grad_norm": 1.1445576596269915,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 17185
    },
    {
      "epoch": 0.17186,
      "grad_norm": 1.3224603315483399,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 17186
    },
    {
      "epoch": 0.17187,
      "grad_norm": 1.2395826243783463,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 17187
    },
    {
      "epoch": 0.17188,
      "grad_norm": 1.2690006660452913,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 17188
    },
    {
      "epoch": 0.17189,
      "grad_norm": 1.2573542334999037,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 17189
    },
    {
      "epoch": 0.1719,
      "grad_norm": 1.2335073132847372,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 17190
    },
    {
      "epoch": 0.17191,
      "grad_norm": 1.1932810784949115,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 17191
    },
    {
      "epoch": 0.17192,
      "grad_norm": 1.4304829083222084,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 17192
    },
    {
      "epoch": 0.17193,
      "grad_norm": 1.1989208824157573,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 17193
    },
    {
      "epoch": 0.17194,
      "grad_norm": 1.3947380618125669,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 17194
    },
    {
      "epoch": 0.17195,
      "grad_norm": 0.9527703239185515,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 17195
    },
    {
      "epoch": 0.17196,
      "grad_norm": 1.342668531453168,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 17196
    },
    {
      "epoch": 0.17197,
      "grad_norm": 1.421452676097791,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 17197
    },
    {
      "epoch": 0.17198,
      "grad_norm": 1.5677601226221192,
      "learning_rate": 0.003,
      "loss": 4.1106,
      "step": 17198
    },
    {
      "epoch": 0.17199,
      "grad_norm": 0.899927555636912,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 17199
    },
    {
      "epoch": 0.172,
      "grad_norm": 1.1429881140819786,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 17200
    },
    {
      "epoch": 0.17201,
      "grad_norm": 1.3964666203361211,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 17201
    },
    {
      "epoch": 0.17202,
      "grad_norm": 1.0194327815481596,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 17202
    },
    {
      "epoch": 0.17203,
      "grad_norm": 1.4128801485233948,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 17203
    },
    {
      "epoch": 0.17204,
      "grad_norm": 1.1642083930368834,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 17204
    },
    {
      "epoch": 0.17205,
      "grad_norm": 1.3433710344584748,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 17205
    },
    {
      "epoch": 0.17206,
      "grad_norm": 1.1485054644045822,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 17206
    },
    {
      "epoch": 0.17207,
      "grad_norm": 1.3944503319228418,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 17207
    },
    {
      "epoch": 0.17208,
      "grad_norm": 1.547361725635885,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 17208
    },
    {
      "epoch": 0.17209,
      "grad_norm": 1.5347040469120532,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 17209
    },
    {
      "epoch": 0.1721,
      "grad_norm": 1.2648440376927654,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 17210
    },
    {
      "epoch": 0.17211,
      "grad_norm": 1.2446087587338501,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 17211
    },
    {
      "epoch": 0.17212,
      "grad_norm": 1.1910938246129608,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 17212
    },
    {
      "epoch": 0.17213,
      "grad_norm": 1.4170422029493879,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 17213
    },
    {
      "epoch": 0.17214,
      "grad_norm": 1.230923562426313,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 17214
    },
    {
      "epoch": 0.17215,
      "grad_norm": 1.3215117269643324,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 17215
    },
    {
      "epoch": 0.17216,
      "grad_norm": 1.332945031735043,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 17216
    },
    {
      "epoch": 0.17217,
      "grad_norm": 1.1035955522319827,
      "learning_rate": 0.003,
      "loss": 4.0184,
      "step": 17217
    },
    {
      "epoch": 0.17218,
      "grad_norm": 1.428788610850809,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 17218
    },
    {
      "epoch": 0.17219,
      "grad_norm": 1.3068129180189931,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 17219
    },
    {
      "epoch": 0.1722,
      "grad_norm": 1.0638078684259753,
      "learning_rate": 0.003,
      "loss": 4.0255,
      "step": 17220
    },
    {
      "epoch": 0.17221,
      "grad_norm": 1.496968304741232,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 17221
    },
    {
      "epoch": 0.17222,
      "grad_norm": 1.1011271093479307,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 17222
    },
    {
      "epoch": 0.17223,
      "grad_norm": 1.3512456024019075,
      "learning_rate": 0.003,
      "loss": 4.0308,
      "step": 17223
    },
    {
      "epoch": 0.17224,
      "grad_norm": 1.1531275939399306,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 17224
    },
    {
      "epoch": 0.17225,
      "grad_norm": 1.4043274764932459,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 17225
    },
    {
      "epoch": 0.17226,
      "grad_norm": 1.1798686365718773,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 17226
    },
    {
      "epoch": 0.17227,
      "grad_norm": 1.4155883410107954,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 17227
    },
    {
      "epoch": 0.17228,
      "grad_norm": 1.261911184142582,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 17228
    },
    {
      "epoch": 0.17229,
      "grad_norm": 1.5439118749060368,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 17229
    },
    {
      "epoch": 0.1723,
      "grad_norm": 1.5607439747288108,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 17230
    },
    {
      "epoch": 0.17231,
      "grad_norm": 1.1242778683582726,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 17231
    },
    {
      "epoch": 0.17232,
      "grad_norm": 1.5294082757902303,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 17232
    },
    {
      "epoch": 0.17233,
      "grad_norm": 1.244232824073628,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 17233
    },
    {
      "epoch": 0.17234,
      "grad_norm": 1.464765045433406,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 17234
    },
    {
      "epoch": 0.17235,
      "grad_norm": 1.212812122055719,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 17235
    },
    {
      "epoch": 0.17236,
      "grad_norm": 1.5073516799941888,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 17236
    },
    {
      "epoch": 0.17237,
      "grad_norm": 0.9852482225652126,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 17237
    },
    {
      "epoch": 0.17238,
      "grad_norm": 1.2345978313177626,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 17238
    },
    {
      "epoch": 0.17239,
      "grad_norm": 1.3266079466508103,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 17239
    },
    {
      "epoch": 0.1724,
      "grad_norm": 1.2165704381666227,
      "learning_rate": 0.003,
      "loss": 4.0105,
      "step": 17240
    },
    {
      "epoch": 0.17241,
      "grad_norm": 1.5382431048618486,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 17241
    },
    {
      "epoch": 0.17242,
      "grad_norm": 1.1256669061556397,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17242
    },
    {
      "epoch": 0.17243,
      "grad_norm": 1.3371283903340336,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 17243
    },
    {
      "epoch": 0.17244,
      "grad_norm": 1.0856568234996853,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 17244
    },
    {
      "epoch": 0.17245,
      "grad_norm": 1.250163433629188,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 17245
    },
    {
      "epoch": 0.17246,
      "grad_norm": 1.4700541215967489,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 17246
    },
    {
      "epoch": 0.17247,
      "grad_norm": 1.2752347955260197,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 17247
    },
    {
      "epoch": 0.17248,
      "grad_norm": 1.4364732398721989,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 17248
    },
    {
      "epoch": 0.17249,
      "grad_norm": 1.1495637776253087,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 17249
    },
    {
      "epoch": 0.1725,
      "grad_norm": 1.583303890796779,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 17250
    },
    {
      "epoch": 0.17251,
      "grad_norm": 1.1828008978179465,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 17251
    },
    {
      "epoch": 0.17252,
      "grad_norm": 1.1962184866477457,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 17252
    },
    {
      "epoch": 0.17253,
      "grad_norm": 1.2827090303428514,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 17253
    },
    {
      "epoch": 0.17254,
      "grad_norm": 1.3991216810491618,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 17254
    },
    {
      "epoch": 0.17255,
      "grad_norm": 1.5051394917977816,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 17255
    },
    {
      "epoch": 0.17256,
      "grad_norm": 1.1584665892286088,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 17256
    },
    {
      "epoch": 0.17257,
      "grad_norm": 1.300102958418158,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 17257
    },
    {
      "epoch": 0.17258,
      "grad_norm": 1.1476915110738537,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17258
    },
    {
      "epoch": 0.17259,
      "grad_norm": 1.314246452505438,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 17259
    },
    {
      "epoch": 0.1726,
      "grad_norm": 1.271661473550511,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 17260
    },
    {
      "epoch": 0.17261,
      "grad_norm": 1.3326567236325095,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 17261
    },
    {
      "epoch": 0.17262,
      "grad_norm": 1.1439022006444373,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 17262
    },
    {
      "epoch": 0.17263,
      "grad_norm": 1.2667710201126365,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 17263
    },
    {
      "epoch": 0.17264,
      "grad_norm": 1.2758170744510648,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 17264
    },
    {
      "epoch": 0.17265,
      "grad_norm": 1.0010394942517953,
      "learning_rate": 0.003,
      "loss": 4.0308,
      "step": 17265
    },
    {
      "epoch": 0.17266,
      "grad_norm": 1.571265103363544,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 17266
    },
    {
      "epoch": 0.17267,
      "grad_norm": 1.2650034695598296,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 17267
    },
    {
      "epoch": 0.17268,
      "grad_norm": 1.0913091722989166,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 17268
    },
    {
      "epoch": 0.17269,
      "grad_norm": 1.5213843839037597,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 17269
    },
    {
      "epoch": 0.1727,
      "grad_norm": 1.3113409721606066,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 17270
    },
    {
      "epoch": 0.17271,
      "grad_norm": 1.3051990704978562,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 17271
    },
    {
      "epoch": 0.17272,
      "grad_norm": 1.205618793219367,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 17272
    },
    {
      "epoch": 0.17273,
      "grad_norm": 1.1300313604332728,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 17273
    },
    {
      "epoch": 0.17274,
      "grad_norm": 1.1243566938219207,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 17274
    },
    {
      "epoch": 0.17275,
      "grad_norm": 1.112047280737149,
      "learning_rate": 0.003,
      "loss": 4.0262,
      "step": 17275
    },
    {
      "epoch": 0.17276,
      "grad_norm": 1.8547947524290764,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 17276
    },
    {
      "epoch": 0.17277,
      "grad_norm": 1.1514749606092396,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 17277
    },
    {
      "epoch": 0.17278,
      "grad_norm": 1.1727422654093491,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 17278
    },
    {
      "epoch": 0.17279,
      "grad_norm": 1.3751968310876361,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 17279
    },
    {
      "epoch": 0.1728,
      "grad_norm": 1.3185583131052647,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 17280
    },
    {
      "epoch": 0.17281,
      "grad_norm": 1.175579160659979,
      "learning_rate": 0.003,
      "loss": 4.0158,
      "step": 17281
    },
    {
      "epoch": 0.17282,
      "grad_norm": 1.3651433359056662,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 17282
    },
    {
      "epoch": 0.17283,
      "grad_norm": 1.202724664414615,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 17283
    },
    {
      "epoch": 0.17284,
      "grad_norm": 1.601234921452785,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17284
    },
    {
      "epoch": 0.17285,
      "grad_norm": 0.9875939541340876,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 17285
    },
    {
      "epoch": 0.17286,
      "grad_norm": 1.280664764636334,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 17286
    },
    {
      "epoch": 0.17287,
      "grad_norm": 1.3142995598924436,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 17287
    },
    {
      "epoch": 0.17288,
      "grad_norm": 1.3170755852035023,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 17288
    },
    {
      "epoch": 0.17289,
      "grad_norm": 1.3429507399872127,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 17289
    },
    {
      "epoch": 0.1729,
      "grad_norm": 1.1784211039042498,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 17290
    },
    {
      "epoch": 0.17291,
      "grad_norm": 1.4020553869097763,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 17291
    },
    {
      "epoch": 0.17292,
      "grad_norm": 1.3454413871152973,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 17292
    },
    {
      "epoch": 0.17293,
      "grad_norm": 1.3043302180960887,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 17293
    },
    {
      "epoch": 0.17294,
      "grad_norm": 1.3963117103349811,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 17294
    },
    {
      "epoch": 0.17295,
      "grad_norm": 1.1466282181055694,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 17295
    },
    {
      "epoch": 0.17296,
      "grad_norm": 1.540582528810366,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 17296
    },
    {
      "epoch": 0.17297,
      "grad_norm": 1.4338787431328484,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 17297
    },
    {
      "epoch": 0.17298,
      "grad_norm": 1.1298772161127753,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 17298
    },
    {
      "epoch": 0.17299,
      "grad_norm": 1.364466493792987,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 17299
    },
    {
      "epoch": 0.173,
      "grad_norm": 1.2171701816028715,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 17300
    },
    {
      "epoch": 0.17301,
      "grad_norm": 1.1607674163362474,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 17301
    },
    {
      "epoch": 0.17302,
      "grad_norm": 1.325897453697002,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 17302
    },
    {
      "epoch": 0.17303,
      "grad_norm": 1.15890586726132,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 17303
    },
    {
      "epoch": 0.17304,
      "grad_norm": 1.2952615224255417,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 17304
    },
    {
      "epoch": 0.17305,
      "grad_norm": 1.140242946901405,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 17305
    },
    {
      "epoch": 0.17306,
      "grad_norm": 1.2200693633923876,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 17306
    },
    {
      "epoch": 0.17307,
      "grad_norm": 1.2683280442429594,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 17307
    },
    {
      "epoch": 0.17308,
      "grad_norm": 1.2786029341686145,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 17308
    },
    {
      "epoch": 0.17309,
      "grad_norm": 1.1671240380588164,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 17309
    },
    {
      "epoch": 0.1731,
      "grad_norm": 1.3699267211144008,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 17310
    },
    {
      "epoch": 0.17311,
      "grad_norm": 1.1161068521276887,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 17311
    },
    {
      "epoch": 0.17312,
      "grad_norm": 1.3884575625854214,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 17312
    },
    {
      "epoch": 0.17313,
      "grad_norm": 1.2709628688699923,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 17313
    },
    {
      "epoch": 0.17314,
      "grad_norm": 1.2587021165172843,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 17314
    },
    {
      "epoch": 0.17315,
      "grad_norm": 1.482888867163177,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 17315
    },
    {
      "epoch": 0.17316,
      "grad_norm": 1.314590397860297,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 17316
    },
    {
      "epoch": 0.17317,
      "grad_norm": 1.3136500728936051,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 17317
    },
    {
      "epoch": 0.17318,
      "grad_norm": 1.3615270441333711,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 17318
    },
    {
      "epoch": 0.17319,
      "grad_norm": 1.3977904928288472,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 17319
    },
    {
      "epoch": 0.1732,
      "grad_norm": 1.4946716511230473,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 17320
    },
    {
      "epoch": 0.17321,
      "grad_norm": 1.1554212838199085,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 17321
    },
    {
      "epoch": 0.17322,
      "grad_norm": 1.1850770640871187,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 17322
    },
    {
      "epoch": 0.17323,
      "grad_norm": 1.286672324486612,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 17323
    },
    {
      "epoch": 0.17324,
      "grad_norm": 1.233534924738266,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 17324
    },
    {
      "epoch": 0.17325,
      "grad_norm": 1.3122657038020409,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 17325
    },
    {
      "epoch": 0.17326,
      "grad_norm": 1.3013893089950082,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 17326
    },
    {
      "epoch": 0.17327,
      "grad_norm": 1.7277467216767701,
      "learning_rate": 0.003,
      "loss": 4.097,
      "step": 17327
    },
    {
      "epoch": 0.17328,
      "grad_norm": 1.1821787172337874,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 17328
    },
    {
      "epoch": 0.17329,
      "grad_norm": 1.6527006005088747,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 17329
    },
    {
      "epoch": 0.1733,
      "grad_norm": 1.1597089388958801,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 17330
    },
    {
      "epoch": 0.17331,
      "grad_norm": 1.160774057476352,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 17331
    },
    {
      "epoch": 0.17332,
      "grad_norm": 1.2581431200567026,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 17332
    },
    {
      "epoch": 0.17333,
      "grad_norm": 1.3622254185376,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 17333
    },
    {
      "epoch": 0.17334,
      "grad_norm": 1.359613278225029,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 17334
    },
    {
      "epoch": 0.17335,
      "grad_norm": 1.0871508787353932,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 17335
    },
    {
      "epoch": 0.17336,
      "grad_norm": 1.476010253662266,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 17336
    },
    {
      "epoch": 0.17337,
      "grad_norm": 1.1018459832652663,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 17337
    },
    {
      "epoch": 0.17338,
      "grad_norm": 1.6347777241101717,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 17338
    },
    {
      "epoch": 0.17339,
      "grad_norm": 1.3550539222364155,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 17339
    },
    {
      "epoch": 0.1734,
      "grad_norm": 1.2307412658433872,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 17340
    },
    {
      "epoch": 0.17341,
      "grad_norm": 1.3116085991874384,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 17341
    },
    {
      "epoch": 0.17342,
      "grad_norm": 1.1839396795080575,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 17342
    },
    {
      "epoch": 0.17343,
      "grad_norm": 1.5878123562511406,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 17343
    },
    {
      "epoch": 0.17344,
      "grad_norm": 0.966311302374749,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 17344
    },
    {
      "epoch": 0.17345,
      "grad_norm": 1.4614186302212233,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 17345
    },
    {
      "epoch": 0.17346,
      "grad_norm": 1.5076469202916865,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 17346
    },
    {
      "epoch": 0.17347,
      "grad_norm": 0.9372154908707657,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 17347
    },
    {
      "epoch": 0.17348,
      "grad_norm": 1.2906133943823377,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 17348
    },
    {
      "epoch": 0.17349,
      "grad_norm": 1.3538707872254896,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 17349
    },
    {
      "epoch": 0.1735,
      "grad_norm": 1.3238868446804868,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 17350
    },
    {
      "epoch": 0.17351,
      "grad_norm": 1.1704739066405059,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 17351
    },
    {
      "epoch": 0.17352,
      "grad_norm": 1.2941053022753086,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 17352
    },
    {
      "epoch": 0.17353,
      "grad_norm": 1.213494266625952,
      "learning_rate": 0.003,
      "loss": 4.0225,
      "step": 17353
    },
    {
      "epoch": 0.17354,
      "grad_norm": 1.161557789178661,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 17354
    },
    {
      "epoch": 0.17355,
      "grad_norm": 1.3474693954183323,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 17355
    },
    {
      "epoch": 0.17356,
      "grad_norm": 1.1487157146444444,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 17356
    },
    {
      "epoch": 0.17357,
      "grad_norm": 1.5917371809698078,
      "learning_rate": 0.003,
      "loss": 4.0989,
      "step": 17357
    },
    {
      "epoch": 0.17358,
      "grad_norm": 0.9987014860498662,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 17358
    },
    {
      "epoch": 0.17359,
      "grad_norm": 1.6232130332722863,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 17359
    },
    {
      "epoch": 0.1736,
      "grad_norm": 1.1533563501761357,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 17360
    },
    {
      "epoch": 0.17361,
      "grad_norm": 1.2258117354232874,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 17361
    },
    {
      "epoch": 0.17362,
      "grad_norm": 1.3591228343764234,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 17362
    },
    {
      "epoch": 0.17363,
      "grad_norm": 1.1575402182752415,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 17363
    },
    {
      "epoch": 0.17364,
      "grad_norm": 1.2584198090000192,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 17364
    },
    {
      "epoch": 0.17365,
      "grad_norm": 1.217932782500489,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 17365
    },
    {
      "epoch": 0.17366,
      "grad_norm": 1.6319796303181,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 17366
    },
    {
      "epoch": 0.17367,
      "grad_norm": 1.2496362335876412,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 17367
    },
    {
      "epoch": 0.17368,
      "grad_norm": 1.3543288489384762,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 17368
    },
    {
      "epoch": 0.17369,
      "grad_norm": 1.227748208441705,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 17369
    },
    {
      "epoch": 0.1737,
      "grad_norm": 1.2128552652264637,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 17370
    },
    {
      "epoch": 0.17371,
      "grad_norm": 1.3151813991304573,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 17371
    },
    {
      "epoch": 0.17372,
      "grad_norm": 1.2675731600786122,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 17372
    },
    {
      "epoch": 0.17373,
      "grad_norm": 1.1910753012986075,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 17373
    },
    {
      "epoch": 0.17374,
      "grad_norm": 1.6257728688556055,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 17374
    },
    {
      "epoch": 0.17375,
      "grad_norm": 1.0985533268317234,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 17375
    },
    {
      "epoch": 0.17376,
      "grad_norm": 1.4441899532090265,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 17376
    },
    {
      "epoch": 0.17377,
      "grad_norm": 1.1473629283807807,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 17377
    },
    {
      "epoch": 0.17378,
      "grad_norm": 1.2819491385872488,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 17378
    },
    {
      "epoch": 0.17379,
      "grad_norm": 1.4537829518027603,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 17379
    },
    {
      "epoch": 0.1738,
      "grad_norm": 1.3293515545404049,
      "learning_rate": 0.003,
      "loss": 4.0243,
      "step": 17380
    },
    {
      "epoch": 0.17381,
      "grad_norm": 1.7194165571937199,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 17381
    },
    {
      "epoch": 0.17382,
      "grad_norm": 1.0295087960436615,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 17382
    },
    {
      "epoch": 0.17383,
      "grad_norm": 1.1044202128867775,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 17383
    },
    {
      "epoch": 0.17384,
      "grad_norm": 1.2696710055506293,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 17384
    },
    {
      "epoch": 0.17385,
      "grad_norm": 1.4853913519470598,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 17385
    },
    {
      "epoch": 0.17386,
      "grad_norm": 1.2943715520245132,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 17386
    },
    {
      "epoch": 0.17387,
      "grad_norm": 1.155838567800402,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 17387
    },
    {
      "epoch": 0.17388,
      "grad_norm": 1.5429276307218556,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 17388
    },
    {
      "epoch": 0.17389,
      "grad_norm": 1.2413990783254136,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 17389
    },
    {
      "epoch": 0.1739,
      "grad_norm": 1.3477300495649724,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 17390
    },
    {
      "epoch": 0.17391,
      "grad_norm": 1.2977172852945469,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 17391
    },
    {
      "epoch": 0.17392,
      "grad_norm": 1.1374129030541535,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 17392
    },
    {
      "epoch": 0.17393,
      "grad_norm": 1.1607938224887557,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 17393
    },
    {
      "epoch": 0.17394,
      "grad_norm": 1.1673671826581336,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 17394
    },
    {
      "epoch": 0.17395,
      "grad_norm": 1.1546886884764822,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 17395
    },
    {
      "epoch": 0.17396,
      "grad_norm": 1.3656784042418233,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 17396
    },
    {
      "epoch": 0.17397,
      "grad_norm": 0.8135727371113464,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 17397
    },
    {
      "epoch": 0.17398,
      "grad_norm": 1.137411890639474,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 17398
    },
    {
      "epoch": 0.17399,
      "grad_norm": 1.6386060306451693,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 17399
    },
    {
      "epoch": 0.174,
      "grad_norm": 1.021912059527306,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 17400
    },
    {
      "epoch": 0.17401,
      "grad_norm": 1.6368228150526618,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 17401
    },
    {
      "epoch": 0.17402,
      "grad_norm": 1.3096375352568765,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17402
    },
    {
      "epoch": 0.17403,
      "grad_norm": 1.15501855236692,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 17403
    },
    {
      "epoch": 0.17404,
      "grad_norm": 1.5085837764292096,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 17404
    },
    {
      "epoch": 0.17405,
      "grad_norm": 1.143097257206991,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 17405
    },
    {
      "epoch": 0.17406,
      "grad_norm": 1.3487692828366185,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 17406
    },
    {
      "epoch": 0.17407,
      "grad_norm": 1.0810910518682053,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 17407
    },
    {
      "epoch": 0.17408,
      "grad_norm": 1.2778119740052092,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 17408
    },
    {
      "epoch": 0.17409,
      "grad_norm": 1.1101934840980172,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 17409
    },
    {
      "epoch": 0.1741,
      "grad_norm": 1.1874662189268035,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 17410
    },
    {
      "epoch": 0.17411,
      "grad_norm": 1.5161881786934777,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 17411
    },
    {
      "epoch": 0.17412,
      "grad_norm": 1.4253812994933417,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 17412
    },
    {
      "epoch": 0.17413,
      "grad_norm": 1.5970659377886456,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 17413
    },
    {
      "epoch": 0.17414,
      "grad_norm": 1.1977208048904462,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17414
    },
    {
      "epoch": 0.17415,
      "grad_norm": 1.2466475381012039,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 17415
    },
    {
      "epoch": 0.17416,
      "grad_norm": 1.4494269543884284,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 17416
    },
    {
      "epoch": 0.17417,
      "grad_norm": 1.304335207124149,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17417
    },
    {
      "epoch": 0.17418,
      "grad_norm": 1.0728577346961645,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 17418
    },
    {
      "epoch": 0.17419,
      "grad_norm": 1.5547921561653366,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 17419
    },
    {
      "epoch": 0.1742,
      "grad_norm": 1.1010696107279216,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 17420
    },
    {
      "epoch": 0.17421,
      "grad_norm": 1.1219877424315068,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 17421
    },
    {
      "epoch": 0.17422,
      "grad_norm": 1.241923058210318,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 17422
    },
    {
      "epoch": 0.17423,
      "grad_norm": 1.4826931444793456,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 17423
    },
    {
      "epoch": 0.17424,
      "grad_norm": 1.2285170108606516,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 17424
    },
    {
      "epoch": 0.17425,
      "grad_norm": 1.181199300063592,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 17425
    },
    {
      "epoch": 0.17426,
      "grad_norm": 1.3961143890285066,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 17426
    },
    {
      "epoch": 0.17427,
      "grad_norm": 1.173246971106792,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 17427
    },
    {
      "epoch": 0.17428,
      "grad_norm": 1.3700077212676505,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 17428
    },
    {
      "epoch": 0.17429,
      "grad_norm": 1.3895862890260624,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 17429
    },
    {
      "epoch": 0.1743,
      "grad_norm": 1.292937448029352,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 17430
    },
    {
      "epoch": 0.17431,
      "grad_norm": 1.3286655907704719,
      "learning_rate": 0.003,
      "loss": 4.028,
      "step": 17431
    },
    {
      "epoch": 0.17432,
      "grad_norm": 1.2366878735177564,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 17432
    },
    {
      "epoch": 0.17433,
      "grad_norm": 1.2354096916732125,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 17433
    },
    {
      "epoch": 0.17434,
      "grad_norm": 1.2531275227176735,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 17434
    },
    {
      "epoch": 0.17435,
      "grad_norm": 1.1650514653874386,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 17435
    },
    {
      "epoch": 0.17436,
      "grad_norm": 1.2550299354752033,
      "learning_rate": 0.003,
      "loss": 4.0232,
      "step": 17436
    },
    {
      "epoch": 0.17437,
      "grad_norm": 1.3582554391865993,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 17437
    },
    {
      "epoch": 0.17438,
      "grad_norm": 1.3617902568728197,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 17438
    },
    {
      "epoch": 0.17439,
      "grad_norm": 1.3004396587625653,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 17439
    },
    {
      "epoch": 0.1744,
      "grad_norm": 1.2032476227166182,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 17440
    },
    {
      "epoch": 0.17441,
      "grad_norm": 1.2918767594577591,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 17441
    },
    {
      "epoch": 0.17442,
      "grad_norm": 1.2137929756564796,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 17442
    },
    {
      "epoch": 0.17443,
      "grad_norm": 1.401955872786344,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 17443
    },
    {
      "epoch": 0.17444,
      "grad_norm": 1.5254161669212798,
      "learning_rate": 0.003,
      "loss": 4.0261,
      "step": 17444
    },
    {
      "epoch": 0.17445,
      "grad_norm": 1.0197218411898092,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 17445
    },
    {
      "epoch": 0.17446,
      "grad_norm": 1.4127761657383924,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 17446
    },
    {
      "epoch": 0.17447,
      "grad_norm": 1.3330025253011795,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 17447
    },
    {
      "epoch": 0.17448,
      "grad_norm": 1.2495163571886592,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 17448
    },
    {
      "epoch": 0.17449,
      "grad_norm": 1.4294119441106288,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 17449
    },
    {
      "epoch": 0.1745,
      "grad_norm": 1.0491530867649907,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 17450
    },
    {
      "epoch": 0.17451,
      "grad_norm": 1.5565773989464897,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 17451
    },
    {
      "epoch": 0.17452,
      "grad_norm": 1.2571825790237903,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 17452
    },
    {
      "epoch": 0.17453,
      "grad_norm": 1.4828859884751822,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 17453
    },
    {
      "epoch": 0.17454,
      "grad_norm": 1.2956449750709225,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 17454
    },
    {
      "epoch": 0.17455,
      "grad_norm": 1.243516393411184,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 17455
    },
    {
      "epoch": 0.17456,
      "grad_norm": 1.319861010512668,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 17456
    },
    {
      "epoch": 0.17457,
      "grad_norm": 1.2945272589556633,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 17457
    },
    {
      "epoch": 0.17458,
      "grad_norm": 1.2834252737184408,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 17458
    },
    {
      "epoch": 0.17459,
      "grad_norm": 1.517112472998707,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17459
    },
    {
      "epoch": 0.1746,
      "grad_norm": 1.0806886971250516,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 17460
    },
    {
      "epoch": 0.17461,
      "grad_norm": 1.3459936523173737,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 17461
    },
    {
      "epoch": 0.17462,
      "grad_norm": 1.1351068594804266,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 17462
    },
    {
      "epoch": 0.17463,
      "grad_norm": 1.3315162335300437,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 17463
    },
    {
      "epoch": 0.17464,
      "grad_norm": 1.1421814018137226,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 17464
    },
    {
      "epoch": 0.17465,
      "grad_norm": 1.4720298962021945,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 17465
    },
    {
      "epoch": 0.17466,
      "grad_norm": 1.0970045489403935,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 17466
    },
    {
      "epoch": 0.17467,
      "grad_norm": 1.4051446872244608,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 17467
    },
    {
      "epoch": 0.17468,
      "grad_norm": 1.1788754409907096,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 17468
    },
    {
      "epoch": 0.17469,
      "grad_norm": 1.4085198117148778,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 17469
    },
    {
      "epoch": 0.1747,
      "grad_norm": 1.0656980683755632,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 17470
    },
    {
      "epoch": 0.17471,
      "grad_norm": 1.3596266943541238,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 17471
    },
    {
      "epoch": 0.17472,
      "grad_norm": 1.4360775955395388,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 17472
    },
    {
      "epoch": 0.17473,
      "grad_norm": 1.1815663699829684,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 17473
    },
    {
      "epoch": 0.17474,
      "grad_norm": 1.3773290285609925,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 17474
    },
    {
      "epoch": 0.17475,
      "grad_norm": 1.138640414247468,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 17475
    },
    {
      "epoch": 0.17476,
      "grad_norm": 1.4774328139223307,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 17476
    },
    {
      "epoch": 0.17477,
      "grad_norm": 1.1258013640931122,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 17477
    },
    {
      "epoch": 0.17478,
      "grad_norm": 1.445371284232178,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 17478
    },
    {
      "epoch": 0.17479,
      "grad_norm": 1.0627794731721976,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 17479
    },
    {
      "epoch": 0.1748,
      "grad_norm": 1.2629567251829825,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 17480
    },
    {
      "epoch": 0.17481,
      "grad_norm": 1.162982937463057,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 17481
    },
    {
      "epoch": 0.17482,
      "grad_norm": 1.2109852386399709,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 17482
    },
    {
      "epoch": 0.17483,
      "grad_norm": 1.351926763850564,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 17483
    },
    {
      "epoch": 0.17484,
      "grad_norm": 1.2675220236375762,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 17484
    },
    {
      "epoch": 0.17485,
      "grad_norm": 1.448801081514365,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 17485
    },
    {
      "epoch": 0.17486,
      "grad_norm": 1.470879245043995,
      "learning_rate": 0.003,
      "loss": 4.0981,
      "step": 17486
    },
    {
      "epoch": 0.17487,
      "grad_norm": 1.5235474751119706,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 17487
    },
    {
      "epoch": 0.17488,
      "grad_norm": 1.2882001150246338,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 17488
    },
    {
      "epoch": 0.17489,
      "grad_norm": 1.1262043232744685,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 17489
    },
    {
      "epoch": 0.1749,
      "grad_norm": 1.2529971851246242,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 17490
    },
    {
      "epoch": 0.17491,
      "grad_norm": 1.3515122585543957,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 17491
    },
    {
      "epoch": 0.17492,
      "grad_norm": 1.1888181326490819,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 17492
    },
    {
      "epoch": 0.17493,
      "grad_norm": 1.269549654881748,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 17493
    },
    {
      "epoch": 0.17494,
      "grad_norm": 1.399664853120194,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 17494
    },
    {
      "epoch": 0.17495,
      "grad_norm": 1.0547626195049762,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 17495
    },
    {
      "epoch": 0.17496,
      "grad_norm": 1.2589039501288364,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 17496
    },
    {
      "epoch": 0.17497,
      "grad_norm": 1.2087009396582926,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 17497
    },
    {
      "epoch": 0.17498,
      "grad_norm": 1.5111210585816903,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 17498
    },
    {
      "epoch": 0.17499,
      "grad_norm": 1.1932866062692538,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 17499
    },
    {
      "epoch": 0.175,
      "grad_norm": 1.3685368613552835,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 17500
    },
    {
      "epoch": 0.17501,
      "grad_norm": 1.4259266355444051,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 17501
    },
    {
      "epoch": 0.17502,
      "grad_norm": 1.0811465802849207,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 17502
    },
    {
      "epoch": 0.17503,
      "grad_norm": 1.3631121796923622,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 17503
    },
    {
      "epoch": 0.17504,
      "grad_norm": 1.3601765579398404,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 17504
    },
    {
      "epoch": 0.17505,
      "grad_norm": 1.3204992282929688,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 17505
    },
    {
      "epoch": 0.17506,
      "grad_norm": 1.6043420025523707,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 17506
    },
    {
      "epoch": 0.17507,
      "grad_norm": 1.1102090085261342,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 17507
    },
    {
      "epoch": 0.17508,
      "grad_norm": 1.3589012557270086,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 17508
    },
    {
      "epoch": 0.17509,
      "grad_norm": 1.4570648815138552,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 17509
    },
    {
      "epoch": 0.1751,
      "grad_norm": 1.2379406823564174,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 17510
    },
    {
      "epoch": 0.17511,
      "grad_norm": 1.2745485508345238,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17511
    },
    {
      "epoch": 0.17512,
      "grad_norm": 1.458311323129371,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 17512
    },
    {
      "epoch": 0.17513,
      "grad_norm": 1.2523363294551801,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 17513
    },
    {
      "epoch": 0.17514,
      "grad_norm": 1.326353253061568,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 17514
    },
    {
      "epoch": 0.17515,
      "grad_norm": 1.2078497675630715,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 17515
    },
    {
      "epoch": 0.17516,
      "grad_norm": 1.332111315017004,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 17516
    },
    {
      "epoch": 0.17517,
      "grad_norm": 1.4681129568264164,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 17517
    },
    {
      "epoch": 0.17518,
      "grad_norm": 0.9898792924352027,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 17518
    },
    {
      "epoch": 0.17519,
      "grad_norm": 1.5452090908593286,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 17519
    },
    {
      "epoch": 0.1752,
      "grad_norm": 1.027115202082818,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 17520
    },
    {
      "epoch": 0.17521,
      "grad_norm": 1.5480272088455356,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 17521
    },
    {
      "epoch": 0.17522,
      "grad_norm": 1.0872498679033396,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 17522
    },
    {
      "epoch": 0.17523,
      "grad_norm": 1.4579650844831544,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 17523
    },
    {
      "epoch": 0.17524,
      "grad_norm": 1.3716290946186658,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 17524
    },
    {
      "epoch": 0.17525,
      "grad_norm": 1.174774209157617,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 17525
    },
    {
      "epoch": 0.17526,
      "grad_norm": 1.3772871583425932,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 17526
    },
    {
      "epoch": 0.17527,
      "grad_norm": 1.34133868900741,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 17527
    },
    {
      "epoch": 0.17528,
      "grad_norm": 1.089014538515785,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 17528
    },
    {
      "epoch": 0.17529,
      "grad_norm": 1.4197266927363068,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 17529
    },
    {
      "epoch": 0.1753,
      "grad_norm": 1.0939073308060439,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 17530
    },
    {
      "epoch": 0.17531,
      "grad_norm": 1.2336453874907791,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 17531
    },
    {
      "epoch": 0.17532,
      "grad_norm": 1.1479276481511065,
      "learning_rate": 0.003,
      "loss": 4.0178,
      "step": 17532
    },
    {
      "epoch": 0.17533,
      "grad_norm": 1.2202607669748253,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 17533
    },
    {
      "epoch": 0.17534,
      "grad_norm": 1.4945469580763797,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 17534
    },
    {
      "epoch": 0.17535,
      "grad_norm": 1.6356180536776386,
      "learning_rate": 0.003,
      "loss": 4.0298,
      "step": 17535
    },
    {
      "epoch": 0.17536,
      "grad_norm": 1.3521367342384314,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 17536
    },
    {
      "epoch": 0.17537,
      "grad_norm": 1.3497434493622424,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 17537
    },
    {
      "epoch": 0.17538,
      "grad_norm": 1.1735820423921637,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 17538
    },
    {
      "epoch": 0.17539,
      "grad_norm": 1.3290898846115236,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 17539
    },
    {
      "epoch": 0.1754,
      "grad_norm": 1.1762529483557118,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 17540
    },
    {
      "epoch": 0.17541,
      "grad_norm": 1.2369001566462126,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 17541
    },
    {
      "epoch": 0.17542,
      "grad_norm": 1.253493920129149,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 17542
    },
    {
      "epoch": 0.17543,
      "grad_norm": 1.1991541022277417,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 17543
    },
    {
      "epoch": 0.17544,
      "grad_norm": 1.6048053140008003,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 17544
    },
    {
      "epoch": 0.17545,
      "grad_norm": 1.080637588554762,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 17545
    },
    {
      "epoch": 0.17546,
      "grad_norm": 1.6465218961255268,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 17546
    },
    {
      "epoch": 0.17547,
      "grad_norm": 1.2059376789891145,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 17547
    },
    {
      "epoch": 0.17548,
      "grad_norm": 1.5033238233306159,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 17548
    },
    {
      "epoch": 0.17549,
      "grad_norm": 1.2903528075058992,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 17549
    },
    {
      "epoch": 0.1755,
      "grad_norm": 1.1825193838204116,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 17550
    },
    {
      "epoch": 0.17551,
      "grad_norm": 1.2435071476516741,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 17551
    },
    {
      "epoch": 0.17552,
      "grad_norm": 1.2351634262921993,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 17552
    },
    {
      "epoch": 0.17553,
      "grad_norm": 1.1599814224863219,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 17553
    },
    {
      "epoch": 0.17554,
      "grad_norm": 1.209609419739831,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 17554
    },
    {
      "epoch": 0.17555,
      "grad_norm": 1.221752704114831,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 17555
    },
    {
      "epoch": 0.17556,
      "grad_norm": 1.7212103012472932,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 17556
    },
    {
      "epoch": 0.17557,
      "grad_norm": 1.3017911530063735,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 17557
    },
    {
      "epoch": 0.17558,
      "grad_norm": 1.485709091804075,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 17558
    },
    {
      "epoch": 0.17559,
      "grad_norm": 1.3049037188814676,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 17559
    },
    {
      "epoch": 0.1756,
      "grad_norm": 1.2577989391627555,
      "learning_rate": 0.003,
      "loss": 4.0292,
      "step": 17560
    },
    {
      "epoch": 0.17561,
      "grad_norm": 1.376837404561379,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 17561
    },
    {
      "epoch": 0.17562,
      "grad_norm": 1.1633686470895166,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 17562
    },
    {
      "epoch": 0.17563,
      "grad_norm": 1.182559838540788,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 17563
    },
    {
      "epoch": 0.17564,
      "grad_norm": 1.3425096746959984,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 17564
    },
    {
      "epoch": 0.17565,
      "grad_norm": 1.133752253129469,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 17565
    },
    {
      "epoch": 0.17566,
      "grad_norm": 1.5169381551979604,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 17566
    },
    {
      "epoch": 0.17567,
      "grad_norm": 1.341228303024416,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 17567
    },
    {
      "epoch": 0.17568,
      "grad_norm": 1.1788620587516216,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 17568
    },
    {
      "epoch": 0.17569,
      "grad_norm": 1.155754217235722,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 17569
    },
    {
      "epoch": 0.1757,
      "grad_norm": 1.2557432658871737,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 17570
    },
    {
      "epoch": 0.17571,
      "grad_norm": 1.2620674018967353,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 17571
    },
    {
      "epoch": 0.17572,
      "grad_norm": 1.2905583879720335,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 17572
    },
    {
      "epoch": 0.17573,
      "grad_norm": 1.275273313168324,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 17573
    },
    {
      "epoch": 0.17574,
      "grad_norm": 1.5488047698359892,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 17574
    },
    {
      "epoch": 0.17575,
      "grad_norm": 1.2913531469165127,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 17575
    },
    {
      "epoch": 0.17576,
      "grad_norm": 1.259871684253289,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 17576
    },
    {
      "epoch": 0.17577,
      "grad_norm": 1.2102978803530382,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 17577
    },
    {
      "epoch": 0.17578,
      "grad_norm": 1.5210035489703877,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 17578
    },
    {
      "epoch": 0.17579,
      "grad_norm": 1.285636618378266,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 17579
    },
    {
      "epoch": 0.1758,
      "grad_norm": 1.392919694020842,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 17580
    },
    {
      "epoch": 0.17581,
      "grad_norm": 1.3840133366277634,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 17581
    },
    {
      "epoch": 0.17582,
      "grad_norm": 1.0871155767026495,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 17582
    },
    {
      "epoch": 0.17583,
      "grad_norm": 1.5645413061585471,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 17583
    },
    {
      "epoch": 0.17584,
      "grad_norm": 1.17321561979706,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 17584
    },
    {
      "epoch": 0.17585,
      "grad_norm": 1.3604032628155835,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 17585
    },
    {
      "epoch": 0.17586,
      "grad_norm": 1.314642451894998,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 17586
    },
    {
      "epoch": 0.17587,
      "grad_norm": 1.3291505728511441,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 17587
    },
    {
      "epoch": 0.17588,
      "grad_norm": 1.4376103152540465,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 17588
    },
    {
      "epoch": 0.17589,
      "grad_norm": 1.3347456845034162,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 17589
    },
    {
      "epoch": 0.1759,
      "grad_norm": 1.363165539937303,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 17590
    },
    {
      "epoch": 0.17591,
      "grad_norm": 1.1446605795386675,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 17591
    },
    {
      "epoch": 0.17592,
      "grad_norm": 1.39653778260535,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 17592
    },
    {
      "epoch": 0.17593,
      "grad_norm": 1.1912570373914466,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 17593
    },
    {
      "epoch": 0.17594,
      "grad_norm": 1.2643293160733424,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 17594
    },
    {
      "epoch": 0.17595,
      "grad_norm": 0.9890806388280463,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 17595
    },
    {
      "epoch": 0.17596,
      "grad_norm": 1.3817387821035372,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 17596
    },
    {
      "epoch": 0.17597,
      "grad_norm": 1.13316950929552,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 17597
    },
    {
      "epoch": 0.17598,
      "grad_norm": 1.7346344042620478,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 17598
    },
    {
      "epoch": 0.17599,
      "grad_norm": 1.1569002155868353,
      "learning_rate": 0.003,
      "loss": 4.0967,
      "step": 17599
    },
    {
      "epoch": 0.176,
      "grad_norm": 1.5841135639006447,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 17600
    },
    {
      "epoch": 0.17601,
      "grad_norm": 1.1700214162493932,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 17601
    },
    {
      "epoch": 0.17602,
      "grad_norm": 1.395655377413905,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 17602
    },
    {
      "epoch": 0.17603,
      "grad_norm": 1.2600613047588067,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 17603
    },
    {
      "epoch": 0.17604,
      "grad_norm": 1.485621223796701,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17604
    },
    {
      "epoch": 0.17605,
      "grad_norm": 1.0934046197953102,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 17605
    },
    {
      "epoch": 0.17606,
      "grad_norm": 1.6813061462712295,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 17606
    },
    {
      "epoch": 0.17607,
      "grad_norm": 1.1496380657690461,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 17607
    },
    {
      "epoch": 0.17608,
      "grad_norm": 1.1253490677517515,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 17608
    },
    {
      "epoch": 0.17609,
      "grad_norm": 1.2947829489405915,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 17609
    },
    {
      "epoch": 0.1761,
      "grad_norm": 1.2533389268660708,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 17610
    },
    {
      "epoch": 0.17611,
      "grad_norm": 1.221058807420232,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 17611
    },
    {
      "epoch": 0.17612,
      "grad_norm": 1.211577910041733,
      "learning_rate": 0.003,
      "loss": 4.0003,
      "step": 17612
    },
    {
      "epoch": 0.17613,
      "grad_norm": 1.4963671852486207,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 17613
    },
    {
      "epoch": 0.17614,
      "grad_norm": 1.3621629216998155,
      "learning_rate": 0.003,
      "loss": 4.0333,
      "step": 17614
    },
    {
      "epoch": 0.17615,
      "grad_norm": 1.1745568317301551,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 17615
    },
    {
      "epoch": 0.17616,
      "grad_norm": 1.4917818846232451,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 17616
    },
    {
      "epoch": 0.17617,
      "grad_norm": 1.4688369182348506,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 17617
    },
    {
      "epoch": 0.17618,
      "grad_norm": 1.037151684296233,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 17618
    },
    {
      "epoch": 0.17619,
      "grad_norm": 1.4524502216009345,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 17619
    },
    {
      "epoch": 0.1762,
      "grad_norm": 1.0697082776084865,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 17620
    },
    {
      "epoch": 0.17621,
      "grad_norm": 1.1625501217376397,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 17621
    },
    {
      "epoch": 0.17622,
      "grad_norm": 1.1469754723606305,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 17622
    },
    {
      "epoch": 0.17623,
      "grad_norm": 1.16858664998892,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 17623
    },
    {
      "epoch": 0.17624,
      "grad_norm": 1.3135664062689731,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 17624
    },
    {
      "epoch": 0.17625,
      "grad_norm": 1.6383406481826261,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 17625
    },
    {
      "epoch": 0.17626,
      "grad_norm": 1.2545188782831724,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 17626
    },
    {
      "epoch": 0.17627,
      "grad_norm": 1.2478847855408735,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 17627
    },
    {
      "epoch": 0.17628,
      "grad_norm": 1.2154447103562211,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 17628
    },
    {
      "epoch": 0.17629,
      "grad_norm": 1.3128872316814053,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 17629
    },
    {
      "epoch": 0.1763,
      "grad_norm": 1.2558917905084745,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 17630
    },
    {
      "epoch": 0.17631,
      "grad_norm": 1.56447314552511,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 17631
    },
    {
      "epoch": 0.17632,
      "grad_norm": 1.2288214495890235,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 17632
    },
    {
      "epoch": 0.17633,
      "grad_norm": 1.461485377819148,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 17633
    },
    {
      "epoch": 0.17634,
      "grad_norm": 1.4210982268515342,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 17634
    },
    {
      "epoch": 0.17635,
      "grad_norm": 1.6796500872140807,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 17635
    },
    {
      "epoch": 0.17636,
      "grad_norm": 1.1161604941055283,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 17636
    },
    {
      "epoch": 0.17637,
      "grad_norm": 1.2688593912531774,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 17637
    },
    {
      "epoch": 0.17638,
      "grad_norm": 1.2239390108100066,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 17638
    },
    {
      "epoch": 0.17639,
      "grad_norm": 1.4543016085223777,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 17639
    },
    {
      "epoch": 0.1764,
      "grad_norm": 1.1311921748288387,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 17640
    },
    {
      "epoch": 0.17641,
      "grad_norm": 1.6222284430220124,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17641
    },
    {
      "epoch": 0.17642,
      "grad_norm": 1.1514824491992248,
      "learning_rate": 0.003,
      "loss": 4.0247,
      "step": 17642
    },
    {
      "epoch": 0.17643,
      "grad_norm": 1.4233439574319937,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 17643
    },
    {
      "epoch": 0.17644,
      "grad_norm": 1.3460528345168452,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 17644
    },
    {
      "epoch": 0.17645,
      "grad_norm": 1.1280697162130136,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 17645
    },
    {
      "epoch": 0.17646,
      "grad_norm": 1.3235388297884105,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 17646
    },
    {
      "epoch": 0.17647,
      "grad_norm": 1.1302885823025965,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 17647
    },
    {
      "epoch": 0.17648,
      "grad_norm": 1.3439873433984786,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 17648
    },
    {
      "epoch": 0.17649,
      "grad_norm": 1.290515056093695,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 17649
    },
    {
      "epoch": 0.1765,
      "grad_norm": 1.4803167997075957,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 17650
    },
    {
      "epoch": 0.17651,
      "grad_norm": 1.1718919994994168,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 17651
    },
    {
      "epoch": 0.17652,
      "grad_norm": 1.3025434998217125,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 17652
    },
    {
      "epoch": 0.17653,
      "grad_norm": 1.1461619738250126,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 17653
    },
    {
      "epoch": 0.17654,
      "grad_norm": 1.2045934682269839,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 17654
    },
    {
      "epoch": 0.17655,
      "grad_norm": 1.2600544087934396,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 17655
    },
    {
      "epoch": 0.17656,
      "grad_norm": 1.2993441694569487,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 17656
    },
    {
      "epoch": 0.17657,
      "grad_norm": 1.58132482735925,
      "learning_rate": 0.003,
      "loss": 4.0258,
      "step": 17657
    },
    {
      "epoch": 0.17658,
      "grad_norm": 1.2216840641376203,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 17658
    },
    {
      "epoch": 0.17659,
      "grad_norm": 1.3101905106329146,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 17659
    },
    {
      "epoch": 0.1766,
      "grad_norm": 1.4183979395592572,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 17660
    },
    {
      "epoch": 0.17661,
      "grad_norm": 1.2422597475901402,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 17661
    },
    {
      "epoch": 0.17662,
      "grad_norm": 1.5392641833011527,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 17662
    },
    {
      "epoch": 0.17663,
      "grad_norm": 0.9302020789539822,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 17663
    },
    {
      "epoch": 0.17664,
      "grad_norm": 1.352504587734451,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 17664
    },
    {
      "epoch": 0.17665,
      "grad_norm": 1.1405003858502711,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 17665
    },
    {
      "epoch": 0.17666,
      "grad_norm": 1.5675115628172653,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 17666
    },
    {
      "epoch": 0.17667,
      "grad_norm": 1.6971132785861853,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 17667
    },
    {
      "epoch": 0.17668,
      "grad_norm": 0.9969641889111963,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 17668
    },
    {
      "epoch": 0.17669,
      "grad_norm": 1.2909894956823855,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 17669
    },
    {
      "epoch": 0.1767,
      "grad_norm": 1.7379145287328,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 17670
    },
    {
      "epoch": 0.17671,
      "grad_norm": 0.9627833816379531,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 17671
    },
    {
      "epoch": 0.17672,
      "grad_norm": 1.0898450590560478,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 17672
    },
    {
      "epoch": 0.17673,
      "grad_norm": 1.248478035290169,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 17673
    },
    {
      "epoch": 0.17674,
      "grad_norm": 1.3593375285238194,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 17674
    },
    {
      "epoch": 0.17675,
      "grad_norm": 1.3523789795245187,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 17675
    },
    {
      "epoch": 0.17676,
      "grad_norm": 1.1310773657139135,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 17676
    },
    {
      "epoch": 0.17677,
      "grad_norm": 1.4442942484105357,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 17677
    },
    {
      "epoch": 0.17678,
      "grad_norm": 1.1969913176518006,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 17678
    },
    {
      "epoch": 0.17679,
      "grad_norm": 1.163658778160288,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 17679
    },
    {
      "epoch": 0.1768,
      "grad_norm": 1.1784312238863694,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 17680
    },
    {
      "epoch": 0.17681,
      "grad_norm": 1.4807219084128371,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 17681
    },
    {
      "epoch": 0.17682,
      "grad_norm": 1.3152398566596446,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 17682
    },
    {
      "epoch": 0.17683,
      "grad_norm": 1.3587381566270007,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 17683
    },
    {
      "epoch": 0.17684,
      "grad_norm": 1.4291855672899187,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 17684
    },
    {
      "epoch": 0.17685,
      "grad_norm": 1.299069865520517,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 17685
    },
    {
      "epoch": 0.17686,
      "grad_norm": 1.423640480642554,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 17686
    },
    {
      "epoch": 0.17687,
      "grad_norm": 1.0631995763892585,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 17687
    },
    {
      "epoch": 0.17688,
      "grad_norm": 1.434347203828858,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 17688
    },
    {
      "epoch": 0.17689,
      "grad_norm": 1.3244904318921638,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 17689
    },
    {
      "epoch": 0.1769,
      "grad_norm": 1.3999105356202497,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 17690
    },
    {
      "epoch": 0.17691,
      "grad_norm": 1.2456988748227091,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 17691
    },
    {
      "epoch": 0.17692,
      "grad_norm": 1.4356245225259319,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 17692
    },
    {
      "epoch": 0.17693,
      "grad_norm": 1.2895322006881962,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 17693
    },
    {
      "epoch": 0.17694,
      "grad_norm": 1.5517959275513893,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 17694
    },
    {
      "epoch": 0.17695,
      "grad_norm": 1.1354139058927026,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 17695
    },
    {
      "epoch": 0.17696,
      "grad_norm": 1.6630191899167326,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17696
    },
    {
      "epoch": 0.17697,
      "grad_norm": 1.067344928888524,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 17697
    },
    {
      "epoch": 0.17698,
      "grad_norm": 1.272823799545181,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 17698
    },
    {
      "epoch": 0.17699,
      "grad_norm": 1.4013209199376062,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 17699
    },
    {
      "epoch": 0.177,
      "grad_norm": 1.1742128784613832,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 17700
    },
    {
      "epoch": 0.17701,
      "grad_norm": 1.2639450997363644,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 17701
    },
    {
      "epoch": 0.17702,
      "grad_norm": 1.3879760637351526,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 17702
    },
    {
      "epoch": 0.17703,
      "grad_norm": 1.1270593312215273,
      "learning_rate": 0.003,
      "loss": 4.0177,
      "step": 17703
    },
    {
      "epoch": 0.17704,
      "grad_norm": 1.3320730940804644,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 17704
    },
    {
      "epoch": 0.17705,
      "grad_norm": 1.280742530407232,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 17705
    },
    {
      "epoch": 0.17706,
      "grad_norm": 1.242758770219724,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 17706
    },
    {
      "epoch": 0.17707,
      "grad_norm": 1.106987527962389,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 17707
    },
    {
      "epoch": 0.17708,
      "grad_norm": 1.4048896644701068,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 17708
    },
    {
      "epoch": 0.17709,
      "grad_norm": 1.4607143220839507,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 17709
    },
    {
      "epoch": 0.1771,
      "grad_norm": 1.2458144205557176,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 17710
    },
    {
      "epoch": 0.17711,
      "grad_norm": 1.2852096577321068,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 17711
    },
    {
      "epoch": 0.17712,
      "grad_norm": 1.6120701938973683,
      "learning_rate": 0.003,
      "loss": 4.0322,
      "step": 17712
    },
    {
      "epoch": 0.17713,
      "grad_norm": 1.1268323134544185,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 17713
    },
    {
      "epoch": 0.17714,
      "grad_norm": 1.3765118008965018,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 17714
    },
    {
      "epoch": 0.17715,
      "grad_norm": 1.2223185178695093,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 17715
    },
    {
      "epoch": 0.17716,
      "grad_norm": 1.1942437661947458,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 17716
    },
    {
      "epoch": 0.17717,
      "grad_norm": 1.348931352784379,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 17717
    },
    {
      "epoch": 0.17718,
      "grad_norm": 0.9868431322378017,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 17718
    },
    {
      "epoch": 0.17719,
      "grad_norm": 1.506214835504423,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 17719
    },
    {
      "epoch": 0.1772,
      "grad_norm": 1.2017136329561915,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 17720
    },
    {
      "epoch": 0.17721,
      "grad_norm": 1.3727715141245602,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 17721
    },
    {
      "epoch": 0.17722,
      "grad_norm": 1.3819746634324437,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 17722
    },
    {
      "epoch": 0.17723,
      "grad_norm": 1.4971091150145805,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 17723
    },
    {
      "epoch": 0.17724,
      "grad_norm": 1.1448220379209995,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17724
    },
    {
      "epoch": 0.17725,
      "grad_norm": 1.3090864820025123,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 17725
    },
    {
      "epoch": 0.17726,
      "grad_norm": 1.416815157664905,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17726
    },
    {
      "epoch": 0.17727,
      "grad_norm": 1.3592864538826421,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 17727
    },
    {
      "epoch": 0.17728,
      "grad_norm": 1.181017492502079,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 17728
    },
    {
      "epoch": 0.17729,
      "grad_norm": 1.416290189676951,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 17729
    },
    {
      "epoch": 0.1773,
      "grad_norm": 1.2872245898560697,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 17730
    },
    {
      "epoch": 0.17731,
      "grad_norm": 1.1322378090274587,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 17731
    },
    {
      "epoch": 0.17732,
      "grad_norm": 1.4674360771390844,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 17732
    },
    {
      "epoch": 0.17733,
      "grad_norm": 1.3912855207092103,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 17733
    },
    {
      "epoch": 0.17734,
      "grad_norm": 1.375396481262623,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 17734
    },
    {
      "epoch": 0.17735,
      "grad_norm": 1.0862521365725795,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 17735
    },
    {
      "epoch": 0.17736,
      "grad_norm": 1.3900815444433379,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 17736
    },
    {
      "epoch": 0.17737,
      "grad_norm": 1.1711284664780262,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 17737
    },
    {
      "epoch": 0.17738,
      "grad_norm": 1.2425651281137042,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 17738
    },
    {
      "epoch": 0.17739,
      "grad_norm": 1.3059184746608743,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 17739
    },
    {
      "epoch": 0.1774,
      "grad_norm": 1.3011870231355265,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 17740
    },
    {
      "epoch": 0.17741,
      "grad_norm": 1.2535436653602747,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 17741
    },
    {
      "epoch": 0.17742,
      "grad_norm": 1.372881049562413,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 17742
    },
    {
      "epoch": 0.17743,
      "grad_norm": 1.737441794190293,
      "learning_rate": 0.003,
      "loss": 4.1109,
      "step": 17743
    },
    {
      "epoch": 0.17744,
      "grad_norm": 1.3626963374777032,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 17744
    },
    {
      "epoch": 0.17745,
      "grad_norm": 1.133048682635455,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 17745
    },
    {
      "epoch": 0.17746,
      "grad_norm": 1.5952755971146844,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 17746
    },
    {
      "epoch": 0.17747,
      "grad_norm": 1.1208706148409617,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 17747
    },
    {
      "epoch": 0.17748,
      "grad_norm": 1.4713550224114142,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 17748
    },
    {
      "epoch": 0.17749,
      "grad_norm": 1.1095058750339268,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 17749
    },
    {
      "epoch": 0.1775,
      "grad_norm": 1.2984318209968637,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 17750
    },
    {
      "epoch": 0.17751,
      "grad_norm": 1.4630332546250533,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 17751
    },
    {
      "epoch": 0.17752,
      "grad_norm": 1.1080067959121638,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 17752
    },
    {
      "epoch": 0.17753,
      "grad_norm": 1.1593676261260744,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 17753
    },
    {
      "epoch": 0.17754,
      "grad_norm": 1.2372894872821043,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 17754
    },
    {
      "epoch": 0.17755,
      "grad_norm": 1.258396401792125,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 17755
    },
    {
      "epoch": 0.17756,
      "grad_norm": 1.0485209910890672,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 17756
    },
    {
      "epoch": 0.17757,
      "grad_norm": 1.462847751511984,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 17757
    },
    {
      "epoch": 0.17758,
      "grad_norm": 1.145811857167575,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 17758
    },
    {
      "epoch": 0.17759,
      "grad_norm": 1.6684204600306993,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 17759
    },
    {
      "epoch": 0.1776,
      "grad_norm": 1.2550512015522435,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 17760
    },
    {
      "epoch": 0.17761,
      "grad_norm": 1.3213866534609735,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 17761
    },
    {
      "epoch": 0.17762,
      "grad_norm": 1.34841225621269,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 17762
    },
    {
      "epoch": 0.17763,
      "grad_norm": 1.4116761442066719,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17763
    },
    {
      "epoch": 0.17764,
      "grad_norm": 1.2992687258706004,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 17764
    },
    {
      "epoch": 0.17765,
      "grad_norm": 1.2747375822536242,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 17765
    },
    {
      "epoch": 0.17766,
      "grad_norm": 1.1038618602071801,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17766
    },
    {
      "epoch": 0.17767,
      "grad_norm": 1.4109619706095022,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 17767
    },
    {
      "epoch": 0.17768,
      "grad_norm": 1.2175900409790084,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 17768
    },
    {
      "epoch": 0.17769,
      "grad_norm": 1.2926353182322496,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 17769
    },
    {
      "epoch": 0.1777,
      "grad_norm": 1.243801243216156,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 17770
    },
    {
      "epoch": 0.17771,
      "grad_norm": 1.2368272296033915,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 17771
    },
    {
      "epoch": 0.17772,
      "grad_norm": 1.2199582482353482,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 17772
    },
    {
      "epoch": 0.17773,
      "grad_norm": 1.1402358760709792,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 17773
    },
    {
      "epoch": 0.17774,
      "grad_norm": 1.6157640077166686,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 17774
    },
    {
      "epoch": 0.17775,
      "grad_norm": 1.0227908232124825,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 17775
    },
    {
      "epoch": 0.17776,
      "grad_norm": 1.4092199269502843,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 17776
    },
    {
      "epoch": 0.17777,
      "grad_norm": 1.2584085745914688,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 17777
    },
    {
      "epoch": 0.17778,
      "grad_norm": 1.4731279422782921,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 17778
    },
    {
      "epoch": 0.17779,
      "grad_norm": 1.4075459756405722,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 17779
    },
    {
      "epoch": 0.1778,
      "grad_norm": 1.1224644816175051,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 17780
    },
    {
      "epoch": 0.17781,
      "grad_norm": 1.4278592701354988,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 17781
    },
    {
      "epoch": 0.17782,
      "grad_norm": 1.4387534531644868,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 17782
    },
    {
      "epoch": 0.17783,
      "grad_norm": 1.2692227755173564,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 17783
    },
    {
      "epoch": 0.17784,
      "grad_norm": 1.3682848068132636,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 17784
    },
    {
      "epoch": 0.17785,
      "grad_norm": 1.169166375284477,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 17785
    },
    {
      "epoch": 0.17786,
      "grad_norm": 1.3169217438370486,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 17786
    },
    {
      "epoch": 0.17787,
      "grad_norm": 1.1376413048380454,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 17787
    },
    {
      "epoch": 0.17788,
      "grad_norm": 1.7124592978411066,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 17788
    },
    {
      "epoch": 0.17789,
      "grad_norm": 1.2745510693179796,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 17789
    },
    {
      "epoch": 0.1779,
      "grad_norm": 1.447901906431428,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 17790
    },
    {
      "epoch": 0.17791,
      "grad_norm": 1.3159462136060966,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 17791
    },
    {
      "epoch": 0.17792,
      "grad_norm": 1.1303918467644893,
      "learning_rate": 0.003,
      "loss": 4.0294,
      "step": 17792
    },
    {
      "epoch": 0.17793,
      "grad_norm": 1.4418297651455436,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 17793
    },
    {
      "epoch": 0.17794,
      "grad_norm": 1.150072467165776,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17794
    },
    {
      "epoch": 0.17795,
      "grad_norm": 1.2882828435216742,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 17795
    },
    {
      "epoch": 0.17796,
      "grad_norm": 1.5300047434012467,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 17796
    },
    {
      "epoch": 0.17797,
      "grad_norm": 1.294882456997206,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 17797
    },
    {
      "epoch": 0.17798,
      "grad_norm": 1.1967947727127546,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 17798
    },
    {
      "epoch": 0.17799,
      "grad_norm": 1.3078744578101413,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 17799
    },
    {
      "epoch": 0.178,
      "grad_norm": 1.2823202298045382,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 17800
    },
    {
      "epoch": 0.17801,
      "grad_norm": 1.138629563832294,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 17801
    },
    {
      "epoch": 0.17802,
      "grad_norm": 1.2476615667142883,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 17802
    },
    {
      "epoch": 0.17803,
      "grad_norm": 1.3380044295600648,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 17803
    },
    {
      "epoch": 0.17804,
      "grad_norm": 1.3769474905486965,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 17804
    },
    {
      "epoch": 0.17805,
      "grad_norm": 1.144291069521152,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 17805
    },
    {
      "epoch": 0.17806,
      "grad_norm": 1.0981636028409751,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 17806
    },
    {
      "epoch": 0.17807,
      "grad_norm": 1.3824357220871897,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 17807
    },
    {
      "epoch": 0.17808,
      "grad_norm": 1.2373087518469619,
      "learning_rate": 0.003,
      "loss": 4.0293,
      "step": 17808
    },
    {
      "epoch": 0.17809,
      "grad_norm": 1.389537481714246,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 17809
    },
    {
      "epoch": 0.1781,
      "grad_norm": 1.1752206957419395,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 17810
    },
    {
      "epoch": 0.17811,
      "grad_norm": 1.509089328615908,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 17811
    },
    {
      "epoch": 0.17812,
      "grad_norm": 1.1544782383342644,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 17812
    },
    {
      "epoch": 0.17813,
      "grad_norm": 1.3880160019556476,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 17813
    },
    {
      "epoch": 0.17814,
      "grad_norm": 1.1144387814214616,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 17814
    },
    {
      "epoch": 0.17815,
      "grad_norm": 1.3763135851316985,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 17815
    },
    {
      "epoch": 0.17816,
      "grad_norm": 1.1987028557509924,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 17816
    },
    {
      "epoch": 0.17817,
      "grad_norm": 1.3212547591704311,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 17817
    },
    {
      "epoch": 0.17818,
      "grad_norm": 1.030098384672589,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 17818
    },
    {
      "epoch": 0.17819,
      "grad_norm": 1.4627642966562506,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 17819
    },
    {
      "epoch": 0.1782,
      "grad_norm": 1.2236767766115737,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 17820
    },
    {
      "epoch": 0.17821,
      "grad_norm": 1.2624109141416808,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 17821
    },
    {
      "epoch": 0.17822,
      "grad_norm": 1.1786888792865688,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 17822
    },
    {
      "epoch": 0.17823,
      "grad_norm": 1.0954014569065735,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 17823
    },
    {
      "epoch": 0.17824,
      "grad_norm": 1.406971076687692,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 17824
    },
    {
      "epoch": 0.17825,
      "grad_norm": 1.191252071858672,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 17825
    },
    {
      "epoch": 0.17826,
      "grad_norm": 1.1932691508802327,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 17826
    },
    {
      "epoch": 0.17827,
      "grad_norm": 1.3445866454330102,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 17827
    },
    {
      "epoch": 0.17828,
      "grad_norm": 0.9395973104612575,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 17828
    },
    {
      "epoch": 0.17829,
      "grad_norm": 1.4964471425613408,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 17829
    },
    {
      "epoch": 0.1783,
      "grad_norm": 1.2770590995250022,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 17830
    },
    {
      "epoch": 0.17831,
      "grad_norm": 1.7767237762098842,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 17831
    },
    {
      "epoch": 0.17832,
      "grad_norm": 1.560627635031259,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 17832
    },
    {
      "epoch": 0.17833,
      "grad_norm": 1.1862234180221531,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 17833
    },
    {
      "epoch": 0.17834,
      "grad_norm": 1.3094702246929275,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 17834
    },
    {
      "epoch": 0.17835,
      "grad_norm": 1.5248087017840872,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 17835
    },
    {
      "epoch": 0.17836,
      "grad_norm": 1.3169564845416029,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 17836
    },
    {
      "epoch": 0.17837,
      "grad_norm": 1.2675087087992958,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 17837
    },
    {
      "epoch": 0.17838,
      "grad_norm": 1.2709712053086606,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 17838
    },
    {
      "epoch": 0.17839,
      "grad_norm": 1.2522036966190369,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 17839
    },
    {
      "epoch": 0.1784,
      "grad_norm": 1.2620829344617028,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 17840
    },
    {
      "epoch": 0.17841,
      "grad_norm": 1.4012932938037403,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 17841
    },
    {
      "epoch": 0.17842,
      "grad_norm": 1.3422000839817314,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 17842
    },
    {
      "epoch": 0.17843,
      "grad_norm": 1.4006989988827396,
      "learning_rate": 0.003,
      "loss": 4.0298,
      "step": 17843
    },
    {
      "epoch": 0.17844,
      "grad_norm": 1.0190924565172015,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 17844
    },
    {
      "epoch": 0.17845,
      "grad_norm": 1.5054749591757886,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 17845
    },
    {
      "epoch": 0.17846,
      "grad_norm": 1.0716919117744008,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 17846
    },
    {
      "epoch": 0.17847,
      "grad_norm": 1.4316572708728408,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 17847
    },
    {
      "epoch": 0.17848,
      "grad_norm": 1.222068982447068,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 17848
    },
    {
      "epoch": 0.17849,
      "grad_norm": 1.7153759550487508,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 17849
    },
    {
      "epoch": 0.1785,
      "grad_norm": 1.2297861316703975,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 17850
    },
    {
      "epoch": 0.17851,
      "grad_norm": 1.4649983039897283,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 17851
    },
    {
      "epoch": 0.17852,
      "grad_norm": 1.4020148957301672,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 17852
    },
    {
      "epoch": 0.17853,
      "grad_norm": 1.2346682112425613,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 17853
    },
    {
      "epoch": 0.17854,
      "grad_norm": 1.106188497254593,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 17854
    },
    {
      "epoch": 0.17855,
      "grad_norm": 1.2493336816231602,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 17855
    },
    {
      "epoch": 0.17856,
      "grad_norm": 1.179738117162747,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 17856
    },
    {
      "epoch": 0.17857,
      "grad_norm": 1.5935205634776302,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 17857
    },
    {
      "epoch": 0.17858,
      "grad_norm": 1.3713885979408045,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 17858
    },
    {
      "epoch": 0.17859,
      "grad_norm": 1.2860808614893842,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 17859
    },
    {
      "epoch": 0.1786,
      "grad_norm": 1.1659456901295933,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 17860
    },
    {
      "epoch": 0.17861,
      "grad_norm": 1.2412258951354143,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 17861
    },
    {
      "epoch": 0.17862,
      "grad_norm": 1.2206348088811467,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 17862
    },
    {
      "epoch": 0.17863,
      "grad_norm": 1.2714996956301137,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 17863
    },
    {
      "epoch": 0.17864,
      "grad_norm": 1.2282390076542682,
      "learning_rate": 0.003,
      "loss": 4.0906,
      "step": 17864
    },
    {
      "epoch": 0.17865,
      "grad_norm": 1.2259074729816142,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 17865
    },
    {
      "epoch": 0.17866,
      "grad_norm": 1.3227112430040862,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17866
    },
    {
      "epoch": 0.17867,
      "grad_norm": 1.2388192275713554,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 17867
    },
    {
      "epoch": 0.17868,
      "grad_norm": 1.36726409734838,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 17868
    },
    {
      "epoch": 0.17869,
      "grad_norm": 1.2070589878757991,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 17869
    },
    {
      "epoch": 0.1787,
      "grad_norm": 1.3158716539193773,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 17870
    },
    {
      "epoch": 0.17871,
      "grad_norm": 1.1728661622004941,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 17871
    },
    {
      "epoch": 0.17872,
      "grad_norm": 1.2427782053645458,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 17872
    },
    {
      "epoch": 0.17873,
      "grad_norm": 1.4042019950491103,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 17873
    },
    {
      "epoch": 0.17874,
      "grad_norm": 1.5812638478605645,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 17874
    },
    {
      "epoch": 0.17875,
      "grad_norm": 1.3564451180031354,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 17875
    },
    {
      "epoch": 0.17876,
      "grad_norm": 1.2726006566024621,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 17876
    },
    {
      "epoch": 0.17877,
      "grad_norm": 1.286756604906063,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 17877
    },
    {
      "epoch": 0.17878,
      "grad_norm": 1.3630425772043697,
      "learning_rate": 0.003,
      "loss": 4.0273,
      "step": 17878
    },
    {
      "epoch": 0.17879,
      "grad_norm": 1.3858094253305993,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 17879
    },
    {
      "epoch": 0.1788,
      "grad_norm": 1.0266998705945807,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 17880
    },
    {
      "epoch": 0.17881,
      "grad_norm": 1.5182973033573133,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 17881
    },
    {
      "epoch": 0.17882,
      "grad_norm": 0.9844433945176064,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 17882
    },
    {
      "epoch": 0.17883,
      "grad_norm": 1.501822512948681,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 17883
    },
    {
      "epoch": 0.17884,
      "grad_norm": 1.3255113384277992,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 17884
    },
    {
      "epoch": 0.17885,
      "grad_norm": 1.198685716640698,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 17885
    },
    {
      "epoch": 0.17886,
      "grad_norm": 1.3021750977825552,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 17886
    },
    {
      "epoch": 0.17887,
      "grad_norm": 1.3747917696887602,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 17887
    },
    {
      "epoch": 0.17888,
      "grad_norm": 1.2841788620965964,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 17888
    },
    {
      "epoch": 0.17889,
      "grad_norm": 1.1959348659836255,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 17889
    },
    {
      "epoch": 0.1789,
      "grad_norm": 1.6015235031736559,
      "learning_rate": 0.003,
      "loss": 4.0973,
      "step": 17890
    },
    {
      "epoch": 0.17891,
      "grad_norm": 1.0814896382172312,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 17891
    },
    {
      "epoch": 0.17892,
      "grad_norm": 1.447675696751703,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 17892
    },
    {
      "epoch": 0.17893,
      "grad_norm": 1.223293511998367,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 17893
    },
    {
      "epoch": 0.17894,
      "grad_norm": 1.4438945880341931,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 17894
    },
    {
      "epoch": 0.17895,
      "grad_norm": 1.2821740281097793,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 17895
    },
    {
      "epoch": 0.17896,
      "grad_norm": 1.3215919068687485,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 17896
    },
    {
      "epoch": 0.17897,
      "grad_norm": 1.3140664733236573,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 17897
    },
    {
      "epoch": 0.17898,
      "grad_norm": 1.1907256660849348,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 17898
    },
    {
      "epoch": 0.17899,
      "grad_norm": 1.3430593063476148,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 17899
    },
    {
      "epoch": 0.179,
      "grad_norm": 1.286488117104526,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 17900
    },
    {
      "epoch": 0.17901,
      "grad_norm": 1.0695589964441206,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 17901
    },
    {
      "epoch": 0.17902,
      "grad_norm": 1.3224104997066575,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 17902
    },
    {
      "epoch": 0.17903,
      "grad_norm": 1.0988145150369284,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 17903
    },
    {
      "epoch": 0.17904,
      "grad_norm": 1.5340834757275605,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 17904
    },
    {
      "epoch": 0.17905,
      "grad_norm": 1.0051242394701418,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 17905
    },
    {
      "epoch": 0.17906,
      "grad_norm": 1.7306028260938866,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 17906
    },
    {
      "epoch": 0.17907,
      "grad_norm": 1.3411396241899942,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 17907
    },
    {
      "epoch": 0.17908,
      "grad_norm": 1.1706475143882584,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 17908
    },
    {
      "epoch": 0.17909,
      "grad_norm": 1.3571135714578095,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 17909
    },
    {
      "epoch": 0.1791,
      "grad_norm": 1.1324149438723727,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 17910
    },
    {
      "epoch": 0.17911,
      "grad_norm": 1.320902031192732,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 17911
    },
    {
      "epoch": 0.17912,
      "grad_norm": 1.2994613905078916,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 17912
    },
    {
      "epoch": 0.17913,
      "grad_norm": 1.3291750239753504,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 17913
    },
    {
      "epoch": 0.17914,
      "grad_norm": 1.172412070111314,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 17914
    },
    {
      "epoch": 0.17915,
      "grad_norm": 1.5141488612900191,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 17915
    },
    {
      "epoch": 0.17916,
      "grad_norm": 1.2874531966763487,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 17916
    },
    {
      "epoch": 0.17917,
      "grad_norm": 1.7777555042118187,
      "learning_rate": 0.003,
      "loss": 4.0982,
      "step": 17917
    },
    {
      "epoch": 0.17918,
      "grad_norm": 0.9335876254074895,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 17918
    },
    {
      "epoch": 0.17919,
      "grad_norm": 1.2730438456312825,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 17919
    },
    {
      "epoch": 0.1792,
      "grad_norm": 1.4749867341353853,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 17920
    },
    {
      "epoch": 0.17921,
      "grad_norm": 1.3918863739506826,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 17921
    },
    {
      "epoch": 0.17922,
      "grad_norm": 1.3448413037065396,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 17922
    },
    {
      "epoch": 0.17923,
      "grad_norm": 1.4183796265762036,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 17923
    },
    {
      "epoch": 0.17924,
      "grad_norm": 0.9052872201481025,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 17924
    },
    {
      "epoch": 0.17925,
      "grad_norm": 1.3920930596029024,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 17925
    },
    {
      "epoch": 0.17926,
      "grad_norm": 1.2048862742257411,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 17926
    },
    {
      "epoch": 0.17927,
      "grad_norm": 1.5643594677964943,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 17927
    },
    {
      "epoch": 0.17928,
      "grad_norm": 1.3310583454415035,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 17928
    },
    {
      "epoch": 0.17929,
      "grad_norm": 1.4820607233171916,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 17929
    },
    {
      "epoch": 0.1793,
      "grad_norm": 1.1100563858479453,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 17930
    },
    {
      "epoch": 0.17931,
      "grad_norm": 1.3002000108662704,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 17931
    },
    {
      "epoch": 0.17932,
      "grad_norm": 1.2290679867347463,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 17932
    },
    {
      "epoch": 0.17933,
      "grad_norm": 1.251852641559448,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 17933
    },
    {
      "epoch": 0.17934,
      "grad_norm": 1.3677317247454568,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 17934
    },
    {
      "epoch": 0.17935,
      "grad_norm": 1.2576152143696935,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 17935
    },
    {
      "epoch": 0.17936,
      "grad_norm": 1.1526586197354032,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 17936
    },
    {
      "epoch": 0.17937,
      "grad_norm": 1.3564977878604803,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 17937
    },
    {
      "epoch": 0.17938,
      "grad_norm": 1.297570645938877,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 17938
    },
    {
      "epoch": 0.17939,
      "grad_norm": 1.2538252403226684,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 17939
    },
    {
      "epoch": 0.1794,
      "grad_norm": 1.4210528432483376,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 17940
    },
    {
      "epoch": 0.17941,
      "grad_norm": 1.29711148921775,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 17941
    },
    {
      "epoch": 0.17942,
      "grad_norm": 1.257774089270801,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 17942
    },
    {
      "epoch": 0.17943,
      "grad_norm": 1.3189931995828559,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 17943
    },
    {
      "epoch": 0.17944,
      "grad_norm": 1.5827618271151869,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 17944
    },
    {
      "epoch": 0.17945,
      "grad_norm": 1.221743354547214,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 17945
    },
    {
      "epoch": 0.17946,
      "grad_norm": 1.4848743350684563,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 17946
    },
    {
      "epoch": 0.17947,
      "grad_norm": 1.096557597019516,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 17947
    },
    {
      "epoch": 0.17948,
      "grad_norm": 1.4094750081051672,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 17948
    },
    {
      "epoch": 0.17949,
      "grad_norm": 1.1307403812560006,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 17949
    },
    {
      "epoch": 0.1795,
      "grad_norm": 1.2544747953601025,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 17950
    },
    {
      "epoch": 0.17951,
      "grad_norm": 1.0712320237750856,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 17951
    },
    {
      "epoch": 0.17952,
      "grad_norm": 1.457389803142145,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 17952
    },
    {
      "epoch": 0.17953,
      "grad_norm": 1.2058223116981568,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 17953
    },
    {
      "epoch": 0.17954,
      "grad_norm": 1.3707871230133384,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 17954
    },
    {
      "epoch": 0.17955,
      "grad_norm": 1.4545497481766416,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 17955
    },
    {
      "epoch": 0.17956,
      "grad_norm": 1.5444783073178818,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 17956
    },
    {
      "epoch": 0.17957,
      "grad_norm": 1.2718200438362388,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 17957
    },
    {
      "epoch": 0.17958,
      "grad_norm": 1.3896879812577554,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 17958
    },
    {
      "epoch": 0.17959,
      "grad_norm": 1.1996278596396401,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 17959
    },
    {
      "epoch": 0.1796,
      "grad_norm": 1.276239698502112,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 17960
    },
    {
      "epoch": 0.17961,
      "grad_norm": 1.1378747572368832,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 17961
    },
    {
      "epoch": 0.17962,
      "grad_norm": 1.2135863102985003,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 17962
    },
    {
      "epoch": 0.17963,
      "grad_norm": 1.3228436626369233,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 17963
    },
    {
      "epoch": 0.17964,
      "grad_norm": 1.3254583377258655,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 17964
    },
    {
      "epoch": 0.17965,
      "grad_norm": 1.3328612987006105,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 17965
    },
    {
      "epoch": 0.17966,
      "grad_norm": 1.560116283233121,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 17966
    },
    {
      "epoch": 0.17967,
      "grad_norm": 1.315203801181062,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 17967
    },
    {
      "epoch": 0.17968,
      "grad_norm": 1.4446968555287374,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 17968
    },
    {
      "epoch": 0.17969,
      "grad_norm": 1.2675016047828012,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 17969
    },
    {
      "epoch": 0.1797,
      "grad_norm": 1.4689414019046123,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 17970
    },
    {
      "epoch": 0.17971,
      "grad_norm": 1.2425913769483934,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 17971
    },
    {
      "epoch": 0.17972,
      "grad_norm": 1.103654538103382,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 17972
    },
    {
      "epoch": 0.17973,
      "grad_norm": 1.3610423718319988,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 17973
    },
    {
      "epoch": 0.17974,
      "grad_norm": 1.2302631753766913,
      "learning_rate": 0.003,
      "loss": 4.0196,
      "step": 17974
    },
    {
      "epoch": 0.17975,
      "grad_norm": 1.608979570575041,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 17975
    },
    {
      "epoch": 0.17976,
      "grad_norm": 1.0055260741776726,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 17976
    },
    {
      "epoch": 0.17977,
      "grad_norm": 1.431267475756132,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 17977
    },
    {
      "epoch": 0.17978,
      "grad_norm": 1.2389833984104677,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 17978
    },
    {
      "epoch": 0.17979,
      "grad_norm": 1.228422133067524,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 17979
    },
    {
      "epoch": 0.1798,
      "grad_norm": 1.2010273619214957,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 17980
    },
    {
      "epoch": 0.17981,
      "grad_norm": 1.2308198217935225,
      "learning_rate": 0.003,
      "loss": 3.9907,
      "step": 17981
    },
    {
      "epoch": 0.17982,
      "grad_norm": 1.323924689196548,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 17982
    },
    {
      "epoch": 0.17983,
      "grad_norm": 1.1901092782957963,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 17983
    },
    {
      "epoch": 0.17984,
      "grad_norm": 1.2999788528968683,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 17984
    },
    {
      "epoch": 0.17985,
      "grad_norm": 1.2620420541392119,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 17985
    },
    {
      "epoch": 0.17986,
      "grad_norm": 1.1734711827379214,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 17986
    },
    {
      "epoch": 0.17987,
      "grad_norm": 1.3278895027861535,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 17987
    },
    {
      "epoch": 0.17988,
      "grad_norm": 1.3501424373966742,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 17988
    },
    {
      "epoch": 0.17989,
      "grad_norm": 1.1354312105079478,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 17989
    },
    {
      "epoch": 0.1799,
      "grad_norm": 1.329559896487889,
      "learning_rate": 0.003,
      "loss": 4.0954,
      "step": 17990
    },
    {
      "epoch": 0.17991,
      "grad_norm": 1.4349224235104836,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 17991
    },
    {
      "epoch": 0.17992,
      "grad_norm": 1.356964151271892,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 17992
    },
    {
      "epoch": 0.17993,
      "grad_norm": 1.5299973250772203,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 17993
    },
    {
      "epoch": 0.17994,
      "grad_norm": 1.3453109033511206,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 17994
    },
    {
      "epoch": 0.17995,
      "grad_norm": 1.4934737887896772,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 17995
    },
    {
      "epoch": 0.17996,
      "grad_norm": 1.2869454407391248,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 17996
    },
    {
      "epoch": 0.17997,
      "grad_norm": 1.1682327411401674,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 17997
    },
    {
      "epoch": 0.17998,
      "grad_norm": 1.5491421161896177,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 17998
    },
    {
      "epoch": 0.17999,
      "grad_norm": 1.0293246193267205,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 17999
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5678489562289901,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18000
    },
    {
      "epoch": 0.18001,
      "grad_norm": 1.1834753468106012,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 18001
    },
    {
      "epoch": 0.18002,
      "grad_norm": 1.6715971464408366,
      "learning_rate": 0.003,
      "loss": 4.0209,
      "step": 18002
    },
    {
      "epoch": 0.18003,
      "grad_norm": 1.4474001820058198,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 18003
    },
    {
      "epoch": 0.18004,
      "grad_norm": 1.1168831694904024,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18004
    },
    {
      "epoch": 0.18005,
      "grad_norm": 1.3800372860497383,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 18005
    },
    {
      "epoch": 0.18006,
      "grad_norm": 1.2425220773680559,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 18006
    },
    {
      "epoch": 0.18007,
      "grad_norm": 1.1991129047475015,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 18007
    },
    {
      "epoch": 0.18008,
      "grad_norm": 1.391553308009128,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 18008
    },
    {
      "epoch": 0.18009,
      "grad_norm": 1.3255645061052017,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 18009
    },
    {
      "epoch": 0.1801,
      "grad_norm": 1.2206631167682767,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 18010
    },
    {
      "epoch": 0.18011,
      "grad_norm": 1.1316284763305127,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 18011
    },
    {
      "epoch": 0.18012,
      "grad_norm": 1.2504797489669202,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 18012
    },
    {
      "epoch": 0.18013,
      "grad_norm": 1.2413159854996367,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 18013
    },
    {
      "epoch": 0.18014,
      "grad_norm": 1.2200518278211834,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 18014
    },
    {
      "epoch": 0.18015,
      "grad_norm": 1.261611278686195,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 18015
    },
    {
      "epoch": 0.18016,
      "grad_norm": 1.2414894702918717,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 18016
    },
    {
      "epoch": 0.18017,
      "grad_norm": 1.2697807333924054,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 18017
    },
    {
      "epoch": 0.18018,
      "grad_norm": 1.2761231347528381,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 18018
    },
    {
      "epoch": 0.18019,
      "grad_norm": 1.4608018721023053,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 18019
    },
    {
      "epoch": 0.1802,
      "grad_norm": 1.1142836402680227,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 18020
    },
    {
      "epoch": 0.18021,
      "grad_norm": 1.2716124604703851,
      "learning_rate": 0.003,
      "loss": 4.0894,
      "step": 18021
    },
    {
      "epoch": 0.18022,
      "grad_norm": 1.374731629938203,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 18022
    },
    {
      "epoch": 0.18023,
      "grad_norm": 1.451811766902569,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 18023
    },
    {
      "epoch": 0.18024,
      "grad_norm": 1.4886827226216808,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 18024
    },
    {
      "epoch": 0.18025,
      "grad_norm": 1.195325145255202,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 18025
    },
    {
      "epoch": 0.18026,
      "grad_norm": 1.5663069281027013,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 18026
    },
    {
      "epoch": 0.18027,
      "grad_norm": 1.1654133475363146,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 18027
    },
    {
      "epoch": 0.18028,
      "grad_norm": 1.4276273643981463,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 18028
    },
    {
      "epoch": 0.18029,
      "grad_norm": 1.0649990139058456,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 18029
    },
    {
      "epoch": 0.1803,
      "grad_norm": 1.4174487375068698,
      "learning_rate": 0.003,
      "loss": 4.0924,
      "step": 18030
    },
    {
      "epoch": 0.18031,
      "grad_norm": 1.1292646849660204,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 18031
    },
    {
      "epoch": 0.18032,
      "grad_norm": 1.4413185788727525,
      "learning_rate": 0.003,
      "loss": 4.1304,
      "step": 18032
    },
    {
      "epoch": 0.18033,
      "grad_norm": 1.093605221382924,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 18033
    },
    {
      "epoch": 0.18034,
      "grad_norm": 1.7307527217592507,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 18034
    },
    {
      "epoch": 0.18035,
      "grad_norm": 1.2898895378608528,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 18035
    },
    {
      "epoch": 0.18036,
      "grad_norm": 1.3536138462937466,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 18036
    },
    {
      "epoch": 0.18037,
      "grad_norm": 1.1937189446306715,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 18037
    },
    {
      "epoch": 0.18038,
      "grad_norm": 1.656402603048207,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 18038
    },
    {
      "epoch": 0.18039,
      "grad_norm": 1.2231668587819087,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 18039
    },
    {
      "epoch": 0.1804,
      "grad_norm": 1.3645169199537732,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 18040
    },
    {
      "epoch": 0.18041,
      "grad_norm": 1.4515925031452472,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 18041
    },
    {
      "epoch": 0.18042,
      "grad_norm": 1.3464220796376452,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 18042
    },
    {
      "epoch": 0.18043,
      "grad_norm": 1.2258874359395866,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 18043
    },
    {
      "epoch": 0.18044,
      "grad_norm": 1.4401874470380218,
      "learning_rate": 0.003,
      "loss": 4.109,
      "step": 18044
    },
    {
      "epoch": 0.18045,
      "grad_norm": 1.1510208774316257,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 18045
    },
    {
      "epoch": 0.18046,
      "grad_norm": 1.1959196359638038,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 18046
    },
    {
      "epoch": 0.18047,
      "grad_norm": 1.276908581818575,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 18047
    },
    {
      "epoch": 0.18048,
      "grad_norm": 1.2791496037211918,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 18048
    },
    {
      "epoch": 0.18049,
      "grad_norm": 1.481903853376919,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 18049
    },
    {
      "epoch": 0.1805,
      "grad_norm": 1.1366849161073074,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 18050
    },
    {
      "epoch": 0.18051,
      "grad_norm": 1.66043727102517,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 18051
    },
    {
      "epoch": 0.18052,
      "grad_norm": 1.1072219344214962,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 18052
    },
    {
      "epoch": 0.18053,
      "grad_norm": 1.384367018671037,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 18053
    },
    {
      "epoch": 0.18054,
      "grad_norm": 1.1999621726418512,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 18054
    },
    {
      "epoch": 0.18055,
      "grad_norm": 1.2667827934525069,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 18055
    },
    {
      "epoch": 0.18056,
      "grad_norm": 1.301159822239713,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 18056
    },
    {
      "epoch": 0.18057,
      "grad_norm": 1.3779888030283471,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 18057
    },
    {
      "epoch": 0.18058,
      "grad_norm": 1.3066880666788008,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 18058
    },
    {
      "epoch": 0.18059,
      "grad_norm": 1.0550051003911673,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 18059
    },
    {
      "epoch": 0.1806,
      "grad_norm": 1.4478437157575914,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 18060
    },
    {
      "epoch": 0.18061,
      "grad_norm": 1.1663387043785898,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 18061
    },
    {
      "epoch": 0.18062,
      "grad_norm": 1.4054634477600398,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 18062
    },
    {
      "epoch": 0.18063,
      "grad_norm": 1.2252286810544568,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 18063
    },
    {
      "epoch": 0.18064,
      "grad_norm": 1.517116308704269,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 18064
    },
    {
      "epoch": 0.18065,
      "grad_norm": 1.1897742736690238,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 18065
    },
    {
      "epoch": 0.18066,
      "grad_norm": 1.7271267226074312,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 18066
    },
    {
      "epoch": 0.18067,
      "grad_norm": 1.2051866040119912,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 18067
    },
    {
      "epoch": 0.18068,
      "grad_norm": 1.1611631120066888,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 18068
    },
    {
      "epoch": 0.18069,
      "grad_norm": 1.527347211410476,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 18069
    },
    {
      "epoch": 0.1807,
      "grad_norm": 1.0483082392759384,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 18070
    },
    {
      "epoch": 0.18071,
      "grad_norm": 1.825337932151561,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 18071
    },
    {
      "epoch": 0.18072,
      "grad_norm": 1.3795901853936237,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 18072
    },
    {
      "epoch": 0.18073,
      "grad_norm": 1.2409011222770525,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 18073
    },
    {
      "epoch": 0.18074,
      "grad_norm": 1.2368957540212302,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 18074
    },
    {
      "epoch": 0.18075,
      "grad_norm": 1.2703828465389364,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 18075
    },
    {
      "epoch": 0.18076,
      "grad_norm": 1.2564309191832448,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 18076
    },
    {
      "epoch": 0.18077,
      "grad_norm": 1.2206101110957257,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 18077
    },
    {
      "epoch": 0.18078,
      "grad_norm": 1.407713569678958,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 18078
    },
    {
      "epoch": 0.18079,
      "grad_norm": 1.3661028245510936,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 18079
    },
    {
      "epoch": 0.1808,
      "grad_norm": 1.2258048402516895,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 18080
    },
    {
      "epoch": 0.18081,
      "grad_norm": 1.5843330540638398,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 18081
    },
    {
      "epoch": 0.18082,
      "grad_norm": 0.6936825374337909,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 18082
    },
    {
      "epoch": 0.18083,
      "grad_norm": 1.0248203663148996,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 18083
    },
    {
      "epoch": 0.18084,
      "grad_norm": 1.716197730295827,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 18084
    },
    {
      "epoch": 0.18085,
      "grad_norm": 0.9585141219870723,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 18085
    },
    {
      "epoch": 0.18086,
      "grad_norm": 1.3379443208903115,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 18086
    },
    {
      "epoch": 0.18087,
      "grad_norm": 1.2366463569268664,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 18087
    },
    {
      "epoch": 0.18088,
      "grad_norm": 1.2636307127605448,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 18088
    },
    {
      "epoch": 0.18089,
      "grad_norm": 1.4224497920411832,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 18089
    },
    {
      "epoch": 0.1809,
      "grad_norm": 1.1023749172205495,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 18090
    },
    {
      "epoch": 0.18091,
      "grad_norm": 1.2543315139083586,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 18091
    },
    {
      "epoch": 0.18092,
      "grad_norm": 1.1783772383280644,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 18092
    },
    {
      "epoch": 0.18093,
      "grad_norm": 1.5117848090462052,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 18093
    },
    {
      "epoch": 0.18094,
      "grad_norm": 1.4048702513973914,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 18094
    },
    {
      "epoch": 0.18095,
      "grad_norm": 1.2867382303272386,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 18095
    },
    {
      "epoch": 0.18096,
      "grad_norm": 1.0842882454502578,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 18096
    },
    {
      "epoch": 0.18097,
      "grad_norm": 1.402049280618989,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 18097
    },
    {
      "epoch": 0.18098,
      "grad_norm": 1.2740222906951524,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 18098
    },
    {
      "epoch": 0.18099,
      "grad_norm": 1.4092021306239473,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 18099
    },
    {
      "epoch": 0.181,
      "grad_norm": 1.212300283369773,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 18100
    },
    {
      "epoch": 0.18101,
      "grad_norm": 1.4631334539717482,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 18101
    },
    {
      "epoch": 0.18102,
      "grad_norm": 1.2044697301208913,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 18102
    },
    {
      "epoch": 0.18103,
      "grad_norm": 1.3150892284595763,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 18103
    },
    {
      "epoch": 0.18104,
      "grad_norm": 1.2506959429607367,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 18104
    },
    {
      "epoch": 0.18105,
      "grad_norm": 1.2323593849004972,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 18105
    },
    {
      "epoch": 0.18106,
      "grad_norm": 1.1983625716479727,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 18106
    },
    {
      "epoch": 0.18107,
      "grad_norm": 1.380401888137709,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 18107
    },
    {
      "epoch": 0.18108,
      "grad_norm": 1.4616032112616257,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 18108
    },
    {
      "epoch": 0.18109,
      "grad_norm": 1.2819611007798433,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 18109
    },
    {
      "epoch": 0.1811,
      "grad_norm": 1.1511736124574043,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 18110
    },
    {
      "epoch": 0.18111,
      "grad_norm": 1.6064113458997202,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 18111
    },
    {
      "epoch": 0.18112,
      "grad_norm": 1.0087387971819672,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 18112
    },
    {
      "epoch": 0.18113,
      "grad_norm": 1.6618416921990447,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 18113
    },
    {
      "epoch": 0.18114,
      "grad_norm": 1.2000846536824,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 18114
    },
    {
      "epoch": 0.18115,
      "grad_norm": 1.1682848970633601,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 18115
    },
    {
      "epoch": 0.18116,
      "grad_norm": 1.3772552377987377,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 18116
    },
    {
      "epoch": 0.18117,
      "grad_norm": 1.2928307550029645,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 18117
    },
    {
      "epoch": 0.18118,
      "grad_norm": 1.5027885344937797,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 18118
    },
    {
      "epoch": 0.18119,
      "grad_norm": 1.434871091486078,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 18119
    },
    {
      "epoch": 0.1812,
      "grad_norm": 1.2094159579471924,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 18120
    },
    {
      "epoch": 0.18121,
      "grad_norm": 1.3542369206507303,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 18121
    },
    {
      "epoch": 0.18122,
      "grad_norm": 1.069341226497996,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 18122
    },
    {
      "epoch": 0.18123,
      "grad_norm": 1.270477721298167,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 18123
    },
    {
      "epoch": 0.18124,
      "grad_norm": 1.2497256681751179,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 18124
    },
    {
      "epoch": 0.18125,
      "grad_norm": 1.1922169911959235,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 18125
    },
    {
      "epoch": 0.18126,
      "grad_norm": 1.4041714736757644,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 18126
    },
    {
      "epoch": 0.18127,
      "grad_norm": 1.4731383145523111,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 18127
    },
    {
      "epoch": 0.18128,
      "grad_norm": 1.2021229909504054,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 18128
    },
    {
      "epoch": 0.18129,
      "grad_norm": 1.5091910261960564,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 18129
    },
    {
      "epoch": 0.1813,
      "grad_norm": 1.3231377072043908,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 18130
    },
    {
      "epoch": 0.18131,
      "grad_norm": 1.1523137535135268,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 18131
    },
    {
      "epoch": 0.18132,
      "grad_norm": 1.2015203464085935,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 18132
    },
    {
      "epoch": 0.18133,
      "grad_norm": 1.2494390904610018,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 18133
    },
    {
      "epoch": 0.18134,
      "grad_norm": 1.2541054427205613,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 18134
    },
    {
      "epoch": 0.18135,
      "grad_norm": 1.2488046257842773,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 18135
    },
    {
      "epoch": 0.18136,
      "grad_norm": 1.2661888618752437,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 18136
    },
    {
      "epoch": 0.18137,
      "grad_norm": 1.3085164931949764,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 18137
    },
    {
      "epoch": 0.18138,
      "grad_norm": 1.1491090041473908,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 18138
    },
    {
      "epoch": 0.18139,
      "grad_norm": 1.491569568936032,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 18139
    },
    {
      "epoch": 0.1814,
      "grad_norm": 1.2807156445409174,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 18140
    },
    {
      "epoch": 0.18141,
      "grad_norm": 1.298896890123939,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 18141
    },
    {
      "epoch": 0.18142,
      "grad_norm": 1.4534689097758247,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 18142
    },
    {
      "epoch": 0.18143,
      "grad_norm": 1.2760736538512882,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 18143
    },
    {
      "epoch": 0.18144,
      "grad_norm": 1.2731093061022918,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 18144
    },
    {
      "epoch": 0.18145,
      "grad_norm": 1.397239557622711,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 18145
    },
    {
      "epoch": 0.18146,
      "grad_norm": 1.2370712894447977,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 18146
    },
    {
      "epoch": 0.18147,
      "grad_norm": 1.408516610656114,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 18147
    },
    {
      "epoch": 0.18148,
      "grad_norm": 1.1404495545854807,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 18148
    },
    {
      "epoch": 0.18149,
      "grad_norm": 1.4999510770973097,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 18149
    },
    {
      "epoch": 0.1815,
      "grad_norm": 1.3412249640259764,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 18150
    },
    {
      "epoch": 0.18151,
      "grad_norm": 1.340655032055312,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 18151
    },
    {
      "epoch": 0.18152,
      "grad_norm": 1.3430861191611438,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 18152
    },
    {
      "epoch": 0.18153,
      "grad_norm": 1.2229632330161644,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 18153
    },
    {
      "epoch": 0.18154,
      "grad_norm": 1.3686194210829088,
      "learning_rate": 0.003,
      "loss": 4.0385,
      "step": 18154
    },
    {
      "epoch": 0.18155,
      "grad_norm": 1.3695351862872984,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 18155
    },
    {
      "epoch": 0.18156,
      "grad_norm": 1.1962950950615079,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 18156
    },
    {
      "epoch": 0.18157,
      "grad_norm": 1.3791216928398669,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 18157
    },
    {
      "epoch": 0.18158,
      "grad_norm": 1.4249186981770756,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 18158
    },
    {
      "epoch": 0.18159,
      "grad_norm": 1.6859917691018185,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 18159
    },
    {
      "epoch": 0.1816,
      "grad_norm": 1.0514003415436335,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 18160
    },
    {
      "epoch": 0.18161,
      "grad_norm": 1.3370164144149683,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 18161
    },
    {
      "epoch": 0.18162,
      "grad_norm": 1.4367597693541223,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 18162
    },
    {
      "epoch": 0.18163,
      "grad_norm": 1.1688501705480174,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 18163
    },
    {
      "epoch": 0.18164,
      "grad_norm": 1.4386363032319445,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 18164
    },
    {
      "epoch": 0.18165,
      "grad_norm": 1.0009150657523624,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 18165
    },
    {
      "epoch": 0.18166,
      "grad_norm": 1.6000360556528737,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 18166
    },
    {
      "epoch": 0.18167,
      "grad_norm": 1.0258430483937688,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 18167
    },
    {
      "epoch": 0.18168,
      "grad_norm": 1.3458346581546479,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 18168
    },
    {
      "epoch": 0.18169,
      "grad_norm": 1.6714686561058865,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 18169
    },
    {
      "epoch": 0.1817,
      "grad_norm": 1.4872920520624644,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 18170
    },
    {
      "epoch": 0.18171,
      "grad_norm": 1.0951495742612987,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 18171
    },
    {
      "epoch": 0.18172,
      "grad_norm": 1.1112247590858297,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 18172
    },
    {
      "epoch": 0.18173,
      "grad_norm": 1.5872827478046938,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 18173
    },
    {
      "epoch": 0.18174,
      "grad_norm": 1.340632879143833,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 18174
    },
    {
      "epoch": 0.18175,
      "grad_norm": 1.12270157417685,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 18175
    },
    {
      "epoch": 0.18176,
      "grad_norm": 1.2548336496990686,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 18176
    },
    {
      "epoch": 0.18177,
      "grad_norm": 1.3348772544620668,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 18177
    },
    {
      "epoch": 0.18178,
      "grad_norm": 1.4976886591580911,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 18178
    },
    {
      "epoch": 0.18179,
      "grad_norm": 1.1434711682797878,
      "learning_rate": 0.003,
      "loss": 4.0312,
      "step": 18179
    },
    {
      "epoch": 0.1818,
      "grad_norm": 1.4240373669379667,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 18180
    },
    {
      "epoch": 0.18181,
      "grad_norm": 1.2303593962497674,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 18181
    },
    {
      "epoch": 0.18182,
      "grad_norm": 1.3255765937167197,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 18182
    },
    {
      "epoch": 0.18183,
      "grad_norm": 1.5157103448871325,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 18183
    },
    {
      "epoch": 0.18184,
      "grad_norm": 1.3625922244136983,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 18184
    },
    {
      "epoch": 0.18185,
      "grad_norm": 1.5395208661727047,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 18185
    },
    {
      "epoch": 0.18186,
      "grad_norm": 0.9853905024374693,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 18186
    },
    {
      "epoch": 0.18187,
      "grad_norm": 1.5047072672416895,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 18187
    },
    {
      "epoch": 0.18188,
      "grad_norm": 1.1775613512995033,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 18188
    },
    {
      "epoch": 0.18189,
      "grad_norm": 1.384245652769638,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 18189
    },
    {
      "epoch": 0.1819,
      "grad_norm": 1.333759677855863,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 18190
    },
    {
      "epoch": 0.18191,
      "grad_norm": 1.3372005644077896,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 18191
    },
    {
      "epoch": 0.18192,
      "grad_norm": 1.3535946694206413,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 18192
    },
    {
      "epoch": 0.18193,
      "grad_norm": 1.2468378090575452,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 18193
    },
    {
      "epoch": 0.18194,
      "grad_norm": 1.4532766301040654,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 18194
    },
    {
      "epoch": 0.18195,
      "grad_norm": 1.1686842936325825,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 18195
    },
    {
      "epoch": 0.18196,
      "grad_norm": 1.4565038985790555,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 18196
    },
    {
      "epoch": 0.18197,
      "grad_norm": 1.0958727172186844,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 18197
    },
    {
      "epoch": 0.18198,
      "grad_norm": 1.4138226117913761,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 18198
    },
    {
      "epoch": 0.18199,
      "grad_norm": 1.162239856816722,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 18199
    },
    {
      "epoch": 0.182,
      "grad_norm": 1.2385407734307363,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 18200
    },
    {
      "epoch": 0.18201,
      "grad_norm": 1.2635024331352518,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 18201
    },
    {
      "epoch": 0.18202,
      "grad_norm": 1.3092375521949944,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 18202
    },
    {
      "epoch": 0.18203,
      "grad_norm": 1.4449188880190564,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 18203
    },
    {
      "epoch": 0.18204,
      "grad_norm": 1.3422035962759664,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 18204
    },
    {
      "epoch": 0.18205,
      "grad_norm": 1.4648471640630332,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 18205
    },
    {
      "epoch": 0.18206,
      "grad_norm": 1.2181134644769143,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 18206
    },
    {
      "epoch": 0.18207,
      "grad_norm": 1.2960922502120704,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 18207
    },
    {
      "epoch": 0.18208,
      "grad_norm": 1.179304476592766,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 18208
    },
    {
      "epoch": 0.18209,
      "grad_norm": 1.1808590466178734,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 18209
    },
    {
      "epoch": 0.1821,
      "grad_norm": 1.2501246950972718,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 18210
    },
    {
      "epoch": 0.18211,
      "grad_norm": 1.175661923685183,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 18211
    },
    {
      "epoch": 0.18212,
      "grad_norm": 1.4023298714534649,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 18212
    },
    {
      "epoch": 0.18213,
      "grad_norm": 1.3569060647030573,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 18213
    },
    {
      "epoch": 0.18214,
      "grad_norm": 1.5227566256911091,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 18214
    },
    {
      "epoch": 0.18215,
      "grad_norm": 1.1338972886688918,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 18215
    },
    {
      "epoch": 0.18216,
      "grad_norm": 1.604034654117611,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 18216
    },
    {
      "epoch": 0.18217,
      "grad_norm": 1.1325825049736928,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18217
    },
    {
      "epoch": 0.18218,
      "grad_norm": 1.2664514497000146,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 18218
    },
    {
      "epoch": 0.18219,
      "grad_norm": 1.442785106896768,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 18219
    },
    {
      "epoch": 0.1822,
      "grad_norm": 1.1107805701881037,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18220
    },
    {
      "epoch": 0.18221,
      "grad_norm": 1.4340328138758345,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 18221
    },
    {
      "epoch": 0.18222,
      "grad_norm": 1.135998506553484,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 18222
    },
    {
      "epoch": 0.18223,
      "grad_norm": 1.410665229499254,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 18223
    },
    {
      "epoch": 0.18224,
      "grad_norm": 1.2607020950425902,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 18224
    },
    {
      "epoch": 0.18225,
      "grad_norm": 1.4245556705655618,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 18225
    },
    {
      "epoch": 0.18226,
      "grad_norm": 1.3512642968003892,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 18226
    },
    {
      "epoch": 0.18227,
      "grad_norm": 1.1604604069289992,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 18227
    },
    {
      "epoch": 0.18228,
      "grad_norm": 1.4738669453150621,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 18228
    },
    {
      "epoch": 0.18229,
      "grad_norm": 1.2406082791105169,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 18229
    },
    {
      "epoch": 0.1823,
      "grad_norm": 1.5715096394050443,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 18230
    },
    {
      "epoch": 0.18231,
      "grad_norm": 1.1275615052222625,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 18231
    },
    {
      "epoch": 0.18232,
      "grad_norm": 1.3406386092189289,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 18232
    },
    {
      "epoch": 0.18233,
      "grad_norm": 1.451787714778182,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 18233
    },
    {
      "epoch": 0.18234,
      "grad_norm": 1.5453428043833808,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 18234
    },
    {
      "epoch": 0.18235,
      "grad_norm": 1.2602528537457918,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 18235
    },
    {
      "epoch": 0.18236,
      "grad_norm": 1.368527128648679,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 18236
    },
    {
      "epoch": 0.18237,
      "grad_norm": 1.1857713306589048,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 18237
    },
    {
      "epoch": 0.18238,
      "grad_norm": 1.1543106412537665,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 18238
    },
    {
      "epoch": 0.18239,
      "grad_norm": 1.3016308243620187,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 18239
    },
    {
      "epoch": 0.1824,
      "grad_norm": 1.231102382837433,
      "learning_rate": 0.003,
      "loss": 4.0319,
      "step": 18240
    },
    {
      "epoch": 0.18241,
      "grad_norm": 1.1077947692409051,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 18241
    },
    {
      "epoch": 0.18242,
      "grad_norm": 1.4706854398724694,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 18242
    },
    {
      "epoch": 0.18243,
      "grad_norm": 1.325152785053419,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 18243
    },
    {
      "epoch": 0.18244,
      "grad_norm": 1.6457422299842572,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 18244
    },
    {
      "epoch": 0.18245,
      "grad_norm": 1.1406046734509585,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 18245
    },
    {
      "epoch": 0.18246,
      "grad_norm": 1.5286619780425068,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 18246
    },
    {
      "epoch": 0.18247,
      "grad_norm": 1.3525066251140536,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 18247
    },
    {
      "epoch": 0.18248,
      "grad_norm": 1.2439100214983063,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 18248
    },
    {
      "epoch": 0.18249,
      "grad_norm": 1.154484097160147,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 18249
    },
    {
      "epoch": 0.1825,
      "grad_norm": 1.4881899126741536,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 18250
    },
    {
      "epoch": 0.18251,
      "grad_norm": 1.465279467451196,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 18251
    },
    {
      "epoch": 0.18252,
      "grad_norm": 1.1982241062906864,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 18252
    },
    {
      "epoch": 0.18253,
      "grad_norm": 1.185664148262174,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 18253
    },
    {
      "epoch": 0.18254,
      "grad_norm": 1.4467398087390484,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 18254
    },
    {
      "epoch": 0.18255,
      "grad_norm": 1.1997101263431185,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 18255
    },
    {
      "epoch": 0.18256,
      "grad_norm": 1.449567953387921,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 18256
    },
    {
      "epoch": 0.18257,
      "grad_norm": 1.0942210590155477,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 18257
    },
    {
      "epoch": 0.18258,
      "grad_norm": 1.452957264341418,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 18258
    },
    {
      "epoch": 0.18259,
      "grad_norm": 1.182946141147893,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 18259
    },
    {
      "epoch": 0.1826,
      "grad_norm": 1.3462294868763975,
      "learning_rate": 0.003,
      "loss": 4.0292,
      "step": 18260
    },
    {
      "epoch": 0.18261,
      "grad_norm": 1.3637873523197717,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 18261
    },
    {
      "epoch": 0.18262,
      "grad_norm": 1.16961441182416,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 18262
    },
    {
      "epoch": 0.18263,
      "grad_norm": 1.326741324794611,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 18263
    },
    {
      "epoch": 0.18264,
      "grad_norm": 1.4158659471478507,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 18264
    },
    {
      "epoch": 0.18265,
      "grad_norm": 1.4652413297438658,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 18265
    },
    {
      "epoch": 0.18266,
      "grad_norm": 1.2955548481668866,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 18266
    },
    {
      "epoch": 0.18267,
      "grad_norm": 1.0863372608696646,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 18267
    },
    {
      "epoch": 0.18268,
      "grad_norm": 1.4834843745030142,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 18268
    },
    {
      "epoch": 0.18269,
      "grad_norm": 1.196583431259954,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 18269
    },
    {
      "epoch": 0.1827,
      "grad_norm": 1.7173668137332267,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 18270
    },
    {
      "epoch": 0.18271,
      "grad_norm": 0.9905754971041496,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 18271
    },
    {
      "epoch": 0.18272,
      "grad_norm": 1.3837736524671629,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 18272
    },
    {
      "epoch": 0.18273,
      "grad_norm": 1.189224070759516,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 18273
    },
    {
      "epoch": 0.18274,
      "grad_norm": 1.163539976581659,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 18274
    },
    {
      "epoch": 0.18275,
      "grad_norm": 1.2124049633265934,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 18275
    },
    {
      "epoch": 0.18276,
      "grad_norm": 1.5260532893202114,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 18276
    },
    {
      "epoch": 0.18277,
      "grad_norm": 1.518910956257597,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 18277
    },
    {
      "epoch": 0.18278,
      "grad_norm": 1.400104167115685,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 18278
    },
    {
      "epoch": 0.18279,
      "grad_norm": 1.1300618726814304,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 18279
    },
    {
      "epoch": 0.1828,
      "grad_norm": 1.2951842139848218,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 18280
    },
    {
      "epoch": 0.18281,
      "grad_norm": 1.3267232344034476,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 18281
    },
    {
      "epoch": 0.18282,
      "grad_norm": 1.45139641040153,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 18282
    },
    {
      "epoch": 0.18283,
      "grad_norm": 1.1427113404000755,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 18283
    },
    {
      "epoch": 0.18284,
      "grad_norm": 1.6198624702211222,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 18284
    },
    {
      "epoch": 0.18285,
      "grad_norm": 1.3551202175210069,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 18285
    },
    {
      "epoch": 0.18286,
      "grad_norm": 1.1961224827929666,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 18286
    },
    {
      "epoch": 0.18287,
      "grad_norm": 1.3148573588163222,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 18287
    },
    {
      "epoch": 0.18288,
      "grad_norm": 1.3019818203493474,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 18288
    },
    {
      "epoch": 0.18289,
      "grad_norm": 1.1852351290061478,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 18289
    },
    {
      "epoch": 0.1829,
      "grad_norm": 1.2162453347899769,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 18290
    },
    {
      "epoch": 0.18291,
      "grad_norm": 1.3228929662924833,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 18291
    },
    {
      "epoch": 0.18292,
      "grad_norm": 1.2996681355619524,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 18292
    },
    {
      "epoch": 0.18293,
      "grad_norm": 1.4240841378139744,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 18293
    },
    {
      "epoch": 0.18294,
      "grad_norm": 1.5406457155971125,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 18294
    },
    {
      "epoch": 0.18295,
      "grad_norm": 1.2010498038914397,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 18295
    },
    {
      "epoch": 0.18296,
      "grad_norm": 1.2747007627421347,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 18296
    },
    {
      "epoch": 0.18297,
      "grad_norm": 1.540524874943087,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 18297
    },
    {
      "epoch": 0.18298,
      "grad_norm": 1.199559660824532,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 18298
    },
    {
      "epoch": 0.18299,
      "grad_norm": 1.291984495675658,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 18299
    },
    {
      "epoch": 0.183,
      "grad_norm": 1.2274624100026723,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 18300
    },
    {
      "epoch": 0.18301,
      "grad_norm": 1.3159384603583382,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 18301
    },
    {
      "epoch": 0.18302,
      "grad_norm": 1.151252609813012,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 18302
    },
    {
      "epoch": 0.18303,
      "grad_norm": 1.3941105679982355,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 18303
    },
    {
      "epoch": 0.18304,
      "grad_norm": 1.2919033233838624,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 18304
    },
    {
      "epoch": 0.18305,
      "grad_norm": 1.2959301435902835,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 18305
    },
    {
      "epoch": 0.18306,
      "grad_norm": 1.3290561075222391,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 18306
    },
    {
      "epoch": 0.18307,
      "grad_norm": 1.3431773936191693,
      "learning_rate": 0.003,
      "loss": 4.0264,
      "step": 18307
    },
    {
      "epoch": 0.18308,
      "grad_norm": 1.4850679199683352,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 18308
    },
    {
      "epoch": 0.18309,
      "grad_norm": 1.111470817529982,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 18309
    },
    {
      "epoch": 0.1831,
      "grad_norm": 1.2791218355500877,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 18310
    },
    {
      "epoch": 0.18311,
      "grad_norm": 1.2679741066880428,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 18311
    },
    {
      "epoch": 0.18312,
      "grad_norm": 1.2558317099155634,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 18312
    },
    {
      "epoch": 0.18313,
      "grad_norm": 1.3025506192884788,
      "learning_rate": 0.003,
      "loss": 3.9883,
      "step": 18313
    },
    {
      "epoch": 0.18314,
      "grad_norm": 1.1158220238267107,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 18314
    },
    {
      "epoch": 0.18315,
      "grad_norm": 1.5905909968246834,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 18315
    },
    {
      "epoch": 0.18316,
      "grad_norm": 1.1498620642458874,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 18316
    },
    {
      "epoch": 0.18317,
      "grad_norm": 1.484744737179069,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 18317
    },
    {
      "epoch": 0.18318,
      "grad_norm": 1.5264021135168917,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 18318
    },
    {
      "epoch": 0.18319,
      "grad_norm": 1.1805558057613221,
      "learning_rate": 0.003,
      "loss": 4.013,
      "step": 18319
    },
    {
      "epoch": 0.1832,
      "grad_norm": 1.5571258801998404,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 18320
    },
    {
      "epoch": 0.18321,
      "grad_norm": 1.1898018228480296,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 18321
    },
    {
      "epoch": 0.18322,
      "grad_norm": 1.4159144879762025,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 18322
    },
    {
      "epoch": 0.18323,
      "grad_norm": 1.1410631707793324,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 18323
    },
    {
      "epoch": 0.18324,
      "grad_norm": 1.4012920373946773,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 18324
    },
    {
      "epoch": 0.18325,
      "grad_norm": 1.4299141559672417,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 18325
    },
    {
      "epoch": 0.18326,
      "grad_norm": 1.2930504859850316,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 18326
    },
    {
      "epoch": 0.18327,
      "grad_norm": 1.3468924090170447,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 18327
    },
    {
      "epoch": 0.18328,
      "grad_norm": 1.3180292036073251,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 18328
    },
    {
      "epoch": 0.18329,
      "grad_norm": 1.320514777498404,
      "learning_rate": 0.003,
      "loss": 4.0299,
      "step": 18329
    },
    {
      "epoch": 0.1833,
      "grad_norm": 1.4404597215645119,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 18330
    },
    {
      "epoch": 0.18331,
      "grad_norm": 1.0605172973346007,
      "learning_rate": 0.003,
      "loss": 4.0298,
      "step": 18331
    },
    {
      "epoch": 0.18332,
      "grad_norm": 1.4843437599893439,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 18332
    },
    {
      "epoch": 0.18333,
      "grad_norm": 1.1693918793582336,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 18333
    },
    {
      "epoch": 0.18334,
      "grad_norm": 1.5217297023948761,
      "learning_rate": 0.003,
      "loss": 4.0309,
      "step": 18334
    },
    {
      "epoch": 0.18335,
      "grad_norm": 1.1593004710167287,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 18335
    },
    {
      "epoch": 0.18336,
      "grad_norm": 1.3327810915410703,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 18336
    },
    {
      "epoch": 0.18337,
      "grad_norm": 1.213999401958382,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 18337
    },
    {
      "epoch": 0.18338,
      "grad_norm": 1.2807554556088505,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 18338
    },
    {
      "epoch": 0.18339,
      "grad_norm": 1.1927103956160532,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 18339
    },
    {
      "epoch": 0.1834,
      "grad_norm": 1.17816569156864,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 18340
    },
    {
      "epoch": 0.18341,
      "grad_norm": 1.3627814653918315,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 18341
    },
    {
      "epoch": 0.18342,
      "grad_norm": 1.2234631781988194,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 18342
    },
    {
      "epoch": 0.18343,
      "grad_norm": 1.3372321269836562,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 18343
    },
    {
      "epoch": 0.18344,
      "grad_norm": 1.2661334492041663,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 18344
    },
    {
      "epoch": 0.18345,
      "grad_norm": 1.3393651557191961,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 18345
    },
    {
      "epoch": 0.18346,
      "grad_norm": 1.300894564841833,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 18346
    },
    {
      "epoch": 0.18347,
      "grad_norm": 1.3276680430228749,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 18347
    },
    {
      "epoch": 0.18348,
      "grad_norm": 1.3937152352869309,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 18348
    },
    {
      "epoch": 0.18349,
      "grad_norm": 1.2502815585549163,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 18349
    },
    {
      "epoch": 0.1835,
      "grad_norm": 1.3517760524979265,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 18350
    },
    {
      "epoch": 0.18351,
      "grad_norm": 1.1511040155399297,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 18351
    },
    {
      "epoch": 0.18352,
      "grad_norm": 1.5658496591292388,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 18352
    },
    {
      "epoch": 0.18353,
      "grad_norm": 1.1934301437246606,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 18353
    },
    {
      "epoch": 0.18354,
      "grad_norm": 1.3428249291928236,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 18354
    },
    {
      "epoch": 0.18355,
      "grad_norm": 1.262184086853639,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18355
    },
    {
      "epoch": 0.18356,
      "grad_norm": 1.1838306132491332,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 18356
    },
    {
      "epoch": 0.18357,
      "grad_norm": 1.3972135551754903,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 18357
    },
    {
      "epoch": 0.18358,
      "grad_norm": 1.2564781647962338,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 18358
    },
    {
      "epoch": 0.18359,
      "grad_norm": 1.4244018335933282,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 18359
    },
    {
      "epoch": 0.1836,
      "grad_norm": 1.5316620076526126,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 18360
    },
    {
      "epoch": 0.18361,
      "grad_norm": 1.2795945610462196,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 18361
    },
    {
      "epoch": 0.18362,
      "grad_norm": 1.1518190277974463,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 18362
    },
    {
      "epoch": 0.18363,
      "grad_norm": 1.530199581234219,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 18363
    },
    {
      "epoch": 0.18364,
      "grad_norm": 1.231297952448577,
      "learning_rate": 0.003,
      "loss": 4.0986,
      "step": 18364
    },
    {
      "epoch": 0.18365,
      "grad_norm": 1.3570466325513808,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 18365
    },
    {
      "epoch": 0.18366,
      "grad_norm": 1.108459820222994,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 18366
    },
    {
      "epoch": 0.18367,
      "grad_norm": 1.4759435295772985,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 18367
    },
    {
      "epoch": 0.18368,
      "grad_norm": 1.2730892852219353,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 18368
    },
    {
      "epoch": 0.18369,
      "grad_norm": 1.2393159631536057,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 18369
    },
    {
      "epoch": 0.1837,
      "grad_norm": 1.5601089691527303,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 18370
    },
    {
      "epoch": 0.18371,
      "grad_norm": 1.2743664874142249,
      "learning_rate": 0.003,
      "loss": 4.0312,
      "step": 18371
    },
    {
      "epoch": 0.18372,
      "grad_norm": 1.2013851385814807,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 18372
    },
    {
      "epoch": 0.18373,
      "grad_norm": 1.4056078532157135,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 18373
    },
    {
      "epoch": 0.18374,
      "grad_norm": 1.435369760830309,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 18374
    },
    {
      "epoch": 0.18375,
      "grad_norm": 1.2704736230239198,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 18375
    },
    {
      "epoch": 0.18376,
      "grad_norm": 1.4623748469123223,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 18376
    },
    {
      "epoch": 0.18377,
      "grad_norm": 1.051727732073009,
      "learning_rate": 0.003,
      "loss": 4.0329,
      "step": 18377
    },
    {
      "epoch": 0.18378,
      "grad_norm": 1.3896777833282445,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 18378
    },
    {
      "epoch": 0.18379,
      "grad_norm": 1.3228853385275277,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 18379
    },
    {
      "epoch": 0.1838,
      "grad_norm": 1.4608524000156693,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 18380
    },
    {
      "epoch": 0.18381,
      "grad_norm": 1.4029575412483193,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 18381
    },
    {
      "epoch": 0.18382,
      "grad_norm": 1.5158907956137597,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 18382
    },
    {
      "epoch": 0.18383,
      "grad_norm": 1.0716546311592323,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 18383
    },
    {
      "epoch": 0.18384,
      "grad_norm": 1.4981090955511132,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 18384
    },
    {
      "epoch": 0.18385,
      "grad_norm": 1.0268919429165768,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 18385
    },
    {
      "epoch": 0.18386,
      "grad_norm": 1.4229816452549553,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 18386
    },
    {
      "epoch": 0.18387,
      "grad_norm": 1.2366046979296248,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 18387
    },
    {
      "epoch": 0.18388,
      "grad_norm": 1.5639521377941439,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 18388
    },
    {
      "epoch": 0.18389,
      "grad_norm": 1.1573115439932287,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 18389
    },
    {
      "epoch": 0.1839,
      "grad_norm": 1.533474200163487,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 18390
    },
    {
      "epoch": 0.18391,
      "grad_norm": 1.3131162909328664,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 18391
    },
    {
      "epoch": 0.18392,
      "grad_norm": 1.4200948057489893,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 18392
    },
    {
      "epoch": 0.18393,
      "grad_norm": 1.4495701534876568,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 18393
    },
    {
      "epoch": 0.18394,
      "grad_norm": 1.0922084287919938,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 18394
    },
    {
      "epoch": 0.18395,
      "grad_norm": 1.3477790387176505,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 18395
    },
    {
      "epoch": 0.18396,
      "grad_norm": 1.2410173223451075,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 18396
    },
    {
      "epoch": 0.18397,
      "grad_norm": 1.5249482865632253,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 18397
    },
    {
      "epoch": 0.18398,
      "grad_norm": 1.1072053712305903,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 18398
    },
    {
      "epoch": 0.18399,
      "grad_norm": 1.5476793058041969,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 18399
    },
    {
      "epoch": 0.184,
      "grad_norm": 1.1607239907993379,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 18400
    },
    {
      "epoch": 0.18401,
      "grad_norm": 1.4989909112745787,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 18401
    },
    {
      "epoch": 0.18402,
      "grad_norm": 0.9660817032706551,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 18402
    },
    {
      "epoch": 0.18403,
      "grad_norm": 1.274004084475861,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 18403
    },
    {
      "epoch": 0.18404,
      "grad_norm": 1.2982193710720116,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 18404
    },
    {
      "epoch": 0.18405,
      "grad_norm": 1.3122029694472575,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 18405
    },
    {
      "epoch": 0.18406,
      "grad_norm": 1.3780064201413342,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 18406
    },
    {
      "epoch": 0.18407,
      "grad_norm": 1.3424378335998188,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 18407
    },
    {
      "epoch": 0.18408,
      "grad_norm": 1.0271180117058203,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 18408
    },
    {
      "epoch": 0.18409,
      "grad_norm": 1.5800784065971756,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 18409
    },
    {
      "epoch": 0.1841,
      "grad_norm": 1.0228906698663216,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 18410
    },
    {
      "epoch": 0.18411,
      "grad_norm": 1.156292790558681,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 18411
    },
    {
      "epoch": 0.18412,
      "grad_norm": 1.2843566367494446,
      "learning_rate": 0.003,
      "loss": 4.0978,
      "step": 18412
    },
    {
      "epoch": 0.18413,
      "grad_norm": 1.1838378450193543,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 18413
    },
    {
      "epoch": 0.18414,
      "grad_norm": 1.5250702367687015,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 18414
    },
    {
      "epoch": 0.18415,
      "grad_norm": 1.4222352386571093,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 18415
    },
    {
      "epoch": 0.18416,
      "grad_norm": 1.435399268996228,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 18416
    },
    {
      "epoch": 0.18417,
      "grad_norm": 1.0218622741098493,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 18417
    },
    {
      "epoch": 0.18418,
      "grad_norm": 1.3939864132572706,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 18418
    },
    {
      "epoch": 0.18419,
      "grad_norm": 1.090328698455138,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 18419
    },
    {
      "epoch": 0.1842,
      "grad_norm": 1.3533141590388846,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 18420
    },
    {
      "epoch": 0.18421,
      "grad_norm": 1.5480003377160603,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 18421
    },
    {
      "epoch": 0.18422,
      "grad_norm": 1.3988223043662067,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 18422
    },
    {
      "epoch": 0.18423,
      "grad_norm": 1.389063700021048,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 18423
    },
    {
      "epoch": 0.18424,
      "grad_norm": 1.3197639854120087,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 18424
    },
    {
      "epoch": 0.18425,
      "grad_norm": 1.1066292086722558,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 18425
    },
    {
      "epoch": 0.18426,
      "grad_norm": 1.4317714961252748,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 18426
    },
    {
      "epoch": 0.18427,
      "grad_norm": 1.1234393863334924,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 18427
    },
    {
      "epoch": 0.18428,
      "grad_norm": 1.3241461850043657,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 18428
    },
    {
      "epoch": 0.18429,
      "grad_norm": 1.2278628309733401,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 18429
    },
    {
      "epoch": 0.1843,
      "grad_norm": 1.3376376277392692,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 18430
    },
    {
      "epoch": 0.18431,
      "grad_norm": 1.1514281077049928,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 18431
    },
    {
      "epoch": 0.18432,
      "grad_norm": 1.4278693885956923,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 18432
    },
    {
      "epoch": 0.18433,
      "grad_norm": 1.421603179534481,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 18433
    },
    {
      "epoch": 0.18434,
      "grad_norm": 1.7790992463663307,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 18434
    },
    {
      "epoch": 0.18435,
      "grad_norm": 1.281655460111802,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 18435
    },
    {
      "epoch": 0.18436,
      "grad_norm": 1.447909389488687,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 18436
    },
    {
      "epoch": 0.18437,
      "grad_norm": 1.3215450363410834,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 18437
    },
    {
      "epoch": 0.18438,
      "grad_norm": 1.2994988643640621,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 18438
    },
    {
      "epoch": 0.18439,
      "grad_norm": 1.1530461997369483,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 18439
    },
    {
      "epoch": 0.1844,
      "grad_norm": 1.2723226530047036,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 18440
    },
    {
      "epoch": 0.18441,
      "grad_norm": 1.4190177482257076,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 18441
    },
    {
      "epoch": 0.18442,
      "grad_norm": 1.298334180626103,
      "learning_rate": 0.003,
      "loss": 4.0218,
      "step": 18442
    },
    {
      "epoch": 0.18443,
      "grad_norm": 1.4213127263863718,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 18443
    },
    {
      "epoch": 0.18444,
      "grad_norm": 1.282156090364548,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 18444
    },
    {
      "epoch": 0.18445,
      "grad_norm": 1.2201086731789128,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 18445
    },
    {
      "epoch": 0.18446,
      "grad_norm": 1.4416977833642208,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 18446
    },
    {
      "epoch": 0.18447,
      "grad_norm": 1.1922971767874007,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 18447
    },
    {
      "epoch": 0.18448,
      "grad_norm": 1.782110662174641,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 18448
    },
    {
      "epoch": 0.18449,
      "grad_norm": 1.042098559036565,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 18449
    },
    {
      "epoch": 0.1845,
      "grad_norm": 1.6867508076955657,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 18450
    },
    {
      "epoch": 0.18451,
      "grad_norm": 1.171477974688329,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 18451
    },
    {
      "epoch": 0.18452,
      "grad_norm": 1.6833010403008586,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 18452
    },
    {
      "epoch": 0.18453,
      "grad_norm": 1.1108276265202692,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 18453
    },
    {
      "epoch": 0.18454,
      "grad_norm": 1.2513031676322688,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 18454
    },
    {
      "epoch": 0.18455,
      "grad_norm": 1.2370626784842842,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 18455
    },
    {
      "epoch": 0.18456,
      "grad_norm": 1.3506690505066392,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 18456
    },
    {
      "epoch": 0.18457,
      "grad_norm": 1.514909151715256,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 18457
    },
    {
      "epoch": 0.18458,
      "grad_norm": 1.3048506316734978,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 18458
    },
    {
      "epoch": 0.18459,
      "grad_norm": 1.1903146478199034,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 18459
    },
    {
      "epoch": 0.1846,
      "grad_norm": 1.2739001027820747,
      "learning_rate": 0.003,
      "loss": 4.0288,
      "step": 18460
    },
    {
      "epoch": 0.18461,
      "grad_norm": 1.262827186533425,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 18461
    },
    {
      "epoch": 0.18462,
      "grad_norm": 1.4505564797482764,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 18462
    },
    {
      "epoch": 0.18463,
      "grad_norm": 1.2961279196648388,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 18463
    },
    {
      "epoch": 0.18464,
      "grad_norm": 1.10546721769061,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 18464
    },
    {
      "epoch": 0.18465,
      "grad_norm": 1.313201999660418,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 18465
    },
    {
      "epoch": 0.18466,
      "grad_norm": 1.1669156833746952,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 18466
    },
    {
      "epoch": 0.18467,
      "grad_norm": 1.534547969859101,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 18467
    },
    {
      "epoch": 0.18468,
      "grad_norm": 1.1393395730416378,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 18468
    },
    {
      "epoch": 0.18469,
      "grad_norm": 1.5335779758703452,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 18469
    },
    {
      "epoch": 0.1847,
      "grad_norm": 1.4055567287610196,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 18470
    },
    {
      "epoch": 0.18471,
      "grad_norm": 1.3167880081135488,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 18471
    },
    {
      "epoch": 0.18472,
      "grad_norm": 1.2576717359061849,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 18472
    },
    {
      "epoch": 0.18473,
      "grad_norm": 1.3957171929089696,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 18473
    },
    {
      "epoch": 0.18474,
      "grad_norm": 1.400310244317996,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 18474
    },
    {
      "epoch": 0.18475,
      "grad_norm": 1.4945282965314968,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 18475
    },
    {
      "epoch": 0.18476,
      "grad_norm": 1.145145941074851,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 18476
    },
    {
      "epoch": 0.18477,
      "grad_norm": 1.455469946244908,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 18477
    },
    {
      "epoch": 0.18478,
      "grad_norm": 1.3604896246438607,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 18478
    },
    {
      "epoch": 0.18479,
      "grad_norm": 1.1812863009354766,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 18479
    },
    {
      "epoch": 0.1848,
      "grad_norm": 1.3283765497707292,
      "learning_rate": 0.003,
      "loss": 4.0244,
      "step": 18480
    },
    {
      "epoch": 0.18481,
      "grad_norm": 1.428316991360103,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 18481
    },
    {
      "epoch": 0.18482,
      "grad_norm": 1.3072437728678792,
      "learning_rate": 0.003,
      "loss": 4.0255,
      "step": 18482
    },
    {
      "epoch": 0.18483,
      "grad_norm": 1.0899128603018826,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 18483
    },
    {
      "epoch": 0.18484,
      "grad_norm": 1.3918421419823224,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 18484
    },
    {
      "epoch": 0.18485,
      "grad_norm": 1.288039295583636,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 18485
    },
    {
      "epoch": 0.18486,
      "grad_norm": 1.3961753464614879,
      "learning_rate": 0.003,
      "loss": 4.0229,
      "step": 18486
    },
    {
      "epoch": 0.18487,
      "grad_norm": 1.416549086470837,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 18487
    },
    {
      "epoch": 0.18488,
      "grad_norm": 1.110424938616956,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 18488
    },
    {
      "epoch": 0.18489,
      "grad_norm": 1.3155622848496273,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 18489
    },
    {
      "epoch": 0.1849,
      "grad_norm": 1.2342935085466868,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 18490
    },
    {
      "epoch": 0.18491,
      "grad_norm": 1.327027657773121,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 18491
    },
    {
      "epoch": 0.18492,
      "grad_norm": 1.277790745038652,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 18492
    },
    {
      "epoch": 0.18493,
      "grad_norm": 1.252452693824744,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 18493
    },
    {
      "epoch": 0.18494,
      "grad_norm": 1.2126657596476098,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 18494
    },
    {
      "epoch": 0.18495,
      "grad_norm": 1.124775468288146,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 18495
    },
    {
      "epoch": 0.18496,
      "grad_norm": 1.3963422558678007,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 18496
    },
    {
      "epoch": 0.18497,
      "grad_norm": 1.3112801504123868,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 18497
    },
    {
      "epoch": 0.18498,
      "grad_norm": 1.7519857710163944,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 18498
    },
    {
      "epoch": 0.18499,
      "grad_norm": 1.194584371996459,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 18499
    },
    {
      "epoch": 0.185,
      "grad_norm": 1.3195094913957386,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 18500
    },
    {
      "epoch": 0.18501,
      "grad_norm": 1.3210369721715793,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 18501
    },
    {
      "epoch": 0.18502,
      "grad_norm": 1.3695833326453506,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 18502
    },
    {
      "epoch": 0.18503,
      "grad_norm": 1.2099011060705465,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 18503
    },
    {
      "epoch": 0.18504,
      "grad_norm": 1.3185094202816177,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 18504
    },
    {
      "epoch": 0.18505,
      "grad_norm": 1.2158771291760275,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 18505
    },
    {
      "epoch": 0.18506,
      "grad_norm": 1.429753701083564,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 18506
    },
    {
      "epoch": 0.18507,
      "grad_norm": 1.3421314391793797,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 18507
    },
    {
      "epoch": 0.18508,
      "grad_norm": 1.22920004751544,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 18508
    },
    {
      "epoch": 0.18509,
      "grad_norm": 1.7533299233045962,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 18509
    },
    {
      "epoch": 0.1851,
      "grad_norm": 1.1870119764968476,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 18510
    },
    {
      "epoch": 0.18511,
      "grad_norm": 1.7634954177335502,
      "learning_rate": 0.003,
      "loss": 4.1127,
      "step": 18511
    },
    {
      "epoch": 0.18512,
      "grad_norm": 1.1433215341938388,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 18512
    },
    {
      "epoch": 0.18513,
      "grad_norm": 1.1818368377792434,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 18513
    },
    {
      "epoch": 0.18514,
      "grad_norm": 1.3936303987837908,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 18514
    },
    {
      "epoch": 0.18515,
      "grad_norm": 1.4106979166568008,
      "learning_rate": 0.003,
      "loss": 4.026,
      "step": 18515
    },
    {
      "epoch": 0.18516,
      "grad_norm": 1.272932681366381,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 18516
    },
    {
      "epoch": 0.18517,
      "grad_norm": 1.2137932700928198,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 18517
    },
    {
      "epoch": 0.18518,
      "grad_norm": 1.5125772549658068,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 18518
    },
    {
      "epoch": 0.18519,
      "grad_norm": 1.1143423364661758,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 18519
    },
    {
      "epoch": 0.1852,
      "grad_norm": 1.3823758749490027,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 18520
    },
    {
      "epoch": 0.18521,
      "grad_norm": 1.2381834575554078,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 18521
    },
    {
      "epoch": 0.18522,
      "grad_norm": 1.2819900778799418,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 18522
    },
    {
      "epoch": 0.18523,
      "grad_norm": 1.5474119030645124,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 18523
    },
    {
      "epoch": 0.18524,
      "grad_norm": 1.5018361450590505,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 18524
    },
    {
      "epoch": 0.18525,
      "grad_norm": 1.2413311633679378,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 18525
    },
    {
      "epoch": 0.18526,
      "grad_norm": 1.1928737595391086,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 18526
    },
    {
      "epoch": 0.18527,
      "grad_norm": 1.131070084402993,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 18527
    },
    {
      "epoch": 0.18528,
      "grad_norm": 1.3469415033017138,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 18528
    },
    {
      "epoch": 0.18529,
      "grad_norm": 1.136975179840116,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 18529
    },
    {
      "epoch": 0.1853,
      "grad_norm": 1.4877184864763433,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 18530
    },
    {
      "epoch": 0.18531,
      "grad_norm": 1.3011975456050204,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 18531
    },
    {
      "epoch": 0.18532,
      "grad_norm": 1.2775776488164456,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 18532
    },
    {
      "epoch": 0.18533,
      "grad_norm": 1.4722603050631387,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 18533
    },
    {
      "epoch": 0.18534,
      "grad_norm": 1.343931279868763,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 18534
    },
    {
      "epoch": 0.18535,
      "grad_norm": 1.1299003794550169,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 18535
    },
    {
      "epoch": 0.18536,
      "grad_norm": 1.7634204482223623,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 18536
    },
    {
      "epoch": 0.18537,
      "grad_norm": 1.1402924195024549,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 18537
    },
    {
      "epoch": 0.18538,
      "grad_norm": 1.347476089176172,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 18538
    },
    {
      "epoch": 0.18539,
      "grad_norm": 1.1904275311794337,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 18539
    },
    {
      "epoch": 0.1854,
      "grad_norm": 1.2355404913474857,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 18540
    },
    {
      "epoch": 0.18541,
      "grad_norm": 1.3796686835983873,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 18541
    },
    {
      "epoch": 0.18542,
      "grad_norm": 1.347006219996438,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 18542
    },
    {
      "epoch": 0.18543,
      "grad_norm": 1.3349654294153273,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 18543
    },
    {
      "epoch": 0.18544,
      "grad_norm": 1.3986597306105117,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 18544
    },
    {
      "epoch": 0.18545,
      "grad_norm": 1.3750003043430603,
      "learning_rate": 0.003,
      "loss": 4.0272,
      "step": 18545
    },
    {
      "epoch": 0.18546,
      "grad_norm": 1.300556246872616,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 18546
    },
    {
      "epoch": 0.18547,
      "grad_norm": 1.4428540809432855,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 18547
    },
    {
      "epoch": 0.18548,
      "grad_norm": 1.2882429176637629,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 18548
    },
    {
      "epoch": 0.18549,
      "grad_norm": 1.159611227275313,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 18549
    },
    {
      "epoch": 0.1855,
      "grad_norm": 1.2285333893294648,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 18550
    },
    {
      "epoch": 0.18551,
      "grad_norm": 1.0752692148588605,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 18551
    },
    {
      "epoch": 0.18552,
      "grad_norm": 1.4897134426884537,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 18552
    },
    {
      "epoch": 0.18553,
      "grad_norm": 1.2080444956794303,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 18553
    },
    {
      "epoch": 0.18554,
      "grad_norm": 1.3374987105954266,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 18554
    },
    {
      "epoch": 0.18555,
      "grad_norm": 1.155311641991209,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 18555
    },
    {
      "epoch": 0.18556,
      "grad_norm": 1.5116105928324368,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 18556
    },
    {
      "epoch": 0.18557,
      "grad_norm": 1.4892136206380056,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 18557
    },
    {
      "epoch": 0.18558,
      "grad_norm": 1.6181035145542664,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 18558
    },
    {
      "epoch": 0.18559,
      "grad_norm": 1.0805838366203129,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 18559
    },
    {
      "epoch": 0.1856,
      "grad_norm": 1.2551320772028751,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 18560
    },
    {
      "epoch": 0.18561,
      "grad_norm": 1.5086740960076293,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 18561
    },
    {
      "epoch": 0.18562,
      "grad_norm": 1.2445156477052786,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 18562
    },
    {
      "epoch": 0.18563,
      "grad_norm": 1.7085190677221334,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 18563
    },
    {
      "epoch": 0.18564,
      "grad_norm": 1.069580666816677,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 18564
    },
    {
      "epoch": 0.18565,
      "grad_norm": 1.5056137508436014,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 18565
    },
    {
      "epoch": 0.18566,
      "grad_norm": 1.375282844606368,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 18566
    },
    {
      "epoch": 0.18567,
      "grad_norm": 1.015088299986784,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 18567
    },
    {
      "epoch": 0.18568,
      "grad_norm": 1.4160477479350546,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 18568
    },
    {
      "epoch": 0.18569,
      "grad_norm": 1.1523439236916797,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 18569
    },
    {
      "epoch": 0.1857,
      "grad_norm": 1.288278721419816,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 18570
    },
    {
      "epoch": 0.18571,
      "grad_norm": 1.4018639203313916,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 18571
    },
    {
      "epoch": 0.18572,
      "grad_norm": 1.2945508993220207,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 18572
    },
    {
      "epoch": 0.18573,
      "grad_norm": 1.4638142568059527,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 18573
    },
    {
      "epoch": 0.18574,
      "grad_norm": 1.375968230740118,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 18574
    },
    {
      "epoch": 0.18575,
      "grad_norm": 1.2933957765850859,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 18575
    },
    {
      "epoch": 0.18576,
      "grad_norm": 1.1955482168446068,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 18576
    },
    {
      "epoch": 0.18577,
      "grad_norm": 1.2236811418253042,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 18577
    },
    {
      "epoch": 0.18578,
      "grad_norm": 1.494000248261918,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 18578
    },
    {
      "epoch": 0.18579,
      "grad_norm": 1.2922934257170442,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 18579
    },
    {
      "epoch": 0.1858,
      "grad_norm": 1.2921553459066577,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 18580
    },
    {
      "epoch": 0.18581,
      "grad_norm": 1.3782924421301486,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 18581
    },
    {
      "epoch": 0.18582,
      "grad_norm": 1.1801858551405555,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 18582
    },
    {
      "epoch": 0.18583,
      "grad_norm": 1.2915391929830393,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 18583
    },
    {
      "epoch": 0.18584,
      "grad_norm": 1.3198335526433973,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 18584
    },
    {
      "epoch": 0.18585,
      "grad_norm": 1.3653107136423173,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 18585
    },
    {
      "epoch": 0.18586,
      "grad_norm": 1.3187905420756678,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 18586
    },
    {
      "epoch": 0.18587,
      "grad_norm": 1.192836291864276,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 18587
    },
    {
      "epoch": 0.18588,
      "grad_norm": 1.4228266801449778,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 18588
    },
    {
      "epoch": 0.18589,
      "grad_norm": 1.272825202985905,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 18589
    },
    {
      "epoch": 0.1859,
      "grad_norm": 1.4866614615421272,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 18590
    },
    {
      "epoch": 0.18591,
      "grad_norm": 1.1458157948550125,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 18591
    },
    {
      "epoch": 0.18592,
      "grad_norm": 1.5460701241205963,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 18592
    },
    {
      "epoch": 0.18593,
      "grad_norm": 1.0751417400123864,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 18593
    },
    {
      "epoch": 0.18594,
      "grad_norm": 1.4584768764361886,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 18594
    },
    {
      "epoch": 0.18595,
      "grad_norm": 1.1815796913379357,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 18595
    },
    {
      "epoch": 0.18596,
      "grad_norm": 1.3602598493256992,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 18596
    },
    {
      "epoch": 0.18597,
      "grad_norm": 1.2208729022936036,
      "learning_rate": 0.003,
      "loss": 4.0917,
      "step": 18597
    },
    {
      "epoch": 0.18598,
      "grad_norm": 1.3645593853664908,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 18598
    },
    {
      "epoch": 0.18599,
      "grad_norm": 1.1340468689586132,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 18599
    },
    {
      "epoch": 0.186,
      "grad_norm": 1.282165997408662,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 18600
    },
    {
      "epoch": 0.18601,
      "grad_norm": 1.2813171509229273,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 18601
    },
    {
      "epoch": 0.18602,
      "grad_norm": 1.4850002532379016,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 18602
    },
    {
      "epoch": 0.18603,
      "grad_norm": 1.626791547634331,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 18603
    },
    {
      "epoch": 0.18604,
      "grad_norm": 1.6681258479888328,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 18604
    },
    {
      "epoch": 0.18605,
      "grad_norm": 1.1555768713912762,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 18605
    },
    {
      "epoch": 0.18606,
      "grad_norm": 1.2016352473617748,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 18606
    },
    {
      "epoch": 0.18607,
      "grad_norm": 1.4890133528694773,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 18607
    },
    {
      "epoch": 0.18608,
      "grad_norm": 1.0252813859102472,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 18608
    },
    {
      "epoch": 0.18609,
      "grad_norm": 1.5067240132439041,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 18609
    },
    {
      "epoch": 0.1861,
      "grad_norm": 1.207825429153814,
      "learning_rate": 0.003,
      "loss": 4.0915,
      "step": 18610
    },
    {
      "epoch": 0.18611,
      "grad_norm": 1.6342571112613051,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 18611
    },
    {
      "epoch": 0.18612,
      "grad_norm": 1.1209411418605804,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 18612
    },
    {
      "epoch": 0.18613,
      "grad_norm": 1.6259316754158002,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 18613
    },
    {
      "epoch": 0.18614,
      "grad_norm": 1.4766417740346707,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 18614
    },
    {
      "epoch": 0.18615,
      "grad_norm": 1.1620482832574794,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 18615
    },
    {
      "epoch": 0.18616,
      "grad_norm": 1.4598453729402767,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 18616
    },
    {
      "epoch": 0.18617,
      "grad_norm": 1.189093300061794,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 18617
    },
    {
      "epoch": 0.18618,
      "grad_norm": 1.2489389933885495,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 18618
    },
    {
      "epoch": 0.18619,
      "grad_norm": 1.1206001012677298,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 18619
    },
    {
      "epoch": 0.1862,
      "grad_norm": 1.362284327011221,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 18620
    },
    {
      "epoch": 0.18621,
      "grad_norm": 1.109117796194937,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 18621
    },
    {
      "epoch": 0.18622,
      "grad_norm": 1.458631584685427,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 18622
    },
    {
      "epoch": 0.18623,
      "grad_norm": 1.2468584062896135,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 18623
    },
    {
      "epoch": 0.18624,
      "grad_norm": 1.5887506899891826,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 18624
    },
    {
      "epoch": 0.18625,
      "grad_norm": 1.2933164217148045,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 18625
    },
    {
      "epoch": 0.18626,
      "grad_norm": 1.2241022509890256,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 18626
    },
    {
      "epoch": 0.18627,
      "grad_norm": 1.4697934921800258,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 18627
    },
    {
      "epoch": 0.18628,
      "grad_norm": 1.0401361110722684,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 18628
    },
    {
      "epoch": 0.18629,
      "grad_norm": 1.637268932483506,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 18629
    },
    {
      "epoch": 0.1863,
      "grad_norm": 0.9642316107689074,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 18630
    },
    {
      "epoch": 0.18631,
      "grad_norm": 1.676448425875956,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 18631
    },
    {
      "epoch": 0.18632,
      "grad_norm": 1.3233058532473072,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 18632
    },
    {
      "epoch": 0.18633,
      "grad_norm": 1.4546212115883517,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 18633
    },
    {
      "epoch": 0.18634,
      "grad_norm": 1.1710395395526325,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 18634
    },
    {
      "epoch": 0.18635,
      "grad_norm": 1.3333854885532885,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 18635
    },
    {
      "epoch": 0.18636,
      "grad_norm": 1.637573612476902,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 18636
    },
    {
      "epoch": 0.18637,
      "grad_norm": 1.0850780365833652,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 18637
    },
    {
      "epoch": 0.18638,
      "grad_norm": 1.386893023915246,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 18638
    },
    {
      "epoch": 0.18639,
      "grad_norm": 1.2265831060490398,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 18639
    },
    {
      "epoch": 0.1864,
      "grad_norm": 1.3292331057595888,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 18640
    },
    {
      "epoch": 0.18641,
      "grad_norm": 1.1445977131120186,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 18641
    },
    {
      "epoch": 0.18642,
      "grad_norm": 1.2961460433247391,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 18642
    },
    {
      "epoch": 0.18643,
      "grad_norm": 1.2967886889101183,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 18643
    },
    {
      "epoch": 0.18644,
      "grad_norm": 1.4426551743402376,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 18644
    },
    {
      "epoch": 0.18645,
      "grad_norm": 1.0748150099012093,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 18645
    },
    {
      "epoch": 0.18646,
      "grad_norm": 1.2475923740165284,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 18646
    },
    {
      "epoch": 0.18647,
      "grad_norm": 1.3745529718226575,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 18647
    },
    {
      "epoch": 0.18648,
      "grad_norm": 1.092272608985131,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 18648
    },
    {
      "epoch": 0.18649,
      "grad_norm": 1.5267651586159152,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 18649
    },
    {
      "epoch": 0.1865,
      "grad_norm": 1.1513082707972615,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 18650
    },
    {
      "epoch": 0.18651,
      "grad_norm": 1.393832333231486,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 18651
    },
    {
      "epoch": 0.18652,
      "grad_norm": 1.2085993270070554,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 18652
    },
    {
      "epoch": 0.18653,
      "grad_norm": 1.1364757032034836,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 18653
    },
    {
      "epoch": 0.18654,
      "grad_norm": 1.4409042604239815,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 18654
    },
    {
      "epoch": 0.18655,
      "grad_norm": 1.3481757205484133,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 18655
    },
    {
      "epoch": 0.18656,
      "grad_norm": 1.7729058779742146,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 18656
    },
    {
      "epoch": 0.18657,
      "grad_norm": 1.2680266619449934,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 18657
    },
    {
      "epoch": 0.18658,
      "grad_norm": 1.5111394644392886,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 18658
    },
    {
      "epoch": 0.18659,
      "grad_norm": 1.2973492131163422,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 18659
    },
    {
      "epoch": 0.1866,
      "grad_norm": 1.492565466964901,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 18660
    },
    {
      "epoch": 0.18661,
      "grad_norm": 1.2836233918295634,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 18661
    },
    {
      "epoch": 0.18662,
      "grad_norm": 1.2841777141938255,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 18662
    },
    {
      "epoch": 0.18663,
      "grad_norm": 1.3870139131389745,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 18663
    },
    {
      "epoch": 0.18664,
      "grad_norm": 1.099793394818637,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 18664
    },
    {
      "epoch": 0.18665,
      "grad_norm": 1.644613665438987,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 18665
    },
    {
      "epoch": 0.18666,
      "grad_norm": 1.0836969998654264,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 18666
    },
    {
      "epoch": 0.18667,
      "grad_norm": 1.47740999718926,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 18667
    },
    {
      "epoch": 0.18668,
      "grad_norm": 1.088842733935222,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 18668
    },
    {
      "epoch": 0.18669,
      "grad_norm": 1.2804939696105566,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 18669
    },
    {
      "epoch": 0.1867,
      "grad_norm": 1.2110539378980032,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 18670
    },
    {
      "epoch": 0.18671,
      "grad_norm": 1.2941641843745262,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 18671
    },
    {
      "epoch": 0.18672,
      "grad_norm": 1.8119945201752838,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 18672
    },
    {
      "epoch": 0.18673,
      "grad_norm": 1.1421809216299648,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 18673
    },
    {
      "epoch": 0.18674,
      "grad_norm": 1.3000509660383397,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 18674
    },
    {
      "epoch": 0.18675,
      "grad_norm": 1.2559014844496346,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 18675
    },
    {
      "epoch": 0.18676,
      "grad_norm": 1.6155641495060489,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 18676
    },
    {
      "epoch": 0.18677,
      "grad_norm": 1.257363389798075,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 18677
    },
    {
      "epoch": 0.18678,
      "grad_norm": 1.3257530109944569,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 18678
    },
    {
      "epoch": 0.18679,
      "grad_norm": 1.314530369143493,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 18679
    },
    {
      "epoch": 0.1868,
      "grad_norm": 1.4568661866191923,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 18680
    },
    {
      "epoch": 0.18681,
      "grad_norm": 1.3720227968576448,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 18681
    },
    {
      "epoch": 0.18682,
      "grad_norm": 1.1044912654278982,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 18682
    },
    {
      "epoch": 0.18683,
      "grad_norm": 1.5181290307684798,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 18683
    },
    {
      "epoch": 0.18684,
      "grad_norm": 1.1483907499672095,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 18684
    },
    {
      "epoch": 0.18685,
      "grad_norm": 1.3808132590103672,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 18685
    },
    {
      "epoch": 0.18686,
      "grad_norm": 1.155800467326248,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 18686
    },
    {
      "epoch": 0.18687,
      "grad_norm": 1.3512800343171085,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 18687
    },
    {
      "epoch": 0.18688,
      "grad_norm": 1.1024999840102774,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 18688
    },
    {
      "epoch": 0.18689,
      "grad_norm": 1.4931714589099854,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 18689
    },
    {
      "epoch": 0.1869,
      "grad_norm": 1.2462176717919797,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 18690
    },
    {
      "epoch": 0.18691,
      "grad_norm": 1.4843451460985109,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 18691
    },
    {
      "epoch": 0.18692,
      "grad_norm": 1.1248574285383672,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 18692
    },
    {
      "epoch": 0.18693,
      "grad_norm": 1.63210117841959,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 18693
    },
    {
      "epoch": 0.18694,
      "grad_norm": 1.3423671310621568,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 18694
    },
    {
      "epoch": 0.18695,
      "grad_norm": 1.2838447284722103,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 18695
    },
    {
      "epoch": 0.18696,
      "grad_norm": 1.502227661627234,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 18696
    },
    {
      "epoch": 0.18697,
      "grad_norm": 1.0560286188418697,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 18697
    },
    {
      "epoch": 0.18698,
      "grad_norm": 1.521082857669598,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 18698
    },
    {
      "epoch": 0.18699,
      "grad_norm": 1.2431876001136009,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 18699
    },
    {
      "epoch": 0.187,
      "grad_norm": 1.372440298449826,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 18700
    },
    {
      "epoch": 0.18701,
      "grad_norm": 1.1941255846607965,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 18701
    },
    {
      "epoch": 0.18702,
      "grad_norm": 1.3425501143162746,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 18702
    },
    {
      "epoch": 0.18703,
      "grad_norm": 1.2710114960563306,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 18703
    },
    {
      "epoch": 0.18704,
      "grad_norm": 1.1443549589368678,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 18704
    },
    {
      "epoch": 0.18705,
      "grad_norm": 1.2958357679471506,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 18705
    },
    {
      "epoch": 0.18706,
      "grad_norm": 1.082371359462474,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 18706
    },
    {
      "epoch": 0.18707,
      "grad_norm": 1.40762327599745,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 18707
    },
    {
      "epoch": 0.18708,
      "grad_norm": 1.0190726307980469,
      "learning_rate": 0.003,
      "loss": 4.0222,
      "step": 18708
    },
    {
      "epoch": 0.18709,
      "grad_norm": 1.5810885741879948,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 18709
    },
    {
      "epoch": 0.1871,
      "grad_norm": 1.1132235129114374,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 18710
    },
    {
      "epoch": 0.18711,
      "grad_norm": 1.4880544970134246,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 18711
    },
    {
      "epoch": 0.18712,
      "grad_norm": 1.269599237812824,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 18712
    },
    {
      "epoch": 0.18713,
      "grad_norm": 1.3346080176901836,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 18713
    },
    {
      "epoch": 0.18714,
      "grad_norm": 1.3880945710550838,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 18714
    },
    {
      "epoch": 0.18715,
      "grad_norm": 1.2721781497891678,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 18715
    },
    {
      "epoch": 0.18716,
      "grad_norm": 1.388109232318027,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 18716
    },
    {
      "epoch": 0.18717,
      "grad_norm": 1.157295967954611,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 18717
    },
    {
      "epoch": 0.18718,
      "grad_norm": 1.6700084966983302,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 18718
    },
    {
      "epoch": 0.18719,
      "grad_norm": 0.9754324085832572,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 18719
    },
    {
      "epoch": 0.1872,
      "grad_norm": 1.6455341310963458,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 18720
    },
    {
      "epoch": 0.18721,
      "grad_norm": 1.542051687037108,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 18721
    },
    {
      "epoch": 0.18722,
      "grad_norm": 1.2234910918244877,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 18722
    },
    {
      "epoch": 0.18723,
      "grad_norm": 1.3644131296122846,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 18723
    },
    {
      "epoch": 0.18724,
      "grad_norm": 1.189035087310562,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 18724
    },
    {
      "epoch": 0.18725,
      "grad_norm": 1.20247614424438,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 18725
    },
    {
      "epoch": 0.18726,
      "grad_norm": 1.304558388125256,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 18726
    },
    {
      "epoch": 0.18727,
      "grad_norm": 1.2216612061120402,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 18727
    },
    {
      "epoch": 0.18728,
      "grad_norm": 1.5019065755274033,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 18728
    },
    {
      "epoch": 0.18729,
      "grad_norm": 1.2775538064993097,
      "learning_rate": 0.003,
      "loss": 4.1023,
      "step": 18729
    },
    {
      "epoch": 0.1873,
      "grad_norm": 1.217586267111922,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 18730
    },
    {
      "epoch": 0.18731,
      "grad_norm": 1.3627546761119038,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 18731
    },
    {
      "epoch": 0.18732,
      "grad_norm": 1.2637175176188113,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 18732
    },
    {
      "epoch": 0.18733,
      "grad_norm": 1.2592185013033486,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 18733
    },
    {
      "epoch": 0.18734,
      "grad_norm": 1.3109836776164643,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 18734
    },
    {
      "epoch": 0.18735,
      "grad_norm": 1.0865367938066122,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 18735
    },
    {
      "epoch": 0.18736,
      "grad_norm": 1.4847306463982128,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 18736
    },
    {
      "epoch": 0.18737,
      "grad_norm": 1.2358872148068392,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 18737
    },
    {
      "epoch": 0.18738,
      "grad_norm": 1.6299832826388085,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 18738
    },
    {
      "epoch": 0.18739,
      "grad_norm": 1.1126078032126734,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 18739
    },
    {
      "epoch": 0.1874,
      "grad_norm": 1.7625780690960753,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 18740
    },
    {
      "epoch": 0.18741,
      "grad_norm": 1.1070665620858755,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 18741
    },
    {
      "epoch": 0.18742,
      "grad_norm": 1.3184582434762913,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 18742
    },
    {
      "epoch": 0.18743,
      "grad_norm": 1.4756540855700473,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 18743
    },
    {
      "epoch": 0.18744,
      "grad_norm": 1.111862833143605,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 18744
    },
    {
      "epoch": 0.18745,
      "grad_norm": 1.5106703817912401,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 18745
    },
    {
      "epoch": 0.18746,
      "grad_norm": 1.278505679171913,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 18746
    },
    {
      "epoch": 0.18747,
      "grad_norm": 1.178612535544609,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 18747
    },
    {
      "epoch": 0.18748,
      "grad_norm": 1.5664895067817362,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 18748
    },
    {
      "epoch": 0.18749,
      "grad_norm": 1.0697821444616817,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 18749
    },
    {
      "epoch": 0.1875,
      "grad_norm": 1.4680988615828654,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 18750
    },
    {
      "epoch": 0.18751,
      "grad_norm": 1.2208587576394223,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 18751
    },
    {
      "epoch": 0.18752,
      "grad_norm": 1.2564115783363097,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 18752
    },
    {
      "epoch": 0.18753,
      "grad_norm": 1.2503742948372905,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 18753
    },
    {
      "epoch": 0.18754,
      "grad_norm": 1.3008110241640158,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 18754
    },
    {
      "epoch": 0.18755,
      "grad_norm": 1.4017182808098334,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 18755
    },
    {
      "epoch": 0.18756,
      "grad_norm": 1.1919745606184113,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 18756
    },
    {
      "epoch": 0.18757,
      "grad_norm": 1.4383827916124516,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 18757
    },
    {
      "epoch": 0.18758,
      "grad_norm": 1.1495324332570196,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 18758
    },
    {
      "epoch": 0.18759,
      "grad_norm": 1.255199727608403,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 18759
    },
    {
      "epoch": 0.1876,
      "grad_norm": 1.4521720355970977,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 18760
    },
    {
      "epoch": 0.18761,
      "grad_norm": 1.9297282228126031,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 18761
    },
    {
      "epoch": 0.18762,
      "grad_norm": 1.0610060005854558,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 18762
    },
    {
      "epoch": 0.18763,
      "grad_norm": 1.4842916207178807,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 18763
    },
    {
      "epoch": 0.18764,
      "grad_norm": 1.2970477632330522,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 18764
    },
    {
      "epoch": 0.18765,
      "grad_norm": 1.3987190734282884,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 18765
    },
    {
      "epoch": 0.18766,
      "grad_norm": 1.0971797656792996,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 18766
    },
    {
      "epoch": 0.18767,
      "grad_norm": 1.4950079914981427,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 18767
    },
    {
      "epoch": 0.18768,
      "grad_norm": 1.073136158356457,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 18768
    },
    {
      "epoch": 0.18769,
      "grad_norm": 1.4906139086402133,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 18769
    },
    {
      "epoch": 0.1877,
      "grad_norm": 1.0817221701374877,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 18770
    },
    {
      "epoch": 0.18771,
      "grad_norm": 1.5447318999519204,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 18771
    },
    {
      "epoch": 0.18772,
      "grad_norm": 1.399556958314134,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 18772
    },
    {
      "epoch": 0.18773,
      "grad_norm": 1.5172635406192099,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18773
    },
    {
      "epoch": 0.18774,
      "grad_norm": 1.308669597093372,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 18774
    },
    {
      "epoch": 0.18775,
      "grad_norm": 1.3594877497009628,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 18775
    },
    {
      "epoch": 0.18776,
      "grad_norm": 1.3904321936520263,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 18776
    },
    {
      "epoch": 0.18777,
      "grad_norm": 1.1499811214556928,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 18777
    },
    {
      "epoch": 0.18778,
      "grad_norm": 1.3851139884350505,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 18778
    },
    {
      "epoch": 0.18779,
      "grad_norm": 1.2638471704331713,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 18779
    },
    {
      "epoch": 0.1878,
      "grad_norm": 1.094986868905266,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 18780
    },
    {
      "epoch": 0.18781,
      "grad_norm": 1.5541715116868489,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 18781
    },
    {
      "epoch": 0.18782,
      "grad_norm": 1.4316392568341905,
      "learning_rate": 0.003,
      "loss": 4.0227,
      "step": 18782
    },
    {
      "epoch": 0.18783,
      "grad_norm": 1.4398117848926761,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 18783
    },
    {
      "epoch": 0.18784,
      "grad_norm": 1.2397901403540124,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 18784
    },
    {
      "epoch": 0.18785,
      "grad_norm": 1.0714101734777415,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 18785
    },
    {
      "epoch": 0.18786,
      "grad_norm": 1.5671817059944262,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 18786
    },
    {
      "epoch": 0.18787,
      "grad_norm": 0.8931289534617091,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 18787
    },
    {
      "epoch": 0.18788,
      "grad_norm": 1.1129403927682204,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 18788
    },
    {
      "epoch": 0.18789,
      "grad_norm": 1.163285805940526,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 18789
    },
    {
      "epoch": 0.1879,
      "grad_norm": 1.2622111365194049,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 18790
    },
    {
      "epoch": 0.18791,
      "grad_norm": 1.372286907108706,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 18791
    },
    {
      "epoch": 0.18792,
      "grad_norm": 1.302717304034495,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 18792
    },
    {
      "epoch": 0.18793,
      "grad_norm": 1.5837673759866204,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 18793
    },
    {
      "epoch": 0.18794,
      "grad_norm": 1.217831445467684,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 18794
    },
    {
      "epoch": 0.18795,
      "grad_norm": 1.3698513239886787,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 18795
    },
    {
      "epoch": 0.18796,
      "grad_norm": 1.4402458918864507,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 18796
    },
    {
      "epoch": 0.18797,
      "grad_norm": 1.3757000551166048,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 18797
    },
    {
      "epoch": 0.18798,
      "grad_norm": 1.2050680386905128,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 18798
    },
    {
      "epoch": 0.18799,
      "grad_norm": 1.4800546409412252,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 18799
    },
    {
      "epoch": 0.188,
      "grad_norm": 1.3658678067586925,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 18800
    },
    {
      "epoch": 0.18801,
      "grad_norm": 1.1888682259684493,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 18801
    },
    {
      "epoch": 0.18802,
      "grad_norm": 1.2904681428387519,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 18802
    },
    {
      "epoch": 0.18803,
      "grad_norm": 1.2784827205733253,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 18803
    },
    {
      "epoch": 0.18804,
      "grad_norm": 1.2075887275062585,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 18804
    },
    {
      "epoch": 0.18805,
      "grad_norm": 1.1848057160336685,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 18805
    },
    {
      "epoch": 0.18806,
      "grad_norm": 1.406609644595669,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 18806
    },
    {
      "epoch": 0.18807,
      "grad_norm": 1.0692070001217224,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 18807
    },
    {
      "epoch": 0.18808,
      "grad_norm": 1.7056009090567865,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 18808
    },
    {
      "epoch": 0.18809,
      "grad_norm": 1.2059696983682684,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 18809
    },
    {
      "epoch": 0.1881,
      "grad_norm": 1.4356000840083234,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 18810
    },
    {
      "epoch": 0.18811,
      "grad_norm": 1.192883530310836,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 18811
    },
    {
      "epoch": 0.18812,
      "grad_norm": 1.589611812778463,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 18812
    },
    {
      "epoch": 0.18813,
      "grad_norm": 1.1552372944192821,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 18813
    },
    {
      "epoch": 0.18814,
      "grad_norm": 1.266039013736394,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 18814
    },
    {
      "epoch": 0.18815,
      "grad_norm": 1.460033899528154,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 18815
    },
    {
      "epoch": 0.18816,
      "grad_norm": 1.2268797460089496,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 18816
    },
    {
      "epoch": 0.18817,
      "grad_norm": 1.3280622100516484,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 18817
    },
    {
      "epoch": 0.18818,
      "grad_norm": 1.3585101052879869,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 18818
    },
    {
      "epoch": 0.18819,
      "grad_norm": 1.1813963734584156,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 18819
    },
    {
      "epoch": 0.1882,
      "grad_norm": 1.5781017808410744,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 18820
    },
    {
      "epoch": 0.18821,
      "grad_norm": 1.0583102367636024,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 18821
    },
    {
      "epoch": 0.18822,
      "grad_norm": 1.4281895673023899,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 18822
    },
    {
      "epoch": 0.18823,
      "grad_norm": 1.212896029624748,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 18823
    },
    {
      "epoch": 0.18824,
      "grad_norm": 1.173722315984396,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 18824
    },
    {
      "epoch": 0.18825,
      "grad_norm": 1.2548827205409894,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 18825
    },
    {
      "epoch": 0.18826,
      "grad_norm": 1.1790516501163488,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 18826
    },
    {
      "epoch": 0.18827,
      "grad_norm": 1.63894430103324,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 18827
    },
    {
      "epoch": 0.18828,
      "grad_norm": 1.3132756256823293,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 18828
    },
    {
      "epoch": 0.18829,
      "grad_norm": 1.711445563403175,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 18829
    },
    {
      "epoch": 0.1883,
      "grad_norm": 0.9668264349745779,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 18830
    },
    {
      "epoch": 0.18831,
      "grad_norm": 1.1400888366794983,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 18831
    },
    {
      "epoch": 0.18832,
      "grad_norm": 1.5372213519021534,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 18832
    },
    {
      "epoch": 0.18833,
      "grad_norm": 1.310059535276661,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 18833
    },
    {
      "epoch": 0.18834,
      "grad_norm": 1.363509962833014,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 18834
    },
    {
      "epoch": 0.18835,
      "grad_norm": 1.4236206910331353,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 18835
    },
    {
      "epoch": 0.18836,
      "grad_norm": 1.3900855704423931,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 18836
    },
    {
      "epoch": 0.18837,
      "grad_norm": 1.1567946586906692,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 18837
    },
    {
      "epoch": 0.18838,
      "grad_norm": 1.4650855471840627,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 18838
    },
    {
      "epoch": 0.18839,
      "grad_norm": 1.2082181065027349,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 18839
    },
    {
      "epoch": 0.1884,
      "grad_norm": 1.3100334618833482,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 18840
    },
    {
      "epoch": 0.18841,
      "grad_norm": 1.2898106179869984,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 18841
    },
    {
      "epoch": 0.18842,
      "grad_norm": 1.3447100530946252,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 18842
    },
    {
      "epoch": 0.18843,
      "grad_norm": 1.4061202076551023,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 18843
    },
    {
      "epoch": 0.18844,
      "grad_norm": 1.4189622114945963,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 18844
    },
    {
      "epoch": 0.18845,
      "grad_norm": 1.4081270069890108,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 18845
    },
    {
      "epoch": 0.18846,
      "grad_norm": 1.266243240453215,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 18846
    },
    {
      "epoch": 0.18847,
      "grad_norm": 1.3470242607308647,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 18847
    },
    {
      "epoch": 0.18848,
      "grad_norm": 1.3204966376292842,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 18848
    },
    {
      "epoch": 0.18849,
      "grad_norm": 1.5356699983353543,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 18849
    },
    {
      "epoch": 0.1885,
      "grad_norm": 1.0159267477280214,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 18850
    },
    {
      "epoch": 0.18851,
      "grad_norm": 1.4983569056145334,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 18851
    },
    {
      "epoch": 0.18852,
      "grad_norm": 1.2704894434822271,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 18852
    },
    {
      "epoch": 0.18853,
      "grad_norm": 1.3369779797887722,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 18853
    },
    {
      "epoch": 0.18854,
      "grad_norm": 1.3035438538446587,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 18854
    },
    {
      "epoch": 0.18855,
      "grad_norm": 1.220516196000558,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 18855
    },
    {
      "epoch": 0.18856,
      "grad_norm": 1.3808656416840817,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 18856
    },
    {
      "epoch": 0.18857,
      "grad_norm": 1.163717869471003,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 18857
    },
    {
      "epoch": 0.18858,
      "grad_norm": 1.5106764678560334,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 18858
    },
    {
      "epoch": 0.18859,
      "grad_norm": 1.153077200909646,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 18859
    },
    {
      "epoch": 0.1886,
      "grad_norm": 1.603833161245429,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 18860
    },
    {
      "epoch": 0.18861,
      "grad_norm": 1.2340982997403442,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 18861
    },
    {
      "epoch": 0.18862,
      "grad_norm": 1.5173359019104398,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 18862
    },
    {
      "epoch": 0.18863,
      "grad_norm": 1.1831778405636078,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 18863
    },
    {
      "epoch": 0.18864,
      "grad_norm": 1.2135050745339184,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 18864
    },
    {
      "epoch": 0.18865,
      "grad_norm": 1.3365944063600408,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 18865
    },
    {
      "epoch": 0.18866,
      "grad_norm": 1.1609925593180206,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 18866
    },
    {
      "epoch": 0.18867,
      "grad_norm": 1.2892142345985795,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 18867
    },
    {
      "epoch": 0.18868,
      "grad_norm": 1.1249370401528647,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 18868
    },
    {
      "epoch": 0.18869,
      "grad_norm": 1.5062388002670444,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 18869
    },
    {
      "epoch": 0.1887,
      "grad_norm": 1.526121507824576,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 18870
    },
    {
      "epoch": 0.18871,
      "grad_norm": 1.420092814956437,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 18871
    },
    {
      "epoch": 0.18872,
      "grad_norm": 1.2955615793460946,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 18872
    },
    {
      "epoch": 0.18873,
      "grad_norm": 1.3497401636635546,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 18873
    },
    {
      "epoch": 0.18874,
      "grad_norm": 1.3588048814689848,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 18874
    },
    {
      "epoch": 0.18875,
      "grad_norm": 1.1401782635831865,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 18875
    },
    {
      "epoch": 0.18876,
      "grad_norm": 1.5040856671285598,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 18876
    },
    {
      "epoch": 0.18877,
      "grad_norm": 1.153114053739368,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 18877
    },
    {
      "epoch": 0.18878,
      "grad_norm": 1.4876661397458029,
      "learning_rate": 0.003,
      "loss": 4.0385,
      "step": 18878
    },
    {
      "epoch": 0.18879,
      "grad_norm": 1.1516952756780263,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 18879
    },
    {
      "epoch": 0.1888,
      "grad_norm": 1.5103888507625314,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 18880
    },
    {
      "epoch": 0.18881,
      "grad_norm": 1.102260589916007,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 18881
    },
    {
      "epoch": 0.18882,
      "grad_norm": 1.4428875482988193,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 18882
    },
    {
      "epoch": 0.18883,
      "grad_norm": 1.2718188780518804,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 18883
    },
    {
      "epoch": 0.18884,
      "grad_norm": 1.3704925061754005,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 18884
    },
    {
      "epoch": 0.18885,
      "grad_norm": 1.4976345293649658,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 18885
    },
    {
      "epoch": 0.18886,
      "grad_norm": 1.2137579687025972,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 18886
    },
    {
      "epoch": 0.18887,
      "grad_norm": 1.4264884660689312,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 18887
    },
    {
      "epoch": 0.18888,
      "grad_norm": 1.0713536509148665,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 18888
    },
    {
      "epoch": 0.18889,
      "grad_norm": 1.5868036919963195,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 18889
    },
    {
      "epoch": 0.1889,
      "grad_norm": 1.180432126330319,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 18890
    },
    {
      "epoch": 0.18891,
      "grad_norm": 1.521492894557538,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 18891
    },
    {
      "epoch": 0.18892,
      "grad_norm": 1.1078770928910782,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 18892
    },
    {
      "epoch": 0.18893,
      "grad_norm": 1.7356931265094164,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 18893
    },
    {
      "epoch": 0.18894,
      "grad_norm": 1.317253082983154,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 18894
    },
    {
      "epoch": 0.18895,
      "grad_norm": 1.1648419331625475,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 18895
    },
    {
      "epoch": 0.18896,
      "grad_norm": 1.250255263445163,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 18896
    },
    {
      "epoch": 0.18897,
      "grad_norm": 1.4344943793092255,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 18897
    },
    {
      "epoch": 0.18898,
      "grad_norm": 1.3279246217317304,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 18898
    },
    {
      "epoch": 0.18899,
      "grad_norm": 1.1889584260873742,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 18899
    },
    {
      "epoch": 0.189,
      "grad_norm": 1.316494368314,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 18900
    },
    {
      "epoch": 0.18901,
      "grad_norm": 1.2230937412314116,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 18901
    },
    {
      "epoch": 0.18902,
      "grad_norm": 1.3945763924516088,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 18902
    },
    {
      "epoch": 0.18903,
      "grad_norm": 1.2739696174630883,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 18903
    },
    {
      "epoch": 0.18904,
      "grad_norm": 1.5827069235789044,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 18904
    },
    {
      "epoch": 0.18905,
      "grad_norm": 1.3636490487827087,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 18905
    },
    {
      "epoch": 0.18906,
      "grad_norm": 1.3408230077629284,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 18906
    },
    {
      "epoch": 0.18907,
      "grad_norm": 1.2225650484016408,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 18907
    },
    {
      "epoch": 0.18908,
      "grad_norm": 1.4123102155843825,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 18908
    },
    {
      "epoch": 0.18909,
      "grad_norm": 1.3793049252028673,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 18909
    },
    {
      "epoch": 0.1891,
      "grad_norm": 1.3107955629341879,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 18910
    },
    {
      "epoch": 0.18911,
      "grad_norm": 1.228759283452415,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 18911
    },
    {
      "epoch": 0.18912,
      "grad_norm": 1.4299315989229704,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 18912
    },
    {
      "epoch": 0.18913,
      "grad_norm": 1.1500243667154835,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 18913
    },
    {
      "epoch": 0.18914,
      "grad_norm": 1.2391305076804604,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 18914
    },
    {
      "epoch": 0.18915,
      "grad_norm": 1.383468763742884,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 18915
    },
    {
      "epoch": 0.18916,
      "grad_norm": 1.1880879138668168,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 18916
    },
    {
      "epoch": 0.18917,
      "grad_norm": 1.5951028679607262,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 18917
    },
    {
      "epoch": 0.18918,
      "grad_norm": 1.2260422838236318,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 18918
    },
    {
      "epoch": 0.18919,
      "grad_norm": 1.375698077238036,
      "learning_rate": 0.003,
      "loss": 4.0871,
      "step": 18919
    },
    {
      "epoch": 0.1892,
      "grad_norm": 1.262268901110128,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 18920
    },
    {
      "epoch": 0.18921,
      "grad_norm": 1.235518346439444,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 18921
    },
    {
      "epoch": 0.18922,
      "grad_norm": 1.447977057278561,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 18922
    },
    {
      "epoch": 0.18923,
      "grad_norm": 1.261849012928809,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 18923
    },
    {
      "epoch": 0.18924,
      "grad_norm": 1.5017822769392577,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 18924
    },
    {
      "epoch": 0.18925,
      "grad_norm": 1.3941361919123514,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 18925
    },
    {
      "epoch": 0.18926,
      "grad_norm": 1.5368017967326983,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 18926
    },
    {
      "epoch": 0.18927,
      "grad_norm": 1.1538779990492576,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 18927
    },
    {
      "epoch": 0.18928,
      "grad_norm": 1.5444160847420918,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 18928
    },
    {
      "epoch": 0.18929,
      "grad_norm": 0.9385130249989668,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 18929
    },
    {
      "epoch": 0.1893,
      "grad_norm": 1.3564797100958084,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 18930
    },
    {
      "epoch": 0.18931,
      "grad_norm": 1.0335793616368942,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 18931
    },
    {
      "epoch": 0.18932,
      "grad_norm": 1.3010390289418123,
      "learning_rate": 0.003,
      "loss": 4.1056,
      "step": 18932
    },
    {
      "epoch": 0.18933,
      "grad_norm": 1.437018943007172,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 18933
    },
    {
      "epoch": 0.18934,
      "grad_norm": 1.4044142852135095,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 18934
    },
    {
      "epoch": 0.18935,
      "grad_norm": 1.4813851489409189,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 18935
    },
    {
      "epoch": 0.18936,
      "grad_norm": 0.9988193921991353,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 18936
    },
    {
      "epoch": 0.18937,
      "grad_norm": 1.3695145211207798,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 18937
    },
    {
      "epoch": 0.18938,
      "grad_norm": 1.2883443280092062,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 18938
    },
    {
      "epoch": 0.18939,
      "grad_norm": 1.1970784165724995,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 18939
    },
    {
      "epoch": 0.1894,
      "grad_norm": 1.2735363899786794,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 18940
    },
    {
      "epoch": 0.18941,
      "grad_norm": 1.3485978322242238,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 18941
    },
    {
      "epoch": 0.18942,
      "grad_norm": 0.9973824337445716,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 18942
    },
    {
      "epoch": 0.18943,
      "grad_norm": 1.7790042413316978,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 18943
    },
    {
      "epoch": 0.18944,
      "grad_norm": 1.4669139722319338,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 18944
    },
    {
      "epoch": 0.18945,
      "grad_norm": 1.4581928468055323,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 18945
    },
    {
      "epoch": 0.18946,
      "grad_norm": 1.1671636085692663,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 18946
    },
    {
      "epoch": 0.18947,
      "grad_norm": 1.3777204974748212,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 18947
    },
    {
      "epoch": 0.18948,
      "grad_norm": 1.1778477242902337,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 18948
    },
    {
      "epoch": 0.18949,
      "grad_norm": 1.3936343034445549,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 18949
    },
    {
      "epoch": 0.1895,
      "grad_norm": 1.222772542255314,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 18950
    },
    {
      "epoch": 0.18951,
      "grad_norm": 1.2931222145801997,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 18951
    },
    {
      "epoch": 0.18952,
      "grad_norm": 1.0958641638209,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 18952
    },
    {
      "epoch": 0.18953,
      "grad_norm": 1.497273939684197,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 18953
    },
    {
      "epoch": 0.18954,
      "grad_norm": 1.336175191600886,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 18954
    },
    {
      "epoch": 0.18955,
      "grad_norm": 1.5511757924893705,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 18955
    },
    {
      "epoch": 0.18956,
      "grad_norm": 1.12151388342686,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 18956
    },
    {
      "epoch": 0.18957,
      "grad_norm": 1.2714774177311274,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 18957
    },
    {
      "epoch": 0.18958,
      "grad_norm": 1.806335271820712,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 18958
    },
    {
      "epoch": 0.18959,
      "grad_norm": 1.1981557925984667,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 18959
    },
    {
      "epoch": 0.1896,
      "grad_norm": 1.2877466637724115,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 18960
    },
    {
      "epoch": 0.18961,
      "grad_norm": 1.2289050495891696,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 18961
    },
    {
      "epoch": 0.18962,
      "grad_norm": 1.3242121522869408,
      "learning_rate": 0.003,
      "loss": 4.1004,
      "step": 18962
    },
    {
      "epoch": 0.18963,
      "grad_norm": 1.2268292390487603,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 18963
    },
    {
      "epoch": 0.18964,
      "grad_norm": 1.4286199540648732,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 18964
    },
    {
      "epoch": 0.18965,
      "grad_norm": 1.1307354889118555,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 18965
    },
    {
      "epoch": 0.18966,
      "grad_norm": 1.3616380925033746,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 18966
    },
    {
      "epoch": 0.18967,
      "grad_norm": 1.3270887278372299,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 18967
    },
    {
      "epoch": 0.18968,
      "grad_norm": 1.180291146237884,
      "learning_rate": 0.003,
      "loss": 4.0874,
      "step": 18968
    },
    {
      "epoch": 0.18969,
      "grad_norm": 1.4248134973281465,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 18969
    },
    {
      "epoch": 0.1897,
      "grad_norm": 1.857430428677563,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 18970
    },
    {
      "epoch": 0.18971,
      "grad_norm": 1.513573264824216,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 18971
    },
    {
      "epoch": 0.18972,
      "grad_norm": 1.2313827860976658,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 18972
    },
    {
      "epoch": 0.18973,
      "grad_norm": 1.567017247104148,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 18973
    },
    {
      "epoch": 0.18974,
      "grad_norm": 1.0781847216389804,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 18974
    },
    {
      "epoch": 0.18975,
      "grad_norm": 1.2342507716478786,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 18975
    },
    {
      "epoch": 0.18976,
      "grad_norm": 1.4390172817298563,
      "learning_rate": 0.003,
      "loss": 4.0995,
      "step": 18976
    },
    {
      "epoch": 0.18977,
      "grad_norm": 1.119764135955646,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 18977
    },
    {
      "epoch": 0.18978,
      "grad_norm": 1.4875556521970705,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 18978
    },
    {
      "epoch": 0.18979,
      "grad_norm": 1.3524860210818532,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 18979
    },
    {
      "epoch": 0.1898,
      "grad_norm": 1.0635357406263535,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 18980
    },
    {
      "epoch": 0.18981,
      "grad_norm": 1.3099436010284684,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 18981
    },
    {
      "epoch": 0.18982,
      "grad_norm": 1.1859148488860554,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 18982
    },
    {
      "epoch": 0.18983,
      "grad_norm": 1.6161576957549602,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 18983
    },
    {
      "epoch": 0.18984,
      "grad_norm": 1.8139644629208311,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 18984
    },
    {
      "epoch": 0.18985,
      "grad_norm": 1.0784425662683723,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 18985
    },
    {
      "epoch": 0.18986,
      "grad_norm": 1.363183551893584,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 18986
    },
    {
      "epoch": 0.18987,
      "grad_norm": 1.2599305508297793,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 18987
    },
    {
      "epoch": 0.18988,
      "grad_norm": 1.4277124939895625,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 18988
    },
    {
      "epoch": 0.18989,
      "grad_norm": 1.3098668231382307,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 18989
    },
    {
      "epoch": 0.1899,
      "grad_norm": 1.5675414246730872,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 18990
    },
    {
      "epoch": 0.18991,
      "grad_norm": 1.240194226915464,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 18991
    },
    {
      "epoch": 0.18992,
      "grad_norm": 1.1193047959233318,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 18992
    },
    {
      "epoch": 0.18993,
      "grad_norm": 1.420012937103303,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 18993
    },
    {
      "epoch": 0.18994,
      "grad_norm": 1.1123855253231913,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 18994
    },
    {
      "epoch": 0.18995,
      "grad_norm": 1.2576329907124797,
      "learning_rate": 0.003,
      "loss": 4.0096,
      "step": 18995
    },
    {
      "epoch": 0.18996,
      "grad_norm": 1.2610917729669,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 18996
    },
    {
      "epoch": 0.18997,
      "grad_norm": 1.3324235682068313,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 18997
    },
    {
      "epoch": 0.18998,
      "grad_norm": 1.2416667836230513,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 18998
    },
    {
      "epoch": 0.18999,
      "grad_norm": 1.4512750335089,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 18999
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2324685620970226,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 19000
    },
    {
      "epoch": 0.19001,
      "grad_norm": 1.5577462599745568,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 19001
    },
    {
      "epoch": 0.19002,
      "grad_norm": 1.2889729802796936,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 19002
    },
    {
      "epoch": 0.19003,
      "grad_norm": 1.233507256766918,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 19003
    },
    {
      "epoch": 0.19004,
      "grad_norm": 1.1803096529839874,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 19004
    },
    {
      "epoch": 0.19005,
      "grad_norm": 1.1170026976717313,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 19005
    },
    {
      "epoch": 0.19006,
      "grad_norm": 1.325523253601187,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 19006
    },
    {
      "epoch": 0.19007,
      "grad_norm": 1.2934096707555685,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 19007
    },
    {
      "epoch": 0.19008,
      "grad_norm": 1.4935043714232368,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 19008
    },
    {
      "epoch": 0.19009,
      "grad_norm": 1.3216276747957934,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 19009
    },
    {
      "epoch": 0.1901,
      "grad_norm": 1.2591839637952178,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 19010
    },
    {
      "epoch": 0.19011,
      "grad_norm": 1.6761998989258382,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 19011
    },
    {
      "epoch": 0.19012,
      "grad_norm": 1.1244001332035305,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 19012
    },
    {
      "epoch": 0.19013,
      "grad_norm": 1.5815884330810812,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 19013
    },
    {
      "epoch": 0.19014,
      "grad_norm": 1.5778195952980243,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 19014
    },
    {
      "epoch": 0.19015,
      "grad_norm": 1.153662316779612,
      "learning_rate": 0.003,
      "loss": 4.0231,
      "step": 19015
    },
    {
      "epoch": 0.19016,
      "grad_norm": 1.1217013138145722,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 19016
    },
    {
      "epoch": 0.19017,
      "grad_norm": 1.3396995051527274,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 19017
    },
    {
      "epoch": 0.19018,
      "grad_norm": 1.3900040315464735,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 19018
    },
    {
      "epoch": 0.19019,
      "grad_norm": 1.392132345133188,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 19019
    },
    {
      "epoch": 0.1902,
      "grad_norm": 1.4664707424192767,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 19020
    },
    {
      "epoch": 0.19021,
      "grad_norm": 1.0816735042234542,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 19021
    },
    {
      "epoch": 0.19022,
      "grad_norm": 1.222139850489423,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 19022
    },
    {
      "epoch": 0.19023,
      "grad_norm": 1.3113447127507638,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 19023
    },
    {
      "epoch": 0.19024,
      "grad_norm": 1.41514397939476,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 19024
    },
    {
      "epoch": 0.19025,
      "grad_norm": 0.9802130337406161,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 19025
    },
    {
      "epoch": 0.19026,
      "grad_norm": 1.5998213443286031,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 19026
    },
    {
      "epoch": 0.19027,
      "grad_norm": 1.1665135813741723,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 19027
    },
    {
      "epoch": 0.19028,
      "grad_norm": 1.4020630133015164,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 19028
    },
    {
      "epoch": 0.19029,
      "grad_norm": 1.4448346034862,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 19029
    },
    {
      "epoch": 0.1903,
      "grad_norm": 1.0550181708031718,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 19030
    },
    {
      "epoch": 0.19031,
      "grad_norm": 1.495096570441009,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 19031
    },
    {
      "epoch": 0.19032,
      "grad_norm": 1.1932281477677271,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 19032
    },
    {
      "epoch": 0.19033,
      "grad_norm": 1.5091186667250431,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 19033
    },
    {
      "epoch": 0.19034,
      "grad_norm": 1.3843832633303428,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 19034
    },
    {
      "epoch": 0.19035,
      "grad_norm": 1.314486318500889,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 19035
    },
    {
      "epoch": 0.19036,
      "grad_norm": 1.2069363457552786,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 19036
    },
    {
      "epoch": 0.19037,
      "grad_norm": 1.3615790478931433,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 19037
    },
    {
      "epoch": 0.19038,
      "grad_norm": 1.193499137863031,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 19038
    },
    {
      "epoch": 0.19039,
      "grad_norm": 1.2939501145230716,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 19039
    },
    {
      "epoch": 0.1904,
      "grad_norm": 1.2528639930786927,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 19040
    },
    {
      "epoch": 0.19041,
      "grad_norm": 1.471674616815551,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 19041
    },
    {
      "epoch": 0.19042,
      "grad_norm": 1.314049431909399,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19042
    },
    {
      "epoch": 0.19043,
      "grad_norm": 1.3624689771020175,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 19043
    },
    {
      "epoch": 0.19044,
      "grad_norm": 1.4134190169893828,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 19044
    },
    {
      "epoch": 0.19045,
      "grad_norm": 1.0960238181783737,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 19045
    },
    {
      "epoch": 0.19046,
      "grad_norm": 1.5017184629957616,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 19046
    },
    {
      "epoch": 0.19047,
      "grad_norm": 1.232097844273087,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 19047
    },
    {
      "epoch": 0.19048,
      "grad_norm": 1.3400956265663477,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19048
    },
    {
      "epoch": 0.19049,
      "grad_norm": 1.2470828259434086,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 19049
    },
    {
      "epoch": 0.1905,
      "grad_norm": 1.5793037547965862,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 19050
    },
    {
      "epoch": 0.19051,
      "grad_norm": 1.338329176891238,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 19051
    },
    {
      "epoch": 0.19052,
      "grad_norm": 1.1619816182998715,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 19052
    },
    {
      "epoch": 0.19053,
      "grad_norm": 1.386888683759865,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 19053
    },
    {
      "epoch": 0.19054,
      "grad_norm": 1.1400577701730739,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 19054
    },
    {
      "epoch": 0.19055,
      "grad_norm": 1.5485345322855846,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 19055
    },
    {
      "epoch": 0.19056,
      "grad_norm": 1.1029821679021268,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 19056
    },
    {
      "epoch": 0.19057,
      "grad_norm": 1.449446935040205,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 19057
    },
    {
      "epoch": 0.19058,
      "grad_norm": 1.3637159846302798,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 19058
    },
    {
      "epoch": 0.19059,
      "grad_norm": 1.436568727485353,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 19059
    },
    {
      "epoch": 0.1906,
      "grad_norm": 1.0690561626750321,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 19060
    },
    {
      "epoch": 0.19061,
      "grad_norm": 1.602754281179194,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 19061
    },
    {
      "epoch": 0.19062,
      "grad_norm": 1.137154741975126,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 19062
    },
    {
      "epoch": 0.19063,
      "grad_norm": 1.4750047660200645,
      "learning_rate": 0.003,
      "loss": 4.0291,
      "step": 19063
    },
    {
      "epoch": 0.19064,
      "grad_norm": 1.0672597314943837,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 19064
    },
    {
      "epoch": 0.19065,
      "grad_norm": 1.6459411674301923,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 19065
    },
    {
      "epoch": 0.19066,
      "grad_norm": 1.1575056516494613,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 19066
    },
    {
      "epoch": 0.19067,
      "grad_norm": 1.232617210848777,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 19067
    },
    {
      "epoch": 0.19068,
      "grad_norm": 1.3765978890878052,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 19068
    },
    {
      "epoch": 0.19069,
      "grad_norm": 1.4613355184334849,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 19069
    },
    {
      "epoch": 0.1907,
      "grad_norm": 1.3304262962194269,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 19070
    },
    {
      "epoch": 0.19071,
      "grad_norm": 1.3375188461800747,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 19071
    },
    {
      "epoch": 0.19072,
      "grad_norm": 1.5392331002690696,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 19072
    },
    {
      "epoch": 0.19073,
      "grad_norm": 1.4565796580099446,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 19073
    },
    {
      "epoch": 0.19074,
      "grad_norm": 1.2720043502537839,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 19074
    },
    {
      "epoch": 0.19075,
      "grad_norm": 1.3716543617079564,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 19075
    },
    {
      "epoch": 0.19076,
      "grad_norm": 1.3971309426935,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 19076
    },
    {
      "epoch": 0.19077,
      "grad_norm": 1.30982799094468,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 19077
    },
    {
      "epoch": 0.19078,
      "grad_norm": 1.1911141610284892,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 19078
    },
    {
      "epoch": 0.19079,
      "grad_norm": 1.2083749238417725,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 19079
    },
    {
      "epoch": 0.1908,
      "grad_norm": 1.2449723095561924,
      "learning_rate": 0.003,
      "loss": 4.0309,
      "step": 19080
    },
    {
      "epoch": 0.19081,
      "grad_norm": 1.3998045528501801,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 19081
    },
    {
      "epoch": 0.19082,
      "grad_norm": 1.4531284576949606,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 19082
    },
    {
      "epoch": 0.19083,
      "grad_norm": 1.064278230314246,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 19083
    },
    {
      "epoch": 0.19084,
      "grad_norm": 1.5083208229316238,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 19084
    },
    {
      "epoch": 0.19085,
      "grad_norm": 1.5000500657542444,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 19085
    },
    {
      "epoch": 0.19086,
      "grad_norm": 1.0974791363157106,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 19086
    },
    {
      "epoch": 0.19087,
      "grad_norm": 1.4513655729032795,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 19087
    },
    {
      "epoch": 0.19088,
      "grad_norm": 1.2977784897854003,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 19088
    },
    {
      "epoch": 0.19089,
      "grad_norm": 1.1013679680267832,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 19089
    },
    {
      "epoch": 0.1909,
      "grad_norm": 1.4129522572596733,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 19090
    },
    {
      "epoch": 0.19091,
      "grad_norm": 1.207696078897892,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 19091
    },
    {
      "epoch": 0.19092,
      "grad_norm": 1.4232598149015803,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 19092
    },
    {
      "epoch": 0.19093,
      "grad_norm": 1.1972431576898495,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 19093
    },
    {
      "epoch": 0.19094,
      "grad_norm": 1.5183181645421888,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 19094
    },
    {
      "epoch": 0.19095,
      "grad_norm": 1.221872072035716,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 19095
    },
    {
      "epoch": 0.19096,
      "grad_norm": 1.4396452999024958,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 19096
    },
    {
      "epoch": 0.19097,
      "grad_norm": 1.2246180936584108,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 19097
    },
    {
      "epoch": 0.19098,
      "grad_norm": 1.4125094881948392,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 19098
    },
    {
      "epoch": 0.19099,
      "grad_norm": 1.190535711223137,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 19099
    },
    {
      "epoch": 0.191,
      "grad_norm": 1.320284979939749,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 19100
    },
    {
      "epoch": 0.19101,
      "grad_norm": 1.2820570689852735,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 19101
    },
    {
      "epoch": 0.19102,
      "grad_norm": 1.640797367448142,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 19102
    },
    {
      "epoch": 0.19103,
      "grad_norm": 1.7711658876323253,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 19103
    },
    {
      "epoch": 0.19104,
      "grad_norm": 1.1898220312387047,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 19104
    },
    {
      "epoch": 0.19105,
      "grad_norm": 1.242179886486909,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 19105
    },
    {
      "epoch": 0.19106,
      "grad_norm": 1.3358435037354868,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 19106
    },
    {
      "epoch": 0.19107,
      "grad_norm": 1.3206693791196868,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 19107
    },
    {
      "epoch": 0.19108,
      "grad_norm": 1.3219251111714507,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 19108
    },
    {
      "epoch": 0.19109,
      "grad_norm": 1.3692187282500543,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 19109
    },
    {
      "epoch": 0.1911,
      "grad_norm": 1.2516639706729331,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 19110
    },
    {
      "epoch": 0.19111,
      "grad_norm": 1.3579748756501073,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 19111
    },
    {
      "epoch": 0.19112,
      "grad_norm": 1.2901685774005183,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 19112
    },
    {
      "epoch": 0.19113,
      "grad_norm": 1.3695313287734234,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 19113
    },
    {
      "epoch": 0.19114,
      "grad_norm": 1.1943068768193779,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 19114
    },
    {
      "epoch": 0.19115,
      "grad_norm": 1.3324190801436753,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 19115
    },
    {
      "epoch": 0.19116,
      "grad_norm": 1.0481218005906356,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 19116
    },
    {
      "epoch": 0.19117,
      "grad_norm": 1.2175372768402644,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 19117
    },
    {
      "epoch": 0.19118,
      "grad_norm": 1.1851769264419512,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 19118
    },
    {
      "epoch": 0.19119,
      "grad_norm": 1.6243204774157487,
      "learning_rate": 0.003,
      "loss": 4.114,
      "step": 19119
    },
    {
      "epoch": 0.1912,
      "grad_norm": 1.134277528181548,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 19120
    },
    {
      "epoch": 0.19121,
      "grad_norm": 1.3437022589300889,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 19121
    },
    {
      "epoch": 0.19122,
      "grad_norm": 1.3333603930491609,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 19122
    },
    {
      "epoch": 0.19123,
      "grad_norm": 1.2745079718137873,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 19123
    },
    {
      "epoch": 0.19124,
      "grad_norm": 1.1960491080657072,
      "learning_rate": 0.003,
      "loss": 4.0955,
      "step": 19124
    },
    {
      "epoch": 0.19125,
      "grad_norm": 1.3172844462471818,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 19125
    },
    {
      "epoch": 0.19126,
      "grad_norm": 1.4834170903173642,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 19126
    },
    {
      "epoch": 0.19127,
      "grad_norm": 1.812018123397985,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 19127
    },
    {
      "epoch": 0.19128,
      "grad_norm": 1.1098226632687962,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 19128
    },
    {
      "epoch": 0.19129,
      "grad_norm": 1.344914180637586,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 19129
    },
    {
      "epoch": 0.1913,
      "grad_norm": 1.2581708893061798,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 19130
    },
    {
      "epoch": 0.19131,
      "grad_norm": 1.240574980726106,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 19131
    },
    {
      "epoch": 0.19132,
      "grad_norm": 1.6100387528656463,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 19132
    },
    {
      "epoch": 0.19133,
      "grad_norm": 1.242291472166886,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 19133
    },
    {
      "epoch": 0.19134,
      "grad_norm": 1.4652522360740017,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 19134
    },
    {
      "epoch": 0.19135,
      "grad_norm": 1.3220586428121326,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 19135
    },
    {
      "epoch": 0.19136,
      "grad_norm": 1.1979879985446065,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 19136
    },
    {
      "epoch": 0.19137,
      "grad_norm": 1.2245714794934948,
      "learning_rate": 0.003,
      "loss": 4.0312,
      "step": 19137
    },
    {
      "epoch": 0.19138,
      "grad_norm": 1.2533839335343562,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 19138
    },
    {
      "epoch": 0.19139,
      "grad_norm": 1.2597911924182927,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 19139
    },
    {
      "epoch": 0.1914,
      "grad_norm": 1.0086352156745937,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 19140
    },
    {
      "epoch": 0.19141,
      "grad_norm": 1.5899559027790509,
      "learning_rate": 0.003,
      "loss": 4.0244,
      "step": 19141
    },
    {
      "epoch": 0.19142,
      "grad_norm": 1.5939739338876755,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 19142
    },
    {
      "epoch": 0.19143,
      "grad_norm": 1.3287884363381772,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 19143
    },
    {
      "epoch": 0.19144,
      "grad_norm": 1.3587858504048773,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 19144
    },
    {
      "epoch": 0.19145,
      "grad_norm": 1.3404937183657344,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 19145
    },
    {
      "epoch": 0.19146,
      "grad_norm": 1.270922752836181,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 19146
    },
    {
      "epoch": 0.19147,
      "grad_norm": 1.2147556481989903,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 19147
    },
    {
      "epoch": 0.19148,
      "grad_norm": 1.3497960230796922,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 19148
    },
    {
      "epoch": 0.19149,
      "grad_norm": 1.256723566268377,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 19149
    },
    {
      "epoch": 0.1915,
      "grad_norm": 1.3643278964494576,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 19150
    },
    {
      "epoch": 0.19151,
      "grad_norm": 1.2185081050635174,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 19151
    },
    {
      "epoch": 0.19152,
      "grad_norm": 1.317530622128689,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 19152
    },
    {
      "epoch": 0.19153,
      "grad_norm": 1.2245685697754942,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 19153
    },
    {
      "epoch": 0.19154,
      "grad_norm": 1.2319047829270333,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 19154
    },
    {
      "epoch": 0.19155,
      "grad_norm": 1.3030123854433888,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 19155
    },
    {
      "epoch": 0.19156,
      "grad_norm": 1.1758000665998163,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 19156
    },
    {
      "epoch": 0.19157,
      "grad_norm": 1.3547531756373572,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 19157
    },
    {
      "epoch": 0.19158,
      "grad_norm": 1.4880671874548606,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 19158
    },
    {
      "epoch": 0.19159,
      "grad_norm": 1.2509190691711998,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 19159
    },
    {
      "epoch": 0.1916,
      "grad_norm": 1.3289551859332054,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 19160
    },
    {
      "epoch": 0.19161,
      "grad_norm": 1.5236236955712956,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 19161
    },
    {
      "epoch": 0.19162,
      "grad_norm": 1.3370152625573213,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 19162
    },
    {
      "epoch": 0.19163,
      "grad_norm": 1.6548746772311307,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 19163
    },
    {
      "epoch": 0.19164,
      "grad_norm": 1.5183876875667157,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 19164
    },
    {
      "epoch": 0.19165,
      "grad_norm": 1.3947780331195114,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 19165
    },
    {
      "epoch": 0.19166,
      "grad_norm": 1.2883469168643833,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 19166
    },
    {
      "epoch": 0.19167,
      "grad_norm": 1.2956180033322826,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 19167
    },
    {
      "epoch": 0.19168,
      "grad_norm": 1.3990904680365333,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 19168
    },
    {
      "epoch": 0.19169,
      "grad_norm": 1.1364576534154611,
      "learning_rate": 0.003,
      "loss": 4.0179,
      "step": 19169
    },
    {
      "epoch": 0.1917,
      "grad_norm": 1.6642795045391565,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 19170
    },
    {
      "epoch": 0.19171,
      "grad_norm": 1.2925604382023206,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 19171
    },
    {
      "epoch": 0.19172,
      "grad_norm": 1.3464205038024264,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 19172
    },
    {
      "epoch": 0.19173,
      "grad_norm": 1.3202152463158157,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 19173
    },
    {
      "epoch": 0.19174,
      "grad_norm": 1.382927722314029,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 19174
    },
    {
      "epoch": 0.19175,
      "grad_norm": 1.422565007939475,
      "learning_rate": 0.003,
      "loss": 4.0987,
      "step": 19175
    },
    {
      "epoch": 0.19176,
      "grad_norm": 1.677057405662981,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 19176
    },
    {
      "epoch": 0.19177,
      "grad_norm": 1.2214924954573159,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 19177
    },
    {
      "epoch": 0.19178,
      "grad_norm": 1.3088438877680733,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 19178
    },
    {
      "epoch": 0.19179,
      "grad_norm": 1.5170282665335622,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 19179
    },
    {
      "epoch": 0.1918,
      "grad_norm": 1.2040190205765118,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 19180
    },
    {
      "epoch": 0.19181,
      "grad_norm": 1.3822570608840876,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 19181
    },
    {
      "epoch": 0.19182,
      "grad_norm": 1.172484815976945,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 19182
    },
    {
      "epoch": 0.19183,
      "grad_norm": 1.274388763595249,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 19183
    },
    {
      "epoch": 0.19184,
      "grad_norm": 1.2924607758107507,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 19184
    },
    {
      "epoch": 0.19185,
      "grad_norm": 1.2128103145138238,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 19185
    },
    {
      "epoch": 0.19186,
      "grad_norm": 1.3739857604766152,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 19186
    },
    {
      "epoch": 0.19187,
      "grad_norm": 1.1002419192622068,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 19187
    },
    {
      "epoch": 0.19188,
      "grad_norm": 1.297982852715103,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 19188
    },
    {
      "epoch": 0.19189,
      "grad_norm": 1.3339832824292281,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 19189
    },
    {
      "epoch": 0.1919,
      "grad_norm": 1.0348721776220091,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 19190
    },
    {
      "epoch": 0.19191,
      "grad_norm": 1.6570408605160454,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 19191
    },
    {
      "epoch": 0.19192,
      "grad_norm": 1.3945724289140948,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 19192
    },
    {
      "epoch": 0.19193,
      "grad_norm": 1.5912266691413073,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 19193
    },
    {
      "epoch": 0.19194,
      "grad_norm": 1.4058863699195427,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 19194
    },
    {
      "epoch": 0.19195,
      "grad_norm": 1.356988434286127,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 19195
    },
    {
      "epoch": 0.19196,
      "grad_norm": 1.1528145797759937,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 19196
    },
    {
      "epoch": 0.19197,
      "grad_norm": 1.5981012584383123,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 19197
    },
    {
      "epoch": 0.19198,
      "grad_norm": 0.9630358323900087,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 19198
    },
    {
      "epoch": 0.19199,
      "grad_norm": 1.5678787468061148,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 19199
    },
    {
      "epoch": 0.192,
      "grad_norm": 1.162720578822619,
      "learning_rate": 0.003,
      "loss": 4.1084,
      "step": 19200
    },
    {
      "epoch": 0.19201,
      "grad_norm": 1.554253879167807,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 19201
    },
    {
      "epoch": 0.19202,
      "grad_norm": 1.3562060036479024,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 19202
    },
    {
      "epoch": 0.19203,
      "grad_norm": 1.572129714906069,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 19203
    },
    {
      "epoch": 0.19204,
      "grad_norm": 1.108132358920365,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 19204
    },
    {
      "epoch": 0.19205,
      "grad_norm": 1.4562272560833824,
      "learning_rate": 0.003,
      "loss": 4.0931,
      "step": 19205
    },
    {
      "epoch": 0.19206,
      "grad_norm": 1.1248392481966405,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 19206
    },
    {
      "epoch": 0.19207,
      "grad_norm": 1.347347178032185,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 19207
    },
    {
      "epoch": 0.19208,
      "grad_norm": 1.2569962037026983,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 19208
    },
    {
      "epoch": 0.19209,
      "grad_norm": 1.3461303832239604,
      "learning_rate": 0.003,
      "loss": 4.0902,
      "step": 19209
    },
    {
      "epoch": 0.1921,
      "grad_norm": 0.9850114287517585,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 19210
    },
    {
      "epoch": 0.19211,
      "grad_norm": 1.4005865113973734,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 19211
    },
    {
      "epoch": 0.19212,
      "grad_norm": 1.3014172439729073,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 19212
    },
    {
      "epoch": 0.19213,
      "grad_norm": 1.3378102945952632,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 19213
    },
    {
      "epoch": 0.19214,
      "grad_norm": 1.3952909056018472,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 19214
    },
    {
      "epoch": 0.19215,
      "grad_norm": 1.611663758916327,
      "learning_rate": 0.003,
      "loss": 4.0941,
      "step": 19215
    },
    {
      "epoch": 0.19216,
      "grad_norm": 0.8963660316350982,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 19216
    },
    {
      "epoch": 0.19217,
      "grad_norm": 1.1968197183576728,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 19217
    },
    {
      "epoch": 0.19218,
      "grad_norm": 1.288873508937206,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 19218
    },
    {
      "epoch": 0.19219,
      "grad_norm": 1.2116427947856974,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 19219
    },
    {
      "epoch": 0.1922,
      "grad_norm": 1.40961153431515,
      "learning_rate": 0.003,
      "loss": 4.0313,
      "step": 19220
    },
    {
      "epoch": 0.19221,
      "grad_norm": 1.3174703939528276,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 19221
    },
    {
      "epoch": 0.19222,
      "grad_norm": 1.1947735338993128,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 19222
    },
    {
      "epoch": 0.19223,
      "grad_norm": 1.3706485111315745,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 19223
    },
    {
      "epoch": 0.19224,
      "grad_norm": 1.2645512304572817,
      "learning_rate": 0.003,
      "loss": 4.0328,
      "step": 19224
    },
    {
      "epoch": 0.19225,
      "grad_norm": 1.1543151542958139,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 19225
    },
    {
      "epoch": 0.19226,
      "grad_norm": 1.291486822731222,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 19226
    },
    {
      "epoch": 0.19227,
      "grad_norm": 1.5269001763632155,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 19227
    },
    {
      "epoch": 0.19228,
      "grad_norm": 1.3782331321295478,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 19228
    },
    {
      "epoch": 0.19229,
      "grad_norm": 1.3237198432938115,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 19229
    },
    {
      "epoch": 0.1923,
      "grad_norm": 1.323968945621076,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 19230
    },
    {
      "epoch": 0.19231,
      "grad_norm": 1.3349763890268658,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 19231
    },
    {
      "epoch": 0.19232,
      "grad_norm": 1.3661924478328742,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 19232
    },
    {
      "epoch": 0.19233,
      "grad_norm": 1.194365790073095,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 19233
    },
    {
      "epoch": 0.19234,
      "grad_norm": 1.3965787017862838,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 19234
    },
    {
      "epoch": 0.19235,
      "grad_norm": 1.196036206103378,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 19235
    },
    {
      "epoch": 0.19236,
      "grad_norm": 1.3711815439105595,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 19236
    },
    {
      "epoch": 0.19237,
      "grad_norm": 1.141165924450159,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 19237
    },
    {
      "epoch": 0.19238,
      "grad_norm": 1.308345863022952,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 19238
    },
    {
      "epoch": 0.19239,
      "grad_norm": 1.3249441038885172,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 19239
    },
    {
      "epoch": 0.1924,
      "grad_norm": 1.3966021263135535,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 19240
    },
    {
      "epoch": 0.19241,
      "grad_norm": 1.3663512273311271,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 19241
    },
    {
      "epoch": 0.19242,
      "grad_norm": 1.2032887696702281,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 19242
    },
    {
      "epoch": 0.19243,
      "grad_norm": 1.3339164408073845,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 19243
    },
    {
      "epoch": 0.19244,
      "grad_norm": 1.4303815180109472,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 19244
    },
    {
      "epoch": 0.19245,
      "grad_norm": 1.423193016673088,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 19245
    },
    {
      "epoch": 0.19246,
      "grad_norm": 1.4514350895399684,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 19246
    },
    {
      "epoch": 0.19247,
      "grad_norm": 1.495226265941209,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 19247
    },
    {
      "epoch": 0.19248,
      "grad_norm": 1.2397736348773147,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 19248
    },
    {
      "epoch": 0.19249,
      "grad_norm": 1.374416195938836,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 19249
    },
    {
      "epoch": 0.1925,
      "grad_norm": 1.3189947596094718,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 19250
    },
    {
      "epoch": 0.19251,
      "grad_norm": 1.3593757512777345,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 19251
    },
    {
      "epoch": 0.19252,
      "grad_norm": 1.2451545340567312,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 19252
    },
    {
      "epoch": 0.19253,
      "grad_norm": 1.4531214541051583,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 19253
    },
    {
      "epoch": 0.19254,
      "grad_norm": 1.1835021545730156,
      "learning_rate": 0.003,
      "loss": 4.1012,
      "step": 19254
    },
    {
      "epoch": 0.19255,
      "grad_norm": 1.576806190726499,
      "learning_rate": 0.003,
      "loss": 4.0852,
      "step": 19255
    },
    {
      "epoch": 0.19256,
      "grad_norm": 0.9618777802716235,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 19256
    },
    {
      "epoch": 0.19257,
      "grad_norm": 1.3588331222703336,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 19257
    },
    {
      "epoch": 0.19258,
      "grad_norm": 1.2878287437826246,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 19258
    },
    {
      "epoch": 0.19259,
      "grad_norm": 1.4055216921584257,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 19259
    },
    {
      "epoch": 0.1926,
      "grad_norm": 1.3051453569308915,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 19260
    },
    {
      "epoch": 0.19261,
      "grad_norm": 1.326885316634315,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 19261
    },
    {
      "epoch": 0.19262,
      "grad_norm": 1.1061915072445085,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 19262
    },
    {
      "epoch": 0.19263,
      "grad_norm": 1.5654195215504978,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 19263
    },
    {
      "epoch": 0.19264,
      "grad_norm": 1.0952973511418131,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 19264
    },
    {
      "epoch": 0.19265,
      "grad_norm": 1.6292694563212933,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 19265
    },
    {
      "epoch": 0.19266,
      "grad_norm": 1.0953996900593994,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 19266
    },
    {
      "epoch": 0.19267,
      "grad_norm": 1.3387421529983907,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 19267
    },
    {
      "epoch": 0.19268,
      "grad_norm": 1.5660995372853055,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 19268
    },
    {
      "epoch": 0.19269,
      "grad_norm": 1.2087119276592098,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 19269
    },
    {
      "epoch": 0.1927,
      "grad_norm": 1.3580668250597665,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 19270
    },
    {
      "epoch": 0.19271,
      "grad_norm": 1.3348444888748765,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 19271
    },
    {
      "epoch": 0.19272,
      "grad_norm": 1.2520883322083953,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 19272
    },
    {
      "epoch": 0.19273,
      "grad_norm": 1.1815566217193947,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 19273
    },
    {
      "epoch": 0.19274,
      "grad_norm": 1.3356315303300998,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 19274
    },
    {
      "epoch": 0.19275,
      "grad_norm": 1.2447575707031089,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 19275
    },
    {
      "epoch": 0.19276,
      "grad_norm": 1.4291321612692254,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 19276
    },
    {
      "epoch": 0.19277,
      "grad_norm": 1.1091724810120003,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 19277
    },
    {
      "epoch": 0.19278,
      "grad_norm": 1.8999785592313487,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 19278
    },
    {
      "epoch": 0.19279,
      "grad_norm": 1.0056310117229874,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 19279
    },
    {
      "epoch": 0.1928,
      "grad_norm": 1.5151507602005392,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 19280
    },
    {
      "epoch": 0.19281,
      "grad_norm": 1.4595941770553433,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 19281
    },
    {
      "epoch": 0.19282,
      "grad_norm": 1.537815087375208,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 19282
    },
    {
      "epoch": 0.19283,
      "grad_norm": 1.1238184726071319,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 19283
    },
    {
      "epoch": 0.19284,
      "grad_norm": 1.2056169917291322,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 19284
    },
    {
      "epoch": 0.19285,
      "grad_norm": 1.4513606586793888,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 19285
    },
    {
      "epoch": 0.19286,
      "grad_norm": 1.2977929469861313,
      "learning_rate": 0.003,
      "loss": 4.033,
      "step": 19286
    },
    {
      "epoch": 0.19287,
      "grad_norm": 1.2631869724978644,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 19287
    },
    {
      "epoch": 0.19288,
      "grad_norm": 1.4007130177214029,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 19288
    },
    {
      "epoch": 0.19289,
      "grad_norm": 1.1353526144337664,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 19289
    },
    {
      "epoch": 0.1929,
      "grad_norm": 1.212149995014371,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 19290
    },
    {
      "epoch": 0.19291,
      "grad_norm": 1.3859392455790678,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 19291
    },
    {
      "epoch": 0.19292,
      "grad_norm": 1.2598101853711283,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 19292
    },
    {
      "epoch": 0.19293,
      "grad_norm": 1.3911589489772473,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 19293
    },
    {
      "epoch": 0.19294,
      "grad_norm": 1.210170004902706,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 19294
    },
    {
      "epoch": 0.19295,
      "grad_norm": 1.299972482005869,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 19295
    },
    {
      "epoch": 0.19296,
      "grad_norm": 1.2627534993763714,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 19296
    },
    {
      "epoch": 0.19297,
      "grad_norm": 1.3443064312058892,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 19297
    },
    {
      "epoch": 0.19298,
      "grad_norm": 1.666208868220041,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 19298
    },
    {
      "epoch": 0.19299,
      "grad_norm": 1.113952270737019,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 19299
    },
    {
      "epoch": 0.193,
      "grad_norm": 1.7041944102215842,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 19300
    },
    {
      "epoch": 0.19301,
      "grad_norm": 1.6375690792017714,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 19301
    },
    {
      "epoch": 0.19302,
      "grad_norm": 1.1871622671795783,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 19302
    },
    {
      "epoch": 0.19303,
      "grad_norm": 1.2523017718203313,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 19303
    },
    {
      "epoch": 0.19304,
      "grad_norm": 1.3298620440401583,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 19304
    },
    {
      "epoch": 0.19305,
      "grad_norm": 1.3646092971100463,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 19305
    },
    {
      "epoch": 0.19306,
      "grad_norm": 1.1620948348169835,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 19306
    },
    {
      "epoch": 0.19307,
      "grad_norm": 1.4016613215927611,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 19307
    },
    {
      "epoch": 0.19308,
      "grad_norm": 1.1736152267860105,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 19308
    },
    {
      "epoch": 0.19309,
      "grad_norm": 1.3182358837677548,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 19309
    },
    {
      "epoch": 0.1931,
      "grad_norm": 1.4501204798766651,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 19310
    },
    {
      "epoch": 0.19311,
      "grad_norm": 1.1434160852236295,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 19311
    },
    {
      "epoch": 0.19312,
      "grad_norm": 1.431722256758653,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 19312
    },
    {
      "epoch": 0.19313,
      "grad_norm": 1.29689321893418,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 19313
    },
    {
      "epoch": 0.19314,
      "grad_norm": 1.608081179380934,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 19314
    },
    {
      "epoch": 0.19315,
      "grad_norm": 1.1555466880702878,
      "learning_rate": 0.003,
      "loss": 4.029,
      "step": 19315
    },
    {
      "epoch": 0.19316,
      "grad_norm": 1.3106279749947531,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 19316
    },
    {
      "epoch": 0.19317,
      "grad_norm": 1.2154730846642248,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 19317
    },
    {
      "epoch": 0.19318,
      "grad_norm": 1.3811202154259432,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 19318
    },
    {
      "epoch": 0.19319,
      "grad_norm": 1.2952358672295878,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 19319
    },
    {
      "epoch": 0.1932,
      "grad_norm": 1.349704612179986,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 19320
    },
    {
      "epoch": 0.19321,
      "grad_norm": 1.4391108882127606,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 19321
    },
    {
      "epoch": 0.19322,
      "grad_norm": 1.0698517431090664,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 19322
    },
    {
      "epoch": 0.19323,
      "grad_norm": 1.5377972878630841,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 19323
    },
    {
      "epoch": 0.19324,
      "grad_norm": 1.1620482234533458,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 19324
    },
    {
      "epoch": 0.19325,
      "grad_norm": 1.706295803169076,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 19325
    },
    {
      "epoch": 0.19326,
      "grad_norm": 1.4627113037011565,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 19326
    },
    {
      "epoch": 0.19327,
      "grad_norm": 1.2952479122134402,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 19327
    },
    {
      "epoch": 0.19328,
      "grad_norm": 1.3647261235036796,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 19328
    },
    {
      "epoch": 0.19329,
      "grad_norm": 1.398775878899619,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19329
    },
    {
      "epoch": 0.1933,
      "grad_norm": 0.9840400767359754,
      "learning_rate": 0.003,
      "loss": 4.0235,
      "step": 19330
    },
    {
      "epoch": 0.19331,
      "grad_norm": 1.3387790757069982,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 19331
    },
    {
      "epoch": 0.19332,
      "grad_norm": 1.1478228412156144,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 19332
    },
    {
      "epoch": 0.19333,
      "grad_norm": 1.363465300325813,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 19333
    },
    {
      "epoch": 0.19334,
      "grad_norm": 1.2228351145593437,
      "learning_rate": 0.003,
      "loss": 4.023,
      "step": 19334
    },
    {
      "epoch": 0.19335,
      "grad_norm": 1.5490623373968835,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 19335
    },
    {
      "epoch": 0.19336,
      "grad_norm": 1.2547671310467585,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 19336
    },
    {
      "epoch": 0.19337,
      "grad_norm": 1.378056199478073,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 19337
    },
    {
      "epoch": 0.19338,
      "grad_norm": 1.2951080135198598,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 19338
    },
    {
      "epoch": 0.19339,
      "grad_norm": 1.4443763625994586,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 19339
    },
    {
      "epoch": 0.1934,
      "grad_norm": 1.0753716014890167,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 19340
    },
    {
      "epoch": 0.19341,
      "grad_norm": 1.3421729672704792,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 19341
    },
    {
      "epoch": 0.19342,
      "grad_norm": 1.3002098104094193,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 19342
    },
    {
      "epoch": 0.19343,
      "grad_norm": 1.1639841138101032,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 19343
    },
    {
      "epoch": 0.19344,
      "grad_norm": 1.5256197722062088,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 19344
    },
    {
      "epoch": 0.19345,
      "grad_norm": 1.1923686050416904,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 19345
    },
    {
      "epoch": 0.19346,
      "grad_norm": 1.4160350775090675,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 19346
    },
    {
      "epoch": 0.19347,
      "grad_norm": 1.1720319120661895,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 19347
    },
    {
      "epoch": 0.19348,
      "grad_norm": 1.3593873642570558,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 19348
    },
    {
      "epoch": 0.19349,
      "grad_norm": 1.3505122712820288,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 19349
    },
    {
      "epoch": 0.1935,
      "grad_norm": 1.3306040730637425,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 19350
    },
    {
      "epoch": 0.19351,
      "grad_norm": 1.655727960969945,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19351
    },
    {
      "epoch": 0.19352,
      "grad_norm": 1.2126789166642862,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 19352
    },
    {
      "epoch": 0.19353,
      "grad_norm": 1.5285402732283524,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 19353
    },
    {
      "epoch": 0.19354,
      "grad_norm": 1.1052693417942343,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 19354
    },
    {
      "epoch": 0.19355,
      "grad_norm": 1.5586897787894143,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 19355
    },
    {
      "epoch": 0.19356,
      "grad_norm": 0.9362571491901476,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 19356
    },
    {
      "epoch": 0.19357,
      "grad_norm": 1.1845455160555813,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 19357
    },
    {
      "epoch": 0.19358,
      "grad_norm": 1.4794767646486355,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 19358
    },
    {
      "epoch": 0.19359,
      "grad_norm": 1.143734784255008,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 19359
    },
    {
      "epoch": 0.1936,
      "grad_norm": 1.4012925676613213,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 19360
    },
    {
      "epoch": 0.19361,
      "grad_norm": 1.4033702307588178,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 19361
    },
    {
      "epoch": 0.19362,
      "grad_norm": 1.3481368007485075,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 19362
    },
    {
      "epoch": 0.19363,
      "grad_norm": 1.4076040383533015,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 19363
    },
    {
      "epoch": 0.19364,
      "grad_norm": 1.1996152385665022,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 19364
    },
    {
      "epoch": 0.19365,
      "grad_norm": 1.4580495886259732,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 19365
    },
    {
      "epoch": 0.19366,
      "grad_norm": 1.521856191699455,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 19366
    },
    {
      "epoch": 0.19367,
      "grad_norm": 1.631340300360941,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 19367
    },
    {
      "epoch": 0.19368,
      "grad_norm": 1.3336661687622993,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 19368
    },
    {
      "epoch": 0.19369,
      "grad_norm": 1.345784656866648,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 19369
    },
    {
      "epoch": 0.1937,
      "grad_norm": 1.2206180867788792,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 19370
    },
    {
      "epoch": 0.19371,
      "grad_norm": 1.1654276116493354,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 19371
    },
    {
      "epoch": 0.19372,
      "grad_norm": 1.3810446201500377,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 19372
    },
    {
      "epoch": 0.19373,
      "grad_norm": 1.373336874783953,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 19373
    },
    {
      "epoch": 0.19374,
      "grad_norm": 1.629303583339713,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 19374
    },
    {
      "epoch": 0.19375,
      "grad_norm": 1.0735248207499326,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 19375
    },
    {
      "epoch": 0.19376,
      "grad_norm": 1.3276649318294715,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 19376
    },
    {
      "epoch": 0.19377,
      "grad_norm": 1.3805067134683795,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 19377
    },
    {
      "epoch": 0.19378,
      "grad_norm": 1.1123481119523868,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 19378
    },
    {
      "epoch": 0.19379,
      "grad_norm": 1.2822520851622643,
      "learning_rate": 0.003,
      "loss": 4.0269,
      "step": 19379
    },
    {
      "epoch": 0.1938,
      "grad_norm": 1.2518157359945186,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 19380
    },
    {
      "epoch": 0.19381,
      "grad_norm": 1.3211605719658674,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 19381
    },
    {
      "epoch": 0.19382,
      "grad_norm": 1.2031288178131152,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 19382
    },
    {
      "epoch": 0.19383,
      "grad_norm": 1.534044515509392,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 19383
    },
    {
      "epoch": 0.19384,
      "grad_norm": 1.2372061620324233,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 19384
    },
    {
      "epoch": 0.19385,
      "grad_norm": 1.4711169660183832,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 19385
    },
    {
      "epoch": 0.19386,
      "grad_norm": 1.1991312259387814,
      "learning_rate": 0.003,
      "loss": 4.0266,
      "step": 19386
    },
    {
      "epoch": 0.19387,
      "grad_norm": 1.417263892946054,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 19387
    },
    {
      "epoch": 0.19388,
      "grad_norm": 1.3587734414119863,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 19388
    },
    {
      "epoch": 0.19389,
      "grad_norm": 1.2021437211625692,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 19389
    },
    {
      "epoch": 0.1939,
      "grad_norm": 1.3861074777535454,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 19390
    },
    {
      "epoch": 0.19391,
      "grad_norm": 1.1842067172210315,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 19391
    },
    {
      "epoch": 0.19392,
      "grad_norm": 1.433987285114199,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 19392
    },
    {
      "epoch": 0.19393,
      "grad_norm": 1.4103147808509588,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 19393
    },
    {
      "epoch": 0.19394,
      "grad_norm": 1.7161650832967164,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 19394
    },
    {
      "epoch": 0.19395,
      "grad_norm": 1.177453349617905,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 19395
    },
    {
      "epoch": 0.19396,
      "grad_norm": 1.4484004379686284,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 19396
    },
    {
      "epoch": 0.19397,
      "grad_norm": 1.2794772010946776,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 19397
    },
    {
      "epoch": 0.19398,
      "grad_norm": 1.0829439718799576,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 19398
    },
    {
      "epoch": 0.19399,
      "grad_norm": 1.6847222832453004,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 19399
    },
    {
      "epoch": 0.194,
      "grad_norm": 1.1198867425614099,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 19400
    },
    {
      "epoch": 0.19401,
      "grad_norm": 1.5339102604902262,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 19401
    },
    {
      "epoch": 0.19402,
      "grad_norm": 1.4360483324581836,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 19402
    },
    {
      "epoch": 0.19403,
      "grad_norm": 1.1690678118895501,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 19403
    },
    {
      "epoch": 0.19404,
      "grad_norm": 1.4040136868442783,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 19404
    },
    {
      "epoch": 0.19405,
      "grad_norm": 1.1443562702751413,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 19405
    },
    {
      "epoch": 0.19406,
      "grad_norm": 1.3270118186775146,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 19406
    },
    {
      "epoch": 0.19407,
      "grad_norm": 1.2485030701960447,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 19407
    },
    {
      "epoch": 0.19408,
      "grad_norm": 1.2257617847502387,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 19408
    },
    {
      "epoch": 0.19409,
      "grad_norm": 1.339782896099073,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 19409
    },
    {
      "epoch": 0.1941,
      "grad_norm": 1.1653312443080106,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 19410
    },
    {
      "epoch": 0.19411,
      "grad_norm": 1.3190368669455002,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 19411
    },
    {
      "epoch": 0.19412,
      "grad_norm": 1.2566581938102939,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 19412
    },
    {
      "epoch": 0.19413,
      "grad_norm": 1.086513711853439,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 19413
    },
    {
      "epoch": 0.19414,
      "grad_norm": 1.3678932310645313,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 19414
    },
    {
      "epoch": 0.19415,
      "grad_norm": 1.8391709363358586,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 19415
    },
    {
      "epoch": 0.19416,
      "grad_norm": 1.319014463307184,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 19416
    },
    {
      "epoch": 0.19417,
      "grad_norm": 1.8220178704496939,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 19417
    },
    {
      "epoch": 0.19418,
      "grad_norm": 1.124723875317562,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 19418
    },
    {
      "epoch": 0.19419,
      "grad_norm": 1.19884607098138,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 19419
    },
    {
      "epoch": 0.1942,
      "grad_norm": 1.449264556093074,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 19420
    },
    {
      "epoch": 0.19421,
      "grad_norm": 1.3191374682439083,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 19421
    },
    {
      "epoch": 0.19422,
      "grad_norm": 1.3981067550540194,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 19422
    },
    {
      "epoch": 0.19423,
      "grad_norm": 1.3127266074302606,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 19423
    },
    {
      "epoch": 0.19424,
      "grad_norm": 1.2563177666759349,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 19424
    },
    {
      "epoch": 0.19425,
      "grad_norm": 1.2782899915280543,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 19425
    },
    {
      "epoch": 0.19426,
      "grad_norm": 1.182685010153037,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 19426
    },
    {
      "epoch": 0.19427,
      "grad_norm": 1.407694968111235,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 19427
    },
    {
      "epoch": 0.19428,
      "grad_norm": 1.2142362395422055,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 19428
    },
    {
      "epoch": 0.19429,
      "grad_norm": 1.3321695542747696,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 19429
    },
    {
      "epoch": 0.1943,
      "grad_norm": 1.3455226385908636,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 19430
    },
    {
      "epoch": 0.19431,
      "grad_norm": 1.1975306942264474,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 19431
    },
    {
      "epoch": 0.19432,
      "grad_norm": 1.306164180647245,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 19432
    },
    {
      "epoch": 0.19433,
      "grad_norm": 1.2038121182877317,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 19433
    },
    {
      "epoch": 0.19434,
      "grad_norm": 1.3100820335537733,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 19434
    },
    {
      "epoch": 0.19435,
      "grad_norm": 1.51000847595959,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 19435
    },
    {
      "epoch": 0.19436,
      "grad_norm": 1.5370351507869693,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 19436
    },
    {
      "epoch": 0.19437,
      "grad_norm": 1.1469400430869083,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 19437
    },
    {
      "epoch": 0.19438,
      "grad_norm": 1.5209830775292177,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 19438
    },
    {
      "epoch": 0.19439,
      "grad_norm": 1.1656555651542801,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 19439
    },
    {
      "epoch": 0.1944,
      "grad_norm": 1.3849667723604768,
      "learning_rate": 0.003,
      "loss": 4.0907,
      "step": 19440
    },
    {
      "epoch": 0.19441,
      "grad_norm": 1.026361630538749,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 19441
    },
    {
      "epoch": 0.19442,
      "grad_norm": 1.4677422327224452,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 19442
    },
    {
      "epoch": 0.19443,
      "grad_norm": 1.1064600363094863,
      "learning_rate": 0.003,
      "loss": 4.087,
      "step": 19443
    },
    {
      "epoch": 0.19444,
      "grad_norm": 1.581247945244656,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 19444
    },
    {
      "epoch": 0.19445,
      "grad_norm": 1.3784402022713331,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 19445
    },
    {
      "epoch": 0.19446,
      "grad_norm": 1.4594290215569299,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 19446
    },
    {
      "epoch": 0.19447,
      "grad_norm": 1.5644238177299843,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 19447
    },
    {
      "epoch": 0.19448,
      "grad_norm": 1.4121667484118812,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 19448
    },
    {
      "epoch": 0.19449,
      "grad_norm": 1.401840224088073,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 19449
    },
    {
      "epoch": 0.1945,
      "grad_norm": 1.3696362388852628,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 19450
    },
    {
      "epoch": 0.19451,
      "grad_norm": 1.5213077490416473,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 19451
    },
    {
      "epoch": 0.19452,
      "grad_norm": 1.0000764143350864,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 19452
    },
    {
      "epoch": 0.19453,
      "grad_norm": 1.4410855592453407,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 19453
    },
    {
      "epoch": 0.19454,
      "grad_norm": 1.3713228817734917,
      "learning_rate": 0.003,
      "loss": 4.0883,
      "step": 19454
    },
    {
      "epoch": 0.19455,
      "grad_norm": 1.3782633140229856,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 19455
    },
    {
      "epoch": 0.19456,
      "grad_norm": 1.137024405499252,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 19456
    },
    {
      "epoch": 0.19457,
      "grad_norm": 1.3749447670302388,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 19457
    },
    {
      "epoch": 0.19458,
      "grad_norm": 1.1077411007998608,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 19458
    },
    {
      "epoch": 0.19459,
      "grad_norm": 1.2389465617647617,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 19459
    },
    {
      "epoch": 0.1946,
      "grad_norm": 1.2397676138930445,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 19460
    },
    {
      "epoch": 0.19461,
      "grad_norm": 1.1978532223077072,
      "learning_rate": 0.003,
      "loss": 4.1113,
      "step": 19461
    },
    {
      "epoch": 0.19462,
      "grad_norm": 1.3514504239036356,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 19462
    },
    {
      "epoch": 0.19463,
      "grad_norm": 1.2523103199057817,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 19463
    },
    {
      "epoch": 0.19464,
      "grad_norm": 1.6969692608474285,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 19464
    },
    {
      "epoch": 0.19465,
      "grad_norm": 0.9414368684036143,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 19465
    },
    {
      "epoch": 0.19466,
      "grad_norm": 1.4974558776391218,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 19466
    },
    {
      "epoch": 0.19467,
      "grad_norm": 1.4638764195062006,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 19467
    },
    {
      "epoch": 0.19468,
      "grad_norm": 1.514719727744151,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 19468
    },
    {
      "epoch": 0.19469,
      "grad_norm": 1.439838514731369,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 19469
    },
    {
      "epoch": 0.1947,
      "grad_norm": 1.203798999719418,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 19470
    },
    {
      "epoch": 0.19471,
      "grad_norm": 1.2416918412369136,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 19471
    },
    {
      "epoch": 0.19472,
      "grad_norm": 1.537366145495566,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 19472
    },
    {
      "epoch": 0.19473,
      "grad_norm": 1.3688593584598943,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 19473
    },
    {
      "epoch": 0.19474,
      "grad_norm": 1.3891744221027942,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 19474
    },
    {
      "epoch": 0.19475,
      "grad_norm": 1.263524168556553,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 19475
    },
    {
      "epoch": 0.19476,
      "grad_norm": 1.3439616213955559,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 19476
    },
    {
      "epoch": 0.19477,
      "grad_norm": 1.3250794091095022,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 19477
    },
    {
      "epoch": 0.19478,
      "grad_norm": 1.3995346362050427,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 19478
    },
    {
      "epoch": 0.19479,
      "grad_norm": 1.2639749929102022,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 19479
    },
    {
      "epoch": 0.1948,
      "grad_norm": 1.5526913027825195,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 19480
    },
    {
      "epoch": 0.19481,
      "grad_norm": 1.1375459439711735,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19481
    },
    {
      "epoch": 0.19482,
      "grad_norm": 1.3154176386784155,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 19482
    },
    {
      "epoch": 0.19483,
      "grad_norm": 1.4716210996534527,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 19483
    },
    {
      "epoch": 0.19484,
      "grad_norm": 1.3905671527809653,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 19484
    },
    {
      "epoch": 0.19485,
      "grad_norm": 1.3117630558690683,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 19485
    },
    {
      "epoch": 0.19486,
      "grad_norm": 1.5008569960991724,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 19486
    },
    {
      "epoch": 0.19487,
      "grad_norm": 1.12906638228282,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 19487
    },
    {
      "epoch": 0.19488,
      "grad_norm": 1.3657365714985676,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 19488
    },
    {
      "epoch": 0.19489,
      "grad_norm": 1.3018201857887008,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 19489
    },
    {
      "epoch": 0.1949,
      "grad_norm": 1.3983987668216904,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 19490
    },
    {
      "epoch": 0.19491,
      "grad_norm": 1.481011022539193,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 19491
    },
    {
      "epoch": 0.19492,
      "grad_norm": 1.1813655279877215,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 19492
    },
    {
      "epoch": 0.19493,
      "grad_norm": 1.4416587610837648,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 19493
    },
    {
      "epoch": 0.19494,
      "grad_norm": 1.1131622561587362,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 19494
    },
    {
      "epoch": 0.19495,
      "grad_norm": 1.4616149240882121,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 19495
    },
    {
      "epoch": 0.19496,
      "grad_norm": 1.0756483327498663,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 19496
    },
    {
      "epoch": 0.19497,
      "grad_norm": 1.1718359588897174,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 19497
    },
    {
      "epoch": 0.19498,
      "grad_norm": 1.4083804540585427,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 19498
    },
    {
      "epoch": 0.19499,
      "grad_norm": 1.2897914243706128,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 19499
    },
    {
      "epoch": 0.195,
      "grad_norm": 1.554466135239505,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 19500
    },
    {
      "epoch": 0.19501,
      "grad_norm": 1.167251633086409,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 19501
    },
    {
      "epoch": 0.19502,
      "grad_norm": 1.759459231435133,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 19502
    },
    {
      "epoch": 0.19503,
      "grad_norm": 1.0433124182408695,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 19503
    },
    {
      "epoch": 0.19504,
      "grad_norm": 1.4082628477633647,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 19504
    },
    {
      "epoch": 0.19505,
      "grad_norm": 1.3823904406544583,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 19505
    },
    {
      "epoch": 0.19506,
      "grad_norm": 1.0638797553557129,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 19506
    },
    {
      "epoch": 0.19507,
      "grad_norm": 1.6891751096651972,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 19507
    },
    {
      "epoch": 0.19508,
      "grad_norm": 0.9416716412099737,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 19508
    },
    {
      "epoch": 0.19509,
      "grad_norm": 1.5765004581747462,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 19509
    },
    {
      "epoch": 0.1951,
      "grad_norm": 1.3089842021211013,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 19510
    },
    {
      "epoch": 0.19511,
      "grad_norm": 1.341192332501336,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 19511
    },
    {
      "epoch": 0.19512,
      "grad_norm": 1.7025869243643506,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 19512
    },
    {
      "epoch": 0.19513,
      "grad_norm": 1.3170221606778398,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 19513
    },
    {
      "epoch": 0.19514,
      "grad_norm": 1.2350694969711091,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 19514
    },
    {
      "epoch": 0.19515,
      "grad_norm": 1.290375283368854,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 19515
    },
    {
      "epoch": 0.19516,
      "grad_norm": 1.3103270446228672,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 19516
    },
    {
      "epoch": 0.19517,
      "grad_norm": 1.2217186690684785,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 19517
    },
    {
      "epoch": 0.19518,
      "grad_norm": 1.3932532373420963,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 19518
    },
    {
      "epoch": 0.19519,
      "grad_norm": 1.6149310376647918,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 19519
    },
    {
      "epoch": 0.1952,
      "grad_norm": 1.1468617142101982,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 19520
    },
    {
      "epoch": 0.19521,
      "grad_norm": 1.3430904754624255,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 19521
    },
    {
      "epoch": 0.19522,
      "grad_norm": 1.4246909909100243,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 19522
    },
    {
      "epoch": 0.19523,
      "grad_norm": 1.1588283886075517,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 19523
    },
    {
      "epoch": 0.19524,
      "grad_norm": 1.3071282399894546,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 19524
    },
    {
      "epoch": 0.19525,
      "grad_norm": 1.5768030414564296,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 19525
    },
    {
      "epoch": 0.19526,
      "grad_norm": 1.1685958781565342,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 19526
    },
    {
      "epoch": 0.19527,
      "grad_norm": 1.313093450897761,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 19527
    },
    {
      "epoch": 0.19528,
      "grad_norm": 1.2742119703672696,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 19528
    },
    {
      "epoch": 0.19529,
      "grad_norm": 1.7290547434394015,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 19529
    },
    {
      "epoch": 0.1953,
      "grad_norm": 1.0529733408830393,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 19530
    },
    {
      "epoch": 0.19531,
      "grad_norm": 1.5482101859966972,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 19531
    },
    {
      "epoch": 0.19532,
      "grad_norm": 1.2989756814178277,
      "learning_rate": 0.003,
      "loss": 4.1079,
      "step": 19532
    },
    {
      "epoch": 0.19533,
      "grad_norm": 1.1864556980596295,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 19533
    },
    {
      "epoch": 0.19534,
      "grad_norm": 1.1906409640998392,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 19534
    },
    {
      "epoch": 0.19535,
      "grad_norm": 1.3279386097222499,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 19535
    },
    {
      "epoch": 0.19536,
      "grad_norm": 1.3249211366390132,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 19536
    },
    {
      "epoch": 0.19537,
      "grad_norm": 1.2708672308772355,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 19537
    },
    {
      "epoch": 0.19538,
      "grad_norm": 1.3315379121969504,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 19538
    },
    {
      "epoch": 0.19539,
      "grad_norm": 1.2429310832605476,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 19539
    },
    {
      "epoch": 0.1954,
      "grad_norm": 1.5054210239848758,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 19540
    },
    {
      "epoch": 0.19541,
      "grad_norm": 1.0475191411451963,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 19541
    },
    {
      "epoch": 0.19542,
      "grad_norm": 1.4284508801628617,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 19542
    },
    {
      "epoch": 0.19543,
      "grad_norm": 1.0933611515009474,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 19543
    },
    {
      "epoch": 0.19544,
      "grad_norm": 1.661660218431094,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19544
    },
    {
      "epoch": 0.19545,
      "grad_norm": 1.2199410776915833,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 19545
    },
    {
      "epoch": 0.19546,
      "grad_norm": 1.6905546886575713,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 19546
    },
    {
      "epoch": 0.19547,
      "grad_norm": 1.29296218202824,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 19547
    },
    {
      "epoch": 0.19548,
      "grad_norm": 1.2837712876533047,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 19548
    },
    {
      "epoch": 0.19549,
      "grad_norm": 1.4703643136338311,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 19549
    },
    {
      "epoch": 0.1955,
      "grad_norm": 1.2700676522488055,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 19550
    },
    {
      "epoch": 0.19551,
      "grad_norm": 1.1109703899269674,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 19551
    },
    {
      "epoch": 0.19552,
      "grad_norm": 1.2673166776363936,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 19552
    },
    {
      "epoch": 0.19553,
      "grad_norm": 1.4214656677819935,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 19553
    },
    {
      "epoch": 0.19554,
      "grad_norm": 1.2423078438651163,
      "learning_rate": 0.003,
      "loss": 4.0245,
      "step": 19554
    },
    {
      "epoch": 0.19555,
      "grad_norm": 1.4398209252606844,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 19555
    },
    {
      "epoch": 0.19556,
      "grad_norm": 1.4883629305181807,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 19556
    },
    {
      "epoch": 0.19557,
      "grad_norm": 1.297674226675014,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 19557
    },
    {
      "epoch": 0.19558,
      "grad_norm": 1.3053307952640294,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 19558
    },
    {
      "epoch": 0.19559,
      "grad_norm": 1.4310611165329026,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 19559
    },
    {
      "epoch": 0.1956,
      "grad_norm": 1.5197492036910145,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 19560
    },
    {
      "epoch": 0.19561,
      "grad_norm": 1.1766594132642922,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 19561
    },
    {
      "epoch": 0.19562,
      "grad_norm": 1.2995395067215239,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 19562
    },
    {
      "epoch": 0.19563,
      "grad_norm": 1.2983980671464603,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 19563
    },
    {
      "epoch": 0.19564,
      "grad_norm": 1.3436105722348386,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 19564
    },
    {
      "epoch": 0.19565,
      "grad_norm": 1.3194744920133412,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 19565
    },
    {
      "epoch": 0.19566,
      "grad_norm": 1.2171903493614817,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 19566
    },
    {
      "epoch": 0.19567,
      "grad_norm": 1.6401387912898335,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 19567
    },
    {
      "epoch": 0.19568,
      "grad_norm": 1.1441649428245726,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 19568
    },
    {
      "epoch": 0.19569,
      "grad_norm": 1.251626956814284,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 19569
    },
    {
      "epoch": 0.1957,
      "grad_norm": 1.3502503512367485,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 19570
    },
    {
      "epoch": 0.19571,
      "grad_norm": 1.2426582395162038,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 19571
    },
    {
      "epoch": 0.19572,
      "grad_norm": 1.2413271686787717,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 19572
    },
    {
      "epoch": 0.19573,
      "grad_norm": 1.1949950162191312,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 19573
    },
    {
      "epoch": 0.19574,
      "grad_norm": 1.288221508446464,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 19574
    },
    {
      "epoch": 0.19575,
      "grad_norm": 1.504647978943427,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 19575
    },
    {
      "epoch": 0.19576,
      "grad_norm": 1.2783626987122165,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 19576
    },
    {
      "epoch": 0.19577,
      "grad_norm": 1.5688010739625575,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 19577
    },
    {
      "epoch": 0.19578,
      "grad_norm": 1.1532961025165187,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 19578
    },
    {
      "epoch": 0.19579,
      "grad_norm": 1.48442655140837,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 19579
    },
    {
      "epoch": 0.1958,
      "grad_norm": 1.1852336573135187,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 19580
    },
    {
      "epoch": 0.19581,
      "grad_norm": 1.4192082820545153,
      "learning_rate": 0.003,
      "loss": 4.0178,
      "step": 19581
    },
    {
      "epoch": 0.19582,
      "grad_norm": 1.2840117706091172,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 19582
    },
    {
      "epoch": 0.19583,
      "grad_norm": 1.2979628805575405,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 19583
    },
    {
      "epoch": 0.19584,
      "grad_norm": 1.0485564253954378,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 19584
    },
    {
      "epoch": 0.19585,
      "grad_norm": 1.698992243899307,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 19585
    },
    {
      "epoch": 0.19586,
      "grad_norm": 1.2007570479118221,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 19586
    },
    {
      "epoch": 0.19587,
      "grad_norm": 1.299143804009154,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 19587
    },
    {
      "epoch": 0.19588,
      "grad_norm": 1.4096556135900993,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 19588
    },
    {
      "epoch": 0.19589,
      "grad_norm": 1.2516873856225414,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 19589
    },
    {
      "epoch": 0.1959,
      "grad_norm": 1.3511913389802903,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 19590
    },
    {
      "epoch": 0.19591,
      "grad_norm": 1.5757957682280774,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 19591
    },
    {
      "epoch": 0.19592,
      "grad_norm": 1.3050377336538197,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 19592
    },
    {
      "epoch": 0.19593,
      "grad_norm": 1.3278524136963044,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 19593
    },
    {
      "epoch": 0.19594,
      "grad_norm": 1.2983540617735474,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 19594
    },
    {
      "epoch": 0.19595,
      "grad_norm": 1.461744079705968,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 19595
    },
    {
      "epoch": 0.19596,
      "grad_norm": 1.1097298106864095,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 19596
    },
    {
      "epoch": 0.19597,
      "grad_norm": 1.5665459522554137,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 19597
    },
    {
      "epoch": 0.19598,
      "grad_norm": 1.167197822058204,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 19598
    },
    {
      "epoch": 0.19599,
      "grad_norm": 1.23200999485626,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 19599
    },
    {
      "epoch": 0.196,
      "grad_norm": 1.5284174897054683,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 19600
    },
    {
      "epoch": 0.19601,
      "grad_norm": 1.5221817934907782,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 19601
    },
    {
      "epoch": 0.19602,
      "grad_norm": 1.243043726637542,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 19602
    },
    {
      "epoch": 0.19603,
      "grad_norm": 1.358168188861275,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 19603
    },
    {
      "epoch": 0.19604,
      "grad_norm": 1.3824418445757143,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 19604
    },
    {
      "epoch": 0.19605,
      "grad_norm": 1.4568231810713086,
      "learning_rate": 0.003,
      "loss": 4.0241,
      "step": 19605
    },
    {
      "epoch": 0.19606,
      "grad_norm": 1.4945389488733762,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 19606
    },
    {
      "epoch": 0.19607,
      "grad_norm": 1.1173944068333428,
      "learning_rate": 0.003,
      "loss": 4.0271,
      "step": 19607
    },
    {
      "epoch": 0.19608,
      "grad_norm": 1.4386483359475473,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 19608
    },
    {
      "epoch": 0.19609,
      "grad_norm": 1.1587209555220592,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 19609
    },
    {
      "epoch": 0.1961,
      "grad_norm": 1.3855977084226376,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 19610
    },
    {
      "epoch": 0.19611,
      "grad_norm": 1.191095261316692,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 19611
    },
    {
      "epoch": 0.19612,
      "grad_norm": 1.329141360808254,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 19612
    },
    {
      "epoch": 0.19613,
      "grad_norm": 1.2914757409323085,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 19613
    },
    {
      "epoch": 0.19614,
      "grad_norm": 1.1990703428141272,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 19614
    },
    {
      "epoch": 0.19615,
      "grad_norm": 1.8458551392426072,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 19615
    },
    {
      "epoch": 0.19616,
      "grad_norm": 1.2531170952353852,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 19616
    },
    {
      "epoch": 0.19617,
      "grad_norm": 1.457888411021175,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 19617
    },
    {
      "epoch": 0.19618,
      "grad_norm": 1.2062617937354843,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 19618
    },
    {
      "epoch": 0.19619,
      "grad_norm": 1.2088629579297447,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 19619
    },
    {
      "epoch": 0.1962,
      "grad_norm": 1.3785785348043016,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 19620
    },
    {
      "epoch": 0.19621,
      "grad_norm": 1.1938751154673777,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 19621
    },
    {
      "epoch": 0.19622,
      "grad_norm": 1.405148392569504,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 19622
    },
    {
      "epoch": 0.19623,
      "grad_norm": 1.4096251360863021,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 19623
    },
    {
      "epoch": 0.19624,
      "grad_norm": 1.2746598690678428,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 19624
    },
    {
      "epoch": 0.19625,
      "grad_norm": 1.307968715382792,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 19625
    },
    {
      "epoch": 0.19626,
      "grad_norm": 1.3487640965516128,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 19626
    },
    {
      "epoch": 0.19627,
      "grad_norm": 1.3014966503765235,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 19627
    },
    {
      "epoch": 0.19628,
      "grad_norm": 1.5196558968221845,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 19628
    },
    {
      "epoch": 0.19629,
      "grad_norm": 1.351611078017904,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 19629
    },
    {
      "epoch": 0.1963,
      "grad_norm": 1.1982555578363527,
      "learning_rate": 0.003,
      "loss": 4.0201,
      "step": 19630
    },
    {
      "epoch": 0.19631,
      "grad_norm": 1.50858883084661,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 19631
    },
    {
      "epoch": 0.19632,
      "grad_norm": 1.209088040951843,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 19632
    },
    {
      "epoch": 0.19633,
      "grad_norm": 1.1706012946828515,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 19633
    },
    {
      "epoch": 0.19634,
      "grad_norm": 1.3884683601648247,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 19634
    },
    {
      "epoch": 0.19635,
      "grad_norm": 1.4084716436490483,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 19635
    },
    {
      "epoch": 0.19636,
      "grad_norm": 1.2012068473180846,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 19636
    },
    {
      "epoch": 0.19637,
      "grad_norm": 1.1904377831067214,
      "learning_rate": 0.003,
      "loss": 4.0312,
      "step": 19637
    },
    {
      "epoch": 0.19638,
      "grad_norm": 1.5549648250763126,
      "learning_rate": 0.003,
      "loss": 4.0106,
      "step": 19638
    },
    {
      "epoch": 0.19639,
      "grad_norm": 1.2924911454223567,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 19639
    },
    {
      "epoch": 0.1964,
      "grad_norm": 1.436073449635221,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 19640
    },
    {
      "epoch": 0.19641,
      "grad_norm": 1.2398135512071407,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 19641
    },
    {
      "epoch": 0.19642,
      "grad_norm": 1.4762790518015563,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 19642
    },
    {
      "epoch": 0.19643,
      "grad_norm": 1.3289287324093857,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 19643
    },
    {
      "epoch": 0.19644,
      "grad_norm": 1.415053931855098,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 19644
    },
    {
      "epoch": 0.19645,
      "grad_norm": 1.3670122111536873,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 19645
    },
    {
      "epoch": 0.19646,
      "grad_norm": 1.2611243225559388,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 19646
    },
    {
      "epoch": 0.19647,
      "grad_norm": 1.290674837100807,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 19647
    },
    {
      "epoch": 0.19648,
      "grad_norm": 1.3510109434285988,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 19648
    },
    {
      "epoch": 0.19649,
      "grad_norm": 1.2111258497810031,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 19649
    },
    {
      "epoch": 0.1965,
      "grad_norm": 1.626658891057458,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 19650
    },
    {
      "epoch": 0.19651,
      "grad_norm": 1.3400385588297339,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 19651
    },
    {
      "epoch": 0.19652,
      "grad_norm": 1.3719636497684822,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 19652
    },
    {
      "epoch": 0.19653,
      "grad_norm": 1.2347975873177621,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 19653
    },
    {
      "epoch": 0.19654,
      "grad_norm": 1.447584926720447,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 19654
    },
    {
      "epoch": 0.19655,
      "grad_norm": 1.1502211212205573,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 19655
    },
    {
      "epoch": 0.19656,
      "grad_norm": 1.5177566977359307,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 19656
    },
    {
      "epoch": 0.19657,
      "grad_norm": 1.6310376806213571,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 19657
    },
    {
      "epoch": 0.19658,
      "grad_norm": 1.1472209213061497,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 19658
    },
    {
      "epoch": 0.19659,
      "grad_norm": 1.4783673540103026,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 19659
    },
    {
      "epoch": 0.1966,
      "grad_norm": 1.442430981905889,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 19660
    },
    {
      "epoch": 0.19661,
      "grad_norm": 1.358558532685666,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 19661
    },
    {
      "epoch": 0.19662,
      "grad_norm": 1.2797478679943666,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 19662
    },
    {
      "epoch": 0.19663,
      "grad_norm": 1.1363377398858079,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 19663
    },
    {
      "epoch": 0.19664,
      "grad_norm": 1.3989413774376376,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 19664
    },
    {
      "epoch": 0.19665,
      "grad_norm": 1.1939159607281624,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 19665
    },
    {
      "epoch": 0.19666,
      "grad_norm": 1.2299607667675085,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 19666
    },
    {
      "epoch": 0.19667,
      "grad_norm": 1.4127656745930717,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 19667
    },
    {
      "epoch": 0.19668,
      "grad_norm": 1.5367893456607715,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 19668
    },
    {
      "epoch": 0.19669,
      "grad_norm": 1.3147698787709958,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 19669
    },
    {
      "epoch": 0.1967,
      "grad_norm": 1.218187691026266,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 19670
    },
    {
      "epoch": 0.19671,
      "grad_norm": 1.6882610127488684,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 19671
    },
    {
      "epoch": 0.19672,
      "grad_norm": 0.9715104515027442,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 19672
    },
    {
      "epoch": 0.19673,
      "grad_norm": 1.6056393856497801,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 19673
    },
    {
      "epoch": 0.19674,
      "grad_norm": 1.1057824576585424,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 19674
    },
    {
      "epoch": 0.19675,
      "grad_norm": 1.5481958238517912,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 19675
    },
    {
      "epoch": 0.19676,
      "grad_norm": 1.1711742364191884,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 19676
    },
    {
      "epoch": 0.19677,
      "grad_norm": 1.7711689647842495,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 19677
    },
    {
      "epoch": 0.19678,
      "grad_norm": 1.219879003445908,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 19678
    },
    {
      "epoch": 0.19679,
      "grad_norm": 1.226992949333215,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 19679
    },
    {
      "epoch": 0.1968,
      "grad_norm": 1.5091792519248097,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 19680
    },
    {
      "epoch": 0.19681,
      "grad_norm": 1.1502942125089162,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 19681
    },
    {
      "epoch": 0.19682,
      "grad_norm": 1.369165848780662,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 19682
    },
    {
      "epoch": 0.19683,
      "grad_norm": 1.50610209047457,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 19683
    },
    {
      "epoch": 0.19684,
      "grad_norm": 1.3913243831111002,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 19684
    },
    {
      "epoch": 0.19685,
      "grad_norm": 1.489411249466416,
      "learning_rate": 0.003,
      "loss": 4.026,
      "step": 19685
    },
    {
      "epoch": 0.19686,
      "grad_norm": 1.2988468071078383,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 19686
    },
    {
      "epoch": 0.19687,
      "grad_norm": 1.2912204537705885,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 19687
    },
    {
      "epoch": 0.19688,
      "grad_norm": 1.3233081118225472,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 19688
    },
    {
      "epoch": 0.19689,
      "grad_norm": 1.3256641452369216,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 19689
    },
    {
      "epoch": 0.1969,
      "grad_norm": 1.2660930257667473,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 19690
    },
    {
      "epoch": 0.19691,
      "grad_norm": 1.1883754721295503,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 19691
    },
    {
      "epoch": 0.19692,
      "grad_norm": 1.376985420827303,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 19692
    },
    {
      "epoch": 0.19693,
      "grad_norm": 1.3150305005470482,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 19693
    },
    {
      "epoch": 0.19694,
      "grad_norm": 1.47217745519008,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 19694
    },
    {
      "epoch": 0.19695,
      "grad_norm": 1.1481972427979865,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 19695
    },
    {
      "epoch": 0.19696,
      "grad_norm": 1.321599317780327,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 19696
    },
    {
      "epoch": 0.19697,
      "grad_norm": 1.6388995502727124,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 19697
    },
    {
      "epoch": 0.19698,
      "grad_norm": 1.2484343120322956,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 19698
    },
    {
      "epoch": 0.19699,
      "grad_norm": 1.5216813620088356,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 19699
    },
    {
      "epoch": 0.197,
      "grad_norm": 1.0916232273133109,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 19700
    },
    {
      "epoch": 0.19701,
      "grad_norm": 1.1873914219973216,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 19701
    },
    {
      "epoch": 0.19702,
      "grad_norm": 1.4738247803097095,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 19702
    },
    {
      "epoch": 0.19703,
      "grad_norm": 1.1357491362604857,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 19703
    },
    {
      "epoch": 0.19704,
      "grad_norm": 1.4484137053164607,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 19704
    },
    {
      "epoch": 0.19705,
      "grad_norm": 1.3439057007523725,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 19705
    },
    {
      "epoch": 0.19706,
      "grad_norm": 1.4023196364405632,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 19706
    },
    {
      "epoch": 0.19707,
      "grad_norm": 1.2544428278487003,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 19707
    },
    {
      "epoch": 0.19708,
      "grad_norm": 1.224674119818485,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 19708
    },
    {
      "epoch": 0.19709,
      "grad_norm": 1.604438625568975,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 19709
    },
    {
      "epoch": 0.1971,
      "grad_norm": 1.2311551502434666,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 19710
    },
    {
      "epoch": 0.19711,
      "grad_norm": 1.227087896947134,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 19711
    },
    {
      "epoch": 0.19712,
      "grad_norm": 1.4217940509714437,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 19712
    },
    {
      "epoch": 0.19713,
      "grad_norm": 1.3413863733209068,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 19713
    },
    {
      "epoch": 0.19714,
      "grad_norm": 1.4078100967643121,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 19714
    },
    {
      "epoch": 0.19715,
      "grad_norm": 1.2338926081565247,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 19715
    },
    {
      "epoch": 0.19716,
      "grad_norm": 1.326378839883387,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 19716
    },
    {
      "epoch": 0.19717,
      "grad_norm": 1.3322748801981528,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 19717
    },
    {
      "epoch": 0.19718,
      "grad_norm": 1.374045360430272,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 19718
    },
    {
      "epoch": 0.19719,
      "grad_norm": 1.2535400024902488,
      "learning_rate": 0.003,
      "loss": 4.0136,
      "step": 19719
    },
    {
      "epoch": 0.1972,
      "grad_norm": 1.7216545683033468,
      "learning_rate": 0.003,
      "loss": 4.0289,
      "step": 19720
    },
    {
      "epoch": 0.19721,
      "grad_norm": 1.143211416994933,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 19721
    },
    {
      "epoch": 0.19722,
      "grad_norm": 1.347843799576987,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 19722
    },
    {
      "epoch": 0.19723,
      "grad_norm": 1.2263043437640626,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 19723
    },
    {
      "epoch": 0.19724,
      "grad_norm": 1.4235896139380093,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 19724
    },
    {
      "epoch": 0.19725,
      "grad_norm": 1.249363935710837,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 19725
    },
    {
      "epoch": 0.19726,
      "grad_norm": 1.3621808175319952,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 19726
    },
    {
      "epoch": 0.19727,
      "grad_norm": 1.2027545345421449,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 19727
    },
    {
      "epoch": 0.19728,
      "grad_norm": 1.3819118273551108,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 19728
    },
    {
      "epoch": 0.19729,
      "grad_norm": 1.3139500700926068,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 19729
    },
    {
      "epoch": 0.1973,
      "grad_norm": 1.2412752229393393,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 19730
    },
    {
      "epoch": 0.19731,
      "grad_norm": 1.2513695591662786,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 19731
    },
    {
      "epoch": 0.19732,
      "grad_norm": 1.6091074837307349,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 19732
    },
    {
      "epoch": 0.19733,
      "grad_norm": 1.1823045345431846,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 19733
    },
    {
      "epoch": 0.19734,
      "grad_norm": 1.5925556387548963,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 19734
    },
    {
      "epoch": 0.19735,
      "grad_norm": 0.9643452763097676,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 19735
    },
    {
      "epoch": 0.19736,
      "grad_norm": 1.4569350256120528,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 19736
    },
    {
      "epoch": 0.19737,
      "grad_norm": 1.3691898829958777,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 19737
    },
    {
      "epoch": 0.19738,
      "grad_norm": 1.0934636594109788,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 19738
    },
    {
      "epoch": 0.19739,
      "grad_norm": 1.4965763361831188,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 19739
    },
    {
      "epoch": 0.1974,
      "grad_norm": 1.2917382252185112,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 19740
    },
    {
      "epoch": 0.19741,
      "grad_norm": 1.3573544639844923,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 19741
    },
    {
      "epoch": 0.19742,
      "grad_norm": 1.1576171567288007,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 19742
    },
    {
      "epoch": 0.19743,
      "grad_norm": 1.333131141716199,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 19743
    },
    {
      "epoch": 0.19744,
      "grad_norm": 1.3814560633215616,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 19744
    },
    {
      "epoch": 0.19745,
      "grad_norm": 1.1487887883235977,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 19745
    },
    {
      "epoch": 0.19746,
      "grad_norm": 1.3776024966932388,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 19746
    },
    {
      "epoch": 0.19747,
      "grad_norm": 1.3859940421713073,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 19747
    },
    {
      "epoch": 0.19748,
      "grad_norm": 1.5127485072971667,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 19748
    },
    {
      "epoch": 0.19749,
      "grad_norm": 1.1785057150237257,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 19749
    },
    {
      "epoch": 0.1975,
      "grad_norm": 1.589956479162368,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 19750
    },
    {
      "epoch": 0.19751,
      "grad_norm": 1.4266342589366159,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 19751
    },
    {
      "epoch": 0.19752,
      "grad_norm": 1.3121486648783165,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 19752
    },
    {
      "epoch": 0.19753,
      "grad_norm": 1.4121421014664186,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 19753
    },
    {
      "epoch": 0.19754,
      "grad_norm": 1.0924639918405121,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 19754
    },
    {
      "epoch": 0.19755,
      "grad_norm": 1.487611705583233,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 19755
    },
    {
      "epoch": 0.19756,
      "grad_norm": 1.2796519124302443,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 19756
    },
    {
      "epoch": 0.19757,
      "grad_norm": 1.6484103057642017,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 19757
    },
    {
      "epoch": 0.19758,
      "grad_norm": 1.311605176822461,
      "learning_rate": 0.003,
      "loss": 4.103,
      "step": 19758
    },
    {
      "epoch": 0.19759,
      "grad_norm": 1.2889955881539013,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 19759
    },
    {
      "epoch": 0.1976,
      "grad_norm": 1.4292082948589133,
      "learning_rate": 0.003,
      "loss": 4.1089,
      "step": 19760
    },
    {
      "epoch": 0.19761,
      "grad_norm": 1.3230632841660548,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 19761
    },
    {
      "epoch": 0.19762,
      "grad_norm": 1.3688875268505394,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 19762
    },
    {
      "epoch": 0.19763,
      "grad_norm": 1.2823318554850982,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 19763
    },
    {
      "epoch": 0.19764,
      "grad_norm": 1.2166216140103328,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 19764
    },
    {
      "epoch": 0.19765,
      "grad_norm": 1.2852363984667483,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 19765
    },
    {
      "epoch": 0.19766,
      "grad_norm": 1.2898294841391993,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 19766
    },
    {
      "epoch": 0.19767,
      "grad_norm": 1.2367526470670498,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 19767
    },
    {
      "epoch": 0.19768,
      "grad_norm": 1.3154865687307928,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 19768
    },
    {
      "epoch": 0.19769,
      "grad_norm": 1.2603167654919585,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 19769
    },
    {
      "epoch": 0.1977,
      "grad_norm": 1.3007040435706034,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 19770
    },
    {
      "epoch": 0.19771,
      "grad_norm": 1.4114993924752812,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 19771
    },
    {
      "epoch": 0.19772,
      "grad_norm": 1.3096137540244503,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 19772
    },
    {
      "epoch": 0.19773,
      "grad_norm": 1.4355194693826703,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 19773
    },
    {
      "epoch": 0.19774,
      "grad_norm": 1.4518512466537297,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 19774
    },
    {
      "epoch": 0.19775,
      "grad_norm": 1.6062944604381495,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 19775
    },
    {
      "epoch": 0.19776,
      "grad_norm": 1.4836222015556304,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 19776
    },
    {
      "epoch": 0.19777,
      "grad_norm": 1.3167608342443442,
      "learning_rate": 0.003,
      "loss": 4.0283,
      "step": 19777
    },
    {
      "epoch": 0.19778,
      "grad_norm": 1.3800670965071138,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 19778
    },
    {
      "epoch": 0.19779,
      "grad_norm": 1.2467694158107294,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 19779
    },
    {
      "epoch": 0.1978,
      "grad_norm": 1.1383855194669683,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 19780
    },
    {
      "epoch": 0.19781,
      "grad_norm": 1.4457824499750687,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 19781
    },
    {
      "epoch": 0.19782,
      "grad_norm": 1.1524489184514475,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 19782
    },
    {
      "epoch": 0.19783,
      "grad_norm": 1.4835000980643374,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 19783
    },
    {
      "epoch": 0.19784,
      "grad_norm": 1.1729540045494726,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 19784
    },
    {
      "epoch": 0.19785,
      "grad_norm": 1.5597271638773154,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 19785
    },
    {
      "epoch": 0.19786,
      "grad_norm": 1.3320044397616282,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 19786
    },
    {
      "epoch": 0.19787,
      "grad_norm": 1.4422976842356665,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 19787
    },
    {
      "epoch": 0.19788,
      "grad_norm": 1.3253455241649499,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 19788
    },
    {
      "epoch": 0.19789,
      "grad_norm": 1.128730509489487,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 19789
    },
    {
      "epoch": 0.1979,
      "grad_norm": 1.5220058756348567,
      "learning_rate": 0.003,
      "loss": 4.0313,
      "step": 19790
    },
    {
      "epoch": 0.19791,
      "grad_norm": 1.0193149807952373,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 19791
    },
    {
      "epoch": 0.19792,
      "grad_norm": 1.7698679856468362,
      "learning_rate": 0.003,
      "loss": 4.0972,
      "step": 19792
    },
    {
      "epoch": 0.19793,
      "grad_norm": 1.1620624717348689,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 19793
    },
    {
      "epoch": 0.19794,
      "grad_norm": 1.3691058623543266,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 19794
    },
    {
      "epoch": 0.19795,
      "grad_norm": 1.3043787042526258,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 19795
    },
    {
      "epoch": 0.19796,
      "grad_norm": 1.4206097790958214,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 19796
    },
    {
      "epoch": 0.19797,
      "grad_norm": 1.236986530855312,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 19797
    },
    {
      "epoch": 0.19798,
      "grad_norm": 1.3913813954888499,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 19798
    },
    {
      "epoch": 0.19799,
      "grad_norm": 1.131110863878918,
      "learning_rate": 0.003,
      "loss": 4.0161,
      "step": 19799
    },
    {
      "epoch": 0.198,
      "grad_norm": 1.4867324186283586,
      "learning_rate": 0.003,
      "loss": 4.0977,
      "step": 19800
    },
    {
      "epoch": 0.19801,
      "grad_norm": 1.419459874682956,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 19801
    },
    {
      "epoch": 0.19802,
      "grad_norm": 1.502929630810888,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 19802
    },
    {
      "epoch": 0.19803,
      "grad_norm": 1.35980660698995,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 19803
    },
    {
      "epoch": 0.19804,
      "grad_norm": 1.4439911170495572,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 19804
    },
    {
      "epoch": 0.19805,
      "grad_norm": 1.3834374075151001,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 19805
    },
    {
      "epoch": 0.19806,
      "grad_norm": 1.1914514918607964,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 19806
    },
    {
      "epoch": 0.19807,
      "grad_norm": 1.290229652438182,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 19807
    },
    {
      "epoch": 0.19808,
      "grad_norm": 1.2694731438203108,
      "learning_rate": 0.003,
      "loss": 4.111,
      "step": 19808
    },
    {
      "epoch": 0.19809,
      "grad_norm": 1.3692766120784927,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 19809
    },
    {
      "epoch": 0.1981,
      "grad_norm": 1.1628895391517091,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 19810
    },
    {
      "epoch": 0.19811,
      "grad_norm": 1.424318988673435,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 19811
    },
    {
      "epoch": 0.19812,
      "grad_norm": 1.4851767350452065,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 19812
    },
    {
      "epoch": 0.19813,
      "grad_norm": 1.4167736869370577,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 19813
    },
    {
      "epoch": 0.19814,
      "grad_norm": 1.470995925875908,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 19814
    },
    {
      "epoch": 0.19815,
      "grad_norm": 1.2856835824921131,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 19815
    },
    {
      "epoch": 0.19816,
      "grad_norm": 1.286169394231622,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 19816
    },
    {
      "epoch": 0.19817,
      "grad_norm": 1.1940010585749086,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 19817
    },
    {
      "epoch": 0.19818,
      "grad_norm": 1.695886874437397,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 19818
    },
    {
      "epoch": 0.19819,
      "grad_norm": 1.0606100857710243,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 19819
    },
    {
      "epoch": 0.1982,
      "grad_norm": 1.3438230427854372,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 19820
    },
    {
      "epoch": 0.19821,
      "grad_norm": 1.2955036264556594,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 19821
    },
    {
      "epoch": 0.19822,
      "grad_norm": 1.3098240891287187,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 19822
    },
    {
      "epoch": 0.19823,
      "grad_norm": 1.3909821017152115,
      "learning_rate": 0.003,
      "loss": 4.0261,
      "step": 19823
    },
    {
      "epoch": 0.19824,
      "grad_norm": 1.3030669319903594,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 19824
    },
    {
      "epoch": 0.19825,
      "grad_norm": 1.2377995859660755,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 19825
    },
    {
      "epoch": 0.19826,
      "grad_norm": 1.138853085824324,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 19826
    },
    {
      "epoch": 0.19827,
      "grad_norm": 1.5446729952809068,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 19827
    },
    {
      "epoch": 0.19828,
      "grad_norm": 1.1720485330017334,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 19828
    },
    {
      "epoch": 0.19829,
      "grad_norm": 1.4631446086855109,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 19829
    },
    {
      "epoch": 0.1983,
      "grad_norm": 1.27759947935634,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 19830
    },
    {
      "epoch": 0.19831,
      "grad_norm": 1.3392163108408761,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 19831
    },
    {
      "epoch": 0.19832,
      "grad_norm": 1.5133191924485834,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 19832
    },
    {
      "epoch": 0.19833,
      "grad_norm": 1.293294990560124,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 19833
    },
    {
      "epoch": 0.19834,
      "grad_norm": 1.6444223465147898,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 19834
    },
    {
      "epoch": 0.19835,
      "grad_norm": 1.1045129064447632,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 19835
    },
    {
      "epoch": 0.19836,
      "grad_norm": 1.3823524291233054,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 19836
    },
    {
      "epoch": 0.19837,
      "grad_norm": 1.3886344862669628,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 19837
    },
    {
      "epoch": 0.19838,
      "grad_norm": 1.310600756697459,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 19838
    },
    {
      "epoch": 0.19839,
      "grad_norm": 1.1322081362500147,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 19839
    },
    {
      "epoch": 0.1984,
      "grad_norm": 1.6079438090906444,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 19840
    },
    {
      "epoch": 0.19841,
      "grad_norm": 1.1272952523817377,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 19841
    },
    {
      "epoch": 0.19842,
      "grad_norm": 1.5750339813943295,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 19842
    },
    {
      "epoch": 0.19843,
      "grad_norm": 1.2076596936468764,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 19843
    },
    {
      "epoch": 0.19844,
      "grad_norm": 1.3084089414826487,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 19844
    },
    {
      "epoch": 0.19845,
      "grad_norm": 1.2248463472094597,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 19845
    },
    {
      "epoch": 0.19846,
      "grad_norm": 1.3091388485192232,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 19846
    },
    {
      "epoch": 0.19847,
      "grad_norm": 1.2908395318229668,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 19847
    },
    {
      "epoch": 0.19848,
      "grad_norm": 1.477025855204809,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 19848
    },
    {
      "epoch": 0.19849,
      "grad_norm": 1.1826392681432139,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 19849
    },
    {
      "epoch": 0.1985,
      "grad_norm": 1.165434565753471,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 19850
    },
    {
      "epoch": 0.19851,
      "grad_norm": 1.498316728735124,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 19851
    },
    {
      "epoch": 0.19852,
      "grad_norm": 1.140564156512752,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 19852
    },
    {
      "epoch": 0.19853,
      "grad_norm": 1.3172906843177683,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 19853
    },
    {
      "epoch": 0.19854,
      "grad_norm": 1.2591174364083555,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 19854
    },
    {
      "epoch": 0.19855,
      "grad_norm": 1.209338093553082,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 19855
    },
    {
      "epoch": 0.19856,
      "grad_norm": 1.2883915670849042,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 19856
    },
    {
      "epoch": 0.19857,
      "grad_norm": 1.4796169194131135,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 19857
    },
    {
      "epoch": 0.19858,
      "grad_norm": 1.3598468046038348,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 19858
    },
    {
      "epoch": 0.19859,
      "grad_norm": 1.4518947651612657,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 19859
    },
    {
      "epoch": 0.1986,
      "grad_norm": 1.2877903450248325,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 19860
    },
    {
      "epoch": 0.19861,
      "grad_norm": 1.3644906622993722,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 19861
    },
    {
      "epoch": 0.19862,
      "grad_norm": 1.3852920489859368,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 19862
    },
    {
      "epoch": 0.19863,
      "grad_norm": 1.2485641830190142,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 19863
    },
    {
      "epoch": 0.19864,
      "grad_norm": 1.5908750559888494,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 19864
    },
    {
      "epoch": 0.19865,
      "grad_norm": 1.0458613828677508,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 19865
    },
    {
      "epoch": 0.19866,
      "grad_norm": 1.5942201250800003,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 19866
    },
    {
      "epoch": 0.19867,
      "grad_norm": 1.4677573902918055,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 19867
    },
    {
      "epoch": 0.19868,
      "grad_norm": 1.6739373855135689,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 19868
    },
    {
      "epoch": 0.19869,
      "grad_norm": 1.2004253899879886,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 19869
    },
    {
      "epoch": 0.1987,
      "grad_norm": 1.1947915365252808,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 19870
    },
    {
      "epoch": 0.19871,
      "grad_norm": 1.3714915182793268,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 19871
    },
    {
      "epoch": 0.19872,
      "grad_norm": 1.303417716010588,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 19872
    },
    {
      "epoch": 0.19873,
      "grad_norm": 1.593425253178839,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 19873
    },
    {
      "epoch": 0.19874,
      "grad_norm": 1.4799633983016245,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 19874
    },
    {
      "epoch": 0.19875,
      "grad_norm": 1.1939553210421372,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 19875
    },
    {
      "epoch": 0.19876,
      "grad_norm": 1.4457963114860786,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 19876
    },
    {
      "epoch": 0.19877,
      "grad_norm": 1.0782302938928832,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 19877
    },
    {
      "epoch": 0.19878,
      "grad_norm": 1.5576792869516773,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 19878
    },
    {
      "epoch": 0.19879,
      "grad_norm": 0.9795673626634734,
      "learning_rate": 0.003,
      "loss": 4.0218,
      "step": 19879
    },
    {
      "epoch": 0.1988,
      "grad_norm": 1.643192358745065,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 19880
    },
    {
      "epoch": 0.19881,
      "grad_norm": 1.2499734613264617,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 19881
    },
    {
      "epoch": 0.19882,
      "grad_norm": 1.3012922937892841,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 19882
    },
    {
      "epoch": 0.19883,
      "grad_norm": 1.681481905552871,
      "learning_rate": 0.003,
      "loss": 4.0914,
      "step": 19883
    },
    {
      "epoch": 0.19884,
      "grad_norm": 1.179966341422776,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 19884
    },
    {
      "epoch": 0.19885,
      "grad_norm": 1.6502341088662849,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 19885
    },
    {
      "epoch": 0.19886,
      "grad_norm": 1.4260864819610708,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 19886
    },
    {
      "epoch": 0.19887,
      "grad_norm": 1.5303099595712761,
      "learning_rate": 0.003,
      "loss": 4.0998,
      "step": 19887
    },
    {
      "epoch": 0.19888,
      "grad_norm": 1.3805490167116559,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 19888
    },
    {
      "epoch": 0.19889,
      "grad_norm": 1.1876169220641908,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 19889
    },
    {
      "epoch": 0.1989,
      "grad_norm": 1.3678792103686264,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 19890
    },
    {
      "epoch": 0.19891,
      "grad_norm": 1.2240000440553933,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 19891
    },
    {
      "epoch": 0.19892,
      "grad_norm": 1.5087633924656867,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 19892
    },
    {
      "epoch": 0.19893,
      "grad_norm": 1.0544259075986604,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 19893
    },
    {
      "epoch": 0.19894,
      "grad_norm": 1.678084831806876,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 19894
    },
    {
      "epoch": 0.19895,
      "grad_norm": 1.2068767730490293,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 19895
    },
    {
      "epoch": 0.19896,
      "grad_norm": 1.1556900429965884,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 19896
    },
    {
      "epoch": 0.19897,
      "grad_norm": 1.39828851238043,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 19897
    },
    {
      "epoch": 0.19898,
      "grad_norm": 1.1750349442371004,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 19898
    },
    {
      "epoch": 0.19899,
      "grad_norm": 1.7125131283056145,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 19899
    },
    {
      "epoch": 0.199,
      "grad_norm": 1.0903171062975907,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 19900
    },
    {
      "epoch": 0.19901,
      "grad_norm": 1.4278968872918296,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 19901
    },
    {
      "epoch": 0.19902,
      "grad_norm": 1.2769568163873293,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 19902
    },
    {
      "epoch": 0.19903,
      "grad_norm": 1.320969115171914,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 19903
    },
    {
      "epoch": 0.19904,
      "grad_norm": 1.4373934007265843,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 19904
    },
    {
      "epoch": 0.19905,
      "grad_norm": 1.3337403406343205,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 19905
    },
    {
      "epoch": 0.19906,
      "grad_norm": 1.425015606638654,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 19906
    },
    {
      "epoch": 0.19907,
      "grad_norm": 1.364814039900154,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 19907
    },
    {
      "epoch": 0.19908,
      "grad_norm": 1.4000606683834802,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 19908
    },
    {
      "epoch": 0.19909,
      "grad_norm": 1.349820791493697,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 19909
    },
    {
      "epoch": 0.1991,
      "grad_norm": 1.2101690084272794,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 19910
    },
    {
      "epoch": 0.19911,
      "grad_norm": 1.328024974881233,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 19911
    },
    {
      "epoch": 0.19912,
      "grad_norm": 1.5214374495245637,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 19912
    },
    {
      "epoch": 0.19913,
      "grad_norm": 1.2319966053358085,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 19913
    },
    {
      "epoch": 0.19914,
      "grad_norm": 1.624354784976297,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 19914
    },
    {
      "epoch": 0.19915,
      "grad_norm": 1.1680021177833477,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 19915
    },
    {
      "epoch": 0.19916,
      "grad_norm": 1.4351009845349718,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 19916
    },
    {
      "epoch": 0.19917,
      "grad_norm": 1.1904991323865397,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 19917
    },
    {
      "epoch": 0.19918,
      "grad_norm": 1.3005765935111615,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 19918
    },
    {
      "epoch": 0.19919,
      "grad_norm": 1.5136706664777948,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 19919
    },
    {
      "epoch": 0.1992,
      "grad_norm": 1.3631547180491872,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 19920
    },
    {
      "epoch": 0.19921,
      "grad_norm": 1.3177597213688967,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 19921
    },
    {
      "epoch": 0.19922,
      "grad_norm": 1.3627078741513534,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 19922
    },
    {
      "epoch": 0.19923,
      "grad_norm": 1.7090650718346874,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 19923
    },
    {
      "epoch": 0.19924,
      "grad_norm": 1.3016585879344675,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 19924
    },
    {
      "epoch": 0.19925,
      "grad_norm": 1.321937135135949,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 19925
    },
    {
      "epoch": 0.19926,
      "grad_norm": 1.129513312386755,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 19926
    },
    {
      "epoch": 0.19927,
      "grad_norm": 1.3347963016520732,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 19927
    },
    {
      "epoch": 0.19928,
      "grad_norm": 1.1853099035095687,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 19928
    },
    {
      "epoch": 0.19929,
      "grad_norm": 1.335338652852557,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 19929
    },
    {
      "epoch": 0.1993,
      "grad_norm": 1.2792498490298683,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 19930
    },
    {
      "epoch": 0.19931,
      "grad_norm": 1.5528370483779794,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 19931
    },
    {
      "epoch": 0.19932,
      "grad_norm": 1.3599477542356222,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 19932
    },
    {
      "epoch": 0.19933,
      "grad_norm": 1.1688347790743336,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 19933
    },
    {
      "epoch": 0.19934,
      "grad_norm": 1.3690338688134256,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 19934
    },
    {
      "epoch": 0.19935,
      "grad_norm": 1.4862042578677468,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 19935
    },
    {
      "epoch": 0.19936,
      "grad_norm": 1.6043547474328443,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 19936
    },
    {
      "epoch": 0.19937,
      "grad_norm": 1.1050656290813126,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 19937
    },
    {
      "epoch": 0.19938,
      "grad_norm": 1.4952547269417187,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 19938
    },
    {
      "epoch": 0.19939,
      "grad_norm": 1.194603360033594,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 19939
    },
    {
      "epoch": 0.1994,
      "grad_norm": 1.665348114353188,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 19940
    },
    {
      "epoch": 0.19941,
      "grad_norm": 1.1753350624452628,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 19941
    },
    {
      "epoch": 0.19942,
      "grad_norm": 1.5595213126425294,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 19942
    },
    {
      "epoch": 0.19943,
      "grad_norm": 1.2802671819523601,
      "learning_rate": 0.003,
      "loss": 4.0815,
      "step": 19943
    },
    {
      "epoch": 0.19944,
      "grad_norm": 1.3149789591004202,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 19944
    },
    {
      "epoch": 0.19945,
      "grad_norm": 1.2048649365192403,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 19945
    },
    {
      "epoch": 0.19946,
      "grad_norm": 1.4394319122791803,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 19946
    },
    {
      "epoch": 0.19947,
      "grad_norm": 1.2301309484363132,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 19947
    },
    {
      "epoch": 0.19948,
      "grad_norm": 1.4508399657719824,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 19948
    },
    {
      "epoch": 0.19949,
      "grad_norm": 1.22998463320884,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 19949
    },
    {
      "epoch": 0.1995,
      "grad_norm": 1.600550233121151,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 19950
    },
    {
      "epoch": 0.19951,
      "grad_norm": 1.1852360744658244,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 19951
    },
    {
      "epoch": 0.19952,
      "grad_norm": 1.2761907028108868,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 19952
    },
    {
      "epoch": 0.19953,
      "grad_norm": 1.2186659520098557,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 19953
    },
    {
      "epoch": 0.19954,
      "grad_norm": 1.7912991145575592,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 19954
    },
    {
      "epoch": 0.19955,
      "grad_norm": 0.9624001380157303,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 19955
    },
    {
      "epoch": 0.19956,
      "grad_norm": 1.5260929370008876,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 19956
    },
    {
      "epoch": 0.19957,
      "grad_norm": 1.2051167624010655,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 19957
    },
    {
      "epoch": 0.19958,
      "grad_norm": 1.468663024710126,
      "learning_rate": 0.003,
      "loss": 4.0882,
      "step": 19958
    },
    {
      "epoch": 0.19959,
      "grad_norm": 1.2840580166998898,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 19959
    },
    {
      "epoch": 0.1996,
      "grad_norm": 1.3296993487934532,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 19960
    },
    {
      "epoch": 0.19961,
      "grad_norm": 1.2756426180152056,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 19961
    },
    {
      "epoch": 0.19962,
      "grad_norm": 1.1883889884680248,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 19962
    },
    {
      "epoch": 0.19963,
      "grad_norm": 1.2891980586589682,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 19963
    },
    {
      "epoch": 0.19964,
      "grad_norm": 1.2864650029059697,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 19964
    },
    {
      "epoch": 0.19965,
      "grad_norm": 1.364135770276665,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 19965
    },
    {
      "epoch": 0.19966,
      "grad_norm": 1.475210178191404,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 19966
    },
    {
      "epoch": 0.19967,
      "grad_norm": 1.180029806466963,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 19967
    },
    {
      "epoch": 0.19968,
      "grad_norm": 1.4155951959307755,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 19968
    },
    {
      "epoch": 0.19969,
      "grad_norm": 1.107023638411012,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 19969
    },
    {
      "epoch": 0.1997,
      "grad_norm": 1.4507236517739721,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 19970
    },
    {
      "epoch": 0.19971,
      "grad_norm": 1.2780514593119168,
      "learning_rate": 0.003,
      "loss": 4.0224,
      "step": 19971
    },
    {
      "epoch": 0.19972,
      "grad_norm": 1.1632612970969403,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 19972
    },
    {
      "epoch": 0.19973,
      "grad_norm": 1.3913863108877422,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 19973
    },
    {
      "epoch": 0.19974,
      "grad_norm": 1.143341976288878,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 19974
    },
    {
      "epoch": 0.19975,
      "grad_norm": 1.3601291769811557,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 19975
    },
    {
      "epoch": 0.19976,
      "grad_norm": 1.1314455011136493,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 19976
    },
    {
      "epoch": 0.19977,
      "grad_norm": 1.6772200213837052,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 19977
    },
    {
      "epoch": 0.19978,
      "grad_norm": 1.165581605438778,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 19978
    },
    {
      "epoch": 0.19979,
      "grad_norm": 1.4402174169262436,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 19979
    },
    {
      "epoch": 0.1998,
      "grad_norm": 1.4146324527494631,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 19980
    },
    {
      "epoch": 0.19981,
      "grad_norm": 1.2198891774791902,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 19981
    },
    {
      "epoch": 0.19982,
      "grad_norm": 1.5295226286415056,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 19982
    },
    {
      "epoch": 0.19983,
      "grad_norm": 1.1364146683710168,
      "learning_rate": 0.003,
      "loss": 4.0988,
      "step": 19983
    },
    {
      "epoch": 0.19984,
      "grad_norm": 1.3098536703130816,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 19984
    },
    {
      "epoch": 0.19985,
      "grad_norm": 1.4299926040866193,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 19985
    },
    {
      "epoch": 0.19986,
      "grad_norm": 1.2957755651925076,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 19986
    },
    {
      "epoch": 0.19987,
      "grad_norm": 1.7681395903891073,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 19987
    },
    {
      "epoch": 0.19988,
      "grad_norm": 1.3033406689832867,
      "learning_rate": 0.003,
      "loss": 4.0134,
      "step": 19988
    },
    {
      "epoch": 0.19989,
      "grad_norm": 1.5434221339459064,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 19989
    },
    {
      "epoch": 0.1999,
      "grad_norm": 1.6724384889930606,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 19990
    },
    {
      "epoch": 0.19991,
      "grad_norm": 1.1969774103251682,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 19991
    },
    {
      "epoch": 0.19992,
      "grad_norm": 1.3784163899054613,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 19992
    },
    {
      "epoch": 0.19993,
      "grad_norm": 1.1385646467829325,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 19993
    },
    {
      "epoch": 0.19994,
      "grad_norm": 1.5850542355195134,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 19994
    },
    {
      "epoch": 0.19995,
      "grad_norm": 0.9839155399999102,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 19995
    },
    {
      "epoch": 0.19996,
      "grad_norm": 1.732207679268672,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 19996
    },
    {
      "epoch": 0.19997,
      "grad_norm": 1.587619606977346,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 19997
    },
    {
      "epoch": 0.19998,
      "grad_norm": 1.3720931438155322,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 19998
    },
    {
      "epoch": 0.19999,
      "grad_norm": 1.189825864203958,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 19999
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0731831679830781,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 20000
    },
    {
      "epoch": 0.20001,
      "grad_norm": 1.3452001360989823,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 20001
    },
    {
      "epoch": 0.20002,
      "grad_norm": 1.2569999204028397,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 20002
    },
    {
      "epoch": 0.20003,
      "grad_norm": 1.1248052065750045,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 20003
    },
    {
      "epoch": 0.20004,
      "grad_norm": 1.3231987893653487,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 20004
    },
    {
      "epoch": 0.20005,
      "grad_norm": 1.1075288655403044,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 20005
    },
    {
      "epoch": 0.20006,
      "grad_norm": 1.76241002894299,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 20006
    },
    {
      "epoch": 0.20007,
      "grad_norm": 1.1364575908920271,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 20007
    },
    {
      "epoch": 0.20008,
      "grad_norm": 1.34353856173137,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 20008
    },
    {
      "epoch": 0.20009,
      "grad_norm": 1.713753754243954,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 20009
    },
    {
      "epoch": 0.2001,
      "grad_norm": 1.2003016467894756,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 20010
    },
    {
      "epoch": 0.20011,
      "grad_norm": 1.3432840288651786,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 20011
    },
    {
      "epoch": 0.20012,
      "grad_norm": 1.2961175098904854,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 20012
    },
    {
      "epoch": 0.20013,
      "grad_norm": 1.2737245750977582,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 20013
    },
    {
      "epoch": 0.20014,
      "grad_norm": 1.3107480433607972,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 20014
    },
    {
      "epoch": 0.20015,
      "grad_norm": 1.2635527281555388,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 20015
    },
    {
      "epoch": 0.20016,
      "grad_norm": 1.3073564491459384,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 20016
    },
    {
      "epoch": 0.20017,
      "grad_norm": 1.5120899435045736,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 20017
    },
    {
      "epoch": 0.20018,
      "grad_norm": 1.4893915911432425,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 20018
    },
    {
      "epoch": 0.20019,
      "grad_norm": 1.0854389602613164,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 20019
    },
    {
      "epoch": 0.2002,
      "grad_norm": 1.3465381909374021,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 20020
    },
    {
      "epoch": 0.20021,
      "grad_norm": 1.192659101793909,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 20021
    },
    {
      "epoch": 0.20022,
      "grad_norm": 1.4715970544341472,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 20022
    },
    {
      "epoch": 0.20023,
      "grad_norm": 1.1248033760576712,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 20023
    },
    {
      "epoch": 0.20024,
      "grad_norm": 1.2968157081760008,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 20024
    },
    {
      "epoch": 0.20025,
      "grad_norm": 1.5507410282592036,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 20025
    },
    {
      "epoch": 0.20026,
      "grad_norm": 1.0551108520430137,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 20026
    },
    {
      "epoch": 0.20027,
      "grad_norm": 1.604496182642803,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 20027
    },
    {
      "epoch": 0.20028,
      "grad_norm": 1.260339308218672,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 20028
    },
    {
      "epoch": 0.20029,
      "grad_norm": 1.4292429714224901,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 20029
    },
    {
      "epoch": 0.2003,
      "grad_norm": 1.332420901476742,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 20030
    },
    {
      "epoch": 0.20031,
      "grad_norm": 1.2147558905404297,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 20031
    },
    {
      "epoch": 0.20032,
      "grad_norm": 1.261006618127135,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 20032
    },
    {
      "epoch": 0.20033,
      "grad_norm": 1.3681351100810146,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 20033
    },
    {
      "epoch": 0.20034,
      "grad_norm": 1.1859421534830465,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 20034
    },
    {
      "epoch": 0.20035,
      "grad_norm": 1.37908980409705,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 20035
    },
    {
      "epoch": 0.20036,
      "grad_norm": 1.1969339621114272,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 20036
    },
    {
      "epoch": 0.20037,
      "grad_norm": 1.4180878572186497,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 20037
    },
    {
      "epoch": 0.20038,
      "grad_norm": 1.2656692876512636,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 20038
    },
    {
      "epoch": 0.20039,
      "grad_norm": 1.8198151291611133,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 20039
    },
    {
      "epoch": 0.2004,
      "grad_norm": 1.0161078553895786,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 20040
    },
    {
      "epoch": 0.20041,
      "grad_norm": 1.355432629523141,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 20041
    },
    {
      "epoch": 0.20042,
      "grad_norm": 1.383629325985083,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 20042
    },
    {
      "epoch": 0.20043,
      "grad_norm": 1.2228534181389905,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 20043
    },
    {
      "epoch": 0.20044,
      "grad_norm": 1.7296168020885307,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 20044
    },
    {
      "epoch": 0.20045,
      "grad_norm": 1.3540347177465948,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 20045
    },
    {
      "epoch": 0.20046,
      "grad_norm": 1.3239312908186236,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 20046
    },
    {
      "epoch": 0.20047,
      "grad_norm": 1.2775701830880886,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 20047
    },
    {
      "epoch": 0.20048,
      "grad_norm": 1.3203763932822852,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 20048
    },
    {
      "epoch": 0.20049,
      "grad_norm": 1.1279264776919082,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 20049
    },
    {
      "epoch": 0.2005,
      "grad_norm": 1.3189293017299164,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 20050
    },
    {
      "epoch": 0.20051,
      "grad_norm": 0.9977932116097673,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 20051
    },
    {
      "epoch": 0.20052,
      "grad_norm": 1.571241154819268,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 20052
    },
    {
      "epoch": 0.20053,
      "grad_norm": 1.1798043208059663,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 20053
    },
    {
      "epoch": 0.20054,
      "grad_norm": 1.4574743878754446,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 20054
    },
    {
      "epoch": 0.20055,
      "grad_norm": 1.4412622360713538,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 20055
    },
    {
      "epoch": 0.20056,
      "grad_norm": 1.4592961399760058,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 20056
    },
    {
      "epoch": 0.20057,
      "grad_norm": 1.3012474854060578,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 20057
    },
    {
      "epoch": 0.20058,
      "grad_norm": 1.5786161375868784,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 20058
    },
    {
      "epoch": 0.20059,
      "grad_norm": 1.1360821546140618,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 20059
    },
    {
      "epoch": 0.2006,
      "grad_norm": 1.5675781861812292,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 20060
    },
    {
      "epoch": 0.20061,
      "grad_norm": 1.0718330140613528,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 20061
    },
    {
      "epoch": 0.20062,
      "grad_norm": 1.5152826138030726,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 20062
    },
    {
      "epoch": 0.20063,
      "grad_norm": 1.2354227782514513,
      "learning_rate": 0.003,
      "loss": 4.0963,
      "step": 20063
    },
    {
      "epoch": 0.20064,
      "grad_norm": 1.3823679920207135,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 20064
    },
    {
      "epoch": 0.20065,
      "grad_norm": 1.4637266564538518,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 20065
    },
    {
      "epoch": 0.20066,
      "grad_norm": 1.1454759865300959,
      "learning_rate": 0.003,
      "loss": 4.0235,
      "step": 20066
    },
    {
      "epoch": 0.20067,
      "grad_norm": 1.6057158909528744,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 20067
    },
    {
      "epoch": 0.20068,
      "grad_norm": 1.168781407246637,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 20068
    },
    {
      "epoch": 0.20069,
      "grad_norm": 1.591690861234611,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 20069
    },
    {
      "epoch": 0.2007,
      "grad_norm": 1.4144978587605654,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 20070
    },
    {
      "epoch": 0.20071,
      "grad_norm": 1.1947853938283506,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 20071
    },
    {
      "epoch": 0.20072,
      "grad_norm": 1.5588563229128116,
      "learning_rate": 0.003,
      "loss": 4.0211,
      "step": 20072
    },
    {
      "epoch": 0.20073,
      "grad_norm": 1.1847033231436441,
      "learning_rate": 0.003,
      "loss": 4.0217,
      "step": 20073
    },
    {
      "epoch": 0.20074,
      "grad_norm": 1.513444662930399,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 20074
    },
    {
      "epoch": 0.20075,
      "grad_norm": 1.10270547997938,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 20075
    },
    {
      "epoch": 0.20076,
      "grad_norm": 1.3780518119497263,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 20076
    },
    {
      "epoch": 0.20077,
      "grad_norm": 1.1144511408052817,
      "learning_rate": 0.003,
      "loss": 4.028,
      "step": 20077
    },
    {
      "epoch": 0.20078,
      "grad_norm": 1.5146472285977763,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 20078
    },
    {
      "epoch": 0.20079,
      "grad_norm": 1.4284349083888075,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 20079
    },
    {
      "epoch": 0.2008,
      "grad_norm": 1.2989598073195823,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 20080
    },
    {
      "epoch": 0.20081,
      "grad_norm": 1.187298984538773,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 20081
    },
    {
      "epoch": 0.20082,
      "grad_norm": 1.5113379046885078,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 20082
    },
    {
      "epoch": 0.20083,
      "grad_norm": 1.1541632019018093,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20083
    },
    {
      "epoch": 0.20084,
      "grad_norm": 1.3668450870562816,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 20084
    },
    {
      "epoch": 0.20085,
      "grad_norm": 1.4203238464339045,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 20085
    },
    {
      "epoch": 0.20086,
      "grad_norm": 1.5895090610935805,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 20086
    },
    {
      "epoch": 0.20087,
      "grad_norm": 1.095686749121987,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 20087
    },
    {
      "epoch": 0.20088,
      "grad_norm": 1.5489389767527166,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 20088
    },
    {
      "epoch": 0.20089,
      "grad_norm": 1.1630214840475503,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 20089
    },
    {
      "epoch": 0.2009,
      "grad_norm": 1.3390333966051802,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 20090
    },
    {
      "epoch": 0.20091,
      "grad_norm": 1.2383088700061902,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 20091
    },
    {
      "epoch": 0.20092,
      "grad_norm": 1.4327085473639651,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 20092
    },
    {
      "epoch": 0.20093,
      "grad_norm": 1.1928407656732276,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 20093
    },
    {
      "epoch": 0.20094,
      "grad_norm": 1.3324033273997509,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 20094
    },
    {
      "epoch": 0.20095,
      "grad_norm": 1.1639919920800843,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 20095
    },
    {
      "epoch": 0.20096,
      "grad_norm": 1.5326473703914898,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 20096
    },
    {
      "epoch": 0.20097,
      "grad_norm": 1.469273855798529,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 20097
    },
    {
      "epoch": 0.20098,
      "grad_norm": 1.4723035892612006,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 20098
    },
    {
      "epoch": 0.20099,
      "grad_norm": 1.233952058675891,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 20099
    },
    {
      "epoch": 0.201,
      "grad_norm": 1.3370880960112455,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 20100
    },
    {
      "epoch": 0.20101,
      "grad_norm": 1.3418940115999278,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 20101
    },
    {
      "epoch": 0.20102,
      "grad_norm": 1.3806154919944444,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 20102
    },
    {
      "epoch": 0.20103,
      "grad_norm": 1.280901208133101,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 20103
    },
    {
      "epoch": 0.20104,
      "grad_norm": 1.1118441129872672,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 20104
    },
    {
      "epoch": 0.20105,
      "grad_norm": 1.3672865430967989,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 20105
    },
    {
      "epoch": 0.20106,
      "grad_norm": 1.262152710425421,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 20106
    },
    {
      "epoch": 0.20107,
      "grad_norm": 1.358097389225668,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 20107
    },
    {
      "epoch": 0.20108,
      "grad_norm": 1.15711306904865,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 20108
    },
    {
      "epoch": 0.20109,
      "grad_norm": 1.3936933683919779,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 20109
    },
    {
      "epoch": 0.2011,
      "grad_norm": 1.2456943349460141,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 20110
    },
    {
      "epoch": 0.20111,
      "grad_norm": 1.4010785359454438,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 20111
    },
    {
      "epoch": 0.20112,
      "grad_norm": 1.2792684280319735,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 20112
    },
    {
      "epoch": 0.20113,
      "grad_norm": 1.1347136211110727,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 20113
    },
    {
      "epoch": 0.20114,
      "grad_norm": 1.423287069113192,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 20114
    },
    {
      "epoch": 0.20115,
      "grad_norm": 1.1749771510898996,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 20115
    },
    {
      "epoch": 0.20116,
      "grad_norm": 1.6274457970602212,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 20116
    },
    {
      "epoch": 0.20117,
      "grad_norm": 1.0200085761280606,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 20117
    },
    {
      "epoch": 0.20118,
      "grad_norm": 2.1369127590205155,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 20118
    },
    {
      "epoch": 0.20119,
      "grad_norm": 1.1187562929696346,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 20119
    },
    {
      "epoch": 0.2012,
      "grad_norm": 1.51805607519156,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 20120
    },
    {
      "epoch": 0.20121,
      "grad_norm": 1.6198867996201227,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 20121
    },
    {
      "epoch": 0.20122,
      "grad_norm": 1.4554496551998943,
      "learning_rate": 0.003,
      "loss": 4.107,
      "step": 20122
    },
    {
      "epoch": 0.20123,
      "grad_norm": 1.2883640374402157,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 20123
    },
    {
      "epoch": 0.20124,
      "grad_norm": 1.620054827451497,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 20124
    },
    {
      "epoch": 0.20125,
      "grad_norm": 1.2614383445274928,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 20125
    },
    {
      "epoch": 0.20126,
      "grad_norm": 1.3521223171358985,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 20126
    },
    {
      "epoch": 0.20127,
      "grad_norm": 1.4850145105267474,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 20127
    },
    {
      "epoch": 0.20128,
      "grad_norm": 1.234214821383813,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 20128
    },
    {
      "epoch": 0.20129,
      "grad_norm": 1.29105799526028,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 20129
    },
    {
      "epoch": 0.2013,
      "grad_norm": 1.224626393915006,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 20130
    },
    {
      "epoch": 0.20131,
      "grad_norm": 1.3799965897460933,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 20131
    },
    {
      "epoch": 0.20132,
      "grad_norm": 1.2221226479357916,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 20132
    },
    {
      "epoch": 0.20133,
      "grad_norm": 1.3700604460052845,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 20133
    },
    {
      "epoch": 0.20134,
      "grad_norm": 1.3710883696203422,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 20134
    },
    {
      "epoch": 0.20135,
      "grad_norm": 1.5385608161996616,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 20135
    },
    {
      "epoch": 0.20136,
      "grad_norm": 1.3333863637238177,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 20136
    },
    {
      "epoch": 0.20137,
      "grad_norm": 1.360397105862845,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 20137
    },
    {
      "epoch": 0.20138,
      "grad_norm": 1.3658203998710363,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 20138
    },
    {
      "epoch": 0.20139,
      "grad_norm": 1.5664289313768378,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 20139
    },
    {
      "epoch": 0.2014,
      "grad_norm": 1.0017139017476433,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 20140
    },
    {
      "epoch": 0.20141,
      "grad_norm": 1.4446306172886767,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 20141
    },
    {
      "epoch": 0.20142,
      "grad_norm": 1.0921082499045707,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 20142
    },
    {
      "epoch": 0.20143,
      "grad_norm": 1.4709145539263713,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 20143
    },
    {
      "epoch": 0.20144,
      "grad_norm": 0.9861328787696235,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 20144
    },
    {
      "epoch": 0.20145,
      "grad_norm": 1.7704188357606225,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 20145
    },
    {
      "epoch": 0.20146,
      "grad_norm": 1.3164527385060978,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 20146
    },
    {
      "epoch": 0.20147,
      "grad_norm": 1.2712983978275954,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 20147
    },
    {
      "epoch": 0.20148,
      "grad_norm": 1.337966940599898,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 20148
    },
    {
      "epoch": 0.20149,
      "grad_norm": 1.289917031299455,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 20149
    },
    {
      "epoch": 0.2015,
      "grad_norm": 1.3383005726772683,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 20150
    },
    {
      "epoch": 0.20151,
      "grad_norm": 1.419122846015543,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 20151
    },
    {
      "epoch": 0.20152,
      "grad_norm": 1.1713851627259142,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 20152
    },
    {
      "epoch": 0.20153,
      "grad_norm": 1.4930766676175529,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 20153
    },
    {
      "epoch": 0.20154,
      "grad_norm": 1.151981171269532,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 20154
    },
    {
      "epoch": 0.20155,
      "grad_norm": 1.3762473501599417,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 20155
    },
    {
      "epoch": 0.20156,
      "grad_norm": 1.3151396278511758,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 20156
    },
    {
      "epoch": 0.20157,
      "grad_norm": 1.4038548922797431,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 20157
    },
    {
      "epoch": 0.20158,
      "grad_norm": 1.11717849011769,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 20158
    },
    {
      "epoch": 0.20159,
      "grad_norm": 1.1990235231336113,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 20159
    },
    {
      "epoch": 0.2016,
      "grad_norm": 1.2462564898723882,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 20160
    },
    {
      "epoch": 0.20161,
      "grad_norm": 1.4524608490551825,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 20161
    },
    {
      "epoch": 0.20162,
      "grad_norm": 1.2838264373462425,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 20162
    },
    {
      "epoch": 0.20163,
      "grad_norm": 1.3131408545808823,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 20163
    },
    {
      "epoch": 0.20164,
      "grad_norm": 1.3843159691175067,
      "learning_rate": 0.003,
      "loss": 4.0996,
      "step": 20164
    },
    {
      "epoch": 0.20165,
      "grad_norm": 1.5100106083950458,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 20165
    },
    {
      "epoch": 0.20166,
      "grad_norm": 1.419607775957611,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 20166
    },
    {
      "epoch": 0.20167,
      "grad_norm": 1.4835777811373496,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 20167
    },
    {
      "epoch": 0.20168,
      "grad_norm": 1.3865149314656044,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 20168
    },
    {
      "epoch": 0.20169,
      "grad_norm": 1.1829050529432203,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 20169
    },
    {
      "epoch": 0.2017,
      "grad_norm": 1.528905891006897,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 20170
    },
    {
      "epoch": 0.20171,
      "grad_norm": 1.06108764356655,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 20171
    },
    {
      "epoch": 0.20172,
      "grad_norm": 1.5104985676085916,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 20172
    },
    {
      "epoch": 0.20173,
      "grad_norm": 1.2729083111073791,
      "learning_rate": 0.003,
      "loss": 4.0162,
      "step": 20173
    },
    {
      "epoch": 0.20174,
      "grad_norm": 1.3503020296505024,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 20174
    },
    {
      "epoch": 0.20175,
      "grad_norm": 1.4827398120112962,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 20175
    },
    {
      "epoch": 0.20176,
      "grad_norm": 1.2927001966130707,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 20176
    },
    {
      "epoch": 0.20177,
      "grad_norm": 1.4896565621946782,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 20177
    },
    {
      "epoch": 0.20178,
      "grad_norm": 0.9985785220810762,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 20178
    },
    {
      "epoch": 0.20179,
      "grad_norm": 1.4515332738871436,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 20179
    },
    {
      "epoch": 0.2018,
      "grad_norm": 1.1484456461245491,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 20180
    },
    {
      "epoch": 0.20181,
      "grad_norm": 1.543886314020279,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 20181
    },
    {
      "epoch": 0.20182,
      "grad_norm": 1.1660738695265007,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 20182
    },
    {
      "epoch": 0.20183,
      "grad_norm": 1.4055135966478245,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 20183
    },
    {
      "epoch": 0.20184,
      "grad_norm": 1.5541115978355016,
      "learning_rate": 0.003,
      "loss": 4.1015,
      "step": 20184
    },
    {
      "epoch": 0.20185,
      "grad_norm": 1.2421326626388232,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 20185
    },
    {
      "epoch": 0.20186,
      "grad_norm": 1.5524946004265114,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 20186
    },
    {
      "epoch": 0.20187,
      "grad_norm": 1.1019332919822227,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 20187
    },
    {
      "epoch": 0.20188,
      "grad_norm": 1.3527494610479813,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 20188
    },
    {
      "epoch": 0.20189,
      "grad_norm": 1.2439827387107196,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 20189
    },
    {
      "epoch": 0.2019,
      "grad_norm": 1.402906416601852,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 20190
    },
    {
      "epoch": 0.20191,
      "grad_norm": 1.336442867221924,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 20191
    },
    {
      "epoch": 0.20192,
      "grad_norm": 1.1787087646056902,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 20192
    },
    {
      "epoch": 0.20193,
      "grad_norm": 1.5573011368805436,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 20193
    },
    {
      "epoch": 0.20194,
      "grad_norm": 1.3902073137712168,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 20194
    },
    {
      "epoch": 0.20195,
      "grad_norm": 1.4901482909029486,
      "learning_rate": 0.003,
      "loss": 4.1158,
      "step": 20195
    },
    {
      "epoch": 0.20196,
      "grad_norm": 1.3014645418256428,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 20196
    },
    {
      "epoch": 0.20197,
      "grad_norm": 1.3414131733426882,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 20197
    },
    {
      "epoch": 0.20198,
      "grad_norm": 1.1407214861367891,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 20198
    },
    {
      "epoch": 0.20199,
      "grad_norm": 1.340005122470549,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 20199
    },
    {
      "epoch": 0.202,
      "grad_norm": 1.2230698647697285,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 20200
    },
    {
      "epoch": 0.20201,
      "grad_norm": 1.3501677002155696,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 20201
    },
    {
      "epoch": 0.20202,
      "grad_norm": 1.1698990019155124,
      "learning_rate": 0.003,
      "loss": 4.0964,
      "step": 20202
    },
    {
      "epoch": 0.20203,
      "grad_norm": 1.5175614202195813,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 20203
    },
    {
      "epoch": 0.20204,
      "grad_norm": 1.2589719409047873,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 20204
    },
    {
      "epoch": 0.20205,
      "grad_norm": 1.3787637881255508,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 20205
    },
    {
      "epoch": 0.20206,
      "grad_norm": 1.5965370500764264,
      "learning_rate": 0.003,
      "loss": 4.0149,
      "step": 20206
    },
    {
      "epoch": 0.20207,
      "grad_norm": 1.2845703200278211,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 20207
    },
    {
      "epoch": 0.20208,
      "grad_norm": 1.397008593604821,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 20208
    },
    {
      "epoch": 0.20209,
      "grad_norm": 1.2114512364999055,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 20209
    },
    {
      "epoch": 0.2021,
      "grad_norm": 1.5700928033091597,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 20210
    },
    {
      "epoch": 0.20211,
      "grad_norm": 1.0960466247124028,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 20211
    },
    {
      "epoch": 0.20212,
      "grad_norm": 1.5013746356631381,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 20212
    },
    {
      "epoch": 0.20213,
      "grad_norm": 1.181832161284184,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 20213
    },
    {
      "epoch": 0.20214,
      "grad_norm": 1.25126739260201,
      "learning_rate": 0.003,
      "loss": 4.0154,
      "step": 20214
    },
    {
      "epoch": 0.20215,
      "grad_norm": 1.4633945227423286,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 20215
    },
    {
      "epoch": 0.20216,
      "grad_norm": 1.4266661902588709,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 20216
    },
    {
      "epoch": 0.20217,
      "grad_norm": 1.2377326654245766,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 20217
    },
    {
      "epoch": 0.20218,
      "grad_norm": 1.2758062982531768,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 20218
    },
    {
      "epoch": 0.20219,
      "grad_norm": 1.2925038112892158,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 20219
    },
    {
      "epoch": 0.2022,
      "grad_norm": 1.353990795336194,
      "learning_rate": 0.003,
      "loss": 4.0257,
      "step": 20220
    },
    {
      "epoch": 0.20221,
      "grad_norm": 1.5681784043895879,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 20221
    },
    {
      "epoch": 0.20222,
      "grad_norm": 1.4594748170477143,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 20222
    },
    {
      "epoch": 0.20223,
      "grad_norm": 1.1328992001472526,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 20223
    },
    {
      "epoch": 0.20224,
      "grad_norm": 1.6070431944351944,
      "learning_rate": 0.003,
      "loss": 4.0939,
      "step": 20224
    },
    {
      "epoch": 0.20225,
      "grad_norm": 1.3345256871383968,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 20225
    },
    {
      "epoch": 0.20226,
      "grad_norm": 1.2039936279757504,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 20226
    },
    {
      "epoch": 0.20227,
      "grad_norm": 1.4112783524690296,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 20227
    },
    {
      "epoch": 0.20228,
      "grad_norm": 1.0879874829746985,
      "learning_rate": 0.003,
      "loss": 4.0299,
      "step": 20228
    },
    {
      "epoch": 0.20229,
      "grad_norm": 1.406909773014083,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 20229
    },
    {
      "epoch": 0.2023,
      "grad_norm": 1.5073298460839546,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 20230
    },
    {
      "epoch": 0.20231,
      "grad_norm": 1.4725979263205242,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 20231
    },
    {
      "epoch": 0.20232,
      "grad_norm": 1.3097844389355746,
      "learning_rate": 0.003,
      "loss": 4.0206,
      "step": 20232
    },
    {
      "epoch": 0.20233,
      "grad_norm": 1.1974186108290308,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 20233
    },
    {
      "epoch": 0.20234,
      "grad_norm": 1.4660146901812041,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 20234
    },
    {
      "epoch": 0.20235,
      "grad_norm": 1.0145110696085622,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 20235
    },
    {
      "epoch": 0.20236,
      "grad_norm": 1.4059391654413957,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 20236
    },
    {
      "epoch": 0.20237,
      "grad_norm": 1.1952217406063508,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 20237
    },
    {
      "epoch": 0.20238,
      "grad_norm": 1.6725738519887603,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 20238
    },
    {
      "epoch": 0.20239,
      "grad_norm": 1.1162383117767734,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 20239
    },
    {
      "epoch": 0.2024,
      "grad_norm": 1.6663407668463703,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 20240
    },
    {
      "epoch": 0.20241,
      "grad_norm": 1.461105874407114,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 20241
    },
    {
      "epoch": 0.20242,
      "grad_norm": 1.7513584676973888,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 20242
    },
    {
      "epoch": 0.20243,
      "grad_norm": 1.0803860595038803,
      "learning_rate": 0.003,
      "loss": 4.0212,
      "step": 20243
    },
    {
      "epoch": 0.20244,
      "grad_norm": 1.3010710656545241,
      "learning_rate": 0.003,
      "loss": 4.0853,
      "step": 20244
    },
    {
      "epoch": 0.20245,
      "grad_norm": 1.3533895832189775,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 20245
    },
    {
      "epoch": 0.20246,
      "grad_norm": 1.4133369311925101,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 20246
    },
    {
      "epoch": 0.20247,
      "grad_norm": 1.2584835076219352,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 20247
    },
    {
      "epoch": 0.20248,
      "grad_norm": 1.4737037157476898,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 20248
    },
    {
      "epoch": 0.20249,
      "grad_norm": 1.1550526835609105,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 20249
    },
    {
      "epoch": 0.2025,
      "grad_norm": 1.3141195332757687,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 20250
    },
    {
      "epoch": 0.20251,
      "grad_norm": 1.2685653761651985,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 20251
    },
    {
      "epoch": 0.20252,
      "grad_norm": 1.3264547028903149,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 20252
    },
    {
      "epoch": 0.20253,
      "grad_norm": 1.217633255514227,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 20253
    },
    {
      "epoch": 0.20254,
      "grad_norm": 1.25743030026997,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 20254
    },
    {
      "epoch": 0.20255,
      "grad_norm": 1.4676744069862193,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 20255
    },
    {
      "epoch": 0.20256,
      "grad_norm": 1.2844555774592705,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 20256
    },
    {
      "epoch": 0.20257,
      "grad_norm": 1.2091318047712682,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 20257
    },
    {
      "epoch": 0.20258,
      "grad_norm": 1.2943925744024822,
      "learning_rate": 0.003,
      "loss": 4.0228,
      "step": 20258
    },
    {
      "epoch": 0.20259,
      "grad_norm": 1.7294020948011957,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 20259
    },
    {
      "epoch": 0.2026,
      "grad_norm": 1.6668637325189721,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 20260
    },
    {
      "epoch": 0.20261,
      "grad_norm": 1.185731719878189,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 20261
    },
    {
      "epoch": 0.20262,
      "grad_norm": 1.4799478765469893,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 20262
    },
    {
      "epoch": 0.20263,
      "grad_norm": 1.5977095992613186,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 20263
    },
    {
      "epoch": 0.20264,
      "grad_norm": 1.2166178118311102,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 20264
    },
    {
      "epoch": 0.20265,
      "grad_norm": 1.1564296247113104,
      "learning_rate": 0.003,
      "loss": 4.0252,
      "step": 20265
    },
    {
      "epoch": 0.20266,
      "grad_norm": 1.7171201542819816,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 20266
    },
    {
      "epoch": 0.20267,
      "grad_norm": 1.0281637559809484,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 20267
    },
    {
      "epoch": 0.20268,
      "grad_norm": 1.3917162972146064,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 20268
    },
    {
      "epoch": 0.20269,
      "grad_norm": 1.160640192991406,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 20269
    },
    {
      "epoch": 0.2027,
      "grad_norm": 1.4939895821106073,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 20270
    },
    {
      "epoch": 0.20271,
      "grad_norm": 1.1561605112219357,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 20271
    },
    {
      "epoch": 0.20272,
      "grad_norm": 1.308674679046234,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 20272
    },
    {
      "epoch": 0.20273,
      "grad_norm": 1.1854461944313113,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 20273
    },
    {
      "epoch": 0.20274,
      "grad_norm": 1.5477263071366116,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 20274
    },
    {
      "epoch": 0.20275,
      "grad_norm": 1.2058426257177344,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 20275
    },
    {
      "epoch": 0.20276,
      "grad_norm": 1.3685886984542353,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 20276
    },
    {
      "epoch": 0.20277,
      "grad_norm": 1.553754341867869,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 20277
    },
    {
      "epoch": 0.20278,
      "grad_norm": 1.2709523206304474,
      "learning_rate": 0.003,
      "loss": 4.0289,
      "step": 20278
    },
    {
      "epoch": 0.20279,
      "grad_norm": 1.3725987808318894,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 20279
    },
    {
      "epoch": 0.2028,
      "grad_norm": 1.2624941388485218,
      "learning_rate": 0.003,
      "loss": 4.0172,
      "step": 20280
    },
    {
      "epoch": 0.20281,
      "grad_norm": 1.233807790705302,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 20281
    },
    {
      "epoch": 0.20282,
      "grad_norm": 1.2521895323654215,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 20282
    },
    {
      "epoch": 0.20283,
      "grad_norm": 1.4753074404834448,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 20283
    },
    {
      "epoch": 0.20284,
      "grad_norm": 1.4409322853168969,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 20284
    },
    {
      "epoch": 0.20285,
      "grad_norm": 1.504716079962182,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 20285
    },
    {
      "epoch": 0.20286,
      "grad_norm": 1.2126978833429751,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 20286
    },
    {
      "epoch": 0.20287,
      "grad_norm": 1.3542455534211755,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 20287
    },
    {
      "epoch": 0.20288,
      "grad_norm": 1.076493558839879,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 20288
    },
    {
      "epoch": 0.20289,
      "grad_norm": 1.4439817719258055,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 20289
    },
    {
      "epoch": 0.2029,
      "grad_norm": 1.303195151449773,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 20290
    },
    {
      "epoch": 0.20291,
      "grad_norm": 1.4547363156693787,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 20291
    },
    {
      "epoch": 0.20292,
      "grad_norm": 1.3927461978685791,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 20292
    },
    {
      "epoch": 0.20293,
      "grad_norm": 1.2390192112833431,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 20293
    },
    {
      "epoch": 0.20294,
      "grad_norm": 1.3637822144285798,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 20294
    },
    {
      "epoch": 0.20295,
      "grad_norm": 1.242130596796715,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 20295
    },
    {
      "epoch": 0.20296,
      "grad_norm": 1.3901310772605913,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 20296
    },
    {
      "epoch": 0.20297,
      "grad_norm": 1.4037666856257323,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 20297
    },
    {
      "epoch": 0.20298,
      "grad_norm": 1.1758694063275068,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 20298
    },
    {
      "epoch": 0.20299,
      "grad_norm": 1.3623746577047566,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 20299
    },
    {
      "epoch": 0.203,
      "grad_norm": 1.4085414522927322,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 20300
    },
    {
      "epoch": 0.20301,
      "grad_norm": 1.5396625511929816,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 20301
    },
    {
      "epoch": 0.20302,
      "grad_norm": 1.266867217134575,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 20302
    },
    {
      "epoch": 0.20303,
      "grad_norm": 1.5175617191016293,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 20303
    },
    {
      "epoch": 0.20304,
      "grad_norm": 1.1821512523591957,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 20304
    },
    {
      "epoch": 0.20305,
      "grad_norm": 1.4226176539591662,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 20305
    },
    {
      "epoch": 0.20306,
      "grad_norm": 1.050703674193398,
      "learning_rate": 0.003,
      "loss": 4.0308,
      "step": 20306
    },
    {
      "epoch": 0.20307,
      "grad_norm": 1.521939154776105,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 20307
    },
    {
      "epoch": 0.20308,
      "grad_norm": 1.33044787968948,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 20308
    },
    {
      "epoch": 0.20309,
      "grad_norm": 1.4484830287900583,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 20309
    },
    {
      "epoch": 0.2031,
      "grad_norm": 1.5269447970098096,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 20310
    },
    {
      "epoch": 0.20311,
      "grad_norm": 1.4328106234116684,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 20311
    },
    {
      "epoch": 0.20312,
      "grad_norm": 1.178726146404056,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 20312
    },
    {
      "epoch": 0.20313,
      "grad_norm": 1.5048542215378433,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 20313
    },
    {
      "epoch": 0.20314,
      "grad_norm": 1.1369721257827141,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 20314
    },
    {
      "epoch": 0.20315,
      "grad_norm": 1.5531515177048691,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 20315
    },
    {
      "epoch": 0.20316,
      "grad_norm": 1.2364073011949877,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 20316
    },
    {
      "epoch": 0.20317,
      "grad_norm": 1.5190111676246623,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 20317
    },
    {
      "epoch": 0.20318,
      "grad_norm": 1.281037111926793,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 20318
    },
    {
      "epoch": 0.20319,
      "grad_norm": 1.2859070666158015,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 20319
    },
    {
      "epoch": 0.2032,
      "grad_norm": 1.092940737832305,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 20320
    },
    {
      "epoch": 0.20321,
      "grad_norm": 1.3967347723396333,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 20321
    },
    {
      "epoch": 0.20322,
      "grad_norm": 1.2313847457891551,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 20322
    },
    {
      "epoch": 0.20323,
      "grad_norm": 1.574469621303561,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 20323
    },
    {
      "epoch": 0.20324,
      "grad_norm": 1.2035069311206108,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 20324
    },
    {
      "epoch": 0.20325,
      "grad_norm": 1.5439161847441938,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 20325
    },
    {
      "epoch": 0.20326,
      "grad_norm": 1.0239332953056914,
      "learning_rate": 0.003,
      "loss": 4.0103,
      "step": 20326
    },
    {
      "epoch": 0.20327,
      "grad_norm": 1.5932857943493817,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 20327
    },
    {
      "epoch": 0.20328,
      "grad_norm": 1.1140383020616065,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 20328
    },
    {
      "epoch": 0.20329,
      "grad_norm": 1.1682440481298149,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 20329
    },
    {
      "epoch": 0.2033,
      "grad_norm": 1.5287704639769066,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 20330
    },
    {
      "epoch": 0.20331,
      "grad_norm": 1.2779345322703561,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 20331
    },
    {
      "epoch": 0.20332,
      "grad_norm": 1.3815575764425685,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 20332
    },
    {
      "epoch": 0.20333,
      "grad_norm": 1.3061960024303134,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 20333
    },
    {
      "epoch": 0.20334,
      "grad_norm": 1.3891872739275988,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 20334
    },
    {
      "epoch": 0.20335,
      "grad_norm": 1.408257164633742,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 20335
    },
    {
      "epoch": 0.20336,
      "grad_norm": 1.3995418314161383,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 20336
    },
    {
      "epoch": 0.20337,
      "grad_norm": 1.3395419724950246,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 20337
    },
    {
      "epoch": 0.20338,
      "grad_norm": 1.2468307107462475,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 20338
    },
    {
      "epoch": 0.20339,
      "grad_norm": 1.4289246140290377,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 20339
    },
    {
      "epoch": 0.2034,
      "grad_norm": 1.2476588970388962,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 20340
    },
    {
      "epoch": 0.20341,
      "grad_norm": 1.3422524370970599,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 20341
    },
    {
      "epoch": 0.20342,
      "grad_norm": 1.0577462129023343,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 20342
    },
    {
      "epoch": 0.20343,
      "grad_norm": 1.7264542407238392,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 20343
    },
    {
      "epoch": 0.20344,
      "grad_norm": 1.2643432223711861,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 20344
    },
    {
      "epoch": 0.20345,
      "grad_norm": 1.529773062601106,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 20345
    },
    {
      "epoch": 0.20346,
      "grad_norm": 1.4361889453096046,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 20346
    },
    {
      "epoch": 0.20347,
      "grad_norm": 1.499920858801668,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 20347
    },
    {
      "epoch": 0.20348,
      "grad_norm": 1.6738367304447712,
      "learning_rate": 0.003,
      "loss": 4.0847,
      "step": 20348
    },
    {
      "epoch": 0.20349,
      "grad_norm": 1.200930809138132,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 20349
    },
    {
      "epoch": 0.2035,
      "grad_norm": 1.467013383787434,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 20350
    },
    {
      "epoch": 0.20351,
      "grad_norm": 1.5679824072008346,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 20351
    },
    {
      "epoch": 0.20352,
      "grad_norm": 1.127251107861855,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 20352
    },
    {
      "epoch": 0.20353,
      "grad_norm": 1.629146596022005,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 20353
    },
    {
      "epoch": 0.20354,
      "grad_norm": 1.183836849526088,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 20354
    },
    {
      "epoch": 0.20355,
      "grad_norm": 1.228857414739491,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 20355
    },
    {
      "epoch": 0.20356,
      "grad_norm": 1.3416733588472254,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 20356
    },
    {
      "epoch": 0.20357,
      "grad_norm": 1.3372111342595498,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 20357
    },
    {
      "epoch": 0.20358,
      "grad_norm": 1.2538288044204846,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 20358
    },
    {
      "epoch": 0.20359,
      "grad_norm": 1.3959090932702416,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 20359
    },
    {
      "epoch": 0.2036,
      "grad_norm": 1.3694363348608283,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 20360
    },
    {
      "epoch": 0.20361,
      "grad_norm": 1.5034844585728124,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 20361
    },
    {
      "epoch": 0.20362,
      "grad_norm": 1.4239047926456798,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 20362
    },
    {
      "epoch": 0.20363,
      "grad_norm": 1.2520521246135627,
      "learning_rate": 0.003,
      "loss": 4.0194,
      "step": 20363
    },
    {
      "epoch": 0.20364,
      "grad_norm": 1.2412612514439594,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 20364
    },
    {
      "epoch": 0.20365,
      "grad_norm": 1.3948056814887138,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 20365
    },
    {
      "epoch": 0.20366,
      "grad_norm": 1.2792736518242112,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 20366
    },
    {
      "epoch": 0.20367,
      "grad_norm": 1.1054230687849584,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 20367
    },
    {
      "epoch": 0.20368,
      "grad_norm": 1.3011606662823947,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 20368
    },
    {
      "epoch": 0.20369,
      "grad_norm": 1.2803734189697848,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 20369
    },
    {
      "epoch": 0.2037,
      "grad_norm": 1.3831472596596728,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 20370
    },
    {
      "epoch": 0.20371,
      "grad_norm": 1.4680128368690626,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 20371
    },
    {
      "epoch": 0.20372,
      "grad_norm": 1.364108438076807,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 20372
    },
    {
      "epoch": 0.20373,
      "grad_norm": 1.4398845101211322,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 20373
    },
    {
      "epoch": 0.20374,
      "grad_norm": 1.3016675770978412,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 20374
    },
    {
      "epoch": 0.20375,
      "grad_norm": 1.440576518128247,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 20375
    },
    {
      "epoch": 0.20376,
      "grad_norm": 1.222119920685402,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 20376
    },
    {
      "epoch": 0.20377,
      "grad_norm": 1.515461877623012,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 20377
    },
    {
      "epoch": 0.20378,
      "grad_norm": 1.2135329370512657,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 20378
    },
    {
      "epoch": 0.20379,
      "grad_norm": 1.5897915340651947,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 20379
    },
    {
      "epoch": 0.2038,
      "grad_norm": 1.0645381777011922,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 20380
    },
    {
      "epoch": 0.20381,
      "grad_norm": 1.6386763563805693,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 20381
    },
    {
      "epoch": 0.20382,
      "grad_norm": 1.1424167086168668,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 20382
    },
    {
      "epoch": 0.20383,
      "grad_norm": 1.3281632988062138,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 20383
    },
    {
      "epoch": 0.20384,
      "grad_norm": 1.5934718851990406,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 20384
    },
    {
      "epoch": 0.20385,
      "grad_norm": 1.0341667755535273,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20385
    },
    {
      "epoch": 0.20386,
      "grad_norm": 1.4459015184338975,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 20386
    },
    {
      "epoch": 0.20387,
      "grad_norm": 1.3366501262145274,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 20387
    },
    {
      "epoch": 0.20388,
      "grad_norm": 1.3915743056060734,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 20388
    },
    {
      "epoch": 0.20389,
      "grad_norm": 1.346962033954529,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 20389
    },
    {
      "epoch": 0.2039,
      "grad_norm": 1.1552000293677478,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 20390
    },
    {
      "epoch": 0.20391,
      "grad_norm": 1.5605764802292823,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 20391
    },
    {
      "epoch": 0.20392,
      "grad_norm": 1.1987219339139694,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 20392
    },
    {
      "epoch": 0.20393,
      "grad_norm": 1.2454690242319035,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 20393
    },
    {
      "epoch": 0.20394,
      "grad_norm": 1.2256636987965361,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 20394
    },
    {
      "epoch": 0.20395,
      "grad_norm": 1.2867586718472315,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 20395
    },
    {
      "epoch": 0.20396,
      "grad_norm": 1.1965507261055377,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 20396
    },
    {
      "epoch": 0.20397,
      "grad_norm": 1.4592698624576914,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 20397
    },
    {
      "epoch": 0.20398,
      "grad_norm": 1.5161036939976629,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 20398
    },
    {
      "epoch": 0.20399,
      "grad_norm": 1.2383093577717579,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 20399
    },
    {
      "epoch": 0.204,
      "grad_norm": 1.569877039855125,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 20400
    },
    {
      "epoch": 0.20401,
      "grad_norm": 1.1692225184397718,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 20401
    },
    {
      "epoch": 0.20402,
      "grad_norm": 1.5226657623961934,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 20402
    },
    {
      "epoch": 0.20403,
      "grad_norm": 1.1749212004945278,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 20403
    },
    {
      "epoch": 0.20404,
      "grad_norm": 1.6261828343359845,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 20404
    },
    {
      "epoch": 0.20405,
      "grad_norm": 1.2151944554580743,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 20405
    },
    {
      "epoch": 0.20406,
      "grad_norm": 1.2945642192787308,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 20406
    },
    {
      "epoch": 0.20407,
      "grad_norm": 1.4069555500329456,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 20407
    },
    {
      "epoch": 0.20408,
      "grad_norm": 1.3830004642020512,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 20408
    },
    {
      "epoch": 0.20409,
      "grad_norm": 1.4319706958306666,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 20409
    },
    {
      "epoch": 0.2041,
      "grad_norm": 1.1870627800044244,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 20410
    },
    {
      "epoch": 0.20411,
      "grad_norm": 1.3968933589457344,
      "learning_rate": 0.003,
      "loss": 4.0385,
      "step": 20411
    },
    {
      "epoch": 0.20412,
      "grad_norm": 1.2634991756889127,
      "learning_rate": 0.003,
      "loss": 4.0283,
      "step": 20412
    },
    {
      "epoch": 0.20413,
      "grad_norm": 1.3074650979006743,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 20413
    },
    {
      "epoch": 0.20414,
      "grad_norm": 1.4663802224254836,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 20414
    },
    {
      "epoch": 0.20415,
      "grad_norm": 1.1745989685736145,
      "learning_rate": 0.003,
      "loss": 4.0291,
      "step": 20415
    },
    {
      "epoch": 0.20416,
      "grad_norm": 1.7716874707579042,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 20416
    },
    {
      "epoch": 0.20417,
      "grad_norm": 1.1465897204395348,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 20417
    },
    {
      "epoch": 0.20418,
      "grad_norm": 1.4541160183866086,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 20418
    },
    {
      "epoch": 0.20419,
      "grad_norm": 1.300364414594133,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20419
    },
    {
      "epoch": 0.2042,
      "grad_norm": 1.1566286092158102,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 20420
    },
    {
      "epoch": 0.20421,
      "grad_norm": 1.4019025899636917,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20421
    },
    {
      "epoch": 0.20422,
      "grad_norm": 1.209222857411128,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 20422
    },
    {
      "epoch": 0.20423,
      "grad_norm": 1.3496282306334784,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 20423
    },
    {
      "epoch": 0.20424,
      "grad_norm": 1.2433548633857987,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 20424
    },
    {
      "epoch": 0.20425,
      "grad_norm": 1.3455150458622027,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 20425
    },
    {
      "epoch": 0.20426,
      "grad_norm": 1.1363945182804216,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 20426
    },
    {
      "epoch": 0.20427,
      "grad_norm": 1.447084854366867,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 20427
    },
    {
      "epoch": 0.20428,
      "grad_norm": 1.349916594398312,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 20428
    },
    {
      "epoch": 0.20429,
      "grad_norm": 1.4837193509187336,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 20429
    },
    {
      "epoch": 0.2043,
      "grad_norm": 1.215639767974536,
      "learning_rate": 0.003,
      "loss": 4.0893,
      "step": 20430
    },
    {
      "epoch": 0.20431,
      "grad_norm": 1.8112911372440925,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 20431
    },
    {
      "epoch": 0.20432,
      "grad_norm": 1.3158951404821044,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20432
    },
    {
      "epoch": 0.20433,
      "grad_norm": 1.1216513036257185,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 20433
    },
    {
      "epoch": 0.20434,
      "grad_norm": 1.5943047717634269,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 20434
    },
    {
      "epoch": 0.20435,
      "grad_norm": 1.343181365017431,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 20435
    },
    {
      "epoch": 0.20436,
      "grad_norm": 1.3464601497928512,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 20436
    },
    {
      "epoch": 0.20437,
      "grad_norm": 1.4358280375881376,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 20437
    },
    {
      "epoch": 0.20438,
      "grad_norm": 1.4503941072369733,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 20438
    },
    {
      "epoch": 0.20439,
      "grad_norm": 1.145430636880698,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 20439
    },
    {
      "epoch": 0.2044,
      "grad_norm": 1.4974558111213008,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 20440
    },
    {
      "epoch": 0.20441,
      "grad_norm": 1.3053873478983593,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 20441
    },
    {
      "epoch": 0.20442,
      "grad_norm": 1.3672662373830282,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20442
    },
    {
      "epoch": 0.20443,
      "grad_norm": 1.3817825636432517,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 20443
    },
    {
      "epoch": 0.20444,
      "grad_norm": 1.1952694223765734,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 20444
    },
    {
      "epoch": 0.20445,
      "grad_norm": 1.4394061531641331,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 20445
    },
    {
      "epoch": 0.20446,
      "grad_norm": 1.3333207449315254,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 20446
    },
    {
      "epoch": 0.20447,
      "grad_norm": 1.4515805384860727,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 20447
    },
    {
      "epoch": 0.20448,
      "grad_norm": 1.2508524069591107,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 20448
    },
    {
      "epoch": 0.20449,
      "grad_norm": 1.29362758082749,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 20449
    },
    {
      "epoch": 0.2045,
      "grad_norm": 1.4251175560020428,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 20450
    },
    {
      "epoch": 0.20451,
      "grad_norm": 1.3396049861183335,
      "learning_rate": 0.003,
      "loss": 4.0261,
      "step": 20451
    },
    {
      "epoch": 0.20452,
      "grad_norm": 1.271441360928029,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 20452
    },
    {
      "epoch": 0.20453,
      "grad_norm": 1.208530742162209,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 20453
    },
    {
      "epoch": 0.20454,
      "grad_norm": 1.4584881237128287,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 20454
    },
    {
      "epoch": 0.20455,
      "grad_norm": 1.0092589910193805,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 20455
    },
    {
      "epoch": 0.20456,
      "grad_norm": 1.4809515743739938,
      "learning_rate": 0.003,
      "loss": 4.0269,
      "step": 20456
    },
    {
      "epoch": 0.20457,
      "grad_norm": 1.2503919073180787,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 20457
    },
    {
      "epoch": 0.20458,
      "grad_norm": 1.3325286512530745,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 20458
    },
    {
      "epoch": 0.20459,
      "grad_norm": 1.2101473095462296,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 20459
    },
    {
      "epoch": 0.2046,
      "grad_norm": 1.3251864349288356,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 20460
    },
    {
      "epoch": 0.20461,
      "grad_norm": 1.1389725466140332,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 20461
    },
    {
      "epoch": 0.20462,
      "grad_norm": 1.542600041658139,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 20462
    },
    {
      "epoch": 0.20463,
      "grad_norm": 1.3023259888264964,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 20463
    },
    {
      "epoch": 0.20464,
      "grad_norm": 1.4764643574700729,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 20464
    },
    {
      "epoch": 0.20465,
      "grad_norm": 1.1186142088971804,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 20465
    },
    {
      "epoch": 0.20466,
      "grad_norm": 2.0460897971041487,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 20466
    },
    {
      "epoch": 0.20467,
      "grad_norm": 1.12972253606819,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 20467
    },
    {
      "epoch": 0.20468,
      "grad_norm": 1.5297144853412366,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 20468
    },
    {
      "epoch": 0.20469,
      "grad_norm": 1.0782410690750979,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 20469
    },
    {
      "epoch": 0.2047,
      "grad_norm": 1.5580879093447175,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 20470
    },
    {
      "epoch": 0.20471,
      "grad_norm": 1.0258224498191024,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 20471
    },
    {
      "epoch": 0.20472,
      "grad_norm": 1.5514806475100573,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 20472
    },
    {
      "epoch": 0.20473,
      "grad_norm": 1.1637118520595569,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 20473
    },
    {
      "epoch": 0.20474,
      "grad_norm": 1.8429010158726027,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 20474
    },
    {
      "epoch": 0.20475,
      "grad_norm": 1.428374246002246,
      "learning_rate": 0.003,
      "loss": 4.1041,
      "step": 20475
    },
    {
      "epoch": 0.20476,
      "grad_norm": 1.2831870445809457,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 20476
    },
    {
      "epoch": 0.20477,
      "grad_norm": 1.2062427063284045,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 20477
    },
    {
      "epoch": 0.20478,
      "grad_norm": 1.5416075304100199,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 20478
    },
    {
      "epoch": 0.20479,
      "grad_norm": 1.2029394842621126,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 20479
    },
    {
      "epoch": 0.2048,
      "grad_norm": 1.569242237429814,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 20480
    },
    {
      "epoch": 0.20481,
      "grad_norm": 1.2023060552514346,
      "learning_rate": 0.003,
      "loss": 4.0205,
      "step": 20481
    },
    {
      "epoch": 0.20482,
      "grad_norm": 1.3671008704673404,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 20482
    },
    {
      "epoch": 0.20483,
      "grad_norm": 1.2181550284200213,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 20483
    },
    {
      "epoch": 0.20484,
      "grad_norm": 1.2107098831651844,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 20484
    },
    {
      "epoch": 0.20485,
      "grad_norm": 1.2914939061169899,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 20485
    },
    {
      "epoch": 0.20486,
      "grad_norm": 1.6783530589798497,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 20486
    },
    {
      "epoch": 0.20487,
      "grad_norm": 1.485364830617809,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 20487
    },
    {
      "epoch": 0.20488,
      "grad_norm": 1.4099567115333826,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 20488
    },
    {
      "epoch": 0.20489,
      "grad_norm": 1.495916460389477,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 20489
    },
    {
      "epoch": 0.2049,
      "grad_norm": 1.178450710384942,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 20490
    },
    {
      "epoch": 0.20491,
      "grad_norm": 1.6912628484171894,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 20491
    },
    {
      "epoch": 0.20492,
      "grad_norm": 1.123540991095715,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 20492
    },
    {
      "epoch": 0.20493,
      "grad_norm": 1.6529649747407045,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 20493
    },
    {
      "epoch": 0.20494,
      "grad_norm": 1.1509127886685253,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 20494
    },
    {
      "epoch": 0.20495,
      "grad_norm": 1.4774965511998783,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 20495
    },
    {
      "epoch": 0.20496,
      "grad_norm": 1.6530410144546603,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 20496
    },
    {
      "epoch": 0.20497,
      "grad_norm": 0.9380240352308407,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 20497
    },
    {
      "epoch": 0.20498,
      "grad_norm": 1.3264095358263313,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 20498
    },
    {
      "epoch": 0.20499,
      "grad_norm": 1.4263106225259305,
      "learning_rate": 0.003,
      "loss": 4.0984,
      "step": 20499
    },
    {
      "epoch": 0.205,
      "grad_norm": 1.258423524127491,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 20500
    },
    {
      "epoch": 0.20501,
      "grad_norm": 1.0772749489987943,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 20501
    },
    {
      "epoch": 0.20502,
      "grad_norm": 1.6768436558193698,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 20502
    },
    {
      "epoch": 0.20503,
      "grad_norm": 1.3719749181356573,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 20503
    },
    {
      "epoch": 0.20504,
      "grad_norm": 1.3076034976415154,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 20504
    },
    {
      "epoch": 0.20505,
      "grad_norm": 1.503886625040588,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 20505
    },
    {
      "epoch": 0.20506,
      "grad_norm": 1.4609689972531943,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 20506
    },
    {
      "epoch": 0.20507,
      "grad_norm": 1.3549616497833068,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 20507
    },
    {
      "epoch": 0.20508,
      "grad_norm": 1.4939887798151368,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 20508
    },
    {
      "epoch": 0.20509,
      "grad_norm": 1.4217893187549624,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 20509
    },
    {
      "epoch": 0.2051,
      "grad_norm": 1.1819692988912474,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 20510
    },
    {
      "epoch": 0.20511,
      "grad_norm": 1.6665703317991702,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 20511
    },
    {
      "epoch": 0.20512,
      "grad_norm": 1.1581066444434676,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 20512
    },
    {
      "epoch": 0.20513,
      "grad_norm": 1.347662207439549,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 20513
    },
    {
      "epoch": 0.20514,
      "grad_norm": 1.3587040372607169,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 20514
    },
    {
      "epoch": 0.20515,
      "grad_norm": 1.3475403635569718,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 20515
    },
    {
      "epoch": 0.20516,
      "grad_norm": 1.1470955731703492,
      "learning_rate": 0.003,
      "loss": 4.0269,
      "step": 20516
    },
    {
      "epoch": 0.20517,
      "grad_norm": 1.2038765316619247,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 20517
    },
    {
      "epoch": 0.20518,
      "grad_norm": 1.3109237023488451,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 20518
    },
    {
      "epoch": 0.20519,
      "grad_norm": 1.2361177572036106,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 20519
    },
    {
      "epoch": 0.2052,
      "grad_norm": 1.4796222856008112,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 20520
    },
    {
      "epoch": 0.20521,
      "grad_norm": 1.2531185153138025,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 20521
    },
    {
      "epoch": 0.20522,
      "grad_norm": 1.730612273552305,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 20522
    },
    {
      "epoch": 0.20523,
      "grad_norm": 1.2706772945225375,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 20523
    },
    {
      "epoch": 0.20524,
      "grad_norm": 1.1398531348530765,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 20524
    },
    {
      "epoch": 0.20525,
      "grad_norm": 1.3519082464893908,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20525
    },
    {
      "epoch": 0.20526,
      "grad_norm": 1.4760340663790463,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 20526
    },
    {
      "epoch": 0.20527,
      "grad_norm": 1.2208782065493444,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 20527
    },
    {
      "epoch": 0.20528,
      "grad_norm": 1.5059266709517738,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 20528
    },
    {
      "epoch": 0.20529,
      "grad_norm": 1.195595525255474,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 20529
    },
    {
      "epoch": 0.2053,
      "grad_norm": 1.4505639511754715,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 20530
    },
    {
      "epoch": 0.20531,
      "grad_norm": 1.18422918279392,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 20531
    },
    {
      "epoch": 0.20532,
      "grad_norm": 1.4236984485357382,
      "learning_rate": 0.003,
      "loss": 4.0259,
      "step": 20532
    },
    {
      "epoch": 0.20533,
      "grad_norm": 1.8410001515180068,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 20533
    },
    {
      "epoch": 0.20534,
      "grad_norm": 0.8771493260205179,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 20534
    },
    {
      "epoch": 0.20535,
      "grad_norm": 1.3573410739858818,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 20535
    },
    {
      "epoch": 0.20536,
      "grad_norm": 1.3291407846772125,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 20536
    },
    {
      "epoch": 0.20537,
      "grad_norm": 1.2363419084100955,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 20537
    },
    {
      "epoch": 0.20538,
      "grad_norm": 1.2954414879710934,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 20538
    },
    {
      "epoch": 0.20539,
      "grad_norm": 1.244450601937851,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 20539
    },
    {
      "epoch": 0.2054,
      "grad_norm": 1.3704835928868393,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 20540
    },
    {
      "epoch": 0.20541,
      "grad_norm": 1.2790535844257738,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 20541
    },
    {
      "epoch": 0.20542,
      "grad_norm": 1.584307647377643,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 20542
    },
    {
      "epoch": 0.20543,
      "grad_norm": 1.5100470754899271,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 20543
    },
    {
      "epoch": 0.20544,
      "grad_norm": 1.2252937633668737,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 20544
    },
    {
      "epoch": 0.20545,
      "grad_norm": 1.2938591800291355,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 20545
    },
    {
      "epoch": 0.20546,
      "grad_norm": 1.4471883307764137,
      "learning_rate": 0.003,
      "loss": 4.0206,
      "step": 20546
    },
    {
      "epoch": 0.20547,
      "grad_norm": 1.5025869751133516,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 20547
    },
    {
      "epoch": 0.20548,
      "grad_norm": 1.0359233811834558,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 20548
    },
    {
      "epoch": 0.20549,
      "grad_norm": 1.566317033194877,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 20549
    },
    {
      "epoch": 0.2055,
      "grad_norm": 1.2156013760342437,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 20550
    },
    {
      "epoch": 0.20551,
      "grad_norm": 1.5673897332688218,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 20551
    },
    {
      "epoch": 0.20552,
      "grad_norm": 1.333953384074785,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 20552
    },
    {
      "epoch": 0.20553,
      "grad_norm": 1.5399730875589686,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 20553
    },
    {
      "epoch": 0.20554,
      "grad_norm": 1.541981174557002,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 20554
    },
    {
      "epoch": 0.20555,
      "grad_norm": 1.1825785678189993,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 20555
    },
    {
      "epoch": 0.20556,
      "grad_norm": 1.3866878454557428,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 20556
    },
    {
      "epoch": 0.20557,
      "grad_norm": 1.2153274615526608,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 20557
    },
    {
      "epoch": 0.20558,
      "grad_norm": 1.495944994373923,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 20558
    },
    {
      "epoch": 0.20559,
      "grad_norm": 1.1103661783779841,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 20559
    },
    {
      "epoch": 0.2056,
      "grad_norm": 1.408891276822968,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 20560
    },
    {
      "epoch": 0.20561,
      "grad_norm": 1.0712000019832115,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 20561
    },
    {
      "epoch": 0.20562,
      "grad_norm": 1.6283493784005738,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 20562
    },
    {
      "epoch": 0.20563,
      "grad_norm": 1.2292113338108412,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 20563
    },
    {
      "epoch": 0.20564,
      "grad_norm": 1.5663562204334438,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 20564
    },
    {
      "epoch": 0.20565,
      "grad_norm": 1.3567179389875492,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 20565
    },
    {
      "epoch": 0.20566,
      "grad_norm": 1.434969226216492,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 20566
    },
    {
      "epoch": 0.20567,
      "grad_norm": 1.5394148286055918,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 20567
    },
    {
      "epoch": 0.20568,
      "grad_norm": 1.289344470377573,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 20568
    },
    {
      "epoch": 0.20569,
      "grad_norm": 1.3333326599327588,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 20569
    },
    {
      "epoch": 0.2057,
      "grad_norm": 1.3239196660894166,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 20570
    },
    {
      "epoch": 0.20571,
      "grad_norm": 1.1103832018978599,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 20571
    },
    {
      "epoch": 0.20572,
      "grad_norm": 1.5524166451986205,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 20572
    },
    {
      "epoch": 0.20573,
      "grad_norm": 1.0716054149155247,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 20573
    },
    {
      "epoch": 0.20574,
      "grad_norm": 1.4685398356964334,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 20574
    },
    {
      "epoch": 0.20575,
      "grad_norm": 1.1829256757367614,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 20575
    },
    {
      "epoch": 0.20576,
      "grad_norm": 1.5334565864330865,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 20576
    },
    {
      "epoch": 0.20577,
      "grad_norm": 1.5058034805054878,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 20577
    },
    {
      "epoch": 0.20578,
      "grad_norm": 1.3935883252454009,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 20578
    },
    {
      "epoch": 0.20579,
      "grad_norm": 1.4609226679617684,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 20579
    },
    {
      "epoch": 0.2058,
      "grad_norm": 1.359084702387566,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 20580
    },
    {
      "epoch": 0.20581,
      "grad_norm": 0.9425275373109557,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 20581
    },
    {
      "epoch": 0.20582,
      "grad_norm": 1.3324735626357085,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 20582
    },
    {
      "epoch": 0.20583,
      "grad_norm": 1.371511234401514,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 20583
    },
    {
      "epoch": 0.20584,
      "grad_norm": 1.3963853154347534,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 20584
    },
    {
      "epoch": 0.20585,
      "grad_norm": 1.4811417069310304,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 20585
    },
    {
      "epoch": 0.20586,
      "grad_norm": 1.376322539672415,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 20586
    },
    {
      "epoch": 0.20587,
      "grad_norm": 1.038201050690109,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 20587
    },
    {
      "epoch": 0.20588,
      "grad_norm": 1.6625458179834598,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 20588
    },
    {
      "epoch": 0.20589,
      "grad_norm": 1.1475171467965672,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 20589
    },
    {
      "epoch": 0.2059,
      "grad_norm": 1.6370923525247905,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 20590
    },
    {
      "epoch": 0.20591,
      "grad_norm": 1.153261681585598,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 20591
    },
    {
      "epoch": 0.20592,
      "grad_norm": 1.2922002799285761,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 20592
    },
    {
      "epoch": 0.20593,
      "grad_norm": 1.3274509594898747,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 20593
    },
    {
      "epoch": 0.20594,
      "grad_norm": 1.155892313781376,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 20594
    },
    {
      "epoch": 0.20595,
      "grad_norm": 1.3954880176684845,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 20595
    },
    {
      "epoch": 0.20596,
      "grad_norm": 1.2973600396488827,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 20596
    },
    {
      "epoch": 0.20597,
      "grad_norm": 1.5925782701536118,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 20597
    },
    {
      "epoch": 0.20598,
      "grad_norm": 1.4086120310400876,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 20598
    },
    {
      "epoch": 0.20599,
      "grad_norm": 1.4680185440620883,
      "learning_rate": 0.003,
      "loss": 4.1086,
      "step": 20599
    },
    {
      "epoch": 0.206,
      "grad_norm": 1.1989041267118707,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 20600
    },
    {
      "epoch": 0.20601,
      "grad_norm": 1.3284651252197444,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 20601
    },
    {
      "epoch": 0.20602,
      "grad_norm": 1.363074375063444,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 20602
    },
    {
      "epoch": 0.20603,
      "grad_norm": 1.4057815359007353,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 20603
    },
    {
      "epoch": 0.20604,
      "grad_norm": 1.2600489526130207,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 20604
    },
    {
      "epoch": 0.20605,
      "grad_norm": 1.4725082018922724,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 20605
    },
    {
      "epoch": 0.20606,
      "grad_norm": 1.0881565574525953,
      "learning_rate": 0.003,
      "loss": 4.0246,
      "step": 20606
    },
    {
      "epoch": 0.20607,
      "grad_norm": 1.5911806474969958,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 20607
    },
    {
      "epoch": 0.20608,
      "grad_norm": 1.2697992330638663,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 20608
    },
    {
      "epoch": 0.20609,
      "grad_norm": 1.4750340535495794,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 20609
    },
    {
      "epoch": 0.2061,
      "grad_norm": 1.4952831819412025,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 20610
    },
    {
      "epoch": 0.20611,
      "grad_norm": 1.402393234502162,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 20611
    },
    {
      "epoch": 0.20612,
      "grad_norm": 1.3257391433023868,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 20612
    },
    {
      "epoch": 0.20613,
      "grad_norm": 1.3717513598075812,
      "learning_rate": 0.003,
      "loss": 4.0294,
      "step": 20613
    },
    {
      "epoch": 0.20614,
      "grad_norm": 1.4537290474940516,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 20614
    },
    {
      "epoch": 0.20615,
      "grad_norm": 1.5962756732291659,
      "learning_rate": 0.003,
      "loss": 4.0205,
      "step": 20615
    },
    {
      "epoch": 0.20616,
      "grad_norm": 1.5051333484818226,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 20616
    },
    {
      "epoch": 0.20617,
      "grad_norm": 1.214466112433022,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 20617
    },
    {
      "epoch": 0.20618,
      "grad_norm": 1.4469079444714108,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 20618
    },
    {
      "epoch": 0.20619,
      "grad_norm": 1.1397343693052815,
      "learning_rate": 0.003,
      "loss": 4.091,
      "step": 20619
    },
    {
      "epoch": 0.2062,
      "grad_norm": 1.3456803832918454,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 20620
    },
    {
      "epoch": 0.20621,
      "grad_norm": 1.1544122825716558,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 20621
    },
    {
      "epoch": 0.20622,
      "grad_norm": 1.4725408234925452,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 20622
    },
    {
      "epoch": 0.20623,
      "grad_norm": 1.1801597668502901,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 20623
    },
    {
      "epoch": 0.20624,
      "grad_norm": 1.6559749019356147,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 20624
    },
    {
      "epoch": 0.20625,
      "grad_norm": 1.2627663038337578,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 20625
    },
    {
      "epoch": 0.20626,
      "grad_norm": 1.5817579773010755,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 20626
    },
    {
      "epoch": 0.20627,
      "grad_norm": 1.1749488660893088,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 20627
    },
    {
      "epoch": 0.20628,
      "grad_norm": 1.2607651323166291,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 20628
    },
    {
      "epoch": 0.20629,
      "grad_norm": 1.1521351642743436,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 20629
    },
    {
      "epoch": 0.2063,
      "grad_norm": 1.341773679744627,
      "learning_rate": 0.003,
      "loss": 4.0285,
      "step": 20630
    },
    {
      "epoch": 0.20631,
      "grad_norm": 1.2009941594378983,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 20631
    },
    {
      "epoch": 0.20632,
      "grad_norm": 1.153179661594956,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 20632
    },
    {
      "epoch": 0.20633,
      "grad_norm": 1.3616113848885394,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 20633
    },
    {
      "epoch": 0.20634,
      "grad_norm": 1.2296475286632265,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 20634
    },
    {
      "epoch": 0.20635,
      "grad_norm": 1.2699110819449624,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 20635
    },
    {
      "epoch": 0.20636,
      "grad_norm": 1.6414333303584467,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 20636
    },
    {
      "epoch": 0.20637,
      "grad_norm": 1.0484630733656326,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 20637
    },
    {
      "epoch": 0.20638,
      "grad_norm": 1.5160647539076348,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 20638
    },
    {
      "epoch": 0.20639,
      "grad_norm": 1.2813846351739706,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 20639
    },
    {
      "epoch": 0.2064,
      "grad_norm": 1.505955669199115,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 20640
    },
    {
      "epoch": 0.20641,
      "grad_norm": 1.3139142425717492,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20641
    },
    {
      "epoch": 0.20642,
      "grad_norm": 1.3777522538089098,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 20642
    },
    {
      "epoch": 0.20643,
      "grad_norm": 1.6087420201941252,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 20643
    },
    {
      "epoch": 0.20644,
      "grad_norm": 1.1610478132610833,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 20644
    },
    {
      "epoch": 0.20645,
      "grad_norm": 1.3894341858483343,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 20645
    },
    {
      "epoch": 0.20646,
      "grad_norm": 1.5936425142861472,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 20646
    },
    {
      "epoch": 0.20647,
      "grad_norm": 1.3309219203024698,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 20647
    },
    {
      "epoch": 0.20648,
      "grad_norm": 1.531445145082477,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 20648
    },
    {
      "epoch": 0.20649,
      "grad_norm": 1.3212088065800662,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 20649
    },
    {
      "epoch": 0.2065,
      "grad_norm": 1.5004849199510446,
      "learning_rate": 0.003,
      "loss": 4.0254,
      "step": 20650
    },
    {
      "epoch": 0.20651,
      "grad_norm": 1.3039083086214973,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 20651
    },
    {
      "epoch": 0.20652,
      "grad_norm": 1.2837868305916982,
      "learning_rate": 0.003,
      "loss": 4.0001,
      "step": 20652
    },
    {
      "epoch": 0.20653,
      "grad_norm": 1.268203737086418,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 20653
    },
    {
      "epoch": 0.20654,
      "grad_norm": 1.3209765544553898,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 20654
    },
    {
      "epoch": 0.20655,
      "grad_norm": 1.3129637976579085,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 20655
    },
    {
      "epoch": 0.20656,
      "grad_norm": 1.4278382500501035,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 20656
    },
    {
      "epoch": 0.20657,
      "grad_norm": 1.4090868893650896,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 20657
    },
    {
      "epoch": 0.20658,
      "grad_norm": 1.1532438735239938,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 20658
    },
    {
      "epoch": 0.20659,
      "grad_norm": 1.5435740786225585,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 20659
    },
    {
      "epoch": 0.2066,
      "grad_norm": 1.1721951629686629,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 20660
    },
    {
      "epoch": 0.20661,
      "grad_norm": 1.432834732135936,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 20661
    },
    {
      "epoch": 0.20662,
      "grad_norm": 1.3326968054391681,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 20662
    },
    {
      "epoch": 0.20663,
      "grad_norm": 1.2551462438480991,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 20663
    },
    {
      "epoch": 0.20664,
      "grad_norm": 1.6477065949606842,
      "learning_rate": 0.003,
      "loss": 4.0262,
      "step": 20664
    },
    {
      "epoch": 0.20665,
      "grad_norm": 1.2073682442471083,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 20665
    },
    {
      "epoch": 0.20666,
      "grad_norm": 1.5967910925825928,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 20666
    },
    {
      "epoch": 0.20667,
      "grad_norm": 1.1702067473700242,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 20667
    },
    {
      "epoch": 0.20668,
      "grad_norm": 1.4490009490171216,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 20668
    },
    {
      "epoch": 0.20669,
      "grad_norm": 1.4982000874554064,
      "learning_rate": 0.003,
      "loss": 4.0211,
      "step": 20669
    },
    {
      "epoch": 0.2067,
      "grad_norm": 1.30080984292686,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 20670
    },
    {
      "epoch": 0.20671,
      "grad_norm": 1.3677103611908143,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 20671
    },
    {
      "epoch": 0.20672,
      "grad_norm": 1.2772405357651315,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 20672
    },
    {
      "epoch": 0.20673,
      "grad_norm": 1.3588295599384557,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 20673
    },
    {
      "epoch": 0.20674,
      "grad_norm": 1.3946185325560583,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 20674
    },
    {
      "epoch": 0.20675,
      "grad_norm": 1.4712534004184314,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 20675
    },
    {
      "epoch": 0.20676,
      "grad_norm": 1.3305414169625183,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 20676
    },
    {
      "epoch": 0.20677,
      "grad_norm": 1.5275630405565999,
      "learning_rate": 0.003,
      "loss": 4.0234,
      "step": 20677
    },
    {
      "epoch": 0.20678,
      "grad_norm": 1.1417928156030526,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 20678
    },
    {
      "epoch": 0.20679,
      "grad_norm": 1.3328961831124926,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 20679
    },
    {
      "epoch": 0.2068,
      "grad_norm": 1.3266191904489568,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 20680
    },
    {
      "epoch": 0.20681,
      "grad_norm": 1.1604107214621595,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 20681
    },
    {
      "epoch": 0.20682,
      "grad_norm": 1.6785511499365684,
      "learning_rate": 0.003,
      "loss": 4.023,
      "step": 20682
    },
    {
      "epoch": 0.20683,
      "grad_norm": 1.0861493863370346,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 20683
    },
    {
      "epoch": 0.20684,
      "grad_norm": 1.5560365428407226,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 20684
    },
    {
      "epoch": 0.20685,
      "grad_norm": 1.2498265822341013,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 20685
    },
    {
      "epoch": 0.20686,
      "grad_norm": 1.264505530010203,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 20686
    },
    {
      "epoch": 0.20687,
      "grad_norm": 1.2262354105769637,
      "learning_rate": 0.003,
      "loss": 4.0292,
      "step": 20687
    },
    {
      "epoch": 0.20688,
      "grad_norm": 1.3547014565798081,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 20688
    },
    {
      "epoch": 0.20689,
      "grad_norm": 1.3205301215705505,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 20689
    },
    {
      "epoch": 0.2069,
      "grad_norm": 1.4385638550422526,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 20690
    },
    {
      "epoch": 0.20691,
      "grad_norm": 1.254260184322425,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 20691
    },
    {
      "epoch": 0.20692,
      "grad_norm": 1.492638364859185,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 20692
    },
    {
      "epoch": 0.20693,
      "grad_norm": 1.6902384151982752,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 20693
    },
    {
      "epoch": 0.20694,
      "grad_norm": 1.5486125304486253,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 20694
    },
    {
      "epoch": 0.20695,
      "grad_norm": 1.3571206194483856,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 20695
    },
    {
      "epoch": 0.20696,
      "grad_norm": 1.195684554846468,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 20696
    },
    {
      "epoch": 0.20697,
      "grad_norm": 1.2673737123930042,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 20697
    },
    {
      "epoch": 0.20698,
      "grad_norm": 1.4501427918214878,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 20698
    },
    {
      "epoch": 0.20699,
      "grad_norm": 1.4616677033308434,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 20699
    },
    {
      "epoch": 0.207,
      "grad_norm": 1.259657976909807,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 20700
    },
    {
      "epoch": 0.20701,
      "grad_norm": 1.467939418322868,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 20701
    },
    {
      "epoch": 0.20702,
      "grad_norm": 1.2539470183297858,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 20702
    },
    {
      "epoch": 0.20703,
      "grad_norm": 1.244938656450836,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 20703
    },
    {
      "epoch": 0.20704,
      "grad_norm": 1.4363533964769344,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 20704
    },
    {
      "epoch": 0.20705,
      "grad_norm": 1.1799204393485285,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 20705
    },
    {
      "epoch": 0.20706,
      "grad_norm": 1.8304053317183064,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 20706
    },
    {
      "epoch": 0.20707,
      "grad_norm": 0.9245806846188429,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 20707
    },
    {
      "epoch": 0.20708,
      "grad_norm": 1.3670050823744555,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 20708
    },
    {
      "epoch": 0.20709,
      "grad_norm": 1.148663547313698,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 20709
    },
    {
      "epoch": 0.2071,
      "grad_norm": 1.4332837261687998,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 20710
    },
    {
      "epoch": 0.20711,
      "grad_norm": 1.2474709065086116,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 20711
    },
    {
      "epoch": 0.20712,
      "grad_norm": 1.518321959945295,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 20712
    },
    {
      "epoch": 0.20713,
      "grad_norm": 1.2589783945695787,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 20713
    },
    {
      "epoch": 0.20714,
      "grad_norm": 1.4680065605046984,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 20714
    },
    {
      "epoch": 0.20715,
      "grad_norm": 1.3689933685128204,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 20715
    },
    {
      "epoch": 0.20716,
      "grad_norm": 1.1690381414163973,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 20716
    },
    {
      "epoch": 0.20717,
      "grad_norm": 1.430373190789418,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 20717
    },
    {
      "epoch": 0.20718,
      "grad_norm": 1.2445200041778108,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 20718
    },
    {
      "epoch": 0.20719,
      "grad_norm": 1.2360357908003283,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 20719
    },
    {
      "epoch": 0.2072,
      "grad_norm": 1.1518801939778631,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 20720
    },
    {
      "epoch": 0.20721,
      "grad_norm": 1.3540922092959804,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 20721
    },
    {
      "epoch": 0.20722,
      "grad_norm": 1.403886039911295,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 20722
    },
    {
      "epoch": 0.20723,
      "grad_norm": 1.408496319277612,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 20723
    },
    {
      "epoch": 0.20724,
      "grad_norm": 1.53624630820367,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 20724
    },
    {
      "epoch": 0.20725,
      "grad_norm": 1.1028599247352295,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 20725
    },
    {
      "epoch": 0.20726,
      "grad_norm": 1.4912493458663871,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 20726
    },
    {
      "epoch": 0.20727,
      "grad_norm": 1.400145332522727,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 20727
    },
    {
      "epoch": 0.20728,
      "grad_norm": 1.2550730824349292,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 20728
    },
    {
      "epoch": 0.20729,
      "grad_norm": 1.5020230627391862,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 20729
    },
    {
      "epoch": 0.2073,
      "grad_norm": 1.094855880312423,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 20730
    },
    {
      "epoch": 0.20731,
      "grad_norm": 1.4454634777088968,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 20731
    },
    {
      "epoch": 0.20732,
      "grad_norm": 1.1960298178416016,
      "learning_rate": 0.003,
      "loss": 4.011,
      "step": 20732
    },
    {
      "epoch": 0.20733,
      "grad_norm": 1.573553533514854,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 20733
    },
    {
      "epoch": 0.20734,
      "grad_norm": 1.3005374564992995,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 20734
    },
    {
      "epoch": 0.20735,
      "grad_norm": 1.4733758613998844,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 20735
    },
    {
      "epoch": 0.20736,
      "grad_norm": 1.192320375520012,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 20736
    },
    {
      "epoch": 0.20737,
      "grad_norm": 1.441776597098866,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20737
    },
    {
      "epoch": 0.20738,
      "grad_norm": 1.2019891485357035,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 20738
    },
    {
      "epoch": 0.20739,
      "grad_norm": 1.7143393481625193,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 20739
    },
    {
      "epoch": 0.2074,
      "grad_norm": 1.3011650865174773,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 20740
    },
    {
      "epoch": 0.20741,
      "grad_norm": 1.420040265271807,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 20741
    },
    {
      "epoch": 0.20742,
      "grad_norm": 1.271071548410552,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 20742
    },
    {
      "epoch": 0.20743,
      "grad_norm": 1.2725194915570306,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 20743
    },
    {
      "epoch": 0.20744,
      "grad_norm": 1.407554084131202,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 20744
    },
    {
      "epoch": 0.20745,
      "grad_norm": 1.2723492493093125,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 20745
    },
    {
      "epoch": 0.20746,
      "grad_norm": 1.2789554944969392,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 20746
    },
    {
      "epoch": 0.20747,
      "grad_norm": 1.2190173749910922,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 20747
    },
    {
      "epoch": 0.20748,
      "grad_norm": 1.3671246285439873,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 20748
    },
    {
      "epoch": 0.20749,
      "grad_norm": 1.2860610697540353,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 20749
    },
    {
      "epoch": 0.2075,
      "grad_norm": 1.5485327618678362,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 20750
    },
    {
      "epoch": 0.20751,
      "grad_norm": 1.0354250543920769,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 20751
    },
    {
      "epoch": 0.20752,
      "grad_norm": 1.5729608571657345,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 20752
    },
    {
      "epoch": 0.20753,
      "grad_norm": 1.293985990906778,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 20753
    },
    {
      "epoch": 0.20754,
      "grad_norm": 1.383144680757959,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 20754
    },
    {
      "epoch": 0.20755,
      "grad_norm": 1.5258289814465253,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 20755
    },
    {
      "epoch": 0.20756,
      "grad_norm": 1.2284186594380586,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 20756
    },
    {
      "epoch": 0.20757,
      "grad_norm": 1.295884549935487,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 20757
    },
    {
      "epoch": 0.20758,
      "grad_norm": 1.2251701011240375,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 20758
    },
    {
      "epoch": 0.20759,
      "grad_norm": 1.4138209427266042,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 20759
    },
    {
      "epoch": 0.2076,
      "grad_norm": 1.2928386514239687,
      "learning_rate": 0.003,
      "loss": 4.0196,
      "step": 20760
    },
    {
      "epoch": 0.20761,
      "grad_norm": 1.6584917470311489,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 20761
    },
    {
      "epoch": 0.20762,
      "grad_norm": 1.1000944720003125,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20762
    },
    {
      "epoch": 0.20763,
      "grad_norm": 1.5065086778699819,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 20763
    },
    {
      "epoch": 0.20764,
      "grad_norm": 1.2405226321846727,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 20764
    },
    {
      "epoch": 0.20765,
      "grad_norm": 1.5062568526642643,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 20765
    },
    {
      "epoch": 0.20766,
      "grad_norm": 1.2543886727598397,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 20766
    },
    {
      "epoch": 0.20767,
      "grad_norm": 1.4686589157048027,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 20767
    },
    {
      "epoch": 0.20768,
      "grad_norm": 1.0878892800689992,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 20768
    },
    {
      "epoch": 0.20769,
      "grad_norm": 1.2890076109933548,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 20769
    },
    {
      "epoch": 0.2077,
      "grad_norm": 1.2698804207351753,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 20770
    },
    {
      "epoch": 0.20771,
      "grad_norm": 1.2963430756336742,
      "learning_rate": 0.003,
      "loss": 4.0139,
      "step": 20771
    },
    {
      "epoch": 0.20772,
      "grad_norm": 1.5503858672891013,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 20772
    },
    {
      "epoch": 0.20773,
      "grad_norm": 1.447824162200313,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 20773
    },
    {
      "epoch": 0.20774,
      "grad_norm": 1.0698241675703897,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 20774
    },
    {
      "epoch": 0.20775,
      "grad_norm": 1.5921843730582812,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 20775
    },
    {
      "epoch": 0.20776,
      "grad_norm": 1.5676573158126632,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 20776
    },
    {
      "epoch": 0.20777,
      "grad_norm": 1.2337925630873627,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 20777
    },
    {
      "epoch": 0.20778,
      "grad_norm": 1.558918139946969,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 20778
    },
    {
      "epoch": 0.20779,
      "grad_norm": 1.2576777157283208,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 20779
    },
    {
      "epoch": 0.2078,
      "grad_norm": 1.4891119004045967,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 20780
    },
    {
      "epoch": 0.20781,
      "grad_norm": 1.4864117549677052,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 20781
    },
    {
      "epoch": 0.20782,
      "grad_norm": 1.3824042412785735,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 20782
    },
    {
      "epoch": 0.20783,
      "grad_norm": 1.2647212252167144,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 20783
    },
    {
      "epoch": 0.20784,
      "grad_norm": 1.3682686819141703,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 20784
    },
    {
      "epoch": 0.20785,
      "grad_norm": 1.3134092379540208,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 20785
    },
    {
      "epoch": 0.20786,
      "grad_norm": 1.2000088386311523,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 20786
    },
    {
      "epoch": 0.20787,
      "grad_norm": 1.358948748403982,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 20787
    },
    {
      "epoch": 0.20788,
      "grad_norm": 1.2266057165726971,
      "learning_rate": 0.003,
      "loss": 4.0299,
      "step": 20788
    },
    {
      "epoch": 0.20789,
      "grad_norm": 1.4599833485159968,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 20789
    },
    {
      "epoch": 0.2079,
      "grad_norm": 1.4274327253015713,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 20790
    },
    {
      "epoch": 0.20791,
      "grad_norm": 1.4618313253646975,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 20791
    },
    {
      "epoch": 0.20792,
      "grad_norm": 1.3711908786137523,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 20792
    },
    {
      "epoch": 0.20793,
      "grad_norm": 1.498250720442979,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 20793
    },
    {
      "epoch": 0.20794,
      "grad_norm": 1.2535367799442354,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 20794
    },
    {
      "epoch": 0.20795,
      "grad_norm": 1.3419060508276737,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 20795
    },
    {
      "epoch": 0.20796,
      "grad_norm": 1.4115902482667926,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 20796
    },
    {
      "epoch": 0.20797,
      "grad_norm": 1.3850709340925633,
      "learning_rate": 0.003,
      "loss": 4.0084,
      "step": 20797
    },
    {
      "epoch": 0.20798,
      "grad_norm": 1.3207471580223884,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 20798
    },
    {
      "epoch": 0.20799,
      "grad_norm": 1.3568939822112909,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 20799
    },
    {
      "epoch": 0.208,
      "grad_norm": 1.2019262130445871,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 20800
    },
    {
      "epoch": 0.20801,
      "grad_norm": 1.5522301951501158,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 20801
    },
    {
      "epoch": 0.20802,
      "grad_norm": 1.2738694402674298,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 20802
    },
    {
      "epoch": 0.20803,
      "grad_norm": 1.6066304757212384,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 20803
    },
    {
      "epoch": 0.20804,
      "grad_norm": 1.1511055437116582,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 20804
    },
    {
      "epoch": 0.20805,
      "grad_norm": 1.3737968339786775,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 20805
    },
    {
      "epoch": 0.20806,
      "grad_norm": 1.1696700332140544,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 20806
    },
    {
      "epoch": 0.20807,
      "grad_norm": 1.438958584346387,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 20807
    },
    {
      "epoch": 0.20808,
      "grad_norm": 1.3446405494157427,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 20808
    },
    {
      "epoch": 0.20809,
      "grad_norm": 1.509586891470056,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 20809
    },
    {
      "epoch": 0.2081,
      "grad_norm": 1.2164155739319678,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 20810
    },
    {
      "epoch": 0.20811,
      "grad_norm": 1.4657159505696644,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 20811
    },
    {
      "epoch": 0.20812,
      "grad_norm": 1.2897275751829282,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 20812
    },
    {
      "epoch": 0.20813,
      "grad_norm": 1.2342837330568117,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 20813
    },
    {
      "epoch": 0.20814,
      "grad_norm": 1.2818396025311736,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 20814
    },
    {
      "epoch": 0.20815,
      "grad_norm": 1.138639834661771,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 20815
    },
    {
      "epoch": 0.20816,
      "grad_norm": 1.5138052170278105,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 20816
    },
    {
      "epoch": 0.20817,
      "grad_norm": 1.1634977361316547,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 20817
    },
    {
      "epoch": 0.20818,
      "grad_norm": 1.5215420540090645,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 20818
    },
    {
      "epoch": 0.20819,
      "grad_norm": 1.5668058020930633,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 20819
    },
    {
      "epoch": 0.2082,
      "grad_norm": 1.4531488544684097,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 20820
    },
    {
      "epoch": 0.20821,
      "grad_norm": 1.2865371605977516,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 20821
    },
    {
      "epoch": 0.20822,
      "grad_norm": 1.309351243002296,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 20822
    },
    {
      "epoch": 0.20823,
      "grad_norm": 1.4421691511982775,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 20823
    },
    {
      "epoch": 0.20824,
      "grad_norm": 1.2366488864526652,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 20824
    },
    {
      "epoch": 0.20825,
      "grad_norm": 1.5101299369902328,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 20825
    },
    {
      "epoch": 0.20826,
      "grad_norm": 1.0820147383838739,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 20826
    },
    {
      "epoch": 0.20827,
      "grad_norm": 1.5934109137934334,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 20827
    },
    {
      "epoch": 0.20828,
      "grad_norm": 1.109182361190493,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20828
    },
    {
      "epoch": 0.20829,
      "grad_norm": 1.6282571708569145,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 20829
    },
    {
      "epoch": 0.2083,
      "grad_norm": 1.3254657626410726,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 20830
    },
    {
      "epoch": 0.20831,
      "grad_norm": 1.4069406914196951,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 20831
    },
    {
      "epoch": 0.20832,
      "grad_norm": 1.1955776117205292,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 20832
    },
    {
      "epoch": 0.20833,
      "grad_norm": 1.6091706976751523,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 20833
    },
    {
      "epoch": 0.20834,
      "grad_norm": 1.3554567338878298,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 20834
    },
    {
      "epoch": 0.20835,
      "grad_norm": 1.2173982775230165,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 20835
    },
    {
      "epoch": 0.20836,
      "grad_norm": 1.5726560445228621,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 20836
    },
    {
      "epoch": 0.20837,
      "grad_norm": 1.3197429362875017,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 20837
    },
    {
      "epoch": 0.20838,
      "grad_norm": 1.3932221313372262,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 20838
    },
    {
      "epoch": 0.20839,
      "grad_norm": 1.3436690757123135,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 20839
    },
    {
      "epoch": 0.2084,
      "grad_norm": 1.2914233320150121,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 20840
    },
    {
      "epoch": 0.20841,
      "grad_norm": 1.4039904864137214,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 20841
    },
    {
      "epoch": 0.20842,
      "grad_norm": 1.1224930480012258,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 20842
    },
    {
      "epoch": 0.20843,
      "grad_norm": 1.5006419635128865,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 20843
    },
    {
      "epoch": 0.20844,
      "grad_norm": 1.316535683710008,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 20844
    },
    {
      "epoch": 0.20845,
      "grad_norm": 1.6994979771996706,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 20845
    },
    {
      "epoch": 0.20846,
      "grad_norm": 1.4424776773399692,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 20846
    },
    {
      "epoch": 0.20847,
      "grad_norm": 1.292769059462763,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 20847
    },
    {
      "epoch": 0.20848,
      "grad_norm": 1.3931095132806635,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 20848
    },
    {
      "epoch": 0.20849,
      "grad_norm": 1.4169322101511408,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 20849
    },
    {
      "epoch": 0.2085,
      "grad_norm": 1.4162308539627977,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 20850
    },
    {
      "epoch": 0.20851,
      "grad_norm": 1.3169184739063786,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 20851
    },
    {
      "epoch": 0.20852,
      "grad_norm": 1.2926076560780115,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 20852
    },
    {
      "epoch": 0.20853,
      "grad_norm": 1.4132541866511903,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 20853
    },
    {
      "epoch": 0.20854,
      "grad_norm": 1.2636015858679868,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 20854
    },
    {
      "epoch": 0.20855,
      "grad_norm": 1.1150503537596996,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 20855
    },
    {
      "epoch": 0.20856,
      "grad_norm": 1.4027910960977807,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 20856
    },
    {
      "epoch": 0.20857,
      "grad_norm": 1.4297746470245551,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 20857
    },
    {
      "epoch": 0.20858,
      "grad_norm": 1.396782413637501,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 20858
    },
    {
      "epoch": 0.20859,
      "grad_norm": 1.164888722395929,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 20859
    },
    {
      "epoch": 0.2086,
      "grad_norm": 1.4318988786579787,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 20860
    },
    {
      "epoch": 0.20861,
      "grad_norm": 1.3869170185913235,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 20861
    },
    {
      "epoch": 0.20862,
      "grad_norm": 1.459312875995161,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 20862
    },
    {
      "epoch": 0.20863,
      "grad_norm": 1.353994075043403,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 20863
    },
    {
      "epoch": 0.20864,
      "grad_norm": 1.1460913522678278,
      "learning_rate": 0.003,
      "loss": 4.0236,
      "step": 20864
    },
    {
      "epoch": 0.20865,
      "grad_norm": 1.1715799872377994,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 20865
    },
    {
      "epoch": 0.20866,
      "grad_norm": 1.5112689480891477,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 20866
    },
    {
      "epoch": 0.20867,
      "grad_norm": 1.2277638411317997,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 20867
    },
    {
      "epoch": 0.20868,
      "grad_norm": 1.3559174681191846,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 20868
    },
    {
      "epoch": 0.20869,
      "grad_norm": 1.1579036036919177,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 20869
    },
    {
      "epoch": 0.2087,
      "grad_norm": 1.5168868574343963,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20870
    },
    {
      "epoch": 0.20871,
      "grad_norm": 1.192882139239422,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 20871
    },
    {
      "epoch": 0.20872,
      "grad_norm": 1.6397552745256645,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 20872
    },
    {
      "epoch": 0.20873,
      "grad_norm": 1.2998046635660576,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20873
    },
    {
      "epoch": 0.20874,
      "grad_norm": 1.170136001703282,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 20874
    },
    {
      "epoch": 0.20875,
      "grad_norm": 1.7712938964576301,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 20875
    },
    {
      "epoch": 0.20876,
      "grad_norm": 1.304514044762122,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 20876
    },
    {
      "epoch": 0.20877,
      "grad_norm": 1.5311149814383385,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 20877
    },
    {
      "epoch": 0.20878,
      "grad_norm": 1.377141579972561,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 20878
    },
    {
      "epoch": 0.20879,
      "grad_norm": 1.2885018804197217,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 20879
    },
    {
      "epoch": 0.2088,
      "grad_norm": 1.2805009293500598,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 20880
    },
    {
      "epoch": 0.20881,
      "grad_norm": 1.5271926042823611,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 20881
    },
    {
      "epoch": 0.20882,
      "grad_norm": 1.313556059499855,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 20882
    },
    {
      "epoch": 0.20883,
      "grad_norm": 1.1464190081317402,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 20883
    },
    {
      "epoch": 0.20884,
      "grad_norm": 1.549942223937229,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 20884
    },
    {
      "epoch": 0.20885,
      "grad_norm": 1.1265602354973108,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 20885
    },
    {
      "epoch": 0.20886,
      "grad_norm": 1.707473848813536,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 20886
    },
    {
      "epoch": 0.20887,
      "grad_norm": 0.9961645801473272,
      "learning_rate": 0.003,
      "loss": 4.099,
      "step": 20887
    },
    {
      "epoch": 0.20888,
      "grad_norm": 1.538759783989019,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 20888
    },
    {
      "epoch": 0.20889,
      "grad_norm": 1.3896655265307651,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 20889
    },
    {
      "epoch": 0.2089,
      "grad_norm": 1.1829799147541915,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 20890
    },
    {
      "epoch": 0.20891,
      "grad_norm": 1.4199392258570829,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 20891
    },
    {
      "epoch": 0.20892,
      "grad_norm": 1.3251457352384242,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 20892
    },
    {
      "epoch": 0.20893,
      "grad_norm": 1.3720741261766807,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 20893
    },
    {
      "epoch": 0.20894,
      "grad_norm": 1.4397315716619452,
      "learning_rate": 0.003,
      "loss": 4.0221,
      "step": 20894
    },
    {
      "epoch": 0.20895,
      "grad_norm": 1.2600167535017244,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 20895
    },
    {
      "epoch": 0.20896,
      "grad_norm": 1.4193530587315695,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 20896
    },
    {
      "epoch": 0.20897,
      "grad_norm": 1.4613543310948776,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 20897
    },
    {
      "epoch": 0.20898,
      "grad_norm": 1.4334011717950574,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 20898
    },
    {
      "epoch": 0.20899,
      "grad_norm": 1.2649157110995044,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 20899
    },
    {
      "epoch": 0.209,
      "grad_norm": 1.4679698539406412,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 20900
    },
    {
      "epoch": 0.20901,
      "grad_norm": 1.1456737698413446,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 20901
    },
    {
      "epoch": 0.20902,
      "grad_norm": 1.260909399618353,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 20902
    },
    {
      "epoch": 0.20903,
      "grad_norm": 1.3774463340297802,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 20903
    },
    {
      "epoch": 0.20904,
      "grad_norm": 1.3792000896777399,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 20904
    },
    {
      "epoch": 0.20905,
      "grad_norm": 1.3145741389508931,
      "learning_rate": 0.003,
      "loss": 4.0991,
      "step": 20905
    },
    {
      "epoch": 0.20906,
      "grad_norm": 1.488555454898239,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 20906
    },
    {
      "epoch": 0.20907,
      "grad_norm": 0.9177311400044877,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 20907
    },
    {
      "epoch": 0.20908,
      "grad_norm": 1.4736044351887085,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 20908
    },
    {
      "epoch": 0.20909,
      "grad_norm": 1.236906959060174,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 20909
    },
    {
      "epoch": 0.2091,
      "grad_norm": 1.2164246192183843,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 20910
    },
    {
      "epoch": 0.20911,
      "grad_norm": 1.4078924617347541,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 20911
    },
    {
      "epoch": 0.20912,
      "grad_norm": 1.1626258082614278,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 20912
    },
    {
      "epoch": 0.20913,
      "grad_norm": 1.5000031481174514,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 20913
    },
    {
      "epoch": 0.20914,
      "grad_norm": 1.711991182599059,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 20914
    },
    {
      "epoch": 0.20915,
      "grad_norm": 1.1173024910508889,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 20915
    },
    {
      "epoch": 0.20916,
      "grad_norm": 1.6195944409721208,
      "learning_rate": 0.003,
      "loss": 4.0968,
      "step": 20916
    },
    {
      "epoch": 0.20917,
      "grad_norm": 1.1495495916677396,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 20917
    },
    {
      "epoch": 0.20918,
      "grad_norm": 1.3315584679143382,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 20918
    },
    {
      "epoch": 0.20919,
      "grad_norm": 1.3129474249266087,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 20919
    },
    {
      "epoch": 0.2092,
      "grad_norm": 1.45735198601752,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 20920
    },
    {
      "epoch": 0.20921,
      "grad_norm": 1.2008207342408617,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 20921
    },
    {
      "epoch": 0.20922,
      "grad_norm": 1.490154537659906,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 20922
    },
    {
      "epoch": 0.20923,
      "grad_norm": 1.3246868033101271,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 20923
    },
    {
      "epoch": 0.20924,
      "grad_norm": 1.4013961451103512,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 20924
    },
    {
      "epoch": 0.20925,
      "grad_norm": 1.3739218556715787,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 20925
    },
    {
      "epoch": 0.20926,
      "grad_norm": 1.272221983849928,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 20926
    },
    {
      "epoch": 0.20927,
      "grad_norm": 1.5647031482397493,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 20927
    },
    {
      "epoch": 0.20928,
      "grad_norm": 1.2508057694645471,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 20928
    },
    {
      "epoch": 0.20929,
      "grad_norm": 1.3826914101940824,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 20929
    },
    {
      "epoch": 0.2093,
      "grad_norm": 1.548268893130226,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 20930
    },
    {
      "epoch": 0.20931,
      "grad_norm": 1.2446087214126185,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 20931
    },
    {
      "epoch": 0.20932,
      "grad_norm": 1.6262942416468529,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 20932
    },
    {
      "epoch": 0.20933,
      "grad_norm": 1.1852497232306514,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 20933
    },
    {
      "epoch": 0.20934,
      "grad_norm": 1.6696998384812047,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 20934
    },
    {
      "epoch": 0.20935,
      "grad_norm": 1.2181296766904035,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 20935
    },
    {
      "epoch": 0.20936,
      "grad_norm": 1.4900857645434027,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 20936
    },
    {
      "epoch": 0.20937,
      "grad_norm": 1.3021609810644803,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 20937
    },
    {
      "epoch": 0.20938,
      "grad_norm": 1.295584171371703,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 20938
    },
    {
      "epoch": 0.20939,
      "grad_norm": 1.7010666167860795,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 20939
    },
    {
      "epoch": 0.2094,
      "grad_norm": 0.9469238419501262,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 20940
    },
    {
      "epoch": 0.20941,
      "grad_norm": 1.3814181330827642,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 20941
    },
    {
      "epoch": 0.20942,
      "grad_norm": 1.4840720303537958,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 20942
    },
    {
      "epoch": 0.20943,
      "grad_norm": 1.3128491819583785,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 20943
    },
    {
      "epoch": 0.20944,
      "grad_norm": 1.4681052115962732,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 20944
    },
    {
      "epoch": 0.20945,
      "grad_norm": 1.0928041953544825,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 20945
    },
    {
      "epoch": 0.20946,
      "grad_norm": 1.2776202521965,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 20946
    },
    {
      "epoch": 0.20947,
      "grad_norm": 1.2072092071595812,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 20947
    },
    {
      "epoch": 0.20948,
      "grad_norm": 1.4983914960524811,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 20948
    },
    {
      "epoch": 0.20949,
      "grad_norm": 1.3556837860308897,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 20949
    },
    {
      "epoch": 0.2095,
      "grad_norm": 1.1747432769790092,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 20950
    },
    {
      "epoch": 0.20951,
      "grad_norm": 1.4823779753416622,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 20951
    },
    {
      "epoch": 0.20952,
      "grad_norm": 0.9942073008067971,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 20952
    },
    {
      "epoch": 0.20953,
      "grad_norm": 1.6230027795029984,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 20953
    },
    {
      "epoch": 0.20954,
      "grad_norm": 1.1421677685740035,
      "learning_rate": 0.003,
      "loss": 4.0269,
      "step": 20954
    },
    {
      "epoch": 0.20955,
      "grad_norm": 1.6820318386159103,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 20955
    },
    {
      "epoch": 0.20956,
      "grad_norm": 1.3487592414130567,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 20956
    },
    {
      "epoch": 0.20957,
      "grad_norm": 1.3066160841943804,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 20957
    },
    {
      "epoch": 0.20958,
      "grad_norm": 1.1619227542351784,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 20958
    },
    {
      "epoch": 0.20959,
      "grad_norm": 1.4546064403658636,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 20959
    },
    {
      "epoch": 0.2096,
      "grad_norm": 1.1838430608934516,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 20960
    },
    {
      "epoch": 0.20961,
      "grad_norm": 1.55115485747746,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 20961
    },
    {
      "epoch": 0.20962,
      "grad_norm": 1.3047990384928365,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 20962
    },
    {
      "epoch": 0.20963,
      "grad_norm": 1.3312728907537128,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 20963
    },
    {
      "epoch": 0.20964,
      "grad_norm": 1.5045742901254493,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 20964
    },
    {
      "epoch": 0.20965,
      "grad_norm": 1.216675557605364,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 20965
    },
    {
      "epoch": 0.20966,
      "grad_norm": 1.3876853879303457,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 20966
    },
    {
      "epoch": 0.20967,
      "grad_norm": 1.3182662675036918,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 20967
    },
    {
      "epoch": 0.20968,
      "grad_norm": 1.3461464413037225,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 20968
    },
    {
      "epoch": 0.20969,
      "grad_norm": 1.310372816887392,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 20969
    },
    {
      "epoch": 0.2097,
      "grad_norm": 1.4287852852752707,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 20970
    },
    {
      "epoch": 0.20971,
      "grad_norm": 1.2742591333006903,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 20971
    },
    {
      "epoch": 0.20972,
      "grad_norm": 1.410034093862182,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 20972
    },
    {
      "epoch": 0.20973,
      "grad_norm": 1.4844923925773001,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 20973
    },
    {
      "epoch": 0.20974,
      "grad_norm": 1.5172877935597546,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 20974
    },
    {
      "epoch": 0.20975,
      "grad_norm": 1.3044946275124711,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 20975
    },
    {
      "epoch": 0.20976,
      "grad_norm": 1.151846917353467,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 20976
    },
    {
      "epoch": 0.20977,
      "grad_norm": 1.5449323600079314,
      "learning_rate": 0.003,
      "loss": 4.09,
      "step": 20977
    },
    {
      "epoch": 0.20978,
      "grad_norm": 1.1036848639930728,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 20978
    },
    {
      "epoch": 0.20979,
      "grad_norm": 1.4178704958997441,
      "learning_rate": 0.003,
      "loss": 4.0948,
      "step": 20979
    },
    {
      "epoch": 0.2098,
      "grad_norm": 1.491666526743511,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 20980
    },
    {
      "epoch": 0.20981,
      "grad_norm": 1.259111988716843,
      "learning_rate": 0.003,
      "loss": 4.0259,
      "step": 20981
    },
    {
      "epoch": 0.20982,
      "grad_norm": 1.392424981454325,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 20982
    },
    {
      "epoch": 0.20983,
      "grad_norm": 1.3734294209193698,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 20983
    },
    {
      "epoch": 0.20984,
      "grad_norm": 1.2098709163289976,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 20984
    },
    {
      "epoch": 0.20985,
      "grad_norm": 1.5174873420676978,
      "learning_rate": 0.003,
      "loss": 4.0862,
      "step": 20985
    },
    {
      "epoch": 0.20986,
      "grad_norm": 1.109118359260239,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 20986
    },
    {
      "epoch": 0.20987,
      "grad_norm": 1.6151173655333941,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 20987
    },
    {
      "epoch": 0.20988,
      "grad_norm": 1.317078526794028,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 20988
    },
    {
      "epoch": 0.20989,
      "grad_norm": 1.3960094316751979,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 20989
    },
    {
      "epoch": 0.2099,
      "grad_norm": 1.502111690521901,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 20990
    },
    {
      "epoch": 0.20991,
      "grad_norm": 1.2405024142395,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 20991
    },
    {
      "epoch": 0.20992,
      "grad_norm": 1.5708082629079743,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 20992
    },
    {
      "epoch": 0.20993,
      "grad_norm": 1.2786993413574794,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 20993
    },
    {
      "epoch": 0.20994,
      "grad_norm": 1.3277837390722653,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 20994
    },
    {
      "epoch": 0.20995,
      "grad_norm": 1.43136863374496,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 20995
    },
    {
      "epoch": 0.20996,
      "grad_norm": 1.1309822698757983,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 20996
    },
    {
      "epoch": 0.20997,
      "grad_norm": 1.567826172504886,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 20997
    },
    {
      "epoch": 0.20998,
      "grad_norm": 1.0645283417366327,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 20998
    },
    {
      "epoch": 0.20999,
      "grad_norm": 1.5062893700951852,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 20999
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2211254389217334,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 21000
    },
    {
      "epoch": 0.21001,
      "grad_norm": 1.461437401694092,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 21001
    },
    {
      "epoch": 0.21002,
      "grad_norm": 1.422544853215876,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 21002
    },
    {
      "epoch": 0.21003,
      "grad_norm": 1.3366035931727727,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 21003
    },
    {
      "epoch": 0.21004,
      "grad_norm": 1.5459735826737728,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 21004
    },
    {
      "epoch": 0.21005,
      "grad_norm": 1.2461805438471347,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 21005
    },
    {
      "epoch": 0.21006,
      "grad_norm": 1.5147791701333184,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 21006
    },
    {
      "epoch": 0.21007,
      "grad_norm": 1.2901714481509183,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 21007
    },
    {
      "epoch": 0.21008,
      "grad_norm": 1.1429132481952657,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 21008
    },
    {
      "epoch": 0.21009,
      "grad_norm": 1.3931111686626179,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 21009
    },
    {
      "epoch": 0.2101,
      "grad_norm": 1.4620137134793034,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 21010
    },
    {
      "epoch": 0.21011,
      "grad_norm": 1.2342907771891274,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 21011
    },
    {
      "epoch": 0.21012,
      "grad_norm": 1.4617620465492411,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 21012
    },
    {
      "epoch": 0.21013,
      "grad_norm": 1.2649399673357389,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 21013
    },
    {
      "epoch": 0.21014,
      "grad_norm": 1.3435256308506187,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 21014
    },
    {
      "epoch": 0.21015,
      "grad_norm": 1.244101791785602,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 21015
    },
    {
      "epoch": 0.21016,
      "grad_norm": 1.1565299822725221,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 21016
    },
    {
      "epoch": 0.21017,
      "grad_norm": 1.252277784984338,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 21017
    },
    {
      "epoch": 0.21018,
      "grad_norm": 1.4535768408396037,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 21018
    },
    {
      "epoch": 0.21019,
      "grad_norm": 1.316193671135584,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 21019
    },
    {
      "epoch": 0.2102,
      "grad_norm": 1.3432893741725889,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 21020
    },
    {
      "epoch": 0.21021,
      "grad_norm": 1.0648436783358863,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 21021
    },
    {
      "epoch": 0.21022,
      "grad_norm": 1.6892575258933102,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 21022
    },
    {
      "epoch": 0.21023,
      "grad_norm": 1.1008692586381736,
      "learning_rate": 0.003,
      "loss": 4.0258,
      "step": 21023
    },
    {
      "epoch": 0.21024,
      "grad_norm": 1.5431788160399122,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 21024
    },
    {
      "epoch": 0.21025,
      "grad_norm": 1.6087965706302754,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 21025
    },
    {
      "epoch": 0.21026,
      "grad_norm": 1.3207794878231496,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 21026
    },
    {
      "epoch": 0.21027,
      "grad_norm": 1.405276543970181,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 21027
    },
    {
      "epoch": 0.21028,
      "grad_norm": 1.376324143024024,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 21028
    },
    {
      "epoch": 0.21029,
      "grad_norm": 1.365003958090158,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 21029
    },
    {
      "epoch": 0.2103,
      "grad_norm": 1.4058150514705787,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 21030
    },
    {
      "epoch": 0.21031,
      "grad_norm": 1.153189792926779,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 21031
    },
    {
      "epoch": 0.21032,
      "grad_norm": 1.3397695114600154,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 21032
    },
    {
      "epoch": 0.21033,
      "grad_norm": 1.2726878278038598,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 21033
    },
    {
      "epoch": 0.21034,
      "grad_norm": 1.1724902441253162,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 21034
    },
    {
      "epoch": 0.21035,
      "grad_norm": 1.4622195266657432,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 21035
    },
    {
      "epoch": 0.21036,
      "grad_norm": 1.2728210321227869,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 21036
    },
    {
      "epoch": 0.21037,
      "grad_norm": 1.337362962671212,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 21037
    },
    {
      "epoch": 0.21038,
      "grad_norm": 1.6430043034892616,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 21038
    },
    {
      "epoch": 0.21039,
      "grad_norm": 1.6129415272481018,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 21039
    },
    {
      "epoch": 0.2104,
      "grad_norm": 1.4475097526229443,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 21040
    },
    {
      "epoch": 0.21041,
      "grad_norm": 1.1784070418565082,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 21041
    },
    {
      "epoch": 0.21042,
      "grad_norm": 1.622691663690252,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 21042
    },
    {
      "epoch": 0.21043,
      "grad_norm": 1.2557235638469777,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 21043
    },
    {
      "epoch": 0.21044,
      "grad_norm": 1.1002234268770594,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 21044
    },
    {
      "epoch": 0.21045,
      "grad_norm": 1.467101985799226,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 21045
    },
    {
      "epoch": 0.21046,
      "grad_norm": 1.4592275042934686,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 21046
    },
    {
      "epoch": 0.21047,
      "grad_norm": 1.1961366095601582,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 21047
    },
    {
      "epoch": 0.21048,
      "grad_norm": 1.3259790306030865,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 21048
    },
    {
      "epoch": 0.21049,
      "grad_norm": 1.459604071228318,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 21049
    },
    {
      "epoch": 0.2105,
      "grad_norm": 1.3356775644760128,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 21050
    },
    {
      "epoch": 0.21051,
      "grad_norm": 1.2611272734027086,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 21051
    },
    {
      "epoch": 0.21052,
      "grad_norm": 1.4561435107404017,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 21052
    },
    {
      "epoch": 0.21053,
      "grad_norm": 1.3172832097982368,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 21053
    },
    {
      "epoch": 0.21054,
      "grad_norm": 1.5079848509312952,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 21054
    },
    {
      "epoch": 0.21055,
      "grad_norm": 1.0682100137209913,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 21055
    },
    {
      "epoch": 0.21056,
      "grad_norm": 1.3989052113249258,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 21056
    },
    {
      "epoch": 0.21057,
      "grad_norm": 1.2515853111972497,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 21057
    },
    {
      "epoch": 0.21058,
      "grad_norm": 1.8385061177094724,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 21058
    },
    {
      "epoch": 0.21059,
      "grad_norm": 1.2619283793884344,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 21059
    },
    {
      "epoch": 0.2106,
      "grad_norm": 1.1603808791470698,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 21060
    },
    {
      "epoch": 0.21061,
      "grad_norm": 1.5766504515523034,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 21061
    },
    {
      "epoch": 0.21062,
      "grad_norm": 1.0923398235748982,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 21062
    },
    {
      "epoch": 0.21063,
      "grad_norm": 1.3157689007715065,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 21063
    },
    {
      "epoch": 0.21064,
      "grad_norm": 1.2952466469961974,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 21064
    },
    {
      "epoch": 0.21065,
      "grad_norm": 1.361447178361404,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 21065
    },
    {
      "epoch": 0.21066,
      "grad_norm": 1.2877009617325028,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 21066
    },
    {
      "epoch": 0.21067,
      "grad_norm": 1.4508306716090476,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 21067
    },
    {
      "epoch": 0.21068,
      "grad_norm": 1.404906272655368,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 21068
    },
    {
      "epoch": 0.21069,
      "grad_norm": 1.2566162259120792,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 21069
    },
    {
      "epoch": 0.2107,
      "grad_norm": 1.4371513027733627,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 21070
    },
    {
      "epoch": 0.21071,
      "grad_norm": 1.1741644539260434,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 21071
    },
    {
      "epoch": 0.21072,
      "grad_norm": 1.7495812264694768,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 21072
    },
    {
      "epoch": 0.21073,
      "grad_norm": 1.1914029554954333,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 21073
    },
    {
      "epoch": 0.21074,
      "grad_norm": 1.358388855376881,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 21074
    },
    {
      "epoch": 0.21075,
      "grad_norm": 1.1295297359650347,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 21075
    },
    {
      "epoch": 0.21076,
      "grad_norm": 1.422826953405643,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 21076
    },
    {
      "epoch": 0.21077,
      "grad_norm": 1.2238983295066468,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 21077
    },
    {
      "epoch": 0.21078,
      "grad_norm": 1.4376730583978223,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 21078
    },
    {
      "epoch": 0.21079,
      "grad_norm": 1.312652953458203,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 21079
    },
    {
      "epoch": 0.2108,
      "grad_norm": 1.2881762161594188,
      "learning_rate": 0.003,
      "loss": 4.0189,
      "step": 21080
    },
    {
      "epoch": 0.21081,
      "grad_norm": 1.3811842481459293,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 21081
    },
    {
      "epoch": 0.21082,
      "grad_norm": 1.4010350452207294,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 21082
    },
    {
      "epoch": 0.21083,
      "grad_norm": 1.290332298644118,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 21083
    },
    {
      "epoch": 0.21084,
      "grad_norm": 1.489071854621563,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 21084
    },
    {
      "epoch": 0.21085,
      "grad_norm": 1.5024296642566897,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 21085
    },
    {
      "epoch": 0.21086,
      "grad_norm": 1.4062125961366714,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 21086
    },
    {
      "epoch": 0.21087,
      "grad_norm": 1.4251667457024468,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 21087
    },
    {
      "epoch": 0.21088,
      "grad_norm": 1.572228967046733,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 21088
    },
    {
      "epoch": 0.21089,
      "grad_norm": 1.3883350111110444,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 21089
    },
    {
      "epoch": 0.2109,
      "grad_norm": 1.3950958358737215,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 21090
    },
    {
      "epoch": 0.21091,
      "grad_norm": 1.2048175579643006,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 21091
    },
    {
      "epoch": 0.21092,
      "grad_norm": 1.5234295295873461,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 21092
    },
    {
      "epoch": 0.21093,
      "grad_norm": 1.294754161938382,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 21093
    },
    {
      "epoch": 0.21094,
      "grad_norm": 1.3512006946099722,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 21094
    },
    {
      "epoch": 0.21095,
      "grad_norm": 1.3137093691553654,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 21095
    },
    {
      "epoch": 0.21096,
      "grad_norm": 1.4209343670036005,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 21096
    },
    {
      "epoch": 0.21097,
      "grad_norm": 1.1895555560423718,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 21097
    },
    {
      "epoch": 0.21098,
      "grad_norm": 1.336801940566388,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 21098
    },
    {
      "epoch": 0.21099,
      "grad_norm": 1.2758607849242902,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 21099
    },
    {
      "epoch": 0.211,
      "grad_norm": 1.2117337581677836,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 21100
    },
    {
      "epoch": 0.21101,
      "grad_norm": 1.3525527482392288,
      "learning_rate": 0.003,
      "loss": 4.0173,
      "step": 21101
    },
    {
      "epoch": 0.21102,
      "grad_norm": 1.3040393589972092,
      "learning_rate": 0.003,
      "loss": 4.0328,
      "step": 21102
    },
    {
      "epoch": 0.21103,
      "grad_norm": 1.3424780630363968,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 21103
    },
    {
      "epoch": 0.21104,
      "grad_norm": 1.542667444735479,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 21104
    },
    {
      "epoch": 0.21105,
      "grad_norm": 1.6365698353753115,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 21105
    },
    {
      "epoch": 0.21106,
      "grad_norm": 1.2084286816553886,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 21106
    },
    {
      "epoch": 0.21107,
      "grad_norm": 1.5019987836084807,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 21107
    },
    {
      "epoch": 0.21108,
      "grad_norm": 1.318700513318784,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 21108
    },
    {
      "epoch": 0.21109,
      "grad_norm": 1.3066584923179836,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 21109
    },
    {
      "epoch": 0.2111,
      "grad_norm": 1.4399512148451805,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 21110
    },
    {
      "epoch": 0.21111,
      "grad_norm": 1.1041215885148292,
      "learning_rate": 0.003,
      "loss": 4.0288,
      "step": 21111
    },
    {
      "epoch": 0.21112,
      "grad_norm": 1.501694680843476,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 21112
    },
    {
      "epoch": 0.21113,
      "grad_norm": 1.2601037756571198,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 21113
    },
    {
      "epoch": 0.21114,
      "grad_norm": 1.2271546237973472,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 21114
    },
    {
      "epoch": 0.21115,
      "grad_norm": 1.2376999598976492,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 21115
    },
    {
      "epoch": 0.21116,
      "grad_norm": 1.5430782245530037,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 21116
    },
    {
      "epoch": 0.21117,
      "grad_norm": 1.3787545900202918,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 21117
    },
    {
      "epoch": 0.21118,
      "grad_norm": 1.4281319973902007,
      "learning_rate": 0.003,
      "loss": 4.0172,
      "step": 21118
    },
    {
      "epoch": 0.21119,
      "grad_norm": 1.6838838996168861,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 21119
    },
    {
      "epoch": 0.2112,
      "grad_norm": 1.2215779967116982,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 21120
    },
    {
      "epoch": 0.21121,
      "grad_norm": 1.5199216079850038,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 21121
    },
    {
      "epoch": 0.21122,
      "grad_norm": 1.1448715515495345,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 21122
    },
    {
      "epoch": 0.21123,
      "grad_norm": 1.4014754531269893,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 21123
    },
    {
      "epoch": 0.21124,
      "grad_norm": 1.143267669929633,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 21124
    },
    {
      "epoch": 0.21125,
      "grad_norm": 1.507889372373385,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 21125
    },
    {
      "epoch": 0.21126,
      "grad_norm": 1.1177387032574877,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 21126
    },
    {
      "epoch": 0.21127,
      "grad_norm": 1.4170981959098783,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 21127
    },
    {
      "epoch": 0.21128,
      "grad_norm": 1.193182993088027,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 21128
    },
    {
      "epoch": 0.21129,
      "grad_norm": 1.4457888997085144,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 21129
    },
    {
      "epoch": 0.2113,
      "grad_norm": 1.3240888408523435,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 21130
    },
    {
      "epoch": 0.21131,
      "grad_norm": 1.4805545843347245,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 21131
    },
    {
      "epoch": 0.21132,
      "grad_norm": 1.4871800892653317,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 21132
    },
    {
      "epoch": 0.21133,
      "grad_norm": 1.2991302286864286,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 21133
    },
    {
      "epoch": 0.21134,
      "grad_norm": 1.6445450451190515,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 21134
    },
    {
      "epoch": 0.21135,
      "grad_norm": 1.4721111980365607,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 21135
    },
    {
      "epoch": 0.21136,
      "grad_norm": 1.2722739660811953,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 21136
    },
    {
      "epoch": 0.21137,
      "grad_norm": 1.3982694574879595,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 21137
    },
    {
      "epoch": 0.21138,
      "grad_norm": 1.226479278940514,
      "learning_rate": 0.003,
      "loss": 4.0275,
      "step": 21138
    },
    {
      "epoch": 0.21139,
      "grad_norm": 1.3917173367564466,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 21139
    },
    {
      "epoch": 0.2114,
      "grad_norm": 1.1647298403369704,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 21140
    },
    {
      "epoch": 0.21141,
      "grad_norm": 1.575793714531139,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 21141
    },
    {
      "epoch": 0.21142,
      "grad_norm": 1.199875711689336,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 21142
    },
    {
      "epoch": 0.21143,
      "grad_norm": 1.6414373088741725,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 21143
    },
    {
      "epoch": 0.21144,
      "grad_norm": 1.091308331386974,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 21144
    },
    {
      "epoch": 0.21145,
      "grad_norm": 1.5590872193904934,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 21145
    },
    {
      "epoch": 0.21146,
      "grad_norm": 1.2853562163327288,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 21146
    },
    {
      "epoch": 0.21147,
      "grad_norm": 1.5550976773299796,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 21147
    },
    {
      "epoch": 0.21148,
      "grad_norm": 1.114369420327927,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 21148
    },
    {
      "epoch": 0.21149,
      "grad_norm": 1.3614548969594964,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 21149
    },
    {
      "epoch": 0.2115,
      "grad_norm": 1.2148196019739446,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 21150
    },
    {
      "epoch": 0.21151,
      "grad_norm": 1.340682128882772,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 21151
    },
    {
      "epoch": 0.21152,
      "grad_norm": 1.4000709731369123,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 21152
    },
    {
      "epoch": 0.21153,
      "grad_norm": 1.1974593610511264,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 21153
    },
    {
      "epoch": 0.21154,
      "grad_norm": 1.5536190722535004,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 21154
    },
    {
      "epoch": 0.21155,
      "grad_norm": 1.3833146429262424,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 21155
    },
    {
      "epoch": 0.21156,
      "grad_norm": 1.2868018237459313,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 21156
    },
    {
      "epoch": 0.21157,
      "grad_norm": 1.3479488477762227,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 21157
    },
    {
      "epoch": 0.21158,
      "grad_norm": 1.3625397818967828,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 21158
    },
    {
      "epoch": 0.21159,
      "grad_norm": 1.20914255417048,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 21159
    },
    {
      "epoch": 0.2116,
      "grad_norm": 1.3458094476032658,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 21160
    },
    {
      "epoch": 0.21161,
      "grad_norm": 1.24337456561934,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 21161
    },
    {
      "epoch": 0.21162,
      "grad_norm": 1.2620528232172612,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 21162
    },
    {
      "epoch": 0.21163,
      "grad_norm": 1.3810456235036637,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 21163
    },
    {
      "epoch": 0.21164,
      "grad_norm": 1.511592471855777,
      "learning_rate": 0.003,
      "loss": 4.0861,
      "step": 21164
    },
    {
      "epoch": 0.21165,
      "grad_norm": 1.3431865078237681,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 21165
    },
    {
      "epoch": 0.21166,
      "grad_norm": 1.7412655610560623,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 21166
    },
    {
      "epoch": 0.21167,
      "grad_norm": 1.1109768866361736,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 21167
    },
    {
      "epoch": 0.21168,
      "grad_norm": 1.399082597999424,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 21168
    },
    {
      "epoch": 0.21169,
      "grad_norm": 1.1489789410732114,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 21169
    },
    {
      "epoch": 0.2117,
      "grad_norm": 1.524057385496699,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 21170
    },
    {
      "epoch": 0.21171,
      "grad_norm": 1.2092265332397514,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 21171
    },
    {
      "epoch": 0.21172,
      "grad_norm": 1.3782809880096014,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 21172
    },
    {
      "epoch": 0.21173,
      "grad_norm": 1.415031645517347,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 21173
    },
    {
      "epoch": 0.21174,
      "grad_norm": 1.4764342717559167,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 21174
    },
    {
      "epoch": 0.21175,
      "grad_norm": 1.5080392364169124,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 21175
    },
    {
      "epoch": 0.21176,
      "grad_norm": 1.4883776418571604,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 21176
    },
    {
      "epoch": 0.21177,
      "grad_norm": 1.5612744737054056,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 21177
    },
    {
      "epoch": 0.21178,
      "grad_norm": 1.2024716611041812,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 21178
    },
    {
      "epoch": 0.21179,
      "grad_norm": 1.1880680580697465,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 21179
    },
    {
      "epoch": 0.2118,
      "grad_norm": 1.6787856254774396,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 21180
    },
    {
      "epoch": 0.21181,
      "grad_norm": 1.2348843210740865,
      "learning_rate": 0.003,
      "loss": 4.0299,
      "step": 21181
    },
    {
      "epoch": 0.21182,
      "grad_norm": 1.272340607329311,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 21182
    },
    {
      "epoch": 0.21183,
      "grad_norm": 1.4679072254469465,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 21183
    },
    {
      "epoch": 0.21184,
      "grad_norm": 1.3585140016622648,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 21184
    },
    {
      "epoch": 0.21185,
      "grad_norm": 1.3043316251259256,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 21185
    },
    {
      "epoch": 0.21186,
      "grad_norm": 1.1942951148218348,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 21186
    },
    {
      "epoch": 0.21187,
      "grad_norm": 1.5555996789667779,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 21187
    },
    {
      "epoch": 0.21188,
      "grad_norm": 1.2566766113824286,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 21188
    },
    {
      "epoch": 0.21189,
      "grad_norm": 1.3926383314895503,
      "learning_rate": 0.003,
      "loss": 4.029,
      "step": 21189
    },
    {
      "epoch": 0.2119,
      "grad_norm": 1.2036849778099206,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 21190
    },
    {
      "epoch": 0.21191,
      "grad_norm": 1.3813535243352322,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 21191
    },
    {
      "epoch": 0.21192,
      "grad_norm": 1.2470340700518887,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 21192
    },
    {
      "epoch": 0.21193,
      "grad_norm": 1.2335173857517638,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 21193
    },
    {
      "epoch": 0.21194,
      "grad_norm": 1.4050847539015667,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 21194
    },
    {
      "epoch": 0.21195,
      "grad_norm": 1.405184145664891,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 21195
    },
    {
      "epoch": 0.21196,
      "grad_norm": 1.1916257136074022,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 21196
    },
    {
      "epoch": 0.21197,
      "grad_norm": 1.2863503878206486,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 21197
    },
    {
      "epoch": 0.21198,
      "grad_norm": 1.3920618485759595,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 21198
    },
    {
      "epoch": 0.21199,
      "grad_norm": 1.6393826888044547,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 21199
    },
    {
      "epoch": 0.212,
      "grad_norm": 1.5873450028938483,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 21200
    },
    {
      "epoch": 0.21201,
      "grad_norm": 1.276498523591141,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 21201
    },
    {
      "epoch": 0.21202,
      "grad_norm": 1.2141772863673175,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 21202
    },
    {
      "epoch": 0.21203,
      "grad_norm": 1.7306047496665609,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 21203
    },
    {
      "epoch": 0.21204,
      "grad_norm": 1.2487486082239754,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 21204
    },
    {
      "epoch": 0.21205,
      "grad_norm": 1.712695784117531,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 21205
    },
    {
      "epoch": 0.21206,
      "grad_norm": 1.1323199873225769,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 21206
    },
    {
      "epoch": 0.21207,
      "grad_norm": 1.2652313130632524,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 21207
    },
    {
      "epoch": 0.21208,
      "grad_norm": 1.2512745025751575,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 21208
    },
    {
      "epoch": 0.21209,
      "grad_norm": 1.2466650859199238,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 21209
    },
    {
      "epoch": 0.2121,
      "grad_norm": 1.2620167175109485,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 21210
    },
    {
      "epoch": 0.21211,
      "grad_norm": 1.5212220521490063,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 21211
    },
    {
      "epoch": 0.21212,
      "grad_norm": 1.2635532006517542,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 21212
    },
    {
      "epoch": 0.21213,
      "grad_norm": 1.347810331585149,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 21213
    },
    {
      "epoch": 0.21214,
      "grad_norm": 1.6564875740727845,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 21214
    },
    {
      "epoch": 0.21215,
      "grad_norm": 1.5438927653902956,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 21215
    },
    {
      "epoch": 0.21216,
      "grad_norm": 1.208733121779013,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 21216
    },
    {
      "epoch": 0.21217,
      "grad_norm": 1.2433388919071149,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 21217
    },
    {
      "epoch": 0.21218,
      "grad_norm": 1.4277012029638962,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 21218
    },
    {
      "epoch": 0.21219,
      "grad_norm": 1.2923936162093763,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 21219
    },
    {
      "epoch": 0.2122,
      "grad_norm": 1.1617299223503543,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 21220
    },
    {
      "epoch": 0.21221,
      "grad_norm": 1.7625335309163919,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 21221
    },
    {
      "epoch": 0.21222,
      "grad_norm": 1.0722691251519127,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 21222
    },
    {
      "epoch": 0.21223,
      "grad_norm": 1.523851439821383,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 21223
    },
    {
      "epoch": 0.21224,
      "grad_norm": 1.5241769908828782,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 21224
    },
    {
      "epoch": 0.21225,
      "grad_norm": 1.5763055700460948,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 21225
    },
    {
      "epoch": 0.21226,
      "grad_norm": 1.288345375256502,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 21226
    },
    {
      "epoch": 0.21227,
      "grad_norm": 1.3593170373457841,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 21227
    },
    {
      "epoch": 0.21228,
      "grad_norm": 1.3798720304842758,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 21228
    },
    {
      "epoch": 0.21229,
      "grad_norm": 1.11347139995492,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 21229
    },
    {
      "epoch": 0.2123,
      "grad_norm": 1.4881861926379105,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 21230
    },
    {
      "epoch": 0.21231,
      "grad_norm": 1.222191148663261,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 21231
    },
    {
      "epoch": 0.21232,
      "grad_norm": 1.4598446473774203,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 21232
    },
    {
      "epoch": 0.21233,
      "grad_norm": 1.3035913178757261,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 21233
    },
    {
      "epoch": 0.21234,
      "grad_norm": 1.4240272674012957,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 21234
    },
    {
      "epoch": 0.21235,
      "grad_norm": 1.7053319580459514,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 21235
    },
    {
      "epoch": 0.21236,
      "grad_norm": 1.445897432070719,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 21236
    },
    {
      "epoch": 0.21237,
      "grad_norm": 1.3817227194504706,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 21237
    },
    {
      "epoch": 0.21238,
      "grad_norm": 1.2676941447795234,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 21238
    },
    {
      "epoch": 0.21239,
      "grad_norm": 1.5641612851516646,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 21239
    },
    {
      "epoch": 0.2124,
      "grad_norm": 1.1241931151835491,
      "learning_rate": 0.003,
      "loss": 4.029,
      "step": 21240
    },
    {
      "epoch": 0.21241,
      "grad_norm": 1.4868409328389938,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 21241
    },
    {
      "epoch": 0.21242,
      "grad_norm": 1.2405669154949257,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 21242
    },
    {
      "epoch": 0.21243,
      "grad_norm": 1.2224834326951193,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 21243
    },
    {
      "epoch": 0.21244,
      "grad_norm": 1.3189689925487944,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 21244
    },
    {
      "epoch": 0.21245,
      "grad_norm": 1.347220053289232,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 21245
    },
    {
      "epoch": 0.21246,
      "grad_norm": 1.2655707700006256,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 21246
    },
    {
      "epoch": 0.21247,
      "grad_norm": 1.362651105931732,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 21247
    },
    {
      "epoch": 0.21248,
      "grad_norm": 1.4785751933994802,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 21248
    },
    {
      "epoch": 0.21249,
      "grad_norm": 1.2504349836530144,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 21249
    },
    {
      "epoch": 0.2125,
      "grad_norm": 1.2990606380438254,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 21250
    },
    {
      "epoch": 0.21251,
      "grad_norm": 1.42497627382582,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 21251
    },
    {
      "epoch": 0.21252,
      "grad_norm": 1.1735585685930139,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 21252
    },
    {
      "epoch": 0.21253,
      "grad_norm": 1.795106850549211,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 21253
    },
    {
      "epoch": 0.21254,
      "grad_norm": 1.0061555841718746,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 21254
    },
    {
      "epoch": 0.21255,
      "grad_norm": 1.4122369561351449,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 21255
    },
    {
      "epoch": 0.21256,
      "grad_norm": 1.4461580757000376,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 21256
    },
    {
      "epoch": 0.21257,
      "grad_norm": 1.4284828541332442,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 21257
    },
    {
      "epoch": 0.21258,
      "grad_norm": 1.2986806596599783,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 21258
    },
    {
      "epoch": 0.21259,
      "grad_norm": 1.6524854377190041,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 21259
    },
    {
      "epoch": 0.2126,
      "grad_norm": 0.9870173335163619,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 21260
    },
    {
      "epoch": 0.21261,
      "grad_norm": 1.5495893972134074,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 21261
    },
    {
      "epoch": 0.21262,
      "grad_norm": 1.126423706366596,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 21262
    },
    {
      "epoch": 0.21263,
      "grad_norm": 1.3734463560813661,
      "learning_rate": 0.003,
      "loss": 4.0166,
      "step": 21263
    },
    {
      "epoch": 0.21264,
      "grad_norm": 1.176126541735446,
      "learning_rate": 0.003,
      "loss": 4.0248,
      "step": 21264
    },
    {
      "epoch": 0.21265,
      "grad_norm": 1.370153879548395,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 21265
    },
    {
      "epoch": 0.21266,
      "grad_norm": 1.3532518679834689,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 21266
    },
    {
      "epoch": 0.21267,
      "grad_norm": 1.6946670175674308,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 21267
    },
    {
      "epoch": 0.21268,
      "grad_norm": 1.0704729251671854,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 21268
    },
    {
      "epoch": 0.21269,
      "grad_norm": 1.4596383231421428,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 21269
    },
    {
      "epoch": 0.2127,
      "grad_norm": 1.1808133393608502,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 21270
    },
    {
      "epoch": 0.21271,
      "grad_norm": 1.4127453609322973,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 21271
    },
    {
      "epoch": 0.21272,
      "grad_norm": 1.2825462550635216,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 21272
    },
    {
      "epoch": 0.21273,
      "grad_norm": 1.4332865788298836,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 21273
    },
    {
      "epoch": 0.21274,
      "grad_norm": 1.2555266101784204,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 21274
    },
    {
      "epoch": 0.21275,
      "grad_norm": 1.491116264139151,
      "learning_rate": 0.003,
      "loss": 4.0179,
      "step": 21275
    },
    {
      "epoch": 0.21276,
      "grad_norm": 1.357691091796868,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 21276
    },
    {
      "epoch": 0.21277,
      "grad_norm": 1.5137001660115919,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 21277
    },
    {
      "epoch": 0.21278,
      "grad_norm": 1.1125889653319778,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 21278
    },
    {
      "epoch": 0.21279,
      "grad_norm": 1.3304734539024916,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 21279
    },
    {
      "epoch": 0.2128,
      "grad_norm": 1.4679732364278968,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 21280
    },
    {
      "epoch": 0.21281,
      "grad_norm": 1.2033274494571464,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 21281
    },
    {
      "epoch": 0.21282,
      "grad_norm": 1.2190823850731825,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 21282
    },
    {
      "epoch": 0.21283,
      "grad_norm": 1.4288583356987408,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 21283
    },
    {
      "epoch": 0.21284,
      "grad_norm": 1.31602755126889,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 21284
    },
    {
      "epoch": 0.21285,
      "grad_norm": 1.2275327062705983,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 21285
    },
    {
      "epoch": 0.21286,
      "grad_norm": 1.5414860872345424,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 21286
    },
    {
      "epoch": 0.21287,
      "grad_norm": 1.1120019871745512,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 21287
    },
    {
      "epoch": 0.21288,
      "grad_norm": 1.4831050923685816,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 21288
    },
    {
      "epoch": 0.21289,
      "grad_norm": 1.7494424733737561,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 21289
    },
    {
      "epoch": 0.2129,
      "grad_norm": 1.342926814624509,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 21290
    },
    {
      "epoch": 0.21291,
      "grad_norm": 1.598270038306869,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 21291
    },
    {
      "epoch": 0.21292,
      "grad_norm": 1.5961776628467343,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 21292
    },
    {
      "epoch": 0.21293,
      "grad_norm": 0.9433010852092817,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 21293
    },
    {
      "epoch": 0.21294,
      "grad_norm": 1.3925449914517969,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 21294
    },
    {
      "epoch": 0.21295,
      "grad_norm": 1.3515145641464523,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 21295
    },
    {
      "epoch": 0.21296,
      "grad_norm": 1.4166695760130046,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 21296
    },
    {
      "epoch": 0.21297,
      "grad_norm": 1.3591828142618334,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 21297
    },
    {
      "epoch": 0.21298,
      "grad_norm": 1.5153716123119334,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 21298
    },
    {
      "epoch": 0.21299,
      "grad_norm": 1.6452501562883108,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 21299
    },
    {
      "epoch": 0.213,
      "grad_norm": 1.196597809663666,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 21300
    },
    {
      "epoch": 0.21301,
      "grad_norm": 1.4346809471876099,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 21301
    },
    {
      "epoch": 0.21302,
      "grad_norm": 1.2582888762436075,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 21302
    },
    {
      "epoch": 0.21303,
      "grad_norm": 1.583365216904255,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 21303
    },
    {
      "epoch": 0.21304,
      "grad_norm": 1.2243987322696352,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 21304
    },
    {
      "epoch": 0.21305,
      "grad_norm": 1.3782757485019397,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 21305
    },
    {
      "epoch": 0.21306,
      "grad_norm": 1.3712449684226944,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 21306
    },
    {
      "epoch": 0.21307,
      "grad_norm": 1.452408027239848,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 21307
    },
    {
      "epoch": 0.21308,
      "grad_norm": 1.227024123708545,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 21308
    },
    {
      "epoch": 0.21309,
      "grad_norm": 1.2127778579692743,
      "learning_rate": 0.003,
      "loss": 4.0105,
      "step": 21309
    },
    {
      "epoch": 0.2131,
      "grad_norm": 1.3640909424814787,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 21310
    },
    {
      "epoch": 0.21311,
      "grad_norm": 1.4625961453018894,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 21311
    },
    {
      "epoch": 0.21312,
      "grad_norm": 1.4381037832524861,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 21312
    },
    {
      "epoch": 0.21313,
      "grad_norm": 1.1144720293083312,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 21313
    },
    {
      "epoch": 0.21314,
      "grad_norm": 1.6391039384828037,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 21314
    },
    {
      "epoch": 0.21315,
      "grad_norm": 1.249524180399814,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 21315
    },
    {
      "epoch": 0.21316,
      "grad_norm": 1.306337595950138,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 21316
    },
    {
      "epoch": 0.21317,
      "grad_norm": 1.3883837054487016,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 21317
    },
    {
      "epoch": 0.21318,
      "grad_norm": 1.6026302139424504,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 21318
    },
    {
      "epoch": 0.21319,
      "grad_norm": 1.2604444747762127,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 21319
    },
    {
      "epoch": 0.2132,
      "grad_norm": 1.3547506649531902,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 21320
    },
    {
      "epoch": 0.21321,
      "grad_norm": 1.5400351682445699,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 21321
    },
    {
      "epoch": 0.21322,
      "grad_norm": 1.4849351496306467,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 21322
    },
    {
      "epoch": 0.21323,
      "grad_norm": 1.2063234257034268,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 21323
    },
    {
      "epoch": 0.21324,
      "grad_norm": 1.5292223315443574,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 21324
    },
    {
      "epoch": 0.21325,
      "grad_norm": 1.151039941590078,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 21325
    },
    {
      "epoch": 0.21326,
      "grad_norm": 1.3318800995328697,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 21326
    },
    {
      "epoch": 0.21327,
      "grad_norm": 1.2652305263453445,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 21327
    },
    {
      "epoch": 0.21328,
      "grad_norm": 1.3515081434361438,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 21328
    },
    {
      "epoch": 0.21329,
      "grad_norm": 1.2875379433950092,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 21329
    },
    {
      "epoch": 0.2133,
      "grad_norm": 1.3972616385126326,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 21330
    },
    {
      "epoch": 0.21331,
      "grad_norm": 1.4816983405239224,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 21331
    },
    {
      "epoch": 0.21332,
      "grad_norm": 1.503467852555335,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 21332
    },
    {
      "epoch": 0.21333,
      "grad_norm": 1.287556683534024,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 21333
    },
    {
      "epoch": 0.21334,
      "grad_norm": 1.293434698898435,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 21334
    },
    {
      "epoch": 0.21335,
      "grad_norm": 1.2973040929964068,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 21335
    },
    {
      "epoch": 0.21336,
      "grad_norm": 1.4591818538831844,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 21336
    },
    {
      "epoch": 0.21337,
      "grad_norm": 1.31338047027029,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 21337
    },
    {
      "epoch": 0.21338,
      "grad_norm": 1.2076429007725722,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 21338
    },
    {
      "epoch": 0.21339,
      "grad_norm": 1.263751954787633,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 21339
    },
    {
      "epoch": 0.2134,
      "grad_norm": 1.2731326941221517,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 21340
    },
    {
      "epoch": 0.21341,
      "grad_norm": 1.4210042790778448,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 21341
    },
    {
      "epoch": 0.21342,
      "grad_norm": 1.3568925479823957,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 21342
    },
    {
      "epoch": 0.21343,
      "grad_norm": 1.5384252079790888,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 21343
    },
    {
      "epoch": 0.21344,
      "grad_norm": 1.4478375676361275,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 21344
    },
    {
      "epoch": 0.21345,
      "grad_norm": 1.3680354906520085,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 21345
    },
    {
      "epoch": 0.21346,
      "grad_norm": 1.3890887098322335,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 21346
    },
    {
      "epoch": 0.21347,
      "grad_norm": 1.1193721005748065,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 21347
    },
    {
      "epoch": 0.21348,
      "grad_norm": 1.5275977448430993,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 21348
    },
    {
      "epoch": 0.21349,
      "grad_norm": 1.3959001019528146,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 21349
    },
    {
      "epoch": 0.2135,
      "grad_norm": 1.490204399602071,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 21350
    },
    {
      "epoch": 0.21351,
      "grad_norm": 1.1499473326805703,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 21351
    },
    {
      "epoch": 0.21352,
      "grad_norm": 1.727939275779018,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 21352
    },
    {
      "epoch": 0.21353,
      "grad_norm": 1.072972726456787,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 21353
    },
    {
      "epoch": 0.21354,
      "grad_norm": 1.483607518560974,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 21354
    },
    {
      "epoch": 0.21355,
      "grad_norm": 1.562193661576263,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 21355
    },
    {
      "epoch": 0.21356,
      "grad_norm": 1.5136521928411162,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 21356
    },
    {
      "epoch": 0.21357,
      "grad_norm": 1.3266924162009406,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 21357
    },
    {
      "epoch": 0.21358,
      "grad_norm": 1.5459850487255908,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 21358
    },
    {
      "epoch": 0.21359,
      "grad_norm": 1.306340308286527,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 21359
    },
    {
      "epoch": 0.2136,
      "grad_norm": 1.4437227572721356,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 21360
    },
    {
      "epoch": 0.21361,
      "grad_norm": 1.1312331537840155,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 21361
    },
    {
      "epoch": 0.21362,
      "grad_norm": 1.4498115212610185,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 21362
    },
    {
      "epoch": 0.21363,
      "grad_norm": 1.2003047711896486,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 21363
    },
    {
      "epoch": 0.21364,
      "grad_norm": 1.330082599913098,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 21364
    },
    {
      "epoch": 0.21365,
      "grad_norm": 1.5471781546909327,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 21365
    },
    {
      "epoch": 0.21366,
      "grad_norm": 1.4316421630210903,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 21366
    },
    {
      "epoch": 0.21367,
      "grad_norm": 1.594803933302907,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 21367
    },
    {
      "epoch": 0.21368,
      "grad_norm": 1.1617312218503295,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 21368
    },
    {
      "epoch": 0.21369,
      "grad_norm": 1.2293064917378977,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 21369
    },
    {
      "epoch": 0.2137,
      "grad_norm": 1.418555917149502,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 21370
    },
    {
      "epoch": 0.21371,
      "grad_norm": 1.3902348667270346,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 21371
    },
    {
      "epoch": 0.21372,
      "grad_norm": 1.189623983536023,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 21372
    },
    {
      "epoch": 0.21373,
      "grad_norm": 1.5618531870219965,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 21373
    },
    {
      "epoch": 0.21374,
      "grad_norm": 1.065036404388025,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 21374
    },
    {
      "epoch": 0.21375,
      "grad_norm": 1.283965896979844,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 21375
    },
    {
      "epoch": 0.21376,
      "grad_norm": 1.261048313827542,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 21376
    },
    {
      "epoch": 0.21377,
      "grad_norm": 1.4229495898524944,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 21377
    },
    {
      "epoch": 0.21378,
      "grad_norm": 1.3428569523264362,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 21378
    },
    {
      "epoch": 0.21379,
      "grad_norm": 1.5985336646598158,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 21379
    },
    {
      "epoch": 0.2138,
      "grad_norm": 1.5818368985579856,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 21380
    },
    {
      "epoch": 0.21381,
      "grad_norm": 1.4472336461809965,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 21381
    },
    {
      "epoch": 0.21382,
      "grad_norm": 1.2563415448416158,
      "learning_rate": 0.003,
      "loss": 4.0181,
      "step": 21382
    },
    {
      "epoch": 0.21383,
      "grad_norm": 1.4436033902315786,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 21383
    },
    {
      "epoch": 0.21384,
      "grad_norm": 1.2441850730232435,
      "learning_rate": 0.003,
      "loss": 4.0295,
      "step": 21384
    },
    {
      "epoch": 0.21385,
      "grad_norm": 1.5092420928285513,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 21385
    },
    {
      "epoch": 0.21386,
      "grad_norm": 1.1510500409342292,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 21386
    },
    {
      "epoch": 0.21387,
      "grad_norm": 1.3560229720526926,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 21387
    },
    {
      "epoch": 0.21388,
      "grad_norm": 1.3342466552744916,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 21388
    },
    {
      "epoch": 0.21389,
      "grad_norm": 1.3020543951603012,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 21389
    },
    {
      "epoch": 0.2139,
      "grad_norm": 1.608568950125023,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 21390
    },
    {
      "epoch": 0.21391,
      "grad_norm": 1.1353591119174902,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 21391
    },
    {
      "epoch": 0.21392,
      "grad_norm": 1.6950414767063793,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 21392
    },
    {
      "epoch": 0.21393,
      "grad_norm": 1.2387184056870462,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 21393
    },
    {
      "epoch": 0.21394,
      "grad_norm": 1.3613953563392753,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 21394
    },
    {
      "epoch": 0.21395,
      "grad_norm": 1.629304745379985,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 21395
    },
    {
      "epoch": 0.21396,
      "grad_norm": 1.0572390211011273,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 21396
    },
    {
      "epoch": 0.21397,
      "grad_norm": 1.3300832783882734,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 21397
    },
    {
      "epoch": 0.21398,
      "grad_norm": 1.3542246431342029,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 21398
    },
    {
      "epoch": 0.21399,
      "grad_norm": 1.39666540188732,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 21399
    },
    {
      "epoch": 0.214,
      "grad_norm": 1.4361003719188568,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 21400
    },
    {
      "epoch": 0.21401,
      "grad_norm": 1.300783711710657,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 21401
    },
    {
      "epoch": 0.21402,
      "grad_norm": 1.290377078934423,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 21402
    },
    {
      "epoch": 0.21403,
      "grad_norm": 1.6750658808471093,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 21403
    },
    {
      "epoch": 0.21404,
      "grad_norm": 0.9435626847165859,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 21404
    },
    {
      "epoch": 0.21405,
      "grad_norm": 1.4968566182794512,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 21405
    },
    {
      "epoch": 0.21406,
      "grad_norm": 1.2175482207573307,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 21406
    },
    {
      "epoch": 0.21407,
      "grad_norm": 1.577026013030714,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 21407
    },
    {
      "epoch": 0.21408,
      "grad_norm": 1.59911020245519,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 21408
    },
    {
      "epoch": 0.21409,
      "grad_norm": 1.4170936917729833,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 21409
    },
    {
      "epoch": 0.2141,
      "grad_norm": 1.3962209202302078,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 21410
    },
    {
      "epoch": 0.21411,
      "grad_norm": 1.3395683786301085,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 21411
    },
    {
      "epoch": 0.21412,
      "grad_norm": 1.4282438033940026,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 21412
    },
    {
      "epoch": 0.21413,
      "grad_norm": 1.238506016974487,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 21413
    },
    {
      "epoch": 0.21414,
      "grad_norm": 1.3051265514631454,
      "learning_rate": 0.003,
      "loss": 4.0163,
      "step": 21414
    },
    {
      "epoch": 0.21415,
      "grad_norm": 1.181389922210956,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 21415
    },
    {
      "epoch": 0.21416,
      "grad_norm": 1.4710337042861845,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 21416
    },
    {
      "epoch": 0.21417,
      "grad_norm": 1.4680527309778317,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 21417
    },
    {
      "epoch": 0.21418,
      "grad_norm": 1.989767235792791,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 21418
    },
    {
      "epoch": 0.21419,
      "grad_norm": 0.9795602781230998,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 21419
    },
    {
      "epoch": 0.2142,
      "grad_norm": 1.641796412983382,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 21420
    },
    {
      "epoch": 0.21421,
      "grad_norm": 1.1101786702473517,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 21421
    },
    {
      "epoch": 0.21422,
      "grad_norm": 1.2490099939055685,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 21422
    },
    {
      "epoch": 0.21423,
      "grad_norm": 1.4418455337047351,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 21423
    },
    {
      "epoch": 0.21424,
      "grad_norm": 1.3036484430249282,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 21424
    },
    {
      "epoch": 0.21425,
      "grad_norm": 1.26627535365833,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 21425
    },
    {
      "epoch": 0.21426,
      "grad_norm": 1.595267433637214,
      "learning_rate": 0.003,
      "loss": 4.0822,
      "step": 21426
    },
    {
      "epoch": 0.21427,
      "grad_norm": 1.1415584876711913,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 21427
    },
    {
      "epoch": 0.21428,
      "grad_norm": 1.553396948370729,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 21428
    },
    {
      "epoch": 0.21429,
      "grad_norm": 1.3669329270555002,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 21429
    },
    {
      "epoch": 0.2143,
      "grad_norm": 1.545399754488947,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 21430
    },
    {
      "epoch": 0.21431,
      "grad_norm": 1.3305307525785282,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 21431
    },
    {
      "epoch": 0.21432,
      "grad_norm": 1.340621650832546,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 21432
    },
    {
      "epoch": 0.21433,
      "grad_norm": 1.3656141783233655,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 21433
    },
    {
      "epoch": 0.21434,
      "grad_norm": 1.350982127361377,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 21434
    },
    {
      "epoch": 0.21435,
      "grad_norm": 1.602265634975586,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 21435
    },
    {
      "epoch": 0.21436,
      "grad_norm": 1.2430334534062013,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 21436
    },
    {
      "epoch": 0.21437,
      "grad_norm": 1.5552661004651134,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 21437
    },
    {
      "epoch": 0.21438,
      "grad_norm": 1.4210844535959881,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 21438
    },
    {
      "epoch": 0.21439,
      "grad_norm": 1.3154138728796594,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 21439
    },
    {
      "epoch": 0.2144,
      "grad_norm": 1.2269413756659808,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 21440
    },
    {
      "epoch": 0.21441,
      "grad_norm": 1.4585751482741935,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 21441
    },
    {
      "epoch": 0.21442,
      "grad_norm": 1.1810732345864168,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 21442
    },
    {
      "epoch": 0.21443,
      "grad_norm": 1.5843736873721943,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 21443
    },
    {
      "epoch": 0.21444,
      "grad_norm": 1.1867732253924455,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 21444
    },
    {
      "epoch": 0.21445,
      "grad_norm": 1.5217034118480697,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 21445
    },
    {
      "epoch": 0.21446,
      "grad_norm": 1.366518802802389,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 21446
    },
    {
      "epoch": 0.21447,
      "grad_norm": 1.2987407839800102,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 21447
    },
    {
      "epoch": 0.21448,
      "grad_norm": 1.4846561697401763,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 21448
    },
    {
      "epoch": 0.21449,
      "grad_norm": 1.4194658593227094,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 21449
    },
    {
      "epoch": 0.2145,
      "grad_norm": 1.2720148168821652,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 21450
    },
    {
      "epoch": 0.21451,
      "grad_norm": 1.4812306933833717,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 21451
    },
    {
      "epoch": 0.21452,
      "grad_norm": 1.2221215107688335,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 21452
    },
    {
      "epoch": 0.21453,
      "grad_norm": 1.4207095245756889,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 21453
    },
    {
      "epoch": 0.21454,
      "grad_norm": 1.2603927311534966,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 21454
    },
    {
      "epoch": 0.21455,
      "grad_norm": 1.570954415430875,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 21455
    },
    {
      "epoch": 0.21456,
      "grad_norm": 1.3941117180325808,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 21456
    },
    {
      "epoch": 0.21457,
      "grad_norm": 1.2417437923049768,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 21457
    },
    {
      "epoch": 0.21458,
      "grad_norm": 1.7488725029412966,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 21458
    },
    {
      "epoch": 0.21459,
      "grad_norm": 1.2214428170658385,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 21459
    },
    {
      "epoch": 0.2146,
      "grad_norm": 1.2970704147224645,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 21460
    },
    {
      "epoch": 0.21461,
      "grad_norm": 1.2203965175276164,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 21461
    },
    {
      "epoch": 0.21462,
      "grad_norm": 1.50131381958523,
      "learning_rate": 0.003,
      "loss": 4.0876,
      "step": 21462
    },
    {
      "epoch": 0.21463,
      "grad_norm": 1.3875039130467537,
      "learning_rate": 0.003,
      "loss": 4.088,
      "step": 21463
    },
    {
      "epoch": 0.21464,
      "grad_norm": 1.3995200427254,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 21464
    },
    {
      "epoch": 0.21465,
      "grad_norm": 1.1731856716385536,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 21465
    },
    {
      "epoch": 0.21466,
      "grad_norm": 1.3101837349274383,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 21466
    },
    {
      "epoch": 0.21467,
      "grad_norm": 1.3126630773851762,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 21467
    },
    {
      "epoch": 0.21468,
      "grad_norm": 1.4889895647025362,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 21468
    },
    {
      "epoch": 0.21469,
      "grad_norm": 1.3248932495708168,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 21469
    },
    {
      "epoch": 0.2147,
      "grad_norm": 1.74651213400279,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 21470
    },
    {
      "epoch": 0.21471,
      "grad_norm": 1.1062616432931969,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 21471
    },
    {
      "epoch": 0.21472,
      "grad_norm": 1.3867854769699446,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 21472
    },
    {
      "epoch": 0.21473,
      "grad_norm": 1.3543784557814584,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 21473
    },
    {
      "epoch": 0.21474,
      "grad_norm": 1.323656914816879,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 21474
    },
    {
      "epoch": 0.21475,
      "grad_norm": 1.4335241743346856,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 21475
    },
    {
      "epoch": 0.21476,
      "grad_norm": 1.4966983163562062,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 21476
    },
    {
      "epoch": 0.21477,
      "grad_norm": 1.341149323909709,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 21477
    },
    {
      "epoch": 0.21478,
      "grad_norm": 1.4698581882393453,
      "learning_rate": 0.003,
      "loss": 4.0253,
      "step": 21478
    },
    {
      "epoch": 0.21479,
      "grad_norm": 1.2797141996345442,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 21479
    },
    {
      "epoch": 0.2148,
      "grad_norm": 1.3373313242855844,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 21480
    },
    {
      "epoch": 0.21481,
      "grad_norm": 1.2637458978020324,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 21481
    },
    {
      "epoch": 0.21482,
      "grad_norm": 1.2091973333952728,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 21482
    },
    {
      "epoch": 0.21483,
      "grad_norm": 1.2198409255112599,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 21483
    },
    {
      "epoch": 0.21484,
      "grad_norm": 1.5400666985787543,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 21484
    },
    {
      "epoch": 0.21485,
      "grad_norm": 1.2893391802163316,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 21485
    },
    {
      "epoch": 0.21486,
      "grad_norm": 1.607570309213407,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 21486
    },
    {
      "epoch": 0.21487,
      "grad_norm": 1.2989691916592647,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 21487
    },
    {
      "epoch": 0.21488,
      "grad_norm": 1.5647644877402134,
      "learning_rate": 0.003,
      "loss": 4.0943,
      "step": 21488
    },
    {
      "epoch": 0.21489,
      "grad_norm": 1.1874910266671246,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 21489
    },
    {
      "epoch": 0.2149,
      "grad_norm": 1.3350334919767148,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 21490
    },
    {
      "epoch": 0.21491,
      "grad_norm": 1.4462992266957824,
      "learning_rate": 0.003,
      "loss": 4.1005,
      "step": 21491
    },
    {
      "epoch": 0.21492,
      "grad_norm": 1.2328030980297424,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 21492
    },
    {
      "epoch": 0.21493,
      "grad_norm": 1.4937031281531943,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 21493
    },
    {
      "epoch": 0.21494,
      "grad_norm": 1.3037103845090852,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 21494
    },
    {
      "epoch": 0.21495,
      "grad_norm": 1.5194341136682965,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 21495
    },
    {
      "epoch": 0.21496,
      "grad_norm": 1.1070218887949796,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 21496
    },
    {
      "epoch": 0.21497,
      "grad_norm": 1.6035447800632763,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 21497
    },
    {
      "epoch": 0.21498,
      "grad_norm": 1.659426705088667,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 21498
    },
    {
      "epoch": 0.21499,
      "grad_norm": 1.2973210627167238,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 21499
    },
    {
      "epoch": 0.215,
      "grad_norm": 1.1729428250012608,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 21500
    },
    {
      "epoch": 0.21501,
      "grad_norm": 1.4944754198342163,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 21501
    },
    {
      "epoch": 0.21502,
      "grad_norm": 1.183193279992519,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 21502
    },
    {
      "epoch": 0.21503,
      "grad_norm": 1.4494391545255134,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 21503
    },
    {
      "epoch": 0.21504,
      "grad_norm": 1.2729605463634632,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 21504
    },
    {
      "epoch": 0.21505,
      "grad_norm": 1.5520676393873234,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 21505
    },
    {
      "epoch": 0.21506,
      "grad_norm": 1.1765614188953981,
      "learning_rate": 0.003,
      "loss": 4.0216,
      "step": 21506
    },
    {
      "epoch": 0.21507,
      "grad_norm": 1.4259282001743225,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 21507
    },
    {
      "epoch": 0.21508,
      "grad_norm": 1.3792415358934287,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 21508
    },
    {
      "epoch": 0.21509,
      "grad_norm": 1.3407684710741041,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 21509
    },
    {
      "epoch": 0.2151,
      "grad_norm": 1.486477340588729,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 21510
    },
    {
      "epoch": 0.21511,
      "grad_norm": 1.206644971654592,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 21511
    },
    {
      "epoch": 0.21512,
      "grad_norm": 1.3279590824284087,
      "learning_rate": 0.003,
      "loss": 4.0143,
      "step": 21512
    },
    {
      "epoch": 0.21513,
      "grad_norm": 1.4300521784318503,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 21513
    },
    {
      "epoch": 0.21514,
      "grad_norm": 1.5089884939791547,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 21514
    },
    {
      "epoch": 0.21515,
      "grad_norm": 1.2972049813786273,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 21515
    },
    {
      "epoch": 0.21516,
      "grad_norm": 1.4342025494136799,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 21516
    },
    {
      "epoch": 0.21517,
      "grad_norm": 1.3277788729791116,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 21517
    },
    {
      "epoch": 0.21518,
      "grad_norm": 1.473741890578428,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 21518
    },
    {
      "epoch": 0.21519,
      "grad_norm": 1.3307439083792205,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 21519
    },
    {
      "epoch": 0.2152,
      "grad_norm": 1.345427902165864,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 21520
    },
    {
      "epoch": 0.21521,
      "grad_norm": 1.357117971587018,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 21521
    },
    {
      "epoch": 0.21522,
      "grad_norm": 1.377549963602921,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 21522
    },
    {
      "epoch": 0.21523,
      "grad_norm": 1.5544818286222037,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 21523
    },
    {
      "epoch": 0.21524,
      "grad_norm": 1.1835800431228938,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 21524
    },
    {
      "epoch": 0.21525,
      "grad_norm": 1.4275006507812762,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 21525
    },
    {
      "epoch": 0.21526,
      "grad_norm": 1.2213527444171892,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 21526
    },
    {
      "epoch": 0.21527,
      "grad_norm": 1.372827038406382,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 21527
    },
    {
      "epoch": 0.21528,
      "grad_norm": 1.3280158379319613,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 21528
    },
    {
      "epoch": 0.21529,
      "grad_norm": 1.4286487404610395,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 21529
    },
    {
      "epoch": 0.2153,
      "grad_norm": 1.3629753919913346,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 21530
    },
    {
      "epoch": 0.21531,
      "grad_norm": 1.406135201149657,
      "learning_rate": 0.003,
      "loss": 4.0198,
      "step": 21531
    },
    {
      "epoch": 0.21532,
      "grad_norm": 1.1678754441716601,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 21532
    },
    {
      "epoch": 0.21533,
      "grad_norm": 1.6796987843706,
      "learning_rate": 0.003,
      "loss": 4.0259,
      "step": 21533
    },
    {
      "epoch": 0.21534,
      "grad_norm": 1.1519129898138445,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 21534
    },
    {
      "epoch": 0.21535,
      "grad_norm": 1.4082012232589747,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 21535
    },
    {
      "epoch": 0.21536,
      "grad_norm": 1.2124778273475414,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 21536
    },
    {
      "epoch": 0.21537,
      "grad_norm": 1.4000489102837348,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 21537
    },
    {
      "epoch": 0.21538,
      "grad_norm": 1.3097754581549077,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 21538
    },
    {
      "epoch": 0.21539,
      "grad_norm": 1.5649742201426702,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 21539
    },
    {
      "epoch": 0.2154,
      "grad_norm": 1.5444317447677847,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 21540
    },
    {
      "epoch": 0.21541,
      "grad_norm": 1.40103127681567,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 21541
    },
    {
      "epoch": 0.21542,
      "grad_norm": 1.3266392336491186,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 21542
    },
    {
      "epoch": 0.21543,
      "grad_norm": 1.5192252571422,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 21543
    },
    {
      "epoch": 0.21544,
      "grad_norm": 1.1035412439757208,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 21544
    },
    {
      "epoch": 0.21545,
      "grad_norm": 1.3680108284675236,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 21545
    },
    {
      "epoch": 0.21546,
      "grad_norm": 1.2214251694454756,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 21546
    },
    {
      "epoch": 0.21547,
      "grad_norm": 1.1026129668183156,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 21547
    },
    {
      "epoch": 0.21548,
      "grad_norm": 1.5406280041130547,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 21548
    },
    {
      "epoch": 0.21549,
      "grad_norm": 1.1287473590446058,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 21549
    },
    {
      "epoch": 0.2155,
      "grad_norm": 1.4497523057720862,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 21550
    },
    {
      "epoch": 0.21551,
      "grad_norm": 1.234154847974657,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 21551
    },
    {
      "epoch": 0.21552,
      "grad_norm": 1.3648088536447298,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 21552
    },
    {
      "epoch": 0.21553,
      "grad_norm": 1.452237570251351,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 21553
    },
    {
      "epoch": 0.21554,
      "grad_norm": 1.3807680563664566,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 21554
    },
    {
      "epoch": 0.21555,
      "grad_norm": 1.3603024336725196,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 21555
    },
    {
      "epoch": 0.21556,
      "grad_norm": 1.2379192286946088,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 21556
    },
    {
      "epoch": 0.21557,
      "grad_norm": 1.3547139405076842,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 21557
    },
    {
      "epoch": 0.21558,
      "grad_norm": 1.309257262851389,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 21558
    },
    {
      "epoch": 0.21559,
      "grad_norm": 1.3329537015540636,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 21559
    },
    {
      "epoch": 0.2156,
      "grad_norm": 1.6293792640850666,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 21560
    },
    {
      "epoch": 0.21561,
      "grad_norm": 1.2600053610320905,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 21561
    },
    {
      "epoch": 0.21562,
      "grad_norm": 1.419231945908033,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 21562
    },
    {
      "epoch": 0.21563,
      "grad_norm": 1.3226862447354382,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 21563
    },
    {
      "epoch": 0.21564,
      "grad_norm": 1.339565374359852,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 21564
    },
    {
      "epoch": 0.21565,
      "grad_norm": 1.5140768921413992,
      "learning_rate": 0.003,
      "loss": 4.0923,
      "step": 21565
    },
    {
      "epoch": 0.21566,
      "grad_norm": 1.3598017958323683,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 21566
    },
    {
      "epoch": 0.21567,
      "grad_norm": 1.5207335031594,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 21567
    },
    {
      "epoch": 0.21568,
      "grad_norm": 1.3713189965898323,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 21568
    },
    {
      "epoch": 0.21569,
      "grad_norm": 1.490099909551732,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 21569
    },
    {
      "epoch": 0.2157,
      "grad_norm": 1.2359100081394938,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 21570
    },
    {
      "epoch": 0.21571,
      "grad_norm": 1.593856573190763,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 21571
    },
    {
      "epoch": 0.21572,
      "grad_norm": 1.3689124722521349,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 21572
    },
    {
      "epoch": 0.21573,
      "grad_norm": 1.4359420290267144,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 21573
    },
    {
      "epoch": 0.21574,
      "grad_norm": 1.435224559342419,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 21574
    },
    {
      "epoch": 0.21575,
      "grad_norm": 1.3236478364948177,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 21575
    },
    {
      "epoch": 0.21576,
      "grad_norm": 1.3549316145333958,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 21576
    },
    {
      "epoch": 0.21577,
      "grad_norm": 1.1280127914712705,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 21577
    },
    {
      "epoch": 0.21578,
      "grad_norm": 1.4183644314880084,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 21578
    },
    {
      "epoch": 0.21579,
      "grad_norm": 1.3812602074814249,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 21579
    },
    {
      "epoch": 0.2158,
      "grad_norm": 1.3967917065919164,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 21580
    },
    {
      "epoch": 0.21581,
      "grad_norm": 1.4160257381986296,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 21581
    },
    {
      "epoch": 0.21582,
      "grad_norm": 1.4757680075116717,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 21582
    },
    {
      "epoch": 0.21583,
      "grad_norm": 1.5358374207230423,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 21583
    },
    {
      "epoch": 0.21584,
      "grad_norm": 1.5574011790728366,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 21584
    },
    {
      "epoch": 0.21585,
      "grad_norm": 1.1214992400217503,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 21585
    },
    {
      "epoch": 0.21586,
      "grad_norm": 1.6992726371130698,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 21586
    },
    {
      "epoch": 0.21587,
      "grad_norm": 1.1753772550111057,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 21587
    },
    {
      "epoch": 0.21588,
      "grad_norm": 1.3809315586928703,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 21588
    },
    {
      "epoch": 0.21589,
      "grad_norm": 1.2637905063856592,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 21589
    },
    {
      "epoch": 0.2159,
      "grad_norm": 1.4473135353071278,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 21590
    },
    {
      "epoch": 0.21591,
      "grad_norm": 1.3323140250720638,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 21591
    },
    {
      "epoch": 0.21592,
      "grad_norm": 1.422764327356364,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 21592
    },
    {
      "epoch": 0.21593,
      "grad_norm": 1.0475022562105123,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 21593
    },
    {
      "epoch": 0.21594,
      "grad_norm": 1.7374956068843377,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 21594
    },
    {
      "epoch": 0.21595,
      "grad_norm": 1.3549828088244944,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 21595
    },
    {
      "epoch": 0.21596,
      "grad_norm": 1.3795446813878405,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 21596
    },
    {
      "epoch": 0.21597,
      "grad_norm": 1.3527697884769816,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 21597
    },
    {
      "epoch": 0.21598,
      "grad_norm": 1.443118614224528,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 21598
    },
    {
      "epoch": 0.21599,
      "grad_norm": 1.5789008710691697,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 21599
    },
    {
      "epoch": 0.216,
      "grad_norm": 0.9971880523868474,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 21600
    },
    {
      "epoch": 0.21601,
      "grad_norm": 1.6959581327753726,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 21601
    },
    {
      "epoch": 0.21602,
      "grad_norm": 1.1579673391638532,
      "learning_rate": 0.003,
      "loss": 4.0121,
      "step": 21602
    },
    {
      "epoch": 0.21603,
      "grad_norm": 1.430437164664146,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 21603
    },
    {
      "epoch": 0.21604,
      "grad_norm": 1.3046635963118043,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 21604
    },
    {
      "epoch": 0.21605,
      "grad_norm": 1.5287335485794018,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 21605
    },
    {
      "epoch": 0.21606,
      "grad_norm": 1.595175366047199,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 21606
    },
    {
      "epoch": 0.21607,
      "grad_norm": 1.2726329818538589,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 21607
    },
    {
      "epoch": 0.21608,
      "grad_norm": 1.4104958340286844,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 21608
    },
    {
      "epoch": 0.21609,
      "grad_norm": 1.4477869337204952,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 21609
    },
    {
      "epoch": 0.2161,
      "grad_norm": 1.1664696230932665,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 21610
    },
    {
      "epoch": 0.21611,
      "grad_norm": 1.477087454913899,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 21611
    },
    {
      "epoch": 0.21612,
      "grad_norm": 1.3714088012117311,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 21612
    },
    {
      "epoch": 0.21613,
      "grad_norm": 1.4116579449268434,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 21613
    },
    {
      "epoch": 0.21614,
      "grad_norm": 1.3807817479659121,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 21614
    },
    {
      "epoch": 0.21615,
      "grad_norm": 1.3619293959221994,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 21615
    },
    {
      "epoch": 0.21616,
      "grad_norm": 1.3616990062115881,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 21616
    },
    {
      "epoch": 0.21617,
      "grad_norm": 1.3587219297321471,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 21617
    },
    {
      "epoch": 0.21618,
      "grad_norm": 1.434026366402768,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 21618
    },
    {
      "epoch": 0.21619,
      "grad_norm": 1.3645624136467729,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 21619
    },
    {
      "epoch": 0.2162,
      "grad_norm": 1.3203288189146836,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 21620
    },
    {
      "epoch": 0.21621,
      "grad_norm": 1.1929662451051506,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 21621
    },
    {
      "epoch": 0.21622,
      "grad_norm": 1.3993197065686078,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 21622
    },
    {
      "epoch": 0.21623,
      "grad_norm": 1.237388037188455,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 21623
    },
    {
      "epoch": 0.21624,
      "grad_norm": 1.51925038792954,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 21624
    },
    {
      "epoch": 0.21625,
      "grad_norm": 1.297490591505803,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 21625
    },
    {
      "epoch": 0.21626,
      "grad_norm": 1.163973032156284,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 21626
    },
    {
      "epoch": 0.21627,
      "grad_norm": 1.2858400236240677,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 21627
    },
    {
      "epoch": 0.21628,
      "grad_norm": 1.144426443129442,
      "learning_rate": 0.003,
      "loss": 4.0031,
      "step": 21628
    },
    {
      "epoch": 0.21629,
      "grad_norm": 1.8709954681847207,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 21629
    },
    {
      "epoch": 0.2163,
      "grad_norm": 1.0437961753192782,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 21630
    },
    {
      "epoch": 0.21631,
      "grad_norm": 1.7225250200734556,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 21631
    },
    {
      "epoch": 0.21632,
      "grad_norm": 1.2537767937813247,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 21632
    },
    {
      "epoch": 0.21633,
      "grad_norm": 1.3556325394402138,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 21633
    },
    {
      "epoch": 0.21634,
      "grad_norm": 1.2231110664289988,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 21634
    },
    {
      "epoch": 0.21635,
      "grad_norm": 1.2845488504738347,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 21635
    },
    {
      "epoch": 0.21636,
      "grad_norm": 1.4193071665596668,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 21636
    },
    {
      "epoch": 0.21637,
      "grad_norm": 1.249477328209799,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 21637
    },
    {
      "epoch": 0.21638,
      "grad_norm": 1.5906557052491492,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 21638
    },
    {
      "epoch": 0.21639,
      "grad_norm": 1.3354898826295734,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 21639
    },
    {
      "epoch": 0.2164,
      "grad_norm": 1.2081118831734878,
      "learning_rate": 0.003,
      "loss": 4.0291,
      "step": 21640
    },
    {
      "epoch": 0.21641,
      "grad_norm": 1.4248452009874346,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 21641
    },
    {
      "epoch": 0.21642,
      "grad_norm": 1.1965572012589307,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 21642
    },
    {
      "epoch": 0.21643,
      "grad_norm": 1.564765311797686,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 21643
    },
    {
      "epoch": 0.21644,
      "grad_norm": 1.21547877729175,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 21644
    },
    {
      "epoch": 0.21645,
      "grad_norm": 1.5468043681041297,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 21645
    },
    {
      "epoch": 0.21646,
      "grad_norm": 1.2890287381911754,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 21646
    },
    {
      "epoch": 0.21647,
      "grad_norm": 1.5472699496146884,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 21647
    },
    {
      "epoch": 0.21648,
      "grad_norm": 1.2753056416161248,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 21648
    },
    {
      "epoch": 0.21649,
      "grad_norm": 1.3225232253904111,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 21649
    },
    {
      "epoch": 0.2165,
      "grad_norm": 1.2158836680965754,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 21650
    },
    {
      "epoch": 0.21651,
      "grad_norm": 1.2555133553073874,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 21651
    },
    {
      "epoch": 0.21652,
      "grad_norm": 1.4069853995644357,
      "learning_rate": 0.003,
      "loss": 4.0215,
      "step": 21652
    },
    {
      "epoch": 0.21653,
      "grad_norm": 1.1577357999122015,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 21653
    },
    {
      "epoch": 0.21654,
      "grad_norm": 1.3273312106615434,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 21654
    },
    {
      "epoch": 0.21655,
      "grad_norm": 1.4568342365873415,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 21655
    },
    {
      "epoch": 0.21656,
      "grad_norm": 1.722474764085723,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 21656
    },
    {
      "epoch": 0.21657,
      "grad_norm": 1.6570507191190962,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 21657
    },
    {
      "epoch": 0.21658,
      "grad_norm": 1.62144372785492,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 21658
    },
    {
      "epoch": 0.21659,
      "grad_norm": 1.1890932469729452,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 21659
    },
    {
      "epoch": 0.2166,
      "grad_norm": 1.4347718290022433,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 21660
    },
    {
      "epoch": 0.21661,
      "grad_norm": 1.2308231138075292,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 21661
    },
    {
      "epoch": 0.21662,
      "grad_norm": 1.5454350106985368,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 21662
    },
    {
      "epoch": 0.21663,
      "grad_norm": 1.046988003669505,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 21663
    },
    {
      "epoch": 0.21664,
      "grad_norm": 1.8212875291669854,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 21664
    },
    {
      "epoch": 0.21665,
      "grad_norm": 1.2163617415976309,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 21665
    },
    {
      "epoch": 0.21666,
      "grad_norm": 1.1954722007991765,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 21666
    },
    {
      "epoch": 0.21667,
      "grad_norm": 1.643670887358789,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 21667
    },
    {
      "epoch": 0.21668,
      "grad_norm": 1.1819719137691378,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 21668
    },
    {
      "epoch": 0.21669,
      "grad_norm": 1.3883845641539347,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 21669
    },
    {
      "epoch": 0.2167,
      "grad_norm": 1.5432137729169304,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 21670
    },
    {
      "epoch": 0.21671,
      "grad_norm": 1.0801449982316462,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 21671
    },
    {
      "epoch": 0.21672,
      "grad_norm": 1.4209952767389749,
      "learning_rate": 0.003,
      "loss": 4.0218,
      "step": 21672
    },
    {
      "epoch": 0.21673,
      "grad_norm": 1.5866298342325442,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 21673
    },
    {
      "epoch": 0.21674,
      "grad_norm": 1.2681859492203604,
      "learning_rate": 0.003,
      "loss": 4.0188,
      "step": 21674
    },
    {
      "epoch": 0.21675,
      "grad_norm": 1.5297042405494283,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 21675
    },
    {
      "epoch": 0.21676,
      "grad_norm": 1.3899495767717518,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 21676
    },
    {
      "epoch": 0.21677,
      "grad_norm": 1.25932941774649,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 21677
    },
    {
      "epoch": 0.21678,
      "grad_norm": 1.221883870806337,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 21678
    },
    {
      "epoch": 0.21679,
      "grad_norm": 1.3651250096100358,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 21679
    },
    {
      "epoch": 0.2168,
      "grad_norm": 1.1807414188959091,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 21680
    },
    {
      "epoch": 0.21681,
      "grad_norm": 1.5244914859217473,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 21681
    },
    {
      "epoch": 0.21682,
      "grad_norm": 1.1578109569084247,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 21682
    },
    {
      "epoch": 0.21683,
      "grad_norm": 1.6649032118906697,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 21683
    },
    {
      "epoch": 0.21684,
      "grad_norm": 1.2053851226183132,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 21684
    },
    {
      "epoch": 0.21685,
      "grad_norm": 1.5142341801766461,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 21685
    },
    {
      "epoch": 0.21686,
      "grad_norm": 1.322200257992489,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 21686
    },
    {
      "epoch": 0.21687,
      "grad_norm": 1.3231026104256665,
      "learning_rate": 0.003,
      "loss": 4.0312,
      "step": 21687
    },
    {
      "epoch": 0.21688,
      "grad_norm": 1.606823587365431,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 21688
    },
    {
      "epoch": 0.21689,
      "grad_norm": 1.1984459130647076,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 21689
    },
    {
      "epoch": 0.2169,
      "grad_norm": 1.47229251005662,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 21690
    },
    {
      "epoch": 0.21691,
      "grad_norm": 1.4065269584718103,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 21691
    },
    {
      "epoch": 0.21692,
      "grad_norm": 1.1773685317018276,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 21692
    },
    {
      "epoch": 0.21693,
      "grad_norm": 1.5216618999402516,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 21693
    },
    {
      "epoch": 0.21694,
      "grad_norm": 1.2676906916747757,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 21694
    },
    {
      "epoch": 0.21695,
      "grad_norm": 1.5011195436506621,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 21695
    },
    {
      "epoch": 0.21696,
      "grad_norm": 1.115549485955369,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 21696
    },
    {
      "epoch": 0.21697,
      "grad_norm": 1.8018502147911997,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 21697
    },
    {
      "epoch": 0.21698,
      "grad_norm": 1.36356208110744,
      "learning_rate": 0.003,
      "loss": 4.0148,
      "step": 21698
    },
    {
      "epoch": 0.21699,
      "grad_norm": 1.5376830435927542,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 21699
    },
    {
      "epoch": 0.217,
      "grad_norm": 1.3219978376624038,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 21700
    },
    {
      "epoch": 0.21701,
      "grad_norm": 1.461548045741008,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 21701
    },
    {
      "epoch": 0.21702,
      "grad_norm": 1.2124060724442183,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 21702
    },
    {
      "epoch": 0.21703,
      "grad_norm": 1.4795227869985204,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 21703
    },
    {
      "epoch": 0.21704,
      "grad_norm": 1.3709728791071893,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 21704
    },
    {
      "epoch": 0.21705,
      "grad_norm": 1.3277364546039956,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 21705
    },
    {
      "epoch": 0.21706,
      "grad_norm": 1.256696362963927,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 21706
    },
    {
      "epoch": 0.21707,
      "grad_norm": 1.543516684423096,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 21707
    },
    {
      "epoch": 0.21708,
      "grad_norm": 1.2960958415082264,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 21708
    },
    {
      "epoch": 0.21709,
      "grad_norm": 1.3911681816473196,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 21709
    },
    {
      "epoch": 0.2171,
      "grad_norm": 1.2940636888664225,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 21710
    },
    {
      "epoch": 0.21711,
      "grad_norm": 1.4519873731834203,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 21711
    },
    {
      "epoch": 0.21712,
      "grad_norm": 1.3964960397836088,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 21712
    },
    {
      "epoch": 0.21713,
      "grad_norm": 1.1396648944563743,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 21713
    },
    {
      "epoch": 0.21714,
      "grad_norm": 1.2894495531812444,
      "learning_rate": 0.003,
      "loss": 4.0235,
      "step": 21714
    },
    {
      "epoch": 0.21715,
      "grad_norm": 1.3753294352282759,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 21715
    },
    {
      "epoch": 0.21716,
      "grad_norm": 1.2255268338041903,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 21716
    },
    {
      "epoch": 0.21717,
      "grad_norm": 1.6193976605963485,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 21717
    },
    {
      "epoch": 0.21718,
      "grad_norm": 1.1317837812203877,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 21718
    },
    {
      "epoch": 0.21719,
      "grad_norm": 1.5081856943937748,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 21719
    },
    {
      "epoch": 0.2172,
      "grad_norm": 1.3034260129123711,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 21720
    },
    {
      "epoch": 0.21721,
      "grad_norm": 1.5751825772886663,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 21721
    },
    {
      "epoch": 0.21722,
      "grad_norm": 1.187587421016948,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 21722
    },
    {
      "epoch": 0.21723,
      "grad_norm": 1.4427287322645799,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 21723
    },
    {
      "epoch": 0.21724,
      "grad_norm": 1.2827694304521151,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 21724
    },
    {
      "epoch": 0.21725,
      "grad_norm": 1.2441120564389594,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 21725
    },
    {
      "epoch": 0.21726,
      "grad_norm": 1.4617663722466212,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 21726
    },
    {
      "epoch": 0.21727,
      "grad_norm": 1.1876164644093088,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 21727
    },
    {
      "epoch": 0.21728,
      "grad_norm": 1.5713004814183846,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 21728
    },
    {
      "epoch": 0.21729,
      "grad_norm": 1.2712912632821525,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 21729
    },
    {
      "epoch": 0.2173,
      "grad_norm": 1.5295052971950158,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 21730
    },
    {
      "epoch": 0.21731,
      "grad_norm": 1.2662831509869186,
      "learning_rate": 0.003,
      "loss": 4.0976,
      "step": 21731
    },
    {
      "epoch": 0.21732,
      "grad_norm": 1.44190359333691,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 21732
    },
    {
      "epoch": 0.21733,
      "grad_norm": 1.352985815465497,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 21733
    },
    {
      "epoch": 0.21734,
      "grad_norm": 1.273980705478746,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 21734
    },
    {
      "epoch": 0.21735,
      "grad_norm": 1.3687623400104987,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 21735
    },
    {
      "epoch": 0.21736,
      "grad_norm": 1.137617573178497,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 21736
    },
    {
      "epoch": 0.21737,
      "grad_norm": 1.5544173363330809,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 21737
    },
    {
      "epoch": 0.21738,
      "grad_norm": 1.1422182207113591,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 21738
    },
    {
      "epoch": 0.21739,
      "grad_norm": 1.8066904455344126,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 21739
    },
    {
      "epoch": 0.2174,
      "grad_norm": 1.1671789252959586,
      "learning_rate": 0.003,
      "loss": 4.0243,
      "step": 21740
    },
    {
      "epoch": 0.21741,
      "grad_norm": 1.4684617417972043,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 21741
    },
    {
      "epoch": 0.21742,
      "grad_norm": 1.399942284573787,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 21742
    },
    {
      "epoch": 0.21743,
      "grad_norm": 1.16995899577208,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 21743
    },
    {
      "epoch": 0.21744,
      "grad_norm": 1.3906569721727808,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 21744
    },
    {
      "epoch": 0.21745,
      "grad_norm": 1.477492393749993,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 21745
    },
    {
      "epoch": 0.21746,
      "grad_norm": 1.2952139975004873,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 21746
    },
    {
      "epoch": 0.21747,
      "grad_norm": 1.3901444632726925,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 21747
    },
    {
      "epoch": 0.21748,
      "grad_norm": 1.1549286366297145,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 21748
    },
    {
      "epoch": 0.21749,
      "grad_norm": 1.7083361886105854,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 21749
    },
    {
      "epoch": 0.2175,
      "grad_norm": 1.2551477661927604,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 21750
    },
    {
      "epoch": 0.21751,
      "grad_norm": 1.3079787486564436,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 21751
    },
    {
      "epoch": 0.21752,
      "grad_norm": 1.5962397143865705,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 21752
    },
    {
      "epoch": 0.21753,
      "grad_norm": 1.596429040132191,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 21753
    },
    {
      "epoch": 0.21754,
      "grad_norm": 1.2113879246711468,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 21754
    },
    {
      "epoch": 0.21755,
      "grad_norm": 1.6332028029615502,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 21755
    },
    {
      "epoch": 0.21756,
      "grad_norm": 1.6463052448187578,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 21756
    },
    {
      "epoch": 0.21757,
      "grad_norm": 1.137118479217971,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 21757
    },
    {
      "epoch": 0.21758,
      "grad_norm": 1.3635544244652056,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 21758
    },
    {
      "epoch": 0.21759,
      "grad_norm": 1.2322221825771982,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 21759
    },
    {
      "epoch": 0.2176,
      "grad_norm": 1.4530300513984373,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 21760
    },
    {
      "epoch": 0.21761,
      "grad_norm": 1.3612567174320283,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 21761
    },
    {
      "epoch": 0.21762,
      "grad_norm": 1.4689818427730725,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 21762
    },
    {
      "epoch": 0.21763,
      "grad_norm": 1.3369737458324107,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 21763
    },
    {
      "epoch": 0.21764,
      "grad_norm": 1.3175865171698724,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 21764
    },
    {
      "epoch": 0.21765,
      "grad_norm": 1.246510114819442,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 21765
    },
    {
      "epoch": 0.21766,
      "grad_norm": 1.2665709182460367,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 21766
    },
    {
      "epoch": 0.21767,
      "grad_norm": 1.2461678747363165,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 21767
    },
    {
      "epoch": 0.21768,
      "grad_norm": 1.277964271546957,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 21768
    },
    {
      "epoch": 0.21769,
      "grad_norm": 1.4382325065553294,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 21769
    },
    {
      "epoch": 0.2177,
      "grad_norm": 1.3483673137453749,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 21770
    },
    {
      "epoch": 0.21771,
      "grad_norm": 1.5007492645327565,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 21771
    },
    {
      "epoch": 0.21772,
      "grad_norm": 1.4906521742284367,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 21772
    },
    {
      "epoch": 0.21773,
      "grad_norm": 1.3829906249211739,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 21773
    },
    {
      "epoch": 0.21774,
      "grad_norm": 1.3756208190347674,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 21774
    },
    {
      "epoch": 0.21775,
      "grad_norm": 1.187419765112109,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 21775
    },
    {
      "epoch": 0.21776,
      "grad_norm": 1.2847796938167904,
      "learning_rate": 0.003,
      "loss": 4.0315,
      "step": 21776
    },
    {
      "epoch": 0.21777,
      "grad_norm": 1.1787148813225765,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 21777
    },
    {
      "epoch": 0.21778,
      "grad_norm": 1.5384271887768861,
      "learning_rate": 0.003,
      "loss": 4.0228,
      "step": 21778
    },
    {
      "epoch": 0.21779,
      "grad_norm": 1.5211241536809952,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 21779
    },
    {
      "epoch": 0.2178,
      "grad_norm": 1.3126100308265978,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 21780
    },
    {
      "epoch": 0.21781,
      "grad_norm": 1.3140837416202098,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 21781
    },
    {
      "epoch": 0.21782,
      "grad_norm": 1.2506337097241051,
      "learning_rate": 0.003,
      "loss": 4.0194,
      "step": 21782
    },
    {
      "epoch": 0.21783,
      "grad_norm": 1.183549010980319,
      "learning_rate": 0.003,
      "loss": 4.0215,
      "step": 21783
    },
    {
      "epoch": 0.21784,
      "grad_norm": 1.2120193027817996,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 21784
    },
    {
      "epoch": 0.21785,
      "grad_norm": 1.4705008753482647,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 21785
    },
    {
      "epoch": 0.21786,
      "grad_norm": 1.3037580962126534,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 21786
    },
    {
      "epoch": 0.21787,
      "grad_norm": 1.5873941885421359,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 21787
    },
    {
      "epoch": 0.21788,
      "grad_norm": 1.293752445907499,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 21788
    },
    {
      "epoch": 0.21789,
      "grad_norm": 1.6278839822338038,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 21789
    },
    {
      "epoch": 0.2179,
      "grad_norm": 1.2487031038570793,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 21790
    },
    {
      "epoch": 0.21791,
      "grad_norm": 1.497883497136813,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 21791
    },
    {
      "epoch": 0.21792,
      "grad_norm": 1.5202302105632721,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 21792
    },
    {
      "epoch": 0.21793,
      "grad_norm": 1.2269791447971383,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 21793
    },
    {
      "epoch": 0.21794,
      "grad_norm": 1.449344551808254,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 21794
    },
    {
      "epoch": 0.21795,
      "grad_norm": 1.9388279199050673,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 21795
    },
    {
      "epoch": 0.21796,
      "grad_norm": 0.8841590120868,
      "learning_rate": 0.003,
      "loss": 4.0055,
      "step": 21796
    },
    {
      "epoch": 0.21797,
      "grad_norm": 1.4292538185520196,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 21797
    },
    {
      "epoch": 0.21798,
      "grad_norm": 1.3789664901476326,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 21798
    },
    {
      "epoch": 0.21799,
      "grad_norm": 1.3927764454629348,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 21799
    },
    {
      "epoch": 0.218,
      "grad_norm": 1.622972417186285,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 21800
    },
    {
      "epoch": 0.21801,
      "grad_norm": 1.2911279890778193,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 21801
    },
    {
      "epoch": 0.21802,
      "grad_norm": 1.5380703604963368,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 21802
    },
    {
      "epoch": 0.21803,
      "grad_norm": 1.1990859205719844,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 21803
    },
    {
      "epoch": 0.21804,
      "grad_norm": 1.3908968327755922,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 21804
    },
    {
      "epoch": 0.21805,
      "grad_norm": 1.6093157797807323,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 21805
    },
    {
      "epoch": 0.21806,
      "grad_norm": 1.082573623518282,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 21806
    },
    {
      "epoch": 0.21807,
      "grad_norm": 1.3874620059290594,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 21807
    },
    {
      "epoch": 0.21808,
      "grad_norm": 1.2340360635483378,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 21808
    },
    {
      "epoch": 0.21809,
      "grad_norm": 1.48196030694557,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 21809
    },
    {
      "epoch": 0.2181,
      "grad_norm": 1.1829455082845057,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 21810
    },
    {
      "epoch": 0.21811,
      "grad_norm": 1.3656560031304386,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 21811
    },
    {
      "epoch": 0.21812,
      "grad_norm": 1.2827909957331247,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 21812
    },
    {
      "epoch": 0.21813,
      "grad_norm": 1.486494039463018,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 21813
    },
    {
      "epoch": 0.21814,
      "grad_norm": 1.3759143268638585,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 21814
    },
    {
      "epoch": 0.21815,
      "grad_norm": 1.2142174582604783,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 21815
    },
    {
      "epoch": 0.21816,
      "grad_norm": 1.4197772920590874,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 21816
    },
    {
      "epoch": 0.21817,
      "grad_norm": 1.323393658081733,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 21817
    },
    {
      "epoch": 0.21818,
      "grad_norm": 1.5159685668028264,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 21818
    },
    {
      "epoch": 0.21819,
      "grad_norm": 1.51023708455265,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 21819
    },
    {
      "epoch": 0.2182,
      "grad_norm": 1.2629682522797707,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 21820
    },
    {
      "epoch": 0.21821,
      "grad_norm": 1.236949904346332,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 21821
    },
    {
      "epoch": 0.21822,
      "grad_norm": 1.4900879106882725,
      "learning_rate": 0.003,
      "loss": 4.0279,
      "step": 21822
    },
    {
      "epoch": 0.21823,
      "grad_norm": 1.3545475172053465,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 21823
    },
    {
      "epoch": 0.21824,
      "grad_norm": 1.4291612969027583,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 21824
    },
    {
      "epoch": 0.21825,
      "grad_norm": 1.2378582878097866,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 21825
    },
    {
      "epoch": 0.21826,
      "grad_norm": 1.56258273065671,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 21826
    },
    {
      "epoch": 0.21827,
      "grad_norm": 1.486960382792435,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 21827
    },
    {
      "epoch": 0.21828,
      "grad_norm": 1.2067952770408727,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 21828
    },
    {
      "epoch": 0.21829,
      "grad_norm": 1.3937543985084526,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 21829
    },
    {
      "epoch": 0.2183,
      "grad_norm": 1.3803056056357,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 21830
    },
    {
      "epoch": 0.21831,
      "grad_norm": 1.315611588775608,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 21831
    },
    {
      "epoch": 0.21832,
      "grad_norm": 1.4144492685364904,
      "learning_rate": 0.003,
      "loss": 4.0226,
      "step": 21832
    },
    {
      "epoch": 0.21833,
      "grad_norm": 1.2024873578795783,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 21833
    },
    {
      "epoch": 0.21834,
      "grad_norm": 1.2345346730467635,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 21834
    },
    {
      "epoch": 0.21835,
      "grad_norm": 1.590026520429568,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 21835
    },
    {
      "epoch": 0.21836,
      "grad_norm": 1.4489599804000857,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 21836
    },
    {
      "epoch": 0.21837,
      "grad_norm": 1.3948991946463194,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 21837
    },
    {
      "epoch": 0.21838,
      "grad_norm": 1.0839190492419373,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 21838
    },
    {
      "epoch": 0.21839,
      "grad_norm": 1.5310246196483528,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 21839
    },
    {
      "epoch": 0.2184,
      "grad_norm": 1.2972416568653322,
      "learning_rate": 0.003,
      "loss": 4.0197,
      "step": 21840
    },
    {
      "epoch": 0.21841,
      "grad_norm": 1.3518609380233026,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 21841
    },
    {
      "epoch": 0.21842,
      "grad_norm": 1.239134396927293,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 21842
    },
    {
      "epoch": 0.21843,
      "grad_norm": 1.4939191795927445,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 21843
    },
    {
      "epoch": 0.21844,
      "grad_norm": 1.2542783730907288,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 21844
    },
    {
      "epoch": 0.21845,
      "grad_norm": 1.3523336953828662,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 21845
    },
    {
      "epoch": 0.21846,
      "grad_norm": 1.3483181772009003,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 21846
    },
    {
      "epoch": 0.21847,
      "grad_norm": 1.4341188224770438,
      "learning_rate": 0.003,
      "loss": 4.0259,
      "step": 21847
    },
    {
      "epoch": 0.21848,
      "grad_norm": 1.2701861347518604,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 21848
    },
    {
      "epoch": 0.21849,
      "grad_norm": 1.5735458276270475,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 21849
    },
    {
      "epoch": 0.2185,
      "grad_norm": 1.2977797392794104,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 21850
    },
    {
      "epoch": 0.21851,
      "grad_norm": 1.3993707258427626,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 21851
    },
    {
      "epoch": 0.21852,
      "grad_norm": 1.3400140972805528,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 21852
    },
    {
      "epoch": 0.21853,
      "grad_norm": 1.3187261982616778,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 21853
    },
    {
      "epoch": 0.21854,
      "grad_norm": 1.373648276029539,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 21854
    },
    {
      "epoch": 0.21855,
      "grad_norm": 1.2314813262996391,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 21855
    },
    {
      "epoch": 0.21856,
      "grad_norm": 1.4455662452529494,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 21856
    },
    {
      "epoch": 0.21857,
      "grad_norm": 1.1081284462682128,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 21857
    },
    {
      "epoch": 0.21858,
      "grad_norm": 1.6186299276515583,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 21858
    },
    {
      "epoch": 0.21859,
      "grad_norm": 1.342266584936963,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 21859
    },
    {
      "epoch": 0.2186,
      "grad_norm": 1.6222494766013495,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 21860
    },
    {
      "epoch": 0.21861,
      "grad_norm": 1.2756726672574457,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 21861
    },
    {
      "epoch": 0.21862,
      "grad_norm": 1.3800762617091014,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 21862
    },
    {
      "epoch": 0.21863,
      "grad_norm": 1.1219644793746457,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 21863
    },
    {
      "epoch": 0.21864,
      "grad_norm": 1.4807516433685504,
      "learning_rate": 0.003,
      "loss": 4.0167,
      "step": 21864
    },
    {
      "epoch": 0.21865,
      "grad_norm": 1.153569375518149,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 21865
    },
    {
      "epoch": 0.21866,
      "grad_norm": 1.3913084942040324,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 21866
    },
    {
      "epoch": 0.21867,
      "grad_norm": 1.3975902554296735,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 21867
    },
    {
      "epoch": 0.21868,
      "grad_norm": 1.6111752800832773,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 21868
    },
    {
      "epoch": 0.21869,
      "grad_norm": 1.3397324373724762,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 21869
    },
    {
      "epoch": 0.2187,
      "grad_norm": 1.5265125558745145,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 21870
    },
    {
      "epoch": 0.21871,
      "grad_norm": 1.5444100630413469,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 21871
    },
    {
      "epoch": 0.21872,
      "grad_norm": 1.4136571138250893,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 21872
    },
    {
      "epoch": 0.21873,
      "grad_norm": 1.492531221152572,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 21873
    },
    {
      "epoch": 0.21874,
      "grad_norm": 1.3801935048510277,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 21874
    },
    {
      "epoch": 0.21875,
      "grad_norm": 1.3464300138449798,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 21875
    },
    {
      "epoch": 0.21876,
      "grad_norm": 1.3230002208921514,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 21876
    },
    {
      "epoch": 0.21877,
      "grad_norm": 1.3184776967515563,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 21877
    },
    {
      "epoch": 0.21878,
      "grad_norm": 1.3243604189039329,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 21878
    },
    {
      "epoch": 0.21879,
      "grad_norm": 1.350716120331371,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 21879
    },
    {
      "epoch": 0.2188,
      "grad_norm": 1.1887655380988411,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 21880
    },
    {
      "epoch": 0.21881,
      "grad_norm": 1.4732317039088851,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 21881
    },
    {
      "epoch": 0.21882,
      "grad_norm": 1.2481675039437126,
      "learning_rate": 0.003,
      "loss": 4.0319,
      "step": 21882
    },
    {
      "epoch": 0.21883,
      "grad_norm": 1.682173239924703,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 21883
    },
    {
      "epoch": 0.21884,
      "grad_norm": 1.408843425940714,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 21884
    },
    {
      "epoch": 0.21885,
      "grad_norm": 1.3991787530439368,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 21885
    },
    {
      "epoch": 0.21886,
      "grad_norm": 1.351255708264895,
      "learning_rate": 0.003,
      "loss": 4.1131,
      "step": 21886
    },
    {
      "epoch": 0.21887,
      "grad_norm": 1.3910581333259997,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 21887
    },
    {
      "epoch": 0.21888,
      "grad_norm": 1.3253623213565662,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 21888
    },
    {
      "epoch": 0.21889,
      "grad_norm": 1.3207359549564404,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 21889
    },
    {
      "epoch": 0.2189,
      "grad_norm": 1.3103691672670739,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 21890
    },
    {
      "epoch": 0.21891,
      "grad_norm": 1.4143579120442842,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 21891
    },
    {
      "epoch": 0.21892,
      "grad_norm": 1.3769560345949032,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 21892
    },
    {
      "epoch": 0.21893,
      "grad_norm": 1.6479910150324262,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 21893
    },
    {
      "epoch": 0.21894,
      "grad_norm": 1.477498345410568,
      "learning_rate": 0.003,
      "loss": 4.02,
      "step": 21894
    },
    {
      "epoch": 0.21895,
      "grad_norm": 1.6010316292528048,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 21895
    },
    {
      "epoch": 0.21896,
      "grad_norm": 1.0736524070408766,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 21896
    },
    {
      "epoch": 0.21897,
      "grad_norm": 1.5513404345546,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 21897
    },
    {
      "epoch": 0.21898,
      "grad_norm": 1.4060130819583434,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 21898
    },
    {
      "epoch": 0.21899,
      "grad_norm": 1.242919541121017,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 21899
    },
    {
      "epoch": 0.219,
      "grad_norm": 1.419710336740735,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 21900
    },
    {
      "epoch": 0.21901,
      "grad_norm": 1.1142225146782796,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 21901
    },
    {
      "epoch": 0.21902,
      "grad_norm": 1.515604579657156,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 21902
    },
    {
      "epoch": 0.21903,
      "grad_norm": 1.383479495850097,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 21903
    },
    {
      "epoch": 0.21904,
      "grad_norm": 1.3652068263044148,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 21904
    },
    {
      "epoch": 0.21905,
      "grad_norm": 1.4900790425030694,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 21905
    },
    {
      "epoch": 0.21906,
      "grad_norm": 1.193768677638813,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 21906
    },
    {
      "epoch": 0.21907,
      "grad_norm": 1.3165246410732574,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 21907
    },
    {
      "epoch": 0.21908,
      "grad_norm": 1.2852506288954528,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 21908
    },
    {
      "epoch": 0.21909,
      "grad_norm": 1.4322754139851814,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 21909
    },
    {
      "epoch": 0.2191,
      "grad_norm": 1.1934054775387493,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 21910
    },
    {
      "epoch": 0.21911,
      "grad_norm": 1.427492098697218,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 21911
    },
    {
      "epoch": 0.21912,
      "grad_norm": 1.3050800021088695,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 21912
    },
    {
      "epoch": 0.21913,
      "grad_norm": 1.5633344870549826,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 21913
    },
    {
      "epoch": 0.21914,
      "grad_norm": 1.4370186550446122,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 21914
    },
    {
      "epoch": 0.21915,
      "grad_norm": 1.5627544102364597,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 21915
    },
    {
      "epoch": 0.21916,
      "grad_norm": 0.9544544959214196,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 21916
    },
    {
      "epoch": 0.21917,
      "grad_norm": 1.3747323703717953,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 21917
    },
    {
      "epoch": 0.21918,
      "grad_norm": 1.356051125720396,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 21918
    },
    {
      "epoch": 0.21919,
      "grad_norm": 1.2647930960633598,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 21919
    },
    {
      "epoch": 0.2192,
      "grad_norm": 1.5382002387844138,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 21920
    },
    {
      "epoch": 0.21921,
      "grad_norm": 1.318015693485354,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 21921
    },
    {
      "epoch": 0.21922,
      "grad_norm": 1.5612089494631403,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 21922
    },
    {
      "epoch": 0.21923,
      "grad_norm": 1.2109777532096635,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 21923
    },
    {
      "epoch": 0.21924,
      "grad_norm": 1.5172583837326865,
      "learning_rate": 0.003,
      "loss": 4.0946,
      "step": 21924
    },
    {
      "epoch": 0.21925,
      "grad_norm": 1.2389141294335106,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 21925
    },
    {
      "epoch": 0.21926,
      "grad_norm": 1.6481160695458348,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 21926
    },
    {
      "epoch": 0.21927,
      "grad_norm": 1.1259072392279144,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 21927
    },
    {
      "epoch": 0.21928,
      "grad_norm": 1.523706799834442,
      "learning_rate": 0.003,
      "loss": 4.0947,
      "step": 21928
    },
    {
      "epoch": 0.21929,
      "grad_norm": 1.3838361375856523,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 21929
    },
    {
      "epoch": 0.2193,
      "grad_norm": 1.3183290648455395,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 21930
    },
    {
      "epoch": 0.21931,
      "grad_norm": 1.3157787075022724,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 21931
    },
    {
      "epoch": 0.21932,
      "grad_norm": 1.361708506842995,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 21932
    },
    {
      "epoch": 0.21933,
      "grad_norm": 1.5617791748578622,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 21933
    },
    {
      "epoch": 0.21934,
      "grad_norm": 1.2795490857302654,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 21934
    },
    {
      "epoch": 0.21935,
      "grad_norm": 1.4973845919237996,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 21935
    },
    {
      "epoch": 0.21936,
      "grad_norm": 1.2442587511940837,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 21936
    },
    {
      "epoch": 0.21937,
      "grad_norm": 1.62905009543493,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 21937
    },
    {
      "epoch": 0.21938,
      "grad_norm": 1.3083598097255102,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 21938
    },
    {
      "epoch": 0.21939,
      "grad_norm": 1.343908636553416,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 21939
    },
    {
      "epoch": 0.2194,
      "grad_norm": 1.2021113121993336,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 21940
    },
    {
      "epoch": 0.21941,
      "grad_norm": 1.491497563459595,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 21941
    },
    {
      "epoch": 0.21942,
      "grad_norm": 1.1311382475212675,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 21942
    },
    {
      "epoch": 0.21943,
      "grad_norm": 1.4006684967081473,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 21943
    },
    {
      "epoch": 0.21944,
      "grad_norm": 1.2935080502234235,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 21944
    },
    {
      "epoch": 0.21945,
      "grad_norm": 1.3453590992792903,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 21945
    },
    {
      "epoch": 0.21946,
      "grad_norm": 1.340367401947809,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 21946
    },
    {
      "epoch": 0.21947,
      "grad_norm": 1.3381634369619626,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 21947
    },
    {
      "epoch": 0.21948,
      "grad_norm": 1.6031288553416352,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 21948
    },
    {
      "epoch": 0.21949,
      "grad_norm": 1.236788110522683,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 21949
    },
    {
      "epoch": 0.2195,
      "grad_norm": 1.3961868195729772,
      "learning_rate": 0.003,
      "loss": 4.0254,
      "step": 21950
    },
    {
      "epoch": 0.21951,
      "grad_norm": 1.3454923024665244,
      "learning_rate": 0.003,
      "loss": 4.0328,
      "step": 21951
    },
    {
      "epoch": 0.21952,
      "grad_norm": 1.1612319105690718,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 21952
    },
    {
      "epoch": 0.21953,
      "grad_norm": 1.3979619770700362,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 21953
    },
    {
      "epoch": 0.21954,
      "grad_norm": 1.3488336642416268,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 21954
    },
    {
      "epoch": 0.21955,
      "grad_norm": 1.5890297765100796,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 21955
    },
    {
      "epoch": 0.21956,
      "grad_norm": 1.3013044515226226,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 21956
    },
    {
      "epoch": 0.21957,
      "grad_norm": 1.3205711846707797,
      "learning_rate": 0.003,
      "loss": 4.0301,
      "step": 21957
    },
    {
      "epoch": 0.21958,
      "grad_norm": 1.3960458913269216,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 21958
    },
    {
      "epoch": 0.21959,
      "grad_norm": 1.3410994689448847,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 21959
    },
    {
      "epoch": 0.2196,
      "grad_norm": 1.3622950495825286,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 21960
    },
    {
      "epoch": 0.21961,
      "grad_norm": 1.2439812562161565,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 21961
    },
    {
      "epoch": 0.21962,
      "grad_norm": 1.6495809689740977,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 21962
    },
    {
      "epoch": 0.21963,
      "grad_norm": 1.1075823228719488,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 21963
    },
    {
      "epoch": 0.21964,
      "grad_norm": 1.40324106022888,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 21964
    },
    {
      "epoch": 0.21965,
      "grad_norm": 1.2884392999720742,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 21965
    },
    {
      "epoch": 0.21966,
      "grad_norm": 1.3539642794047264,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 21966
    },
    {
      "epoch": 0.21967,
      "grad_norm": 1.4955974362173503,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 21967
    },
    {
      "epoch": 0.21968,
      "grad_norm": 1.5116388453685434,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 21968
    },
    {
      "epoch": 0.21969,
      "grad_norm": 1.3549104667418375,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 21969
    },
    {
      "epoch": 0.2197,
      "grad_norm": 1.4079075297455776,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 21970
    },
    {
      "epoch": 0.21971,
      "grad_norm": 1.453401454507369,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 21971
    },
    {
      "epoch": 0.21972,
      "grad_norm": 1.4250613175756226,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 21972
    },
    {
      "epoch": 0.21973,
      "grad_norm": 1.3831072647546812,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 21973
    },
    {
      "epoch": 0.21974,
      "grad_norm": 1.4164304746590137,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 21974
    },
    {
      "epoch": 0.21975,
      "grad_norm": 1.3363938348699675,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 21975
    },
    {
      "epoch": 0.21976,
      "grad_norm": 1.2854685734366242,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 21976
    },
    {
      "epoch": 0.21977,
      "grad_norm": 1.5134779219680778,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 21977
    },
    {
      "epoch": 0.21978,
      "grad_norm": 1.4153170955539442,
      "learning_rate": 0.003,
      "loss": 4.0898,
      "step": 21978
    },
    {
      "epoch": 0.21979,
      "grad_norm": 1.6458748964455412,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 21979
    },
    {
      "epoch": 0.2198,
      "grad_norm": 1.2724509650276048,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 21980
    },
    {
      "epoch": 0.21981,
      "grad_norm": 1.4023100840793203,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 21981
    },
    {
      "epoch": 0.21982,
      "grad_norm": 1.3912212765186005,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 21982
    },
    {
      "epoch": 0.21983,
      "grad_norm": 1.427322774860121,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 21983
    },
    {
      "epoch": 0.21984,
      "grad_norm": 1.3483426313043931,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 21984
    },
    {
      "epoch": 0.21985,
      "grad_norm": 1.3793512180367091,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 21985
    },
    {
      "epoch": 0.21986,
      "grad_norm": 1.1907907204772645,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 21986
    },
    {
      "epoch": 0.21987,
      "grad_norm": 1.1974306360117373,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 21987
    },
    {
      "epoch": 0.21988,
      "grad_norm": 1.470162481516743,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 21988
    },
    {
      "epoch": 0.21989,
      "grad_norm": 1.1308975327673492,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 21989
    },
    {
      "epoch": 0.2199,
      "grad_norm": 1.5982872045018905,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 21990
    },
    {
      "epoch": 0.21991,
      "grad_norm": 1.3511960661468174,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 21991
    },
    {
      "epoch": 0.21992,
      "grad_norm": 1.2009597118512085,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 21992
    },
    {
      "epoch": 0.21993,
      "grad_norm": 1.3066035698767409,
      "learning_rate": 0.003,
      "loss": 4.0293,
      "step": 21993
    },
    {
      "epoch": 0.21994,
      "grad_norm": 1.4236477958372746,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 21994
    },
    {
      "epoch": 0.21995,
      "grad_norm": 1.4189188999668838,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 21995
    },
    {
      "epoch": 0.21996,
      "grad_norm": 1.5128501434641455,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 21996
    },
    {
      "epoch": 0.21997,
      "grad_norm": 1.172286039870528,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 21997
    },
    {
      "epoch": 0.21998,
      "grad_norm": 1.6729050834879595,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 21998
    },
    {
      "epoch": 0.21999,
      "grad_norm": 0.9473969066854019,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 21999
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3857757584959134,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 22000
    },
    {
      "epoch": 0.22001,
      "grad_norm": 1.3242205618236835,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 22001
    },
    {
      "epoch": 0.22002,
      "grad_norm": 1.4698493681572296,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 22002
    },
    {
      "epoch": 0.22003,
      "grad_norm": 1.2867339428696243,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 22003
    },
    {
      "epoch": 0.22004,
      "grad_norm": 1.3454886959099548,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 22004
    },
    {
      "epoch": 0.22005,
      "grad_norm": 1.253041919375533,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 22005
    },
    {
      "epoch": 0.22006,
      "grad_norm": 1.7307220189692687,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 22006
    },
    {
      "epoch": 0.22007,
      "grad_norm": 1.13504168347129,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 22007
    },
    {
      "epoch": 0.22008,
      "grad_norm": 1.8296563772979586,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 22008
    },
    {
      "epoch": 0.22009,
      "grad_norm": 1.1741589978484182,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 22009
    },
    {
      "epoch": 0.2201,
      "grad_norm": 1.5381026497279788,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 22010
    },
    {
      "epoch": 0.22011,
      "grad_norm": 1.1094107114377345,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 22011
    },
    {
      "epoch": 0.22012,
      "grad_norm": 1.442829519692605,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 22012
    },
    {
      "epoch": 0.22013,
      "grad_norm": 1.631016202833152,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 22013
    },
    {
      "epoch": 0.22014,
      "grad_norm": 1.6106086231447079,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 22014
    },
    {
      "epoch": 0.22015,
      "grad_norm": 1.1814828776053918,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 22015
    },
    {
      "epoch": 0.22016,
      "grad_norm": 1.578015763537259,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 22016
    },
    {
      "epoch": 0.22017,
      "grad_norm": 1.3228701853660656,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 22017
    },
    {
      "epoch": 0.22018,
      "grad_norm": 1.3954939622633813,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 22018
    },
    {
      "epoch": 0.22019,
      "grad_norm": 1.4848139680187202,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 22019
    },
    {
      "epoch": 0.2202,
      "grad_norm": 1.1779000454188744,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 22020
    },
    {
      "epoch": 0.22021,
      "grad_norm": 1.359113863980639,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 22021
    },
    {
      "epoch": 0.22022,
      "grad_norm": 1.6020148930131255,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 22022
    },
    {
      "epoch": 0.22023,
      "grad_norm": 1.080821702668198,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 22023
    },
    {
      "epoch": 0.22024,
      "grad_norm": 1.3793835762293125,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 22024
    },
    {
      "epoch": 0.22025,
      "grad_norm": 1.3996018199293805,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 22025
    },
    {
      "epoch": 0.22026,
      "grad_norm": 1.4541457512077411,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 22026
    },
    {
      "epoch": 0.22027,
      "grad_norm": 1.5036697031980546,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 22027
    },
    {
      "epoch": 0.22028,
      "grad_norm": 1.3927468977215929,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 22028
    },
    {
      "epoch": 0.22029,
      "grad_norm": 1.1116074644616787,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 22029
    },
    {
      "epoch": 0.2203,
      "grad_norm": 1.4137392822592407,
      "learning_rate": 0.003,
      "loss": 4.0791,
      "step": 22030
    },
    {
      "epoch": 0.22031,
      "grad_norm": 1.433020747516023,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 22031
    },
    {
      "epoch": 0.22032,
      "grad_norm": 1.191612135290802,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 22032
    },
    {
      "epoch": 0.22033,
      "grad_norm": 1.4952455930118238,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 22033
    },
    {
      "epoch": 0.22034,
      "grad_norm": 1.2605812991854766,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 22034
    },
    {
      "epoch": 0.22035,
      "grad_norm": 1.7377216985408868,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 22035
    },
    {
      "epoch": 0.22036,
      "grad_norm": 1.9438553629326887,
      "learning_rate": 0.003,
      "loss": 4.0919,
      "step": 22036
    },
    {
      "epoch": 0.22037,
      "grad_norm": 1.0593080689450938,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 22037
    },
    {
      "epoch": 0.22038,
      "grad_norm": 1.6234239889397044,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 22038
    },
    {
      "epoch": 0.22039,
      "grad_norm": 1.4127420233740093,
      "learning_rate": 0.003,
      "loss": 4.1046,
      "step": 22039
    },
    {
      "epoch": 0.2204,
      "grad_norm": 1.4386928226103217,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 22040
    },
    {
      "epoch": 0.22041,
      "grad_norm": 1.3624686217299482,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 22041
    },
    {
      "epoch": 0.22042,
      "grad_norm": 1.3013724831468199,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 22042
    },
    {
      "epoch": 0.22043,
      "grad_norm": 1.2419646753823672,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 22043
    },
    {
      "epoch": 0.22044,
      "grad_norm": 1.4984428244005863,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 22044
    },
    {
      "epoch": 0.22045,
      "grad_norm": 1.5316081336786143,
      "learning_rate": 0.003,
      "loss": 4.0798,
      "step": 22045
    },
    {
      "epoch": 0.22046,
      "grad_norm": 1.4880937306232618,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 22046
    },
    {
      "epoch": 0.22047,
      "grad_norm": 1.1892740904767414,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 22047
    },
    {
      "epoch": 0.22048,
      "grad_norm": 1.483080169944513,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 22048
    },
    {
      "epoch": 0.22049,
      "grad_norm": 1.2408488390797463,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 22049
    },
    {
      "epoch": 0.2205,
      "grad_norm": 1.3882050104985644,
      "learning_rate": 0.003,
      "loss": 4.0303,
      "step": 22050
    },
    {
      "epoch": 0.22051,
      "grad_norm": 1.3989381887034402,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 22051
    },
    {
      "epoch": 0.22052,
      "grad_norm": 1.225212641636643,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 22052
    },
    {
      "epoch": 0.22053,
      "grad_norm": 1.298685388696244,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 22053
    },
    {
      "epoch": 0.22054,
      "grad_norm": 1.2475364898339563,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 22054
    },
    {
      "epoch": 0.22055,
      "grad_norm": 1.2144002309450608,
      "learning_rate": 0.003,
      "loss": 4.017,
      "step": 22055
    },
    {
      "epoch": 0.22056,
      "grad_norm": 1.618483149512431,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 22056
    },
    {
      "epoch": 0.22057,
      "grad_norm": 1.1606438126874004,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 22057
    },
    {
      "epoch": 0.22058,
      "grad_norm": 1.4745123089166658,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 22058
    },
    {
      "epoch": 0.22059,
      "grad_norm": 1.2769331651867426,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 22059
    },
    {
      "epoch": 0.2206,
      "grad_norm": 1.4767858939060083,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 22060
    },
    {
      "epoch": 0.22061,
      "grad_norm": 1.4935033078416218,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 22061
    },
    {
      "epoch": 0.22062,
      "grad_norm": 1.1771267549329605,
      "learning_rate": 0.003,
      "loss": 4.025,
      "step": 22062
    },
    {
      "epoch": 0.22063,
      "grad_norm": 1.3796150280199535,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 22063
    },
    {
      "epoch": 0.22064,
      "grad_norm": 1.3271035855948137,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 22064
    },
    {
      "epoch": 0.22065,
      "grad_norm": 1.1652481329836022,
      "learning_rate": 0.003,
      "loss": 4.0151,
      "step": 22065
    },
    {
      "epoch": 0.22066,
      "grad_norm": 1.2905792226996244,
      "learning_rate": 0.003,
      "loss": 4.0248,
      "step": 22066
    },
    {
      "epoch": 0.22067,
      "grad_norm": 1.2851657372405187,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 22067
    },
    {
      "epoch": 0.22068,
      "grad_norm": 1.5742584223384706,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 22068
    },
    {
      "epoch": 0.22069,
      "grad_norm": 1.338394614444369,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 22069
    },
    {
      "epoch": 0.2207,
      "grad_norm": 1.3004002796280805,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 22070
    },
    {
      "epoch": 0.22071,
      "grad_norm": 1.421639971712515,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 22071
    },
    {
      "epoch": 0.22072,
      "grad_norm": 1.428979973898089,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 22072
    },
    {
      "epoch": 0.22073,
      "grad_norm": 1.668612932878163,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 22073
    },
    {
      "epoch": 0.22074,
      "grad_norm": 1.4643693345898032,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 22074
    },
    {
      "epoch": 0.22075,
      "grad_norm": 1.3369643684227495,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 22075
    },
    {
      "epoch": 0.22076,
      "grad_norm": 1.3383140026879339,
      "learning_rate": 0.003,
      "loss": 4.0892,
      "step": 22076
    },
    {
      "epoch": 0.22077,
      "grad_norm": 1.2131089352572,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 22077
    },
    {
      "epoch": 0.22078,
      "grad_norm": 1.3283572949254057,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 22078
    },
    {
      "epoch": 0.22079,
      "grad_norm": 1.2187737844894437,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 22079
    },
    {
      "epoch": 0.2208,
      "grad_norm": 1.450702948191589,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 22080
    },
    {
      "epoch": 0.22081,
      "grad_norm": 1.372454012378222,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 22081
    },
    {
      "epoch": 0.22082,
      "grad_norm": 1.5001801474665366,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 22082
    },
    {
      "epoch": 0.22083,
      "grad_norm": 1.2899948523986073,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 22083
    },
    {
      "epoch": 0.22084,
      "grad_norm": 1.5862468501136513,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 22084
    },
    {
      "epoch": 0.22085,
      "grad_norm": 1.7794074253172933,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 22085
    },
    {
      "epoch": 0.22086,
      "grad_norm": 1.3740897458861343,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 22086
    },
    {
      "epoch": 0.22087,
      "grad_norm": 1.4937486407431884,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 22087
    },
    {
      "epoch": 0.22088,
      "grad_norm": 1.177792066169784,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 22088
    },
    {
      "epoch": 0.22089,
      "grad_norm": 1.5566010919211382,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 22089
    },
    {
      "epoch": 0.2209,
      "grad_norm": 1.1499788653179301,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 22090
    },
    {
      "epoch": 0.22091,
      "grad_norm": 1.4722548255020058,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 22091
    },
    {
      "epoch": 0.22092,
      "grad_norm": 1.2267218318689077,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 22092
    },
    {
      "epoch": 0.22093,
      "grad_norm": 1.6261563629891562,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 22093
    },
    {
      "epoch": 0.22094,
      "grad_norm": 1.1668655227439646,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 22094
    },
    {
      "epoch": 0.22095,
      "grad_norm": 1.3177353554533442,
      "learning_rate": 0.003,
      "loss": 4.0227,
      "step": 22095
    },
    {
      "epoch": 0.22096,
      "grad_norm": 1.37216423647877,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 22096
    },
    {
      "epoch": 0.22097,
      "grad_norm": 1.6901582244876612,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 22097
    },
    {
      "epoch": 0.22098,
      "grad_norm": 1.093044749690386,
      "learning_rate": 0.003,
      "loss": 4.0238,
      "step": 22098
    },
    {
      "epoch": 0.22099,
      "grad_norm": 1.358581512018892,
      "learning_rate": 0.003,
      "loss": 4.0304,
      "step": 22099
    },
    {
      "epoch": 0.221,
      "grad_norm": 1.4734676480778928,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 22100
    },
    {
      "epoch": 0.22101,
      "grad_norm": 1.1247671678322673,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 22101
    },
    {
      "epoch": 0.22102,
      "grad_norm": 1.5760390934631532,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 22102
    },
    {
      "epoch": 0.22103,
      "grad_norm": 1.4089002766530339,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 22103
    },
    {
      "epoch": 0.22104,
      "grad_norm": 1.3022377509278193,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 22104
    },
    {
      "epoch": 0.22105,
      "grad_norm": 1.3879977222861182,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 22105
    },
    {
      "epoch": 0.22106,
      "grad_norm": 1.2008127244062086,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 22106
    },
    {
      "epoch": 0.22107,
      "grad_norm": 1.3873357243221884,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 22107
    },
    {
      "epoch": 0.22108,
      "grad_norm": 1.1851877743649342,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 22108
    },
    {
      "epoch": 0.22109,
      "grad_norm": 1.7553480872377545,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 22109
    },
    {
      "epoch": 0.2211,
      "grad_norm": 1.3783634274316432,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 22110
    },
    {
      "epoch": 0.22111,
      "grad_norm": 1.6975422750772762,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 22111
    },
    {
      "epoch": 0.22112,
      "grad_norm": 1.2598251750885294,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 22112
    },
    {
      "epoch": 0.22113,
      "grad_norm": 1.3488825181885948,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 22113
    },
    {
      "epoch": 0.22114,
      "grad_norm": 1.1707139578745651,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 22114
    },
    {
      "epoch": 0.22115,
      "grad_norm": 1.4076970079350817,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 22115
    },
    {
      "epoch": 0.22116,
      "grad_norm": 1.2035989898263957,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 22116
    },
    {
      "epoch": 0.22117,
      "grad_norm": 1.313990258923647,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 22117
    },
    {
      "epoch": 0.22118,
      "grad_norm": 1.4596496530806549,
      "learning_rate": 0.003,
      "loss": 4.0945,
      "step": 22118
    },
    {
      "epoch": 0.22119,
      "grad_norm": 1.3887979786596634,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 22119
    },
    {
      "epoch": 0.2212,
      "grad_norm": 1.2892662283743792,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 22120
    },
    {
      "epoch": 0.22121,
      "grad_norm": 1.3677195530909632,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 22121
    },
    {
      "epoch": 0.22122,
      "grad_norm": 1.0895280670903982,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 22122
    },
    {
      "epoch": 0.22123,
      "grad_norm": 1.7090804039595255,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 22123
    },
    {
      "epoch": 0.22124,
      "grad_norm": 1.2604584884079608,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 22124
    },
    {
      "epoch": 0.22125,
      "grad_norm": 1.5893034946385178,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 22125
    },
    {
      "epoch": 0.22126,
      "grad_norm": 1.8171719489776013,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 22126
    },
    {
      "epoch": 0.22127,
      "grad_norm": 1.079785537190597,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 22127
    },
    {
      "epoch": 0.22128,
      "grad_norm": 1.4253921256548443,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 22128
    },
    {
      "epoch": 0.22129,
      "grad_norm": 1.2752312555966736,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 22129
    },
    {
      "epoch": 0.2213,
      "grad_norm": 1.5912898189212894,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 22130
    },
    {
      "epoch": 0.22131,
      "grad_norm": 1.212781183286263,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 22131
    },
    {
      "epoch": 0.22132,
      "grad_norm": 1.6288246243696782,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 22132
    },
    {
      "epoch": 0.22133,
      "grad_norm": 1.364695014188986,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 22133
    },
    {
      "epoch": 0.22134,
      "grad_norm": 1.4164140732085475,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 22134
    },
    {
      "epoch": 0.22135,
      "grad_norm": 1.2534625007837457,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 22135
    },
    {
      "epoch": 0.22136,
      "grad_norm": 1.196897126890033,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 22136
    },
    {
      "epoch": 0.22137,
      "grad_norm": 1.502320147887795,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 22137
    },
    {
      "epoch": 0.22138,
      "grad_norm": 1.3266848797936062,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 22138
    },
    {
      "epoch": 0.22139,
      "grad_norm": 1.5001054181374793,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 22139
    },
    {
      "epoch": 0.2214,
      "grad_norm": 1.473091688110243,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 22140
    },
    {
      "epoch": 0.22141,
      "grad_norm": 1.3799393569486098,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 22141
    },
    {
      "epoch": 0.22142,
      "grad_norm": 1.4247663877479786,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 22142
    },
    {
      "epoch": 0.22143,
      "grad_norm": 1.0853672842292295,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 22143
    },
    {
      "epoch": 0.22144,
      "grad_norm": 1.6092501187145332,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 22144
    },
    {
      "epoch": 0.22145,
      "grad_norm": 1.1489883941066004,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 22145
    },
    {
      "epoch": 0.22146,
      "grad_norm": 1.5132661551487747,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 22146
    },
    {
      "epoch": 0.22147,
      "grad_norm": 1.3046988580626422,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 22147
    },
    {
      "epoch": 0.22148,
      "grad_norm": 1.4132640202686817,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 22148
    },
    {
      "epoch": 0.22149,
      "grad_norm": 1.56435744408823,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 22149
    },
    {
      "epoch": 0.2215,
      "grad_norm": 1.686833161313362,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 22150
    },
    {
      "epoch": 0.22151,
      "grad_norm": 1.3313560145204009,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 22151
    },
    {
      "epoch": 0.22152,
      "grad_norm": 1.3614519489465264,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 22152
    },
    {
      "epoch": 0.22153,
      "grad_norm": 1.4732927145628927,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 22153
    },
    {
      "epoch": 0.22154,
      "grad_norm": 1.3336301761906506,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 22154
    },
    {
      "epoch": 0.22155,
      "grad_norm": 1.2845741404694027,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 22155
    },
    {
      "epoch": 0.22156,
      "grad_norm": 1.362731838707528,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 22156
    },
    {
      "epoch": 0.22157,
      "grad_norm": 1.279897373506415,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 22157
    },
    {
      "epoch": 0.22158,
      "grad_norm": 1.3044229575795832,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 22158
    },
    {
      "epoch": 0.22159,
      "grad_norm": 1.3994874042785792,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 22159
    },
    {
      "epoch": 0.2216,
      "grad_norm": 1.274607452634766,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 22160
    },
    {
      "epoch": 0.22161,
      "grad_norm": 1.2943297130759688,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 22161
    },
    {
      "epoch": 0.22162,
      "grad_norm": 1.362183302837162,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 22162
    },
    {
      "epoch": 0.22163,
      "grad_norm": 1.5660664006583445,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 22163
    },
    {
      "epoch": 0.22164,
      "grad_norm": 1.3757306851918003,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 22164
    },
    {
      "epoch": 0.22165,
      "grad_norm": 1.4412752627541876,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 22165
    },
    {
      "epoch": 0.22166,
      "grad_norm": 1.5658425271051613,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 22166
    },
    {
      "epoch": 0.22167,
      "grad_norm": 1.6523052232881559,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 22167
    },
    {
      "epoch": 0.22168,
      "grad_norm": 1.1567559762437325,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 22168
    },
    {
      "epoch": 0.22169,
      "grad_norm": 1.4328603257986592,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 22169
    },
    {
      "epoch": 0.2217,
      "grad_norm": 1.3470626463872521,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 22170
    },
    {
      "epoch": 0.22171,
      "grad_norm": 1.4284284243004235,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 22171
    },
    {
      "epoch": 0.22172,
      "grad_norm": 1.1893474315038903,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 22172
    },
    {
      "epoch": 0.22173,
      "grad_norm": 1.4289456979469684,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 22173
    },
    {
      "epoch": 0.22174,
      "grad_norm": 1.057297957599464,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 22174
    },
    {
      "epoch": 0.22175,
      "grad_norm": 1.4257501085613518,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 22175
    },
    {
      "epoch": 0.22176,
      "grad_norm": 1.3262331040893989,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 22176
    },
    {
      "epoch": 0.22177,
      "grad_norm": 1.4321319336927019,
      "learning_rate": 0.003,
      "loss": 4.0916,
      "step": 22177
    },
    {
      "epoch": 0.22178,
      "grad_norm": 1.5739569894444765,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 22178
    },
    {
      "epoch": 0.22179,
      "grad_norm": 1.3596252113335465,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 22179
    },
    {
      "epoch": 0.2218,
      "grad_norm": 1.4105314530330597,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 22180
    },
    {
      "epoch": 0.22181,
      "grad_norm": 1.362897752269198,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 22181
    },
    {
      "epoch": 0.22182,
      "grad_norm": 1.4495305691893379,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 22182
    },
    {
      "epoch": 0.22183,
      "grad_norm": 1.3338275962033688,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 22183
    },
    {
      "epoch": 0.22184,
      "grad_norm": 1.3961143833545246,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 22184
    },
    {
      "epoch": 0.22185,
      "grad_norm": 1.2054362280955921,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 22185
    },
    {
      "epoch": 0.22186,
      "grad_norm": 1.4617105421301142,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 22186
    },
    {
      "epoch": 0.22187,
      "grad_norm": 1.3091955764656495,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 22187
    },
    {
      "epoch": 0.22188,
      "grad_norm": 1.3965949404899751,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 22188
    },
    {
      "epoch": 0.22189,
      "grad_norm": 1.7821877666649462,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 22189
    },
    {
      "epoch": 0.2219,
      "grad_norm": 1.295830581139901,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 22190
    },
    {
      "epoch": 0.22191,
      "grad_norm": 1.445223537820084,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 22191
    },
    {
      "epoch": 0.22192,
      "grad_norm": 1.3071233557061357,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 22192
    },
    {
      "epoch": 0.22193,
      "grad_norm": 1.2753165325582236,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 22193
    },
    {
      "epoch": 0.22194,
      "grad_norm": 1.5055299549441128,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 22194
    },
    {
      "epoch": 0.22195,
      "grad_norm": 1.1342693535556096,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 22195
    },
    {
      "epoch": 0.22196,
      "grad_norm": 1.3858983883526306,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 22196
    },
    {
      "epoch": 0.22197,
      "grad_norm": 1.6139655980808818,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 22197
    },
    {
      "epoch": 0.22198,
      "grad_norm": 1.370292636226592,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 22198
    },
    {
      "epoch": 0.22199,
      "grad_norm": 1.4124649114097427,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 22199
    },
    {
      "epoch": 0.222,
      "grad_norm": 1.293709147338817,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 22200
    },
    {
      "epoch": 0.22201,
      "grad_norm": 1.4533704007836947,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 22201
    },
    {
      "epoch": 0.22202,
      "grad_norm": 1.567765777013424,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 22202
    },
    {
      "epoch": 0.22203,
      "grad_norm": 1.279871157610687,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 22203
    },
    {
      "epoch": 0.22204,
      "grad_norm": 1.385971844695112,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 22204
    },
    {
      "epoch": 0.22205,
      "grad_norm": 1.2180530423728437,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 22205
    },
    {
      "epoch": 0.22206,
      "grad_norm": 1.515881036126505,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 22206
    },
    {
      "epoch": 0.22207,
      "grad_norm": 1.305396320762747,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 22207
    },
    {
      "epoch": 0.22208,
      "grad_norm": 1.907422079706361,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 22208
    },
    {
      "epoch": 0.22209,
      "grad_norm": 1.21270690214005,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 22209
    },
    {
      "epoch": 0.2221,
      "grad_norm": 1.3404071885874145,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 22210
    },
    {
      "epoch": 0.22211,
      "grad_norm": 1.2763298267522407,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 22211
    },
    {
      "epoch": 0.22212,
      "grad_norm": 1.4845486426731773,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 22212
    },
    {
      "epoch": 0.22213,
      "grad_norm": 1.1597212035481745,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 22213
    },
    {
      "epoch": 0.22214,
      "grad_norm": 1.1880382563326437,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 22214
    },
    {
      "epoch": 0.22215,
      "grad_norm": 1.4824757540258113,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 22215
    },
    {
      "epoch": 0.22216,
      "grad_norm": 1.3396597534142682,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 22216
    },
    {
      "epoch": 0.22217,
      "grad_norm": 1.4647900070278586,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 22217
    },
    {
      "epoch": 0.22218,
      "grad_norm": 1.1550937275289768,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 22218
    },
    {
      "epoch": 0.22219,
      "grad_norm": 1.4764592357132116,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 22219
    },
    {
      "epoch": 0.2222,
      "grad_norm": 1.7143856305412355,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 22220
    },
    {
      "epoch": 0.22221,
      "grad_norm": 1.5401353576578565,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 22221
    },
    {
      "epoch": 0.22222,
      "grad_norm": 1.3416022332549657,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 22222
    },
    {
      "epoch": 0.22223,
      "grad_norm": 1.4242574036135836,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 22223
    },
    {
      "epoch": 0.22224,
      "grad_norm": 1.1973358150028863,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 22224
    },
    {
      "epoch": 0.22225,
      "grad_norm": 1.2884641971513957,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 22225
    },
    {
      "epoch": 0.22226,
      "grad_norm": 1.424689662013279,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 22226
    },
    {
      "epoch": 0.22227,
      "grad_norm": 1.2250087938369854,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 22227
    },
    {
      "epoch": 0.22228,
      "grad_norm": 1.5751666610904282,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 22228
    },
    {
      "epoch": 0.22229,
      "grad_norm": 1.1494551555423635,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 22229
    },
    {
      "epoch": 0.2223,
      "grad_norm": 1.4688453613121737,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 22230
    },
    {
      "epoch": 0.22231,
      "grad_norm": 1.3699272268665363,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 22231
    },
    {
      "epoch": 0.22232,
      "grad_norm": 1.537492019731281,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 22232
    },
    {
      "epoch": 0.22233,
      "grad_norm": 1.0330617251989942,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 22233
    },
    {
      "epoch": 0.22234,
      "grad_norm": 1.4603664211895149,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 22234
    },
    {
      "epoch": 0.22235,
      "grad_norm": 1.2612609323108521,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 22235
    },
    {
      "epoch": 0.22236,
      "grad_norm": 1.566572150437271,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 22236
    },
    {
      "epoch": 0.22237,
      "grad_norm": 1.1865597068407436,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 22237
    },
    {
      "epoch": 0.22238,
      "grad_norm": 1.5926773568356256,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 22238
    },
    {
      "epoch": 0.22239,
      "grad_norm": 1.3216088389014924,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 22239
    },
    {
      "epoch": 0.2224,
      "grad_norm": 1.3976057640990152,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 22240
    },
    {
      "epoch": 0.22241,
      "grad_norm": 1.1496350226071208,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 22241
    },
    {
      "epoch": 0.22242,
      "grad_norm": 1.1580898237737296,
      "learning_rate": 0.003,
      "loss": 4.0315,
      "step": 22242
    },
    {
      "epoch": 0.22243,
      "grad_norm": 1.558939181904705,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 22243
    },
    {
      "epoch": 0.22244,
      "grad_norm": 1.3480081665211374,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 22244
    },
    {
      "epoch": 0.22245,
      "grad_norm": 1.4832107461145823,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 22245
    },
    {
      "epoch": 0.22246,
      "grad_norm": 1.4375844492231902,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 22246
    },
    {
      "epoch": 0.22247,
      "grad_norm": 1.5835609335873375,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 22247
    },
    {
      "epoch": 0.22248,
      "grad_norm": 1.4205729727127157,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 22248
    },
    {
      "epoch": 0.22249,
      "grad_norm": 1.6034593434402513,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 22249
    },
    {
      "epoch": 0.2225,
      "grad_norm": 1.3755340142441637,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 22250
    },
    {
      "epoch": 0.22251,
      "grad_norm": 1.4787785090773218,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 22251
    },
    {
      "epoch": 0.22252,
      "grad_norm": 1.3389467168011282,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 22252
    },
    {
      "epoch": 0.22253,
      "grad_norm": 1.4218076740248158,
      "learning_rate": 0.003,
      "loss": 4.0186,
      "step": 22253
    },
    {
      "epoch": 0.22254,
      "grad_norm": 1.20919633978248,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 22254
    },
    {
      "epoch": 0.22255,
      "grad_norm": 1.3390517491426153,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 22255
    },
    {
      "epoch": 0.22256,
      "grad_norm": 1.3712688774483852,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 22256
    },
    {
      "epoch": 0.22257,
      "grad_norm": 1.4771106606224886,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 22257
    },
    {
      "epoch": 0.22258,
      "grad_norm": 1.2131879469373439,
      "learning_rate": 0.003,
      "loss": 4.0285,
      "step": 22258
    },
    {
      "epoch": 0.22259,
      "grad_norm": 1.430070864931875,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 22259
    },
    {
      "epoch": 0.2226,
      "grad_norm": 1.1127157666257408,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 22260
    },
    {
      "epoch": 0.22261,
      "grad_norm": 1.4544316597453024,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 22261
    },
    {
      "epoch": 0.22262,
      "grad_norm": 1.1531901204919115,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 22262
    },
    {
      "epoch": 0.22263,
      "grad_norm": 1.5001268775560253,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 22263
    },
    {
      "epoch": 0.22264,
      "grad_norm": 1.388798488222104,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 22264
    },
    {
      "epoch": 0.22265,
      "grad_norm": 1.7432748344689732,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 22265
    },
    {
      "epoch": 0.22266,
      "grad_norm": 1.036013823414377,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 22266
    },
    {
      "epoch": 0.22267,
      "grad_norm": 1.2522459940721222,
      "learning_rate": 0.003,
      "loss": 4.023,
      "step": 22267
    },
    {
      "epoch": 0.22268,
      "grad_norm": 1.600614247711205,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 22268
    },
    {
      "epoch": 0.22269,
      "grad_norm": 1.3074802160740018,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 22269
    },
    {
      "epoch": 0.2227,
      "grad_norm": 1.4003433694119265,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 22270
    },
    {
      "epoch": 0.22271,
      "grad_norm": 1.3913342361356809,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 22271
    },
    {
      "epoch": 0.22272,
      "grad_norm": 1.2703376309876075,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 22272
    },
    {
      "epoch": 0.22273,
      "grad_norm": 1.500351184869441,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 22273
    },
    {
      "epoch": 0.22274,
      "grad_norm": 1.3699969478225038,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 22274
    },
    {
      "epoch": 0.22275,
      "grad_norm": 1.5223426025607947,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 22275
    },
    {
      "epoch": 0.22276,
      "grad_norm": 1.4033689564381397,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 22276
    },
    {
      "epoch": 0.22277,
      "grad_norm": 1.4601187426494533,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 22277
    },
    {
      "epoch": 0.22278,
      "grad_norm": 1.3888147570157965,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 22278
    },
    {
      "epoch": 0.22279,
      "grad_norm": 1.2404973573348714,
      "learning_rate": 0.003,
      "loss": 4.0279,
      "step": 22279
    },
    {
      "epoch": 0.2228,
      "grad_norm": 1.4555403624250252,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 22280
    },
    {
      "epoch": 0.22281,
      "grad_norm": 1.216160029817018,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 22281
    },
    {
      "epoch": 0.22282,
      "grad_norm": 1.4058869811595323,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 22282
    },
    {
      "epoch": 0.22283,
      "grad_norm": 1.2758387158134383,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 22283
    },
    {
      "epoch": 0.22284,
      "grad_norm": 1.2207169266410474,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 22284
    },
    {
      "epoch": 0.22285,
      "grad_norm": 1.412295249924611,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 22285
    },
    {
      "epoch": 0.22286,
      "grad_norm": 1.3732494203503594,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 22286
    },
    {
      "epoch": 0.22287,
      "grad_norm": 1.3692184472505888,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 22287
    },
    {
      "epoch": 0.22288,
      "grad_norm": 1.3759831387998156,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 22288
    },
    {
      "epoch": 0.22289,
      "grad_norm": 1.2595618500573362,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 22289
    },
    {
      "epoch": 0.2229,
      "grad_norm": 1.6814775146119474,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 22290
    },
    {
      "epoch": 0.22291,
      "grad_norm": 1.3953387253409102,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 22291
    },
    {
      "epoch": 0.22292,
      "grad_norm": 1.7317713148633327,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 22292
    },
    {
      "epoch": 0.22293,
      "grad_norm": 1.2844604150024208,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 22293
    },
    {
      "epoch": 0.22294,
      "grad_norm": 1.4291699204264559,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 22294
    },
    {
      "epoch": 0.22295,
      "grad_norm": 1.5150130594599611,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 22295
    },
    {
      "epoch": 0.22296,
      "grad_norm": 1.3082256750627732,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 22296
    },
    {
      "epoch": 0.22297,
      "grad_norm": 1.3875180604880697,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 22297
    },
    {
      "epoch": 0.22298,
      "grad_norm": 1.3343820918708291,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 22298
    },
    {
      "epoch": 0.22299,
      "grad_norm": 1.3183796205639757,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 22299
    },
    {
      "epoch": 0.223,
      "grad_norm": 1.2510060244764296,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 22300
    },
    {
      "epoch": 0.22301,
      "grad_norm": 1.3716672418427478,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 22301
    },
    {
      "epoch": 0.22302,
      "grad_norm": 1.2727298642982772,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 22302
    },
    {
      "epoch": 0.22303,
      "grad_norm": 1.488032564045237,
      "learning_rate": 0.003,
      "loss": 3.9981,
      "step": 22303
    },
    {
      "epoch": 0.22304,
      "grad_norm": 1.4141357281721683,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 22304
    },
    {
      "epoch": 0.22305,
      "grad_norm": 1.3282741557292026,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 22305
    },
    {
      "epoch": 0.22306,
      "grad_norm": 1.6074682935491493,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 22306
    },
    {
      "epoch": 0.22307,
      "grad_norm": 1.393713371586898,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 22307
    },
    {
      "epoch": 0.22308,
      "grad_norm": 1.307301396543611,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 22308
    },
    {
      "epoch": 0.22309,
      "grad_norm": 1.3461150372960098,
      "learning_rate": 0.003,
      "loss": 4.0292,
      "step": 22309
    },
    {
      "epoch": 0.2231,
      "grad_norm": 1.3792478838566686,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 22310
    },
    {
      "epoch": 0.22311,
      "grad_norm": 1.4901498414341812,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 22311
    },
    {
      "epoch": 0.22312,
      "grad_norm": 1.485376306335906,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 22312
    },
    {
      "epoch": 0.22313,
      "grad_norm": 1.4336359621977617,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 22313
    },
    {
      "epoch": 0.22314,
      "grad_norm": 1.4223755734156847,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 22314
    },
    {
      "epoch": 0.22315,
      "grad_norm": 1.2564372412889622,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 22315
    },
    {
      "epoch": 0.22316,
      "grad_norm": 1.2536470066684344,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 22316
    },
    {
      "epoch": 0.22317,
      "grad_norm": 1.5964660266455553,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 22317
    },
    {
      "epoch": 0.22318,
      "grad_norm": 1.1143586140085437,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 22318
    },
    {
      "epoch": 0.22319,
      "grad_norm": 1.8202754570487123,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 22319
    },
    {
      "epoch": 0.2232,
      "grad_norm": 1.1241456027960381,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 22320
    },
    {
      "epoch": 0.22321,
      "grad_norm": 1.4648023777671297,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 22321
    },
    {
      "epoch": 0.22322,
      "grad_norm": 1.4663830323431575,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 22322
    },
    {
      "epoch": 0.22323,
      "grad_norm": 1.1849994982508425,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 22323
    },
    {
      "epoch": 0.22324,
      "grad_norm": 1.371011478385604,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 22324
    },
    {
      "epoch": 0.22325,
      "grad_norm": 1.2628273562560974,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 22325
    },
    {
      "epoch": 0.22326,
      "grad_norm": 1.3872267826245577,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 22326
    },
    {
      "epoch": 0.22327,
      "grad_norm": 1.3685132988827458,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 22327
    },
    {
      "epoch": 0.22328,
      "grad_norm": 1.1028904550243237,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 22328
    },
    {
      "epoch": 0.22329,
      "grad_norm": 1.3618561520281762,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 22329
    },
    {
      "epoch": 0.2233,
      "grad_norm": 1.1445670049550696,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 22330
    },
    {
      "epoch": 0.22331,
      "grad_norm": 1.4022987676702379,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 22331
    },
    {
      "epoch": 0.22332,
      "grad_norm": 1.5599619432209575,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 22332
    },
    {
      "epoch": 0.22333,
      "grad_norm": 1.878794688612716,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 22333
    },
    {
      "epoch": 0.22334,
      "grad_norm": 1.3050525918215132,
      "learning_rate": 0.003,
      "loss": 4.1006,
      "step": 22334
    },
    {
      "epoch": 0.22335,
      "grad_norm": 1.336371350588149,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 22335
    },
    {
      "epoch": 0.22336,
      "grad_norm": 1.208607895788011,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 22336
    },
    {
      "epoch": 0.22337,
      "grad_norm": 1.6457839457667778,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 22337
    },
    {
      "epoch": 0.22338,
      "grad_norm": 1.4127662001557084,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 22338
    },
    {
      "epoch": 0.22339,
      "grad_norm": 1.376876386888607,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 22339
    },
    {
      "epoch": 0.2234,
      "grad_norm": 1.4570261167898284,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 22340
    },
    {
      "epoch": 0.22341,
      "grad_norm": 1.211546957361986,
      "learning_rate": 0.003,
      "loss": 4.0262,
      "step": 22341
    },
    {
      "epoch": 0.22342,
      "grad_norm": 1.3270741919713742,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 22342
    },
    {
      "epoch": 0.22343,
      "grad_norm": 1.397195166385156,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 22343
    },
    {
      "epoch": 0.22344,
      "grad_norm": 1.4763850759105857,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 22344
    },
    {
      "epoch": 0.22345,
      "grad_norm": 1.3501067025222995,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 22345
    },
    {
      "epoch": 0.22346,
      "grad_norm": 1.279130243319669,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 22346
    },
    {
      "epoch": 0.22347,
      "grad_norm": 1.352191370618253,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 22347
    },
    {
      "epoch": 0.22348,
      "grad_norm": 1.4361298838127596,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 22348
    },
    {
      "epoch": 0.22349,
      "grad_norm": 1.2251495326785093,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 22349
    },
    {
      "epoch": 0.2235,
      "grad_norm": 1.3910755839615114,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 22350
    },
    {
      "epoch": 0.22351,
      "grad_norm": 1.3125911961348145,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 22351
    },
    {
      "epoch": 0.22352,
      "grad_norm": 1.5656196258299442,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 22352
    },
    {
      "epoch": 0.22353,
      "grad_norm": 1.1531059389308478,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 22353
    },
    {
      "epoch": 0.22354,
      "grad_norm": 1.4007020106647727,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 22354
    },
    {
      "epoch": 0.22355,
      "grad_norm": 1.2354692843363417,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 22355
    },
    {
      "epoch": 0.22356,
      "grad_norm": 1.590234105863267,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 22356
    },
    {
      "epoch": 0.22357,
      "grad_norm": 1.3509492624833452,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 22357
    },
    {
      "epoch": 0.22358,
      "grad_norm": 1.5275962199582724,
      "learning_rate": 0.003,
      "loss": 4.0959,
      "step": 22358
    },
    {
      "epoch": 0.22359,
      "grad_norm": 1.209076278040971,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 22359
    },
    {
      "epoch": 0.2236,
      "grad_norm": 1.5185901971561602,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 22360
    },
    {
      "epoch": 0.22361,
      "grad_norm": 1.532891108295708,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 22361
    },
    {
      "epoch": 0.22362,
      "grad_norm": 1.5865126216653196,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 22362
    },
    {
      "epoch": 0.22363,
      "grad_norm": 1.270551236014406,
      "learning_rate": 0.003,
      "loss": 4.0306,
      "step": 22363
    },
    {
      "epoch": 0.22364,
      "grad_norm": 1.4329458288291133,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 22364
    },
    {
      "epoch": 0.22365,
      "grad_norm": 1.4096876484606953,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 22365
    },
    {
      "epoch": 0.22366,
      "grad_norm": 1.2142737042399505,
      "learning_rate": 0.003,
      "loss": 4.0186,
      "step": 22366
    },
    {
      "epoch": 0.22367,
      "grad_norm": 1.2614164115511466,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 22367
    },
    {
      "epoch": 0.22368,
      "grad_norm": 1.2051666649693737,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 22368
    },
    {
      "epoch": 0.22369,
      "grad_norm": 1.590599099523706,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 22369
    },
    {
      "epoch": 0.2237,
      "grad_norm": 1.1253766088301476,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 22370
    },
    {
      "epoch": 0.22371,
      "grad_norm": 1.5552956946763759,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 22371
    },
    {
      "epoch": 0.22372,
      "grad_norm": 1.3491888281502122,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 22372
    },
    {
      "epoch": 0.22373,
      "grad_norm": 1.3715023603072742,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 22373
    },
    {
      "epoch": 0.22374,
      "grad_norm": 1.3616540152037222,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 22374
    },
    {
      "epoch": 0.22375,
      "grad_norm": 1.177630374725669,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 22375
    },
    {
      "epoch": 0.22376,
      "grad_norm": 1.3473079774707535,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 22376
    },
    {
      "epoch": 0.22377,
      "grad_norm": 1.3537114757041389,
      "learning_rate": 0.003,
      "loss": 4.0107,
      "step": 22377
    },
    {
      "epoch": 0.22378,
      "grad_norm": 1.5142584430469408,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 22378
    },
    {
      "epoch": 0.22379,
      "grad_norm": 1.4421599606051467,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 22379
    },
    {
      "epoch": 0.2238,
      "grad_norm": 1.5732608949994045,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 22380
    },
    {
      "epoch": 0.22381,
      "grad_norm": 1.2532908478937852,
      "learning_rate": 0.003,
      "loss": 4.0205,
      "step": 22381
    },
    {
      "epoch": 0.22382,
      "grad_norm": 1.4183163010238564,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 22382
    },
    {
      "epoch": 0.22383,
      "grad_norm": 1.3639666346782449,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 22383
    },
    {
      "epoch": 0.22384,
      "grad_norm": 1.1796621129050568,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 22384
    },
    {
      "epoch": 0.22385,
      "grad_norm": 1.500752473050197,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 22385
    },
    {
      "epoch": 0.22386,
      "grad_norm": 1.299806986589891,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 22386
    },
    {
      "epoch": 0.22387,
      "grad_norm": 1.3628678395012956,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 22387
    },
    {
      "epoch": 0.22388,
      "grad_norm": 1.2145588780699879,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 22388
    },
    {
      "epoch": 0.22389,
      "grad_norm": 1.4354220668775866,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 22389
    },
    {
      "epoch": 0.2239,
      "grad_norm": 1.1753083460488114,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 22390
    },
    {
      "epoch": 0.22391,
      "grad_norm": 1.4450015905639986,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 22391
    },
    {
      "epoch": 0.22392,
      "grad_norm": 1.2241119805245964,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 22392
    },
    {
      "epoch": 0.22393,
      "grad_norm": 1.5411689184760144,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 22393
    },
    {
      "epoch": 0.22394,
      "grad_norm": 1.279268771909209,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 22394
    },
    {
      "epoch": 0.22395,
      "grad_norm": 1.7097093223307864,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 22395
    },
    {
      "epoch": 0.22396,
      "grad_norm": 1.4902652883348568,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 22396
    },
    {
      "epoch": 0.22397,
      "grad_norm": 1.6807845293587749,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 22397
    },
    {
      "epoch": 0.22398,
      "grad_norm": 1.463050957154932,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 22398
    },
    {
      "epoch": 0.22399,
      "grad_norm": 1.4815674497276095,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 22399
    },
    {
      "epoch": 0.224,
      "grad_norm": 1.2927791947078966,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 22400
    },
    {
      "epoch": 0.22401,
      "grad_norm": 1.3616976324807482,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 22401
    },
    {
      "epoch": 0.22402,
      "grad_norm": 1.3534396398652109,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 22402
    },
    {
      "epoch": 0.22403,
      "grad_norm": 1.286676318927532,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 22403
    },
    {
      "epoch": 0.22404,
      "grad_norm": 1.3768423147509237,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 22404
    },
    {
      "epoch": 0.22405,
      "grad_norm": 1.1635484876103743,
      "learning_rate": 0.003,
      "loss": 4.0283,
      "step": 22405
    },
    {
      "epoch": 0.22406,
      "grad_norm": 1.3140438382708493,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 22406
    },
    {
      "epoch": 0.22407,
      "grad_norm": 1.3712284265314465,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 22407
    },
    {
      "epoch": 0.22408,
      "grad_norm": 1.1867705031365885,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 22408
    },
    {
      "epoch": 0.22409,
      "grad_norm": 1.1574130541241712,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 22409
    },
    {
      "epoch": 0.2241,
      "grad_norm": 1.2587040600977484,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 22410
    },
    {
      "epoch": 0.22411,
      "grad_norm": 1.3772805242419397,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 22411
    },
    {
      "epoch": 0.22412,
      "grad_norm": 1.4911914513850275,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 22412
    },
    {
      "epoch": 0.22413,
      "grad_norm": 1.2854427393908894,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 22413
    },
    {
      "epoch": 0.22414,
      "grad_norm": 1.5290408026536262,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 22414
    },
    {
      "epoch": 0.22415,
      "grad_norm": 1.7649135826504136,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 22415
    },
    {
      "epoch": 0.22416,
      "grad_norm": 1.1977781109776366,
      "learning_rate": 0.003,
      "loss": 4.0157,
      "step": 22416
    },
    {
      "epoch": 0.22417,
      "grad_norm": 1.293867976503959,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 22417
    },
    {
      "epoch": 0.22418,
      "grad_norm": 1.630642197670863,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 22418
    },
    {
      "epoch": 0.22419,
      "grad_norm": 1.4741383401959558,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 22419
    },
    {
      "epoch": 0.2242,
      "grad_norm": 1.2986315664115111,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 22420
    },
    {
      "epoch": 0.22421,
      "grad_norm": 1.3154813967475658,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 22421
    },
    {
      "epoch": 0.22422,
      "grad_norm": 1.5969126924782886,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 22422
    },
    {
      "epoch": 0.22423,
      "grad_norm": 1.1954027795687365,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 22423
    },
    {
      "epoch": 0.22424,
      "grad_norm": 1.3933097252550717,
      "learning_rate": 0.003,
      "loss": 4.0215,
      "step": 22424
    },
    {
      "epoch": 0.22425,
      "grad_norm": 1.04149405593488,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 22425
    },
    {
      "epoch": 0.22426,
      "grad_norm": 1.6382424532271298,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 22426
    },
    {
      "epoch": 0.22427,
      "grad_norm": 1.2027355428887891,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 22427
    },
    {
      "epoch": 0.22428,
      "grad_norm": 1.525117510391192,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 22428
    },
    {
      "epoch": 0.22429,
      "grad_norm": 1.426170080120505,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 22429
    },
    {
      "epoch": 0.2243,
      "grad_norm": 1.221673236416552,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 22430
    },
    {
      "epoch": 0.22431,
      "grad_norm": 1.5885387264266417,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 22431
    },
    {
      "epoch": 0.22432,
      "grad_norm": 1.2345654548738305,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 22432
    },
    {
      "epoch": 0.22433,
      "grad_norm": 1.5798732787241532,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 22433
    },
    {
      "epoch": 0.22434,
      "grad_norm": 1.3471715133315665,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 22434
    },
    {
      "epoch": 0.22435,
      "grad_norm": 1.3854398618572008,
      "learning_rate": 0.003,
      "loss": 4.0863,
      "step": 22435
    },
    {
      "epoch": 0.22436,
      "grad_norm": 1.3779282433032272,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 22436
    },
    {
      "epoch": 0.22437,
      "grad_norm": 1.5169256570561949,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 22437
    },
    {
      "epoch": 0.22438,
      "grad_norm": 1.4647617156380661,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 22438
    },
    {
      "epoch": 0.22439,
      "grad_norm": 1.2177903916745678,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 22439
    },
    {
      "epoch": 0.2244,
      "grad_norm": 1.4261753366390773,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 22440
    },
    {
      "epoch": 0.22441,
      "grad_norm": 1.1943996436170505,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 22441
    },
    {
      "epoch": 0.22442,
      "grad_norm": 1.650187508331776,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 22442
    },
    {
      "epoch": 0.22443,
      "grad_norm": 1.2273277856836666,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 22443
    },
    {
      "epoch": 0.22444,
      "grad_norm": 1.563436201042193,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 22444
    },
    {
      "epoch": 0.22445,
      "grad_norm": 1.2477425551662131,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 22445
    },
    {
      "epoch": 0.22446,
      "grad_norm": 1.459547516657567,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 22446
    },
    {
      "epoch": 0.22447,
      "grad_norm": 1.237427311213376,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 22447
    },
    {
      "epoch": 0.22448,
      "grad_norm": 1.582809499011086,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 22448
    },
    {
      "epoch": 0.22449,
      "grad_norm": 1.189896586373214,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 22449
    },
    {
      "epoch": 0.2245,
      "grad_norm": 1.9901290265561125,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 22450
    },
    {
      "epoch": 0.22451,
      "grad_norm": 0.9412658196252865,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 22451
    },
    {
      "epoch": 0.22452,
      "grad_norm": 1.518676999335754,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 22452
    },
    {
      "epoch": 0.22453,
      "grad_norm": 1.457530808236886,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 22453
    },
    {
      "epoch": 0.22454,
      "grad_norm": 1.2408992964524617,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 22454
    },
    {
      "epoch": 0.22455,
      "grad_norm": 1.5144145116345065,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 22455
    },
    {
      "epoch": 0.22456,
      "grad_norm": 1.2176355775449714,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 22456
    },
    {
      "epoch": 0.22457,
      "grad_norm": 1.578395937149088,
      "learning_rate": 0.003,
      "loss": 4.0782,
      "step": 22457
    },
    {
      "epoch": 0.22458,
      "grad_norm": 1.5822231270531464,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 22458
    },
    {
      "epoch": 0.22459,
      "grad_norm": 1.3717071898734479,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 22459
    },
    {
      "epoch": 0.2246,
      "grad_norm": 1.6047488710069298,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 22460
    },
    {
      "epoch": 0.22461,
      "grad_norm": 1.1601238559678562,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 22461
    },
    {
      "epoch": 0.22462,
      "grad_norm": 1.3735324716539978,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 22462
    },
    {
      "epoch": 0.22463,
      "grad_norm": 1.1844585024097143,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 22463
    },
    {
      "epoch": 0.22464,
      "grad_norm": 1.5323727852590565,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 22464
    },
    {
      "epoch": 0.22465,
      "grad_norm": 1.2472750170538465,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 22465
    },
    {
      "epoch": 0.22466,
      "grad_norm": 1.5245760113116062,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 22466
    },
    {
      "epoch": 0.22467,
      "grad_norm": 1.304704002247532,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 22467
    },
    {
      "epoch": 0.22468,
      "grad_norm": 1.2921716053001555,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 22468
    },
    {
      "epoch": 0.22469,
      "grad_norm": 1.4888616674466386,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 22469
    },
    {
      "epoch": 0.2247,
      "grad_norm": 1.4069052197621945,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 22470
    },
    {
      "epoch": 0.22471,
      "grad_norm": 1.2099995067052336,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 22471
    },
    {
      "epoch": 0.22472,
      "grad_norm": 1.4435501977018366,
      "learning_rate": 0.003,
      "loss": 4.093,
      "step": 22472
    },
    {
      "epoch": 0.22473,
      "grad_norm": 1.1184531466394783,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 22473
    },
    {
      "epoch": 0.22474,
      "grad_norm": 1.42146086377367,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 22474
    },
    {
      "epoch": 0.22475,
      "grad_norm": 1.2721389239149017,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 22475
    },
    {
      "epoch": 0.22476,
      "grad_norm": 1.5888927772852055,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 22476
    },
    {
      "epoch": 0.22477,
      "grad_norm": 1.2323738375159854,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 22477
    },
    {
      "epoch": 0.22478,
      "grad_norm": 1.8480494916768275,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 22478
    },
    {
      "epoch": 0.22479,
      "grad_norm": 1.039961285327314,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 22479
    },
    {
      "epoch": 0.2248,
      "grad_norm": 1.3240798637812554,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 22480
    },
    {
      "epoch": 0.22481,
      "grad_norm": 1.138999054570129,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 22481
    },
    {
      "epoch": 0.22482,
      "grad_norm": 1.6015792048656434,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 22482
    },
    {
      "epoch": 0.22483,
      "grad_norm": 1.2811650768298755,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 22483
    },
    {
      "epoch": 0.22484,
      "grad_norm": 1.4050836723792486,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 22484
    },
    {
      "epoch": 0.22485,
      "grad_norm": 1.401263027541136,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 22485
    },
    {
      "epoch": 0.22486,
      "grad_norm": 1.3808037000615874,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 22486
    },
    {
      "epoch": 0.22487,
      "grad_norm": 1.0181960800189673,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 22487
    },
    {
      "epoch": 0.22488,
      "grad_norm": 1.4367139726826885,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 22488
    },
    {
      "epoch": 0.22489,
      "grad_norm": 1.2553541203514906,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 22489
    },
    {
      "epoch": 0.2249,
      "grad_norm": 1.61788577800353,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 22490
    },
    {
      "epoch": 0.22491,
      "grad_norm": 1.3568049795287969,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 22491
    },
    {
      "epoch": 0.22492,
      "grad_norm": 1.9470979963846207,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 22492
    },
    {
      "epoch": 0.22493,
      "grad_norm": 0.999144675700383,
      "learning_rate": 0.003,
      "loss": 4.0201,
      "step": 22493
    },
    {
      "epoch": 0.22494,
      "grad_norm": 1.4578098514241924,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 22494
    },
    {
      "epoch": 0.22495,
      "grad_norm": 1.5647267704157142,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 22495
    },
    {
      "epoch": 0.22496,
      "grad_norm": 1.2162677300726918,
      "learning_rate": 0.003,
      "loss": 4.0926,
      "step": 22496
    },
    {
      "epoch": 0.22497,
      "grad_norm": 1.3655534797327769,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 22497
    },
    {
      "epoch": 0.22498,
      "grad_norm": 1.3563680721289868,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 22498
    },
    {
      "epoch": 0.22499,
      "grad_norm": 1.314289618928932,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 22499
    },
    {
      "epoch": 0.225,
      "grad_norm": 1.2412727130259196,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 22500
    },
    {
      "epoch": 0.22501,
      "grad_norm": 1.407664446284243,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 22501
    },
    {
      "epoch": 0.22502,
      "grad_norm": 1.1452879237859683,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 22502
    },
    {
      "epoch": 0.22503,
      "grad_norm": 1.681630374166743,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 22503
    },
    {
      "epoch": 0.22504,
      "grad_norm": 1.5208064356506332,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 22504
    },
    {
      "epoch": 0.22505,
      "grad_norm": 1.9040342674938469,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 22505
    },
    {
      "epoch": 0.22506,
      "grad_norm": 0.9039834512748233,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 22506
    },
    {
      "epoch": 0.22507,
      "grad_norm": 1.1289527494797158,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 22507
    },
    {
      "epoch": 0.22508,
      "grad_norm": 1.4814373394057403,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 22508
    },
    {
      "epoch": 0.22509,
      "grad_norm": 1.251319290489344,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 22509
    },
    {
      "epoch": 0.2251,
      "grad_norm": 1.5562981775401703,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 22510
    },
    {
      "epoch": 0.22511,
      "grad_norm": 1.3051826964869775,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 22511
    },
    {
      "epoch": 0.22512,
      "grad_norm": 1.6229240732536356,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 22512
    },
    {
      "epoch": 0.22513,
      "grad_norm": 1.3042315359192964,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 22513
    },
    {
      "epoch": 0.22514,
      "grad_norm": 1.4675307982498782,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 22514
    },
    {
      "epoch": 0.22515,
      "grad_norm": 1.6252411003222522,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 22515
    },
    {
      "epoch": 0.22516,
      "grad_norm": 1.0812707864662017,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 22516
    },
    {
      "epoch": 0.22517,
      "grad_norm": 1.3767570950385721,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 22517
    },
    {
      "epoch": 0.22518,
      "grad_norm": 1.4763795196532024,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 22518
    },
    {
      "epoch": 0.22519,
      "grad_norm": 1.4993782616115667,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 22519
    },
    {
      "epoch": 0.2252,
      "grad_norm": 1.3537945752575407,
      "learning_rate": 0.003,
      "loss": 4.0225,
      "step": 22520
    },
    {
      "epoch": 0.22521,
      "grad_norm": 1.3059347310012368,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 22521
    },
    {
      "epoch": 0.22522,
      "grad_norm": 1.4442979723184586,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 22522
    },
    {
      "epoch": 0.22523,
      "grad_norm": 1.2281159833260513,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 22523
    },
    {
      "epoch": 0.22524,
      "grad_norm": 1.4702059963582377,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 22524
    },
    {
      "epoch": 0.22525,
      "grad_norm": 1.2684940320670692,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 22525
    },
    {
      "epoch": 0.22526,
      "grad_norm": 1.5278258401588285,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 22526
    },
    {
      "epoch": 0.22527,
      "grad_norm": 1.2524578589278004,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 22527
    },
    {
      "epoch": 0.22528,
      "grad_norm": 1.520965957159605,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 22528
    },
    {
      "epoch": 0.22529,
      "grad_norm": 1.2582154302824615,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 22529
    },
    {
      "epoch": 0.2253,
      "grad_norm": 1.3265404712352202,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 22530
    },
    {
      "epoch": 0.22531,
      "grad_norm": 1.4400352868471908,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 22531
    },
    {
      "epoch": 0.22532,
      "grad_norm": 1.1005549267295203,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 22532
    },
    {
      "epoch": 0.22533,
      "grad_norm": 1.3708248175839755,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 22533
    },
    {
      "epoch": 0.22534,
      "grad_norm": 1.3036441616788939,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 22534
    },
    {
      "epoch": 0.22535,
      "grad_norm": 1.3604752364885113,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 22535
    },
    {
      "epoch": 0.22536,
      "grad_norm": 1.516204836371943,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 22536
    },
    {
      "epoch": 0.22537,
      "grad_norm": 1.3429535542996642,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 22537
    },
    {
      "epoch": 0.22538,
      "grad_norm": 1.5562585708934367,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 22538
    },
    {
      "epoch": 0.22539,
      "grad_norm": 1.36622916218615,
      "learning_rate": 0.003,
      "loss": 4.0348,
      "step": 22539
    },
    {
      "epoch": 0.2254,
      "grad_norm": 1.3929020539285881,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 22540
    },
    {
      "epoch": 0.22541,
      "grad_norm": 1.3343444959931712,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 22541
    },
    {
      "epoch": 0.22542,
      "grad_norm": 1.2985053494629357,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 22542
    },
    {
      "epoch": 0.22543,
      "grad_norm": 1.4792491364252658,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 22543
    },
    {
      "epoch": 0.22544,
      "grad_norm": 1.1959731134436147,
      "learning_rate": 0.003,
      "loss": 4.0315,
      "step": 22544
    },
    {
      "epoch": 0.22545,
      "grad_norm": 1.4724669305686342,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 22545
    },
    {
      "epoch": 0.22546,
      "grad_norm": 1.1761225949887955,
      "learning_rate": 0.003,
      "loss": 4.0217,
      "step": 22546
    },
    {
      "epoch": 0.22547,
      "grad_norm": 1.834610039270939,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 22547
    },
    {
      "epoch": 0.22548,
      "grad_norm": 1.3844490196846349,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 22548
    },
    {
      "epoch": 0.22549,
      "grad_norm": 1.3630463640737251,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 22549
    },
    {
      "epoch": 0.2255,
      "grad_norm": 1.3436021286818365,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 22550
    },
    {
      "epoch": 0.22551,
      "grad_norm": 1.126599897912119,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 22551
    },
    {
      "epoch": 0.22552,
      "grad_norm": 1.3418933257662087,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 22552
    },
    {
      "epoch": 0.22553,
      "grad_norm": 1.5600682091030313,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 22553
    },
    {
      "epoch": 0.22554,
      "grad_norm": 1.27398638381522,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 22554
    },
    {
      "epoch": 0.22555,
      "grad_norm": 1.362818561676482,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 22555
    },
    {
      "epoch": 0.22556,
      "grad_norm": 1.4124436107553786,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 22556
    },
    {
      "epoch": 0.22557,
      "grad_norm": 1.266220853086558,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 22557
    },
    {
      "epoch": 0.22558,
      "grad_norm": 1.426748845952024,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 22558
    },
    {
      "epoch": 0.22559,
      "grad_norm": 1.4317525615950089,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 22559
    },
    {
      "epoch": 0.2256,
      "grad_norm": 1.5982695632874984,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 22560
    },
    {
      "epoch": 0.22561,
      "grad_norm": 1.1566627687428226,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 22561
    },
    {
      "epoch": 0.22562,
      "grad_norm": 1.521382546062163,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 22562
    },
    {
      "epoch": 0.22563,
      "grad_norm": 1.3820456222187198,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 22563
    },
    {
      "epoch": 0.22564,
      "grad_norm": 1.5823811522579028,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 22564
    },
    {
      "epoch": 0.22565,
      "grad_norm": 1.500567266962183,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 22565
    },
    {
      "epoch": 0.22566,
      "grad_norm": 1.3315173635786572,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 22566
    },
    {
      "epoch": 0.22567,
      "grad_norm": 1.282929593176905,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 22567
    },
    {
      "epoch": 0.22568,
      "grad_norm": 1.1853758748353218,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 22568
    },
    {
      "epoch": 0.22569,
      "grad_norm": 1.621094200564941,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 22569
    },
    {
      "epoch": 0.2257,
      "grad_norm": 1.046054275055912,
      "learning_rate": 0.003,
      "loss": 4.0111,
      "step": 22570
    },
    {
      "epoch": 0.22571,
      "grad_norm": 1.6577346130398705,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 22571
    },
    {
      "epoch": 0.22572,
      "grad_norm": 1.2758186688416238,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 22572
    },
    {
      "epoch": 0.22573,
      "grad_norm": 1.353975079659158,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 22573
    },
    {
      "epoch": 0.22574,
      "grad_norm": 1.5463450147980808,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 22574
    },
    {
      "epoch": 0.22575,
      "grad_norm": 1.3217539352653154,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 22575
    },
    {
      "epoch": 0.22576,
      "grad_norm": 1.3559801332602135,
      "learning_rate": 0.003,
      "loss": 4.0252,
      "step": 22576
    },
    {
      "epoch": 0.22577,
      "grad_norm": 1.3706392752172982,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 22577
    },
    {
      "epoch": 0.22578,
      "grad_norm": 1.3520523513731522,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 22578
    },
    {
      "epoch": 0.22579,
      "grad_norm": 1.4081367473851485,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 22579
    },
    {
      "epoch": 0.2258,
      "grad_norm": 1.39604049224214,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 22580
    },
    {
      "epoch": 0.22581,
      "grad_norm": 0.981022986380773,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 22581
    },
    {
      "epoch": 0.22582,
      "grad_norm": 1.47186423594832,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 22582
    },
    {
      "epoch": 0.22583,
      "grad_norm": 1.2225686547530379,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 22583
    },
    {
      "epoch": 0.22584,
      "grad_norm": 1.67045896861936,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 22584
    },
    {
      "epoch": 0.22585,
      "grad_norm": 1.4326038394314937,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 22585
    },
    {
      "epoch": 0.22586,
      "grad_norm": 1.172198451848986,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 22586
    },
    {
      "epoch": 0.22587,
      "grad_norm": 1.7019830866366101,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 22587
    },
    {
      "epoch": 0.22588,
      "grad_norm": 1.4090667334413138,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 22588
    },
    {
      "epoch": 0.22589,
      "grad_norm": 1.3827337695248374,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 22589
    },
    {
      "epoch": 0.2259,
      "grad_norm": 1.396615880672321,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 22590
    },
    {
      "epoch": 0.22591,
      "grad_norm": 1.4282167560265029,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 22591
    },
    {
      "epoch": 0.22592,
      "grad_norm": 1.3580130177571552,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 22592
    },
    {
      "epoch": 0.22593,
      "grad_norm": 1.3110588395587528,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 22593
    },
    {
      "epoch": 0.22594,
      "grad_norm": 1.249002226714932,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 22594
    },
    {
      "epoch": 0.22595,
      "grad_norm": 1.3160504276882146,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 22595
    },
    {
      "epoch": 0.22596,
      "grad_norm": 1.5536352519305991,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 22596
    },
    {
      "epoch": 0.22597,
      "grad_norm": 1.3787725503319284,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 22597
    },
    {
      "epoch": 0.22598,
      "grad_norm": 1.3394153645608453,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 22598
    },
    {
      "epoch": 0.22599,
      "grad_norm": 1.931446622847827,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 22599
    },
    {
      "epoch": 0.226,
      "grad_norm": 1.06935267886322,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 22600
    },
    {
      "epoch": 0.22601,
      "grad_norm": 1.6100699849910967,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 22601
    },
    {
      "epoch": 0.22602,
      "grad_norm": 1.4042224864761799,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 22602
    },
    {
      "epoch": 0.22603,
      "grad_norm": 1.555809248028907,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 22603
    },
    {
      "epoch": 0.22604,
      "grad_norm": 1.3116509073024893,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 22604
    },
    {
      "epoch": 0.22605,
      "grad_norm": 1.2766533829004472,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 22605
    },
    {
      "epoch": 0.22606,
      "grad_norm": 1.3952547284964383,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 22606
    },
    {
      "epoch": 0.22607,
      "grad_norm": 1.2181291883730188,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 22607
    },
    {
      "epoch": 0.22608,
      "grad_norm": 1.2698844722199734,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 22608
    },
    {
      "epoch": 0.22609,
      "grad_norm": 1.2171669532200873,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 22609
    },
    {
      "epoch": 0.2261,
      "grad_norm": 1.2821859204849533,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 22610
    },
    {
      "epoch": 0.22611,
      "grad_norm": 1.697919090405346,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 22611
    },
    {
      "epoch": 0.22612,
      "grad_norm": 1.3176091918110264,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 22612
    },
    {
      "epoch": 0.22613,
      "grad_norm": 1.717003828516172,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 22613
    },
    {
      "epoch": 0.22614,
      "grad_norm": 1.2890734962230233,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 22614
    },
    {
      "epoch": 0.22615,
      "grad_norm": 1.4817984017078425,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 22615
    },
    {
      "epoch": 0.22616,
      "grad_norm": 1.3572368270716264,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 22616
    },
    {
      "epoch": 0.22617,
      "grad_norm": 1.2028491274514344,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 22617
    },
    {
      "epoch": 0.22618,
      "grad_norm": 1.5242588280099894,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 22618
    },
    {
      "epoch": 0.22619,
      "grad_norm": 1.1436578845230336,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 22619
    },
    {
      "epoch": 0.2262,
      "grad_norm": 1.3811672576952223,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 22620
    },
    {
      "epoch": 0.22621,
      "grad_norm": 1.2953398770733349,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 22621
    },
    {
      "epoch": 0.22622,
      "grad_norm": 1.3431482846945377,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 22622
    },
    {
      "epoch": 0.22623,
      "grad_norm": 1.1372747823405238,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 22623
    },
    {
      "epoch": 0.22624,
      "grad_norm": 1.4052776077364333,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 22624
    },
    {
      "epoch": 0.22625,
      "grad_norm": 1.5489455531528675,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 22625
    },
    {
      "epoch": 0.22626,
      "grad_norm": 1.4301500501178745,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 22626
    },
    {
      "epoch": 0.22627,
      "grad_norm": 1.2121269881757493,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 22627
    },
    {
      "epoch": 0.22628,
      "grad_norm": 1.3274725839129087,
      "learning_rate": 0.003,
      "loss": 4.0763,
      "step": 22628
    },
    {
      "epoch": 0.22629,
      "grad_norm": 1.2245562570378739,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 22629
    },
    {
      "epoch": 0.2263,
      "grad_norm": 1.5273451682665296,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 22630
    },
    {
      "epoch": 0.22631,
      "grad_norm": 1.279537679015296,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 22631
    },
    {
      "epoch": 0.22632,
      "grad_norm": 1.3074789891509484,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 22632
    },
    {
      "epoch": 0.22633,
      "grad_norm": 1.2957812732762986,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 22633
    },
    {
      "epoch": 0.22634,
      "grad_norm": 1.3134530453328914,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 22634
    },
    {
      "epoch": 0.22635,
      "grad_norm": 1.4303878963643286,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 22635
    },
    {
      "epoch": 0.22636,
      "grad_norm": 1.4462539730138868,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 22636
    },
    {
      "epoch": 0.22637,
      "grad_norm": 1.348818328916082,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 22637
    },
    {
      "epoch": 0.22638,
      "grad_norm": 1.5418063711053704,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 22638
    },
    {
      "epoch": 0.22639,
      "grad_norm": 1.341868096211138,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 22639
    },
    {
      "epoch": 0.2264,
      "grad_norm": 1.5288327353418887,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 22640
    },
    {
      "epoch": 0.22641,
      "grad_norm": 1.1756331603935857,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 22641
    },
    {
      "epoch": 0.22642,
      "grad_norm": 1.7303641780587524,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 22642
    },
    {
      "epoch": 0.22643,
      "grad_norm": 1.675870297737419,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 22643
    },
    {
      "epoch": 0.22644,
      "grad_norm": 1.2903553844701023,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 22644
    },
    {
      "epoch": 0.22645,
      "grad_norm": 1.3563239081724314,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 22645
    },
    {
      "epoch": 0.22646,
      "grad_norm": 1.4468430507102832,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 22646
    },
    {
      "epoch": 0.22647,
      "grad_norm": 1.3259292173535169,
      "learning_rate": 0.003,
      "loss": 4.0144,
      "step": 22647
    },
    {
      "epoch": 0.22648,
      "grad_norm": 1.4509995946776215,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 22648
    },
    {
      "epoch": 0.22649,
      "grad_norm": 1.4294367517999493,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 22649
    },
    {
      "epoch": 0.2265,
      "grad_norm": 1.3177940217664728,
      "learning_rate": 0.003,
      "loss": 4.0214,
      "step": 22650
    },
    {
      "epoch": 0.22651,
      "grad_norm": 1.2960509928184432,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 22651
    },
    {
      "epoch": 0.22652,
      "grad_norm": 1.372940217189916,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 22652
    },
    {
      "epoch": 0.22653,
      "grad_norm": 1.2565000630216578,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 22653
    },
    {
      "epoch": 0.22654,
      "grad_norm": 1.4910404307810698,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 22654
    },
    {
      "epoch": 0.22655,
      "grad_norm": 1.4885483591490194,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 22655
    },
    {
      "epoch": 0.22656,
      "grad_norm": 1.2593129639972969,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 22656
    },
    {
      "epoch": 0.22657,
      "grad_norm": 1.4690399184727785,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 22657
    },
    {
      "epoch": 0.22658,
      "grad_norm": 1.2480041075236903,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 22658
    },
    {
      "epoch": 0.22659,
      "grad_norm": 1.5116333980188945,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 22659
    },
    {
      "epoch": 0.2266,
      "grad_norm": 1.398692993616762,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 22660
    },
    {
      "epoch": 0.22661,
      "grad_norm": 1.2407464218984163,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 22661
    },
    {
      "epoch": 0.22662,
      "grad_norm": 1.2793070754179567,
      "learning_rate": 0.003,
      "loss": 4.0329,
      "step": 22662
    },
    {
      "epoch": 0.22663,
      "grad_norm": 1.4347225094402987,
      "learning_rate": 0.003,
      "loss": 4.0333,
      "step": 22663
    },
    {
      "epoch": 0.22664,
      "grad_norm": 1.4493947841028387,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 22664
    },
    {
      "epoch": 0.22665,
      "grad_norm": 1.267139810879918,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 22665
    },
    {
      "epoch": 0.22666,
      "grad_norm": 1.466853579717287,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 22666
    },
    {
      "epoch": 0.22667,
      "grad_norm": 1.2689565681150594,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 22667
    },
    {
      "epoch": 0.22668,
      "grad_norm": 1.7870831055293204,
      "learning_rate": 0.003,
      "loss": 4.033,
      "step": 22668
    },
    {
      "epoch": 0.22669,
      "grad_norm": 1.320122758108175,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 22669
    },
    {
      "epoch": 0.2267,
      "grad_norm": 1.5005533778962779,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 22670
    },
    {
      "epoch": 0.22671,
      "grad_norm": 1.2797600664813926,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 22671
    },
    {
      "epoch": 0.22672,
      "grad_norm": 1.4017309372932885,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 22672
    },
    {
      "epoch": 0.22673,
      "grad_norm": 1.471016236023522,
      "learning_rate": 0.003,
      "loss": 4.015,
      "step": 22673
    },
    {
      "epoch": 0.22674,
      "grad_norm": 1.2984861405736587,
      "learning_rate": 0.003,
      "loss": 4.0313,
      "step": 22674
    },
    {
      "epoch": 0.22675,
      "grad_norm": 1.4666226372746194,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 22675
    },
    {
      "epoch": 0.22676,
      "grad_norm": 1.057823960025185,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 22676
    },
    {
      "epoch": 0.22677,
      "grad_norm": 1.6549346121176827,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 22677
    },
    {
      "epoch": 0.22678,
      "grad_norm": 1.1585263518139224,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 22678
    },
    {
      "epoch": 0.22679,
      "grad_norm": 1.4487039629449114,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 22679
    },
    {
      "epoch": 0.2268,
      "grad_norm": 1.325337987181312,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 22680
    },
    {
      "epoch": 0.22681,
      "grad_norm": 1.4040369566590036,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 22681
    },
    {
      "epoch": 0.22682,
      "grad_norm": 1.462985041434129,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 22682
    },
    {
      "epoch": 0.22683,
      "grad_norm": 1.4942923170342381,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 22683
    },
    {
      "epoch": 0.22684,
      "grad_norm": 1.364818885404756,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 22684
    },
    {
      "epoch": 0.22685,
      "grad_norm": 1.3381450883990378,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 22685
    },
    {
      "epoch": 0.22686,
      "grad_norm": 1.4139855403410388,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 22686
    },
    {
      "epoch": 0.22687,
      "grad_norm": 1.283220290729647,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 22687
    },
    {
      "epoch": 0.22688,
      "grad_norm": 1.1335905218354496,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 22688
    },
    {
      "epoch": 0.22689,
      "grad_norm": 1.5542724056246844,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 22689
    },
    {
      "epoch": 0.2269,
      "grad_norm": 1.1854055008264828,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 22690
    },
    {
      "epoch": 0.22691,
      "grad_norm": 1.671151129810973,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 22691
    },
    {
      "epoch": 0.22692,
      "grad_norm": 1.4417314469168165,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 22692
    },
    {
      "epoch": 0.22693,
      "grad_norm": 1.4484208424018798,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 22693
    },
    {
      "epoch": 0.22694,
      "grad_norm": 1.226954351094281,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 22694
    },
    {
      "epoch": 0.22695,
      "grad_norm": 1.4044018349908938,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 22695
    },
    {
      "epoch": 0.22696,
      "grad_norm": 1.2895303456317728,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 22696
    },
    {
      "epoch": 0.22697,
      "grad_norm": 1.3062241257820577,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 22697
    },
    {
      "epoch": 0.22698,
      "grad_norm": 1.2504406893191973,
      "learning_rate": 0.003,
      "loss": 4.0289,
      "step": 22698
    },
    {
      "epoch": 0.22699,
      "grad_norm": 1.3097849809865796,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 22699
    },
    {
      "epoch": 0.227,
      "grad_norm": 1.4801030945943425,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 22700
    },
    {
      "epoch": 0.22701,
      "grad_norm": 1.4569176256875882,
      "learning_rate": 0.003,
      "loss": 4.0819,
      "step": 22701
    },
    {
      "epoch": 0.22702,
      "grad_norm": 1.2775531479377897,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 22702
    },
    {
      "epoch": 0.22703,
      "grad_norm": 1.2828919103497352,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 22703
    },
    {
      "epoch": 0.22704,
      "grad_norm": 1.4993856414685836,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 22704
    },
    {
      "epoch": 0.22705,
      "grad_norm": 1.4399009328303836,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 22705
    },
    {
      "epoch": 0.22706,
      "grad_norm": 1.3271074450231812,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 22706
    },
    {
      "epoch": 0.22707,
      "grad_norm": 1.6489946598338783,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 22707
    },
    {
      "epoch": 0.22708,
      "grad_norm": 1.6158436493310295,
      "learning_rate": 0.003,
      "loss": 4.0164,
      "step": 22708
    },
    {
      "epoch": 0.22709,
      "grad_norm": 1.2037217417560346,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 22709
    },
    {
      "epoch": 0.2271,
      "grad_norm": 1.5531011340860523,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 22710
    },
    {
      "epoch": 0.22711,
      "grad_norm": 1.425885017699002,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 22711
    },
    {
      "epoch": 0.22712,
      "grad_norm": 1.2002858912145136,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 22712
    },
    {
      "epoch": 0.22713,
      "grad_norm": 1.430470037719689,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 22713
    },
    {
      "epoch": 0.22714,
      "grad_norm": 1.636990043396574,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 22714
    },
    {
      "epoch": 0.22715,
      "grad_norm": 1.1099921951607443,
      "learning_rate": 0.003,
      "loss": 4.0313,
      "step": 22715
    },
    {
      "epoch": 0.22716,
      "grad_norm": 1.5754495429131838,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 22716
    },
    {
      "epoch": 0.22717,
      "grad_norm": 1.3833784390319153,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 22717
    },
    {
      "epoch": 0.22718,
      "grad_norm": 1.324141264588012,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 22718
    },
    {
      "epoch": 0.22719,
      "grad_norm": 1.6033767035520747,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 22719
    },
    {
      "epoch": 0.2272,
      "grad_norm": 1.1777043738733637,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 22720
    },
    {
      "epoch": 0.22721,
      "grad_norm": 1.2804792369525233,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 22721
    },
    {
      "epoch": 0.22722,
      "grad_norm": 1.369117070641414,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 22722
    },
    {
      "epoch": 0.22723,
      "grad_norm": 1.3113031825756496,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 22723
    },
    {
      "epoch": 0.22724,
      "grad_norm": 1.505547392760889,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 22724
    },
    {
      "epoch": 0.22725,
      "grad_norm": 1.3722136634024067,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 22725
    },
    {
      "epoch": 0.22726,
      "grad_norm": 1.3270259799031803,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 22726
    },
    {
      "epoch": 0.22727,
      "grad_norm": 1.4597975785251383,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 22727
    },
    {
      "epoch": 0.22728,
      "grad_norm": 1.4347900265936195,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 22728
    },
    {
      "epoch": 0.22729,
      "grad_norm": 1.784177762127377,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 22729
    },
    {
      "epoch": 0.2273,
      "grad_norm": 1.119019016203389,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 22730
    },
    {
      "epoch": 0.22731,
      "grad_norm": 1.298704994703112,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 22731
    },
    {
      "epoch": 0.22732,
      "grad_norm": 1.6145253889696385,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 22732
    },
    {
      "epoch": 0.22733,
      "grad_norm": 1.009332942569693,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 22733
    },
    {
      "epoch": 0.22734,
      "grad_norm": 1.8197940796945007,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 22734
    },
    {
      "epoch": 0.22735,
      "grad_norm": 1.278516691136997,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 22735
    },
    {
      "epoch": 0.22736,
      "grad_norm": 1.3859433272846655,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 22736
    },
    {
      "epoch": 0.22737,
      "grad_norm": 1.5925513615482265,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 22737
    },
    {
      "epoch": 0.22738,
      "grad_norm": 1.21997029965742,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 22738
    },
    {
      "epoch": 0.22739,
      "grad_norm": 1.193074086408873,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 22739
    },
    {
      "epoch": 0.2274,
      "grad_norm": 1.5345763007239461,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 22740
    },
    {
      "epoch": 0.22741,
      "grad_norm": 1.3989942151178685,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 22741
    },
    {
      "epoch": 0.22742,
      "grad_norm": 1.1211151425682409,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 22742
    },
    {
      "epoch": 0.22743,
      "grad_norm": 1.5827480809997463,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 22743
    },
    {
      "epoch": 0.22744,
      "grad_norm": 1.0515175238725416,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 22744
    },
    {
      "epoch": 0.22745,
      "grad_norm": 1.5357721268409588,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 22745
    },
    {
      "epoch": 0.22746,
      "grad_norm": 1.3106788831436889,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 22746
    },
    {
      "epoch": 0.22747,
      "grad_norm": 1.5928894097233948,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 22747
    },
    {
      "epoch": 0.22748,
      "grad_norm": 1.7929971445680843,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 22748
    },
    {
      "epoch": 0.22749,
      "grad_norm": 1.3772959045324669,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 22749
    },
    {
      "epoch": 0.2275,
      "grad_norm": 1.3321748543633538,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 22750
    },
    {
      "epoch": 0.22751,
      "grad_norm": 1.556758363868485,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 22751
    },
    {
      "epoch": 0.22752,
      "grad_norm": 1.3206813624861218,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 22752
    },
    {
      "epoch": 0.22753,
      "grad_norm": 1.2152563756581984,
      "learning_rate": 0.003,
      "loss": 4.015,
      "step": 22753
    },
    {
      "epoch": 0.22754,
      "grad_norm": 1.544337384684816,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 22754
    },
    {
      "epoch": 0.22755,
      "grad_norm": 1.1660014149356033,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 22755
    },
    {
      "epoch": 0.22756,
      "grad_norm": 1.4894832389062866,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 22756
    },
    {
      "epoch": 0.22757,
      "grad_norm": 1.1558412605789987,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 22757
    },
    {
      "epoch": 0.22758,
      "grad_norm": 2.0043005159119542,
      "learning_rate": 0.003,
      "loss": 4.0836,
      "step": 22758
    },
    {
      "epoch": 0.22759,
      "grad_norm": 1.319424743303632,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 22759
    },
    {
      "epoch": 0.2276,
      "grad_norm": 1.414897525724111,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 22760
    },
    {
      "epoch": 0.22761,
      "grad_norm": 1.2681178406931337,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 22761
    },
    {
      "epoch": 0.22762,
      "grad_norm": 1.6161084330047177,
      "learning_rate": 0.003,
      "loss": 4.098,
      "step": 22762
    },
    {
      "epoch": 0.22763,
      "grad_norm": 1.321014957858152,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 22763
    },
    {
      "epoch": 0.22764,
      "grad_norm": 1.4416448047819612,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 22764
    },
    {
      "epoch": 0.22765,
      "grad_norm": 1.3369984308013854,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 22765
    },
    {
      "epoch": 0.22766,
      "grad_norm": 1.2179835583670722,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 22766
    },
    {
      "epoch": 0.22767,
      "grad_norm": 1.8279900236105129,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 22767
    },
    {
      "epoch": 0.22768,
      "grad_norm": 1.213602320810376,
      "learning_rate": 0.003,
      "loss": 4.0873,
      "step": 22768
    },
    {
      "epoch": 0.22769,
      "grad_norm": 1.706127453137258,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 22769
    },
    {
      "epoch": 0.2277,
      "grad_norm": 1.1029359724223382,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 22770
    },
    {
      "epoch": 0.22771,
      "grad_norm": 1.1785664252188008,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 22771
    },
    {
      "epoch": 0.22772,
      "grad_norm": 1.541480008398517,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 22772
    },
    {
      "epoch": 0.22773,
      "grad_norm": 1.2234950246108043,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 22773
    },
    {
      "epoch": 0.22774,
      "grad_norm": 1.4002365824636416,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 22774
    },
    {
      "epoch": 0.22775,
      "grad_norm": 1.192025795982403,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 22775
    },
    {
      "epoch": 0.22776,
      "grad_norm": 1.5961151778908518,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 22776
    },
    {
      "epoch": 0.22777,
      "grad_norm": 1.2336588502501258,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 22777
    },
    {
      "epoch": 0.22778,
      "grad_norm": 1.5534815136535918,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 22778
    },
    {
      "epoch": 0.22779,
      "grad_norm": 1.39564969745224,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 22779
    },
    {
      "epoch": 0.2278,
      "grad_norm": 1.1800592622090422,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 22780
    },
    {
      "epoch": 0.22781,
      "grad_norm": 1.2848512272701038,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 22781
    },
    {
      "epoch": 0.22782,
      "grad_norm": 1.693306261388428,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 22782
    },
    {
      "epoch": 0.22783,
      "grad_norm": 1.1690051634797456,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 22783
    },
    {
      "epoch": 0.22784,
      "grad_norm": 1.6415685237943014,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 22784
    },
    {
      "epoch": 0.22785,
      "grad_norm": 1.0812010529935254,
      "learning_rate": 0.003,
      "loss": 4.0505,
      "step": 22785
    },
    {
      "epoch": 0.22786,
      "grad_norm": 1.4193971463042083,
      "learning_rate": 0.003,
      "loss": 4.0198,
      "step": 22786
    },
    {
      "epoch": 0.22787,
      "grad_norm": 1.2659239259709194,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 22787
    },
    {
      "epoch": 0.22788,
      "grad_norm": 1.291019294829421,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 22788
    },
    {
      "epoch": 0.22789,
      "grad_norm": 1.3837939932139751,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 22789
    },
    {
      "epoch": 0.2279,
      "grad_norm": 1.1401013654455603,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 22790
    },
    {
      "epoch": 0.22791,
      "grad_norm": 1.7155477896803115,
      "learning_rate": 0.003,
      "loss": 4.0848,
      "step": 22791
    },
    {
      "epoch": 0.22792,
      "grad_norm": 1.440121875217581,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 22792
    },
    {
      "epoch": 0.22793,
      "grad_norm": 1.1993406304256435,
      "learning_rate": 0.003,
      "loss": 4.0278,
      "step": 22793
    },
    {
      "epoch": 0.22794,
      "grad_norm": 1.499868426529385,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 22794
    },
    {
      "epoch": 0.22795,
      "grad_norm": 1.5788791898780075,
      "learning_rate": 0.003,
      "loss": 4.1136,
      "step": 22795
    },
    {
      "epoch": 0.22796,
      "grad_norm": 1.3204890194348837,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 22796
    },
    {
      "epoch": 0.22797,
      "grad_norm": 1.3316817258024336,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 22797
    },
    {
      "epoch": 0.22798,
      "grad_norm": 1.5794818947540428,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 22798
    },
    {
      "epoch": 0.22799,
      "grad_norm": 1.3802337991872735,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 22799
    },
    {
      "epoch": 0.228,
      "grad_norm": 1.1524532883825167,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 22800
    },
    {
      "epoch": 0.22801,
      "grad_norm": 1.5056397569716424,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 22801
    },
    {
      "epoch": 0.22802,
      "grad_norm": 1.2668787670242587,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 22802
    },
    {
      "epoch": 0.22803,
      "grad_norm": 1.547512171508173,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 22803
    },
    {
      "epoch": 0.22804,
      "grad_norm": 1.4285910030488986,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 22804
    },
    {
      "epoch": 0.22805,
      "grad_norm": 1.2549136389487845,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 22805
    },
    {
      "epoch": 0.22806,
      "grad_norm": 1.3894638291845278,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 22806
    },
    {
      "epoch": 0.22807,
      "grad_norm": 1.1302487368272556,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 22807
    },
    {
      "epoch": 0.22808,
      "grad_norm": 1.4340720715074096,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 22808
    },
    {
      "epoch": 0.22809,
      "grad_norm": 1.0800651849205487,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 22809
    },
    {
      "epoch": 0.2281,
      "grad_norm": 1.5043339147042853,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 22810
    },
    {
      "epoch": 0.22811,
      "grad_norm": 1.4439286772604305,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 22811
    },
    {
      "epoch": 0.22812,
      "grad_norm": 1.295586844963577,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 22812
    },
    {
      "epoch": 0.22813,
      "grad_norm": 1.256345973930996,
      "learning_rate": 0.003,
      "loss": 4.0296,
      "step": 22813
    },
    {
      "epoch": 0.22814,
      "grad_norm": 1.3967598449398202,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 22814
    },
    {
      "epoch": 0.22815,
      "grad_norm": 1.5467402999818534,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 22815
    },
    {
      "epoch": 0.22816,
      "grad_norm": 1.206409645763743,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 22816
    },
    {
      "epoch": 0.22817,
      "grad_norm": 1.5334760029471426,
      "learning_rate": 0.003,
      "loss": 4.0261,
      "step": 22817
    },
    {
      "epoch": 0.22818,
      "grad_norm": 1.3003110524721266,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 22818
    },
    {
      "epoch": 0.22819,
      "grad_norm": 1.2978256734104014,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 22819
    },
    {
      "epoch": 0.2282,
      "grad_norm": 1.43622507146669,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 22820
    },
    {
      "epoch": 0.22821,
      "grad_norm": 1.3712213718832809,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 22821
    },
    {
      "epoch": 0.22822,
      "grad_norm": 1.417984294755835,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 22822
    },
    {
      "epoch": 0.22823,
      "grad_norm": 1.4077668676520234,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 22823
    },
    {
      "epoch": 0.22824,
      "grad_norm": 1.5500761369323117,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 22824
    },
    {
      "epoch": 0.22825,
      "grad_norm": 1.4119672397602403,
      "learning_rate": 0.003,
      "loss": 4.0299,
      "step": 22825
    },
    {
      "epoch": 0.22826,
      "grad_norm": 1.4647195023239734,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 22826
    },
    {
      "epoch": 0.22827,
      "grad_norm": 1.5921915457413027,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 22827
    },
    {
      "epoch": 0.22828,
      "grad_norm": 1.175582461663063,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 22828
    },
    {
      "epoch": 0.22829,
      "grad_norm": 1.4089405036060674,
      "learning_rate": 0.003,
      "loss": 4.0273,
      "step": 22829
    },
    {
      "epoch": 0.2283,
      "grad_norm": 1.297291736433117,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 22830
    },
    {
      "epoch": 0.22831,
      "grad_norm": 1.4566281352328887,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 22831
    },
    {
      "epoch": 0.22832,
      "grad_norm": 1.2910121694672738,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 22832
    },
    {
      "epoch": 0.22833,
      "grad_norm": 1.4999606670033645,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 22833
    },
    {
      "epoch": 0.22834,
      "grad_norm": 1.2919907599807343,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 22834
    },
    {
      "epoch": 0.22835,
      "grad_norm": 1.3725973490557473,
      "learning_rate": 0.003,
      "loss": 4.0108,
      "step": 22835
    },
    {
      "epoch": 0.22836,
      "grad_norm": 1.4844399503007777,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 22836
    },
    {
      "epoch": 0.22837,
      "grad_norm": 1.2469283102796385,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 22837
    },
    {
      "epoch": 0.22838,
      "grad_norm": 1.5636974982482015,
      "learning_rate": 0.003,
      "loss": 4.1003,
      "step": 22838
    },
    {
      "epoch": 0.22839,
      "grad_norm": 1.1210394448162195,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 22839
    },
    {
      "epoch": 0.2284,
      "grad_norm": 1.3892339988548474,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 22840
    },
    {
      "epoch": 0.22841,
      "grad_norm": 1.4362686688740154,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 22841
    },
    {
      "epoch": 0.22842,
      "grad_norm": 1.4266672801548619,
      "learning_rate": 0.003,
      "loss": 4.0214,
      "step": 22842
    },
    {
      "epoch": 0.22843,
      "grad_norm": 1.5466784780867424,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 22843
    },
    {
      "epoch": 0.22844,
      "grad_norm": 1.2903751481327705,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 22844
    },
    {
      "epoch": 0.22845,
      "grad_norm": 1.4654023642465581,
      "learning_rate": 0.003,
      "loss": 4.0246,
      "step": 22845
    },
    {
      "epoch": 0.22846,
      "grad_norm": 1.346459458656631,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 22846
    },
    {
      "epoch": 0.22847,
      "grad_norm": 1.3280473063892437,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 22847
    },
    {
      "epoch": 0.22848,
      "grad_norm": 1.4291368851027328,
      "learning_rate": 0.003,
      "loss": 4.0869,
      "step": 22848
    },
    {
      "epoch": 0.22849,
      "grad_norm": 1.4043713279974217,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 22849
    },
    {
      "epoch": 0.2285,
      "grad_norm": 1.3779039823011487,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 22850
    },
    {
      "epoch": 0.22851,
      "grad_norm": 1.3587430350694432,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 22851
    },
    {
      "epoch": 0.22852,
      "grad_norm": 1.6487730120652673,
      "learning_rate": 0.003,
      "loss": 4.0206,
      "step": 22852
    },
    {
      "epoch": 0.22853,
      "grad_norm": 1.4694544507663712,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 22853
    },
    {
      "epoch": 0.22854,
      "grad_norm": 1.12884649591588,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 22854
    },
    {
      "epoch": 0.22855,
      "grad_norm": 1.38024726019724,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 22855
    },
    {
      "epoch": 0.22856,
      "grad_norm": 1.5847228529615476,
      "learning_rate": 0.003,
      "loss": 4.0969,
      "step": 22856
    },
    {
      "epoch": 0.22857,
      "grad_norm": 1.3817997015632875,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 22857
    },
    {
      "epoch": 0.22858,
      "grad_norm": 1.2984459309332053,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 22858
    },
    {
      "epoch": 0.22859,
      "grad_norm": 1.3225040141194015,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 22859
    },
    {
      "epoch": 0.2286,
      "grad_norm": 1.3417509700768964,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 22860
    },
    {
      "epoch": 0.22861,
      "grad_norm": 1.379279945212339,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 22861
    },
    {
      "epoch": 0.22862,
      "grad_norm": 1.4159060377000825,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 22862
    },
    {
      "epoch": 0.22863,
      "grad_norm": 1.3739216498139826,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 22863
    },
    {
      "epoch": 0.22864,
      "grad_norm": 1.3711404096302824,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 22864
    },
    {
      "epoch": 0.22865,
      "grad_norm": 1.3548666387373958,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 22865
    },
    {
      "epoch": 0.22866,
      "grad_norm": 1.4234263619205612,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 22866
    },
    {
      "epoch": 0.22867,
      "grad_norm": 1.4163917455959771,
      "learning_rate": 0.003,
      "loss": 4.0303,
      "step": 22867
    },
    {
      "epoch": 0.22868,
      "grad_norm": 1.3975264010223116,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 22868
    },
    {
      "epoch": 0.22869,
      "grad_norm": 1.3400257381170582,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 22869
    },
    {
      "epoch": 0.2287,
      "grad_norm": 1.176772266744342,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 22870
    },
    {
      "epoch": 0.22871,
      "grad_norm": 1.2383970809001525,
      "learning_rate": 0.003,
      "loss": 4.0211,
      "step": 22871
    },
    {
      "epoch": 0.22872,
      "grad_norm": 1.333779816791978,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 22872
    },
    {
      "epoch": 0.22873,
      "grad_norm": 1.3122308334037585,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 22873
    },
    {
      "epoch": 0.22874,
      "grad_norm": 1.364839848854087,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 22874
    },
    {
      "epoch": 0.22875,
      "grad_norm": 1.60895472345263,
      "learning_rate": 0.003,
      "loss": 4.0181,
      "step": 22875
    },
    {
      "epoch": 0.22876,
      "grad_norm": 1.4546016585888952,
      "learning_rate": 0.003,
      "loss": 4.0176,
      "step": 22876
    },
    {
      "epoch": 0.22877,
      "grad_norm": 1.3148719840863317,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 22877
    },
    {
      "epoch": 0.22878,
      "grad_norm": 1.392800427880589,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 22878
    },
    {
      "epoch": 0.22879,
      "grad_norm": 1.2159669449941268,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 22879
    },
    {
      "epoch": 0.2288,
      "grad_norm": 1.3155450912631526,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 22880
    },
    {
      "epoch": 0.22881,
      "grad_norm": 1.3312969299174433,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 22881
    },
    {
      "epoch": 0.22882,
      "grad_norm": 1.3002076627978985,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 22882
    },
    {
      "epoch": 0.22883,
      "grad_norm": 1.2661962155110686,
      "learning_rate": 0.003,
      "loss": 4.0257,
      "step": 22883
    },
    {
      "epoch": 0.22884,
      "grad_norm": 1.3450381125574369,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 22884
    },
    {
      "epoch": 0.22885,
      "grad_norm": 1.339805467192508,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 22885
    },
    {
      "epoch": 0.22886,
      "grad_norm": 1.4809585998745207,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 22886
    },
    {
      "epoch": 0.22887,
      "grad_norm": 1.162420575641933,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 22887
    },
    {
      "epoch": 0.22888,
      "grad_norm": 1.7168073083852065,
      "learning_rate": 0.003,
      "loss": 4.0275,
      "step": 22888
    },
    {
      "epoch": 0.22889,
      "grad_norm": 1.3367044738475202,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 22889
    },
    {
      "epoch": 0.2289,
      "grad_norm": 1.5371849697118831,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 22890
    },
    {
      "epoch": 0.22891,
      "grad_norm": 1.5009762139201366,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 22891
    },
    {
      "epoch": 0.22892,
      "grad_norm": 1.3995811493563683,
      "learning_rate": 0.003,
      "loss": 4.0258,
      "step": 22892
    },
    {
      "epoch": 0.22893,
      "grad_norm": 1.283622543765794,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 22893
    },
    {
      "epoch": 0.22894,
      "grad_norm": 1.3468349419979935,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 22894
    },
    {
      "epoch": 0.22895,
      "grad_norm": 1.408573251621135,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 22895
    },
    {
      "epoch": 0.22896,
      "grad_norm": 1.5397575419788314,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 22896
    },
    {
      "epoch": 0.22897,
      "grad_norm": 1.3886972589073308,
      "learning_rate": 0.003,
      "loss": 4.0189,
      "step": 22897
    },
    {
      "epoch": 0.22898,
      "grad_norm": 1.3145346513690945,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 22898
    },
    {
      "epoch": 0.22899,
      "grad_norm": 1.518352083951177,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 22899
    },
    {
      "epoch": 0.229,
      "grad_norm": 1.1782434367507242,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 22900
    },
    {
      "epoch": 0.22901,
      "grad_norm": 1.5610356730680504,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 22901
    },
    {
      "epoch": 0.22902,
      "grad_norm": 1.614204415348391,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 22902
    },
    {
      "epoch": 0.22903,
      "grad_norm": 1.5406045784687927,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 22903
    },
    {
      "epoch": 0.22904,
      "grad_norm": 1.170225557948686,
      "learning_rate": 0.003,
      "loss": 4.0139,
      "step": 22904
    },
    {
      "epoch": 0.22905,
      "grad_norm": 1.4559568853599187,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 22905
    },
    {
      "epoch": 0.22906,
      "grad_norm": 1.3330053767174777,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 22906
    },
    {
      "epoch": 0.22907,
      "grad_norm": 1.447820799841297,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 22907
    },
    {
      "epoch": 0.22908,
      "grad_norm": 1.5794080995014115,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 22908
    },
    {
      "epoch": 0.22909,
      "grad_norm": 1.3907820618217002,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 22909
    },
    {
      "epoch": 0.2291,
      "grad_norm": 1.425359578922196,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 22910
    },
    {
      "epoch": 0.22911,
      "grad_norm": 1.58183015644073,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 22911
    },
    {
      "epoch": 0.22912,
      "grad_norm": 1.2066358438817404,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 22912
    },
    {
      "epoch": 0.22913,
      "grad_norm": 1.419448658573045,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 22913
    },
    {
      "epoch": 0.22914,
      "grad_norm": 1.175171432981624,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 22914
    },
    {
      "epoch": 0.22915,
      "grad_norm": 1.235510844625249,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 22915
    },
    {
      "epoch": 0.22916,
      "grad_norm": 1.3219877756870189,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 22916
    },
    {
      "epoch": 0.22917,
      "grad_norm": 1.4428557821369838,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 22917
    },
    {
      "epoch": 0.22918,
      "grad_norm": 1.1858581525800913,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 22918
    },
    {
      "epoch": 0.22919,
      "grad_norm": 1.6051683298512007,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 22919
    },
    {
      "epoch": 0.2292,
      "grad_norm": 1.3032162782771097,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 22920
    },
    {
      "epoch": 0.22921,
      "grad_norm": 1.2802020113054942,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 22921
    },
    {
      "epoch": 0.22922,
      "grad_norm": 1.6529037593282858,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 22922
    },
    {
      "epoch": 0.22923,
      "grad_norm": 1.2700712794416136,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 22923
    },
    {
      "epoch": 0.22924,
      "grad_norm": 1.4259915181989866,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 22924
    },
    {
      "epoch": 0.22925,
      "grad_norm": 1.4124012855645303,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 22925
    },
    {
      "epoch": 0.22926,
      "grad_norm": 1.364736100490055,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 22926
    },
    {
      "epoch": 0.22927,
      "grad_norm": 1.6581967917988738,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 22927
    },
    {
      "epoch": 0.22928,
      "grad_norm": 1.0919513674529981,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 22928
    },
    {
      "epoch": 0.22929,
      "grad_norm": 1.606412865866531,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 22929
    },
    {
      "epoch": 0.2293,
      "grad_norm": 1.2609937857636244,
      "learning_rate": 0.003,
      "loss": 4.0262,
      "step": 22930
    },
    {
      "epoch": 0.22931,
      "grad_norm": 1.7529582876795367,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 22931
    },
    {
      "epoch": 0.22932,
      "grad_norm": 1.4139012140387022,
      "learning_rate": 0.003,
      "loss": 4.0258,
      "step": 22932
    },
    {
      "epoch": 0.22933,
      "grad_norm": 1.566417900134426,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 22933
    },
    {
      "epoch": 0.22934,
      "grad_norm": 1.1962950032689532,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 22934
    },
    {
      "epoch": 0.22935,
      "grad_norm": 1.4383831293450509,
      "learning_rate": 0.003,
      "loss": 4.0167,
      "step": 22935
    },
    {
      "epoch": 0.22936,
      "grad_norm": 1.565994399753327,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 22936
    },
    {
      "epoch": 0.22937,
      "grad_norm": 1.0786662983194601,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 22937
    },
    {
      "epoch": 0.22938,
      "grad_norm": 1.4133924513402512,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 22938
    },
    {
      "epoch": 0.22939,
      "grad_norm": 1.084571229371841,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 22939
    },
    {
      "epoch": 0.2294,
      "grad_norm": 1.3965331813781878,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 22940
    },
    {
      "epoch": 0.22941,
      "grad_norm": 1.0325201682266654,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 22941
    },
    {
      "epoch": 0.22942,
      "grad_norm": 1.542890426626316,
      "learning_rate": 0.003,
      "loss": 4.0271,
      "step": 22942
    },
    {
      "epoch": 0.22943,
      "grad_norm": 1.1180506726697423,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 22943
    },
    {
      "epoch": 0.22944,
      "grad_norm": 1.491061165134877,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 22944
    },
    {
      "epoch": 0.22945,
      "grad_norm": 1.1811620917207506,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 22945
    },
    {
      "epoch": 0.22946,
      "grad_norm": 1.4797081927263973,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 22946
    },
    {
      "epoch": 0.22947,
      "grad_norm": 1.6598278116824179,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 22947
    },
    {
      "epoch": 0.22948,
      "grad_norm": 1.4760840712503893,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 22948
    },
    {
      "epoch": 0.22949,
      "grad_norm": 1.7641302893273025,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 22949
    },
    {
      "epoch": 0.2295,
      "grad_norm": 1.5219615885729352,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 22950
    },
    {
      "epoch": 0.22951,
      "grad_norm": 1.0801487313124043,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 22951
    },
    {
      "epoch": 0.22952,
      "grad_norm": 1.7030498238077243,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 22952
    },
    {
      "epoch": 0.22953,
      "grad_norm": 1.228314594691771,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 22953
    },
    {
      "epoch": 0.22954,
      "grad_norm": 1.3193889796150344,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 22954
    },
    {
      "epoch": 0.22955,
      "grad_norm": 1.3953681681989965,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 22955
    },
    {
      "epoch": 0.22956,
      "grad_norm": 1.3948156590524627,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 22956
    },
    {
      "epoch": 0.22957,
      "grad_norm": 1.212636409383338,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 22957
    },
    {
      "epoch": 0.22958,
      "grad_norm": 1.3611995845323064,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 22958
    },
    {
      "epoch": 0.22959,
      "grad_norm": 1.3668827353309014,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 22959
    },
    {
      "epoch": 0.2296,
      "grad_norm": 1.4707145337750223,
      "learning_rate": 0.003,
      "loss": 3.9987,
      "step": 22960
    },
    {
      "epoch": 0.22961,
      "grad_norm": 1.4328658070926283,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 22961
    },
    {
      "epoch": 0.22962,
      "grad_norm": 1.3462902085348085,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 22962
    },
    {
      "epoch": 0.22963,
      "grad_norm": 1.2726388097540406,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 22963
    },
    {
      "epoch": 0.22964,
      "grad_norm": 1.390529523454449,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 22964
    },
    {
      "epoch": 0.22965,
      "grad_norm": 1.341336906374233,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 22965
    },
    {
      "epoch": 0.22966,
      "grad_norm": 1.403688994902358,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 22966
    },
    {
      "epoch": 0.22967,
      "grad_norm": 1.2501517024262214,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 22967
    },
    {
      "epoch": 0.22968,
      "grad_norm": 1.2657217447996258,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 22968
    },
    {
      "epoch": 0.22969,
      "grad_norm": 1.4040568075913151,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 22969
    },
    {
      "epoch": 0.2297,
      "grad_norm": 1.3447201613229363,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 22970
    },
    {
      "epoch": 0.22971,
      "grad_norm": 1.512430617630756,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 22971
    },
    {
      "epoch": 0.22972,
      "grad_norm": 1.1426672822249624,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 22972
    },
    {
      "epoch": 0.22973,
      "grad_norm": 1.6517270229492123,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 22973
    },
    {
      "epoch": 0.22974,
      "grad_norm": 1.4281794579999305,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 22974
    },
    {
      "epoch": 0.22975,
      "grad_norm": 1.3955411968895046,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 22975
    },
    {
      "epoch": 0.22976,
      "grad_norm": 1.615919374855493,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 22976
    },
    {
      "epoch": 0.22977,
      "grad_norm": 1.2222241116553751,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 22977
    },
    {
      "epoch": 0.22978,
      "grad_norm": 1.6735990138092902,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 22978
    },
    {
      "epoch": 0.22979,
      "grad_norm": 1.2673683420332382,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 22979
    },
    {
      "epoch": 0.2298,
      "grad_norm": 1.4814718469317947,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 22980
    },
    {
      "epoch": 0.22981,
      "grad_norm": 1.350122169637812,
      "learning_rate": 0.003,
      "loss": 4.0802,
      "step": 22981
    },
    {
      "epoch": 0.22982,
      "grad_norm": 1.533561415066937,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 22982
    },
    {
      "epoch": 0.22983,
      "grad_norm": 1.2700255494324815,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 22983
    },
    {
      "epoch": 0.22984,
      "grad_norm": 1.4737021321360557,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 22984
    },
    {
      "epoch": 0.22985,
      "grad_norm": 1.174831944097366,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 22985
    },
    {
      "epoch": 0.22986,
      "grad_norm": 1.3997706918456425,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 22986
    },
    {
      "epoch": 0.22987,
      "grad_norm": 1.1127635429152034,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 22987
    },
    {
      "epoch": 0.22988,
      "grad_norm": 1.538052023411139,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 22988
    },
    {
      "epoch": 0.22989,
      "grad_norm": 1.1620311378531678,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 22989
    },
    {
      "epoch": 0.2299,
      "grad_norm": 1.2928872617805722,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 22990
    },
    {
      "epoch": 0.22991,
      "grad_norm": 1.3814882984970216,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 22991
    },
    {
      "epoch": 0.22992,
      "grad_norm": 1.6545724967470796,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 22992
    },
    {
      "epoch": 0.22993,
      "grad_norm": 1.6784809434047772,
      "learning_rate": 0.003,
      "loss": 4.0957,
      "step": 22993
    },
    {
      "epoch": 0.22994,
      "grad_norm": 1.4447138372507262,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 22994
    },
    {
      "epoch": 0.22995,
      "grad_norm": 1.3250502982354588,
      "learning_rate": 0.003,
      "loss": 4.0656,
      "step": 22995
    },
    {
      "epoch": 0.22996,
      "grad_norm": 1.801497709744894,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 22996
    },
    {
      "epoch": 0.22997,
      "grad_norm": 1.254220967590485,
      "learning_rate": 0.003,
      "loss": 4.0201,
      "step": 22997
    },
    {
      "epoch": 0.22998,
      "grad_norm": 1.7448802253970186,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 22998
    },
    {
      "epoch": 0.22999,
      "grad_norm": 1.1891084182034939,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 22999
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4394272234556105,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 23000
    },
    {
      "epoch": 0.23001,
      "grad_norm": 1.3168780938686961,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 23001
    },
    {
      "epoch": 0.23002,
      "grad_norm": 1.6278056183432175,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 23002
    },
    {
      "epoch": 0.23003,
      "grad_norm": 1.2519319632700479,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 23003
    },
    {
      "epoch": 0.23004,
      "grad_norm": 1.262104534876486,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 23004
    },
    {
      "epoch": 0.23005,
      "grad_norm": 1.243282803795832,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 23005
    },
    {
      "epoch": 0.23006,
      "grad_norm": 1.4441124180883653,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 23006
    },
    {
      "epoch": 0.23007,
      "grad_norm": 1.434945131236988,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 23007
    },
    {
      "epoch": 0.23008,
      "grad_norm": 1.4197288535722148,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 23008
    },
    {
      "epoch": 0.23009,
      "grad_norm": 1.3982519755985698,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 23009
    },
    {
      "epoch": 0.2301,
      "grad_norm": 1.469397525657271,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 23010
    },
    {
      "epoch": 0.23011,
      "grad_norm": 1.3102803657496553,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 23011
    },
    {
      "epoch": 0.23012,
      "grad_norm": 1.278005363063069,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 23012
    },
    {
      "epoch": 0.23013,
      "grad_norm": 1.3300809198050145,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 23013
    },
    {
      "epoch": 0.23014,
      "grad_norm": 1.5934259801017507,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 23014
    },
    {
      "epoch": 0.23015,
      "grad_norm": 1.196117130422267,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 23015
    },
    {
      "epoch": 0.23016,
      "grad_norm": 1.6251479607161354,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 23016
    },
    {
      "epoch": 0.23017,
      "grad_norm": 1.4486473889099827,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 23017
    },
    {
      "epoch": 0.23018,
      "grad_norm": 1.3693501972873945,
      "learning_rate": 0.003,
      "loss": 4.0252,
      "step": 23018
    },
    {
      "epoch": 0.23019,
      "grad_norm": 1.27548053234981,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 23019
    },
    {
      "epoch": 0.2302,
      "grad_norm": 1.3200342429436345,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 23020
    },
    {
      "epoch": 0.23021,
      "grad_norm": 1.5248744753660666,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 23021
    },
    {
      "epoch": 0.23022,
      "grad_norm": 1.158348168545915,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 23022
    },
    {
      "epoch": 0.23023,
      "grad_norm": 1.9071166989661332,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 23023
    },
    {
      "epoch": 0.23024,
      "grad_norm": 1.1970788055193202,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 23024
    },
    {
      "epoch": 0.23025,
      "grad_norm": 1.2160555414860683,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 23025
    },
    {
      "epoch": 0.23026,
      "grad_norm": 1.344484298259978,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 23026
    },
    {
      "epoch": 0.23027,
      "grad_norm": 1.5255176934262091,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 23027
    },
    {
      "epoch": 0.23028,
      "grad_norm": 1.1681419830245958,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 23028
    },
    {
      "epoch": 0.23029,
      "grad_norm": 1.394941371974529,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 23029
    },
    {
      "epoch": 0.2303,
      "grad_norm": 1.1122218157565824,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 23030
    },
    {
      "epoch": 0.23031,
      "grad_norm": 1.424333834084624,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 23031
    },
    {
      "epoch": 0.23032,
      "grad_norm": 1.5497192043982402,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 23032
    },
    {
      "epoch": 0.23033,
      "grad_norm": 1.5981620336165951,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 23033
    },
    {
      "epoch": 0.23034,
      "grad_norm": 1.252140877826222,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 23034
    },
    {
      "epoch": 0.23035,
      "grad_norm": 1.401712325898123,
      "learning_rate": 0.003,
      "loss": 4.0665,
      "step": 23035
    },
    {
      "epoch": 0.23036,
      "grad_norm": 1.6003688749238476,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 23036
    },
    {
      "epoch": 0.23037,
      "grad_norm": 1.3118799647869948,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 23037
    },
    {
      "epoch": 0.23038,
      "grad_norm": 1.3595215125288516,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 23038
    },
    {
      "epoch": 0.23039,
      "grad_norm": 1.3915882753467457,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 23039
    },
    {
      "epoch": 0.2304,
      "grad_norm": 1.2549520097406257,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 23040
    },
    {
      "epoch": 0.23041,
      "grad_norm": 1.3439948582731243,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 23041
    },
    {
      "epoch": 0.23042,
      "grad_norm": 1.2951894916424498,
      "learning_rate": 0.003,
      "loss": 4.0081,
      "step": 23042
    },
    {
      "epoch": 0.23043,
      "grad_norm": 1.4622390794486304,
      "learning_rate": 0.003,
      "loss": 4.0306,
      "step": 23043
    },
    {
      "epoch": 0.23044,
      "grad_norm": 1.4544637414603367,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 23044
    },
    {
      "epoch": 0.23045,
      "grad_norm": 1.6228262982357775,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 23045
    },
    {
      "epoch": 0.23046,
      "grad_norm": 1.3595395138767727,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 23046
    },
    {
      "epoch": 0.23047,
      "grad_norm": 1.3348288664332195,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 23047
    },
    {
      "epoch": 0.23048,
      "grad_norm": 1.3961785392373596,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 23048
    },
    {
      "epoch": 0.23049,
      "grad_norm": 1.3894074047128306,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 23049
    },
    {
      "epoch": 0.2305,
      "grad_norm": 1.4851582462848276,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 23050
    },
    {
      "epoch": 0.23051,
      "grad_norm": 1.019450939862282,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 23051
    },
    {
      "epoch": 0.23052,
      "grad_norm": 1.4883474037119848,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 23052
    },
    {
      "epoch": 0.23053,
      "grad_norm": 1.1632807575688127,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 23053
    },
    {
      "epoch": 0.23054,
      "grad_norm": 1.7056925565619752,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 23054
    },
    {
      "epoch": 0.23055,
      "grad_norm": 1.3118817855524907,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 23055
    },
    {
      "epoch": 0.23056,
      "grad_norm": 1.523639956822756,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 23056
    },
    {
      "epoch": 0.23057,
      "grad_norm": 1.5461451176446483,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 23057
    },
    {
      "epoch": 0.23058,
      "grad_norm": 1.314297129018635,
      "learning_rate": 0.003,
      "loss": 4.0182,
      "step": 23058
    },
    {
      "epoch": 0.23059,
      "grad_norm": 1.533021870348019,
      "learning_rate": 0.003,
      "loss": 4.0308,
      "step": 23059
    },
    {
      "epoch": 0.2306,
      "grad_norm": 1.3037588670864102,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 23060
    },
    {
      "epoch": 0.23061,
      "grad_norm": 1.2486908498374083,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 23061
    },
    {
      "epoch": 0.23062,
      "grad_norm": 1.3545675816325136,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 23062
    },
    {
      "epoch": 0.23063,
      "grad_norm": 1.4238703700593431,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 23063
    },
    {
      "epoch": 0.23064,
      "grad_norm": 1.260590410799713,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 23064
    },
    {
      "epoch": 0.23065,
      "grad_norm": 1.5345428707631013,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 23065
    },
    {
      "epoch": 0.23066,
      "grad_norm": 1.2083117870553177,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 23066
    },
    {
      "epoch": 0.23067,
      "grad_norm": 1.6470988351741727,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 23067
    },
    {
      "epoch": 0.23068,
      "grad_norm": 1.1630818070944566,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 23068
    },
    {
      "epoch": 0.23069,
      "grad_norm": 1.6895136508341708,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 23069
    },
    {
      "epoch": 0.2307,
      "grad_norm": 1.25827195763974,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 23070
    },
    {
      "epoch": 0.23071,
      "grad_norm": 1.410811113011283,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 23071
    },
    {
      "epoch": 0.23072,
      "grad_norm": 1.3395227633208335,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 23072
    },
    {
      "epoch": 0.23073,
      "grad_norm": 1.6835649017578809,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 23073
    },
    {
      "epoch": 0.23074,
      "grad_norm": 1.1112159433910216,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 23074
    },
    {
      "epoch": 0.23075,
      "grad_norm": 1.7472991571570606,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 23075
    },
    {
      "epoch": 0.23076,
      "grad_norm": 1.2145662256286143,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 23076
    },
    {
      "epoch": 0.23077,
      "grad_norm": 1.3231062274195922,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 23077
    },
    {
      "epoch": 0.23078,
      "grad_norm": 1.4200853882677293,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 23078
    },
    {
      "epoch": 0.23079,
      "grad_norm": 1.3677150569072831,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 23079
    },
    {
      "epoch": 0.2308,
      "grad_norm": 1.3676110212542516,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 23080
    },
    {
      "epoch": 0.23081,
      "grad_norm": 1.4137468889436755,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 23081
    },
    {
      "epoch": 0.23082,
      "grad_norm": 1.177594064115716,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 23082
    },
    {
      "epoch": 0.23083,
      "grad_norm": 1.5074512855725697,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 23083
    },
    {
      "epoch": 0.23084,
      "grad_norm": 1.3238048485838252,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 23084
    },
    {
      "epoch": 0.23085,
      "grad_norm": 1.4820599142183168,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 23085
    },
    {
      "epoch": 0.23086,
      "grad_norm": 1.3445418917237837,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 23086
    },
    {
      "epoch": 0.23087,
      "grad_norm": 1.2570486310307096,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 23087
    },
    {
      "epoch": 0.23088,
      "grad_norm": 1.436336772648397,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 23088
    },
    {
      "epoch": 0.23089,
      "grad_norm": 1.1637129945506097,
      "learning_rate": 0.003,
      "loss": 4.0237,
      "step": 23089
    },
    {
      "epoch": 0.2309,
      "grad_norm": 1.6038071535423457,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 23090
    },
    {
      "epoch": 0.23091,
      "grad_norm": 1.3679500974508012,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 23091
    },
    {
      "epoch": 0.23092,
      "grad_norm": 1.6065914153722922,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 23092
    },
    {
      "epoch": 0.23093,
      "grad_norm": 1.1361562542425891,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 23093
    },
    {
      "epoch": 0.23094,
      "grad_norm": 1.2749205805039103,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 23094
    },
    {
      "epoch": 0.23095,
      "grad_norm": 1.2988769294460014,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 23095
    },
    {
      "epoch": 0.23096,
      "grad_norm": 1.3538875084098738,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 23096
    },
    {
      "epoch": 0.23097,
      "grad_norm": 1.315085808107658,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 23097
    },
    {
      "epoch": 0.23098,
      "grad_norm": 1.4182471702035666,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 23098
    },
    {
      "epoch": 0.23099,
      "grad_norm": 1.3566694693414556,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 23099
    },
    {
      "epoch": 0.231,
      "grad_norm": 1.2740335308945778,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 23100
    },
    {
      "epoch": 0.23101,
      "grad_norm": 1.3224127108449382,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 23101
    },
    {
      "epoch": 0.23102,
      "grad_norm": 1.2037559306869496,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 23102
    },
    {
      "epoch": 0.23103,
      "grad_norm": 1.9817962812717091,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 23103
    },
    {
      "epoch": 0.23104,
      "grad_norm": 1.2225055196978887,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 23104
    },
    {
      "epoch": 0.23105,
      "grad_norm": 1.5000579479300924,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 23105
    },
    {
      "epoch": 0.23106,
      "grad_norm": 1.5623842054630377,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 23106
    },
    {
      "epoch": 0.23107,
      "grad_norm": 1.1491403271789082,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 23107
    },
    {
      "epoch": 0.23108,
      "grad_norm": 1.2366893852890424,
      "learning_rate": 0.003,
      "loss": 4.0085,
      "step": 23108
    },
    {
      "epoch": 0.23109,
      "grad_norm": 1.3032847184870682,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 23109
    },
    {
      "epoch": 0.2311,
      "grad_norm": 1.2902243588838969,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 23110
    },
    {
      "epoch": 0.23111,
      "grad_norm": 1.589132549648097,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 23111
    },
    {
      "epoch": 0.23112,
      "grad_norm": 1.4206133407013364,
      "learning_rate": 0.003,
      "loss": 4.0961,
      "step": 23112
    },
    {
      "epoch": 0.23113,
      "grad_norm": 1.5505661618510957,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 23113
    },
    {
      "epoch": 0.23114,
      "grad_norm": 1.4278368033703972,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 23114
    },
    {
      "epoch": 0.23115,
      "grad_norm": 1.49234650598739,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 23115
    },
    {
      "epoch": 0.23116,
      "grad_norm": 1.1263272541904785,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 23116
    },
    {
      "epoch": 0.23117,
      "grad_norm": 1.5748913296976446,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 23117
    },
    {
      "epoch": 0.23118,
      "grad_norm": 1.1986532243015542,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 23118
    },
    {
      "epoch": 0.23119,
      "grad_norm": 1.4866078782519772,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 23119
    },
    {
      "epoch": 0.2312,
      "grad_norm": 1.122292274862112,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 23120
    },
    {
      "epoch": 0.23121,
      "grad_norm": 1.554413091342367,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 23121
    },
    {
      "epoch": 0.23122,
      "grad_norm": 1.5887694598592788,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 23122
    },
    {
      "epoch": 0.23123,
      "grad_norm": 1.5626775116592848,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 23123
    },
    {
      "epoch": 0.23124,
      "grad_norm": 1.395390976413349,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 23124
    },
    {
      "epoch": 0.23125,
      "grad_norm": 1.360377731365266,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 23125
    },
    {
      "epoch": 0.23126,
      "grad_norm": 1.437002287152256,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 23126
    },
    {
      "epoch": 0.23127,
      "grad_norm": 1.2982733064086367,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 23127
    },
    {
      "epoch": 0.23128,
      "grad_norm": 1.2165050994472264,
      "learning_rate": 0.003,
      "loss": 4.0222,
      "step": 23128
    },
    {
      "epoch": 0.23129,
      "grad_norm": 1.484541274078872,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 23129
    },
    {
      "epoch": 0.2313,
      "grad_norm": 1.3126693478781408,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 23130
    },
    {
      "epoch": 0.23131,
      "grad_norm": 1.6188912436746827,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 23131
    },
    {
      "epoch": 0.23132,
      "grad_norm": 0.9608496433254575,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 23132
    },
    {
      "epoch": 0.23133,
      "grad_norm": 1.4943965654805949,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 23133
    },
    {
      "epoch": 0.23134,
      "grad_norm": 1.6103602703785342,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 23134
    },
    {
      "epoch": 0.23135,
      "grad_norm": 1.4865506551792285,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 23135
    },
    {
      "epoch": 0.23136,
      "grad_norm": 1.4997999500696977,
      "learning_rate": 0.003,
      "loss": 4.0229,
      "step": 23136
    },
    {
      "epoch": 0.23137,
      "grad_norm": 1.2820657856007582,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 23137
    },
    {
      "epoch": 0.23138,
      "grad_norm": 1.1256573261519152,
      "learning_rate": 0.003,
      "loss": 4.0186,
      "step": 23138
    },
    {
      "epoch": 0.23139,
      "grad_norm": 1.4841915233502543,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 23139
    },
    {
      "epoch": 0.2314,
      "grad_norm": 1.4319827874652815,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 23140
    },
    {
      "epoch": 0.23141,
      "grad_norm": 1.649943338937173,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 23141
    },
    {
      "epoch": 0.23142,
      "grad_norm": 1.2014528257151893,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 23142
    },
    {
      "epoch": 0.23143,
      "grad_norm": 1.581294329507776,
      "learning_rate": 0.003,
      "loss": 4.0194,
      "step": 23143
    },
    {
      "epoch": 0.23144,
      "grad_norm": 1.3873995621357795,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 23144
    },
    {
      "epoch": 0.23145,
      "grad_norm": 1.741841217549336,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 23145
    },
    {
      "epoch": 0.23146,
      "grad_norm": 1.0427242420105038,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 23146
    },
    {
      "epoch": 0.23147,
      "grad_norm": 1.2988860658950285,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 23147
    },
    {
      "epoch": 0.23148,
      "grad_norm": 1.224340615320788,
      "learning_rate": 0.003,
      "loss": 4.0956,
      "step": 23148
    },
    {
      "epoch": 0.23149,
      "grad_norm": 1.602496600517891,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 23149
    },
    {
      "epoch": 0.2315,
      "grad_norm": 1.1586195794603469,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 23150
    },
    {
      "epoch": 0.23151,
      "grad_norm": 1.6108298467371582,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 23151
    },
    {
      "epoch": 0.23152,
      "grad_norm": 1.2344649499198077,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 23152
    },
    {
      "epoch": 0.23153,
      "grad_norm": 1.3800617425480675,
      "learning_rate": 0.003,
      "loss": 4.0881,
      "step": 23153
    },
    {
      "epoch": 0.23154,
      "grad_norm": 1.3758833483827941,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 23154
    },
    {
      "epoch": 0.23155,
      "grad_norm": 1.5336184022458934,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 23155
    },
    {
      "epoch": 0.23156,
      "grad_norm": 1.370913566210212,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 23156
    },
    {
      "epoch": 0.23157,
      "grad_norm": 1.2596585620803824,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 23157
    },
    {
      "epoch": 0.23158,
      "grad_norm": 1.6195621423029691,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 23158
    },
    {
      "epoch": 0.23159,
      "grad_norm": 1.3963011760127697,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 23159
    },
    {
      "epoch": 0.2316,
      "grad_norm": 1.427369476982424,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 23160
    },
    {
      "epoch": 0.23161,
      "grad_norm": 1.6023339723130183,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 23161
    },
    {
      "epoch": 0.23162,
      "grad_norm": 1.3977799439014873,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 23162
    },
    {
      "epoch": 0.23163,
      "grad_norm": 1.1471109711506093,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 23163
    },
    {
      "epoch": 0.23164,
      "grad_norm": 1.4522518793825152,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 23164
    },
    {
      "epoch": 0.23165,
      "grad_norm": 1.2770313310326598,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 23165
    },
    {
      "epoch": 0.23166,
      "grad_norm": 1.5415520221080945,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 23166
    },
    {
      "epoch": 0.23167,
      "grad_norm": 1.192403516683385,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 23167
    },
    {
      "epoch": 0.23168,
      "grad_norm": 1.4369307951722878,
      "learning_rate": 0.003,
      "loss": 4.0215,
      "step": 23168
    },
    {
      "epoch": 0.23169,
      "grad_norm": 1.3428022781460058,
      "learning_rate": 0.003,
      "loss": 4.0112,
      "step": 23169
    },
    {
      "epoch": 0.2317,
      "grad_norm": 1.4573974960234255,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 23170
    },
    {
      "epoch": 0.23171,
      "grad_norm": 1.3559560376745463,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 23171
    },
    {
      "epoch": 0.23172,
      "grad_norm": 1.290824733352766,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 23172
    },
    {
      "epoch": 0.23173,
      "grad_norm": 1.4016706474171015,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 23173
    },
    {
      "epoch": 0.23174,
      "grad_norm": 1.2672525599673197,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 23174
    },
    {
      "epoch": 0.23175,
      "grad_norm": 1.505934042929275,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 23175
    },
    {
      "epoch": 0.23176,
      "grad_norm": 1.4555875781210392,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 23176
    },
    {
      "epoch": 0.23177,
      "grad_norm": 1.4200226637953215,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 23177
    },
    {
      "epoch": 0.23178,
      "grad_norm": 1.3511209540027005,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 23178
    },
    {
      "epoch": 0.23179,
      "grad_norm": 1.535066650779469,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 23179
    },
    {
      "epoch": 0.2318,
      "grad_norm": 1.040862751941025,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 23180
    },
    {
      "epoch": 0.23181,
      "grad_norm": 1.5481846575702471,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 23181
    },
    {
      "epoch": 0.23182,
      "grad_norm": 1.0461838488185702,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 23182
    },
    {
      "epoch": 0.23183,
      "grad_norm": 1.4058362774518611,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 23183
    },
    {
      "epoch": 0.23184,
      "grad_norm": 1.2187833980533604,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 23184
    },
    {
      "epoch": 0.23185,
      "grad_norm": 1.399106467132836,
      "learning_rate": 0.003,
      "loss": 4.0738,
      "step": 23185
    },
    {
      "epoch": 0.23186,
      "grad_norm": 1.4194814086397585,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 23186
    },
    {
      "epoch": 0.23187,
      "grad_norm": 1.7670927996006467,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 23187
    },
    {
      "epoch": 0.23188,
      "grad_norm": 1.3519469406355538,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 23188
    },
    {
      "epoch": 0.23189,
      "grad_norm": 1.5863237688284646,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 23189
    },
    {
      "epoch": 0.2319,
      "grad_norm": 1.419398324412055,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 23190
    },
    {
      "epoch": 0.23191,
      "grad_norm": 1.5093101000095643,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 23191
    },
    {
      "epoch": 0.23192,
      "grad_norm": 1.3126495336212163,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 23192
    },
    {
      "epoch": 0.23193,
      "grad_norm": 1.5041173424379344,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 23193
    },
    {
      "epoch": 0.23194,
      "grad_norm": 1.2081253656400714,
      "learning_rate": 0.003,
      "loss": 4.0165,
      "step": 23194
    },
    {
      "epoch": 0.23195,
      "grad_norm": 1.6018004083847395,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 23195
    },
    {
      "epoch": 0.23196,
      "grad_norm": 1.1038385697305264,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 23196
    },
    {
      "epoch": 0.23197,
      "grad_norm": 1.4140298665465247,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 23197
    },
    {
      "epoch": 0.23198,
      "grad_norm": 1.2455816829302624,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 23198
    },
    {
      "epoch": 0.23199,
      "grad_norm": 1.293493351059705,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 23199
    },
    {
      "epoch": 0.232,
      "grad_norm": 1.4399840994518005,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 23200
    },
    {
      "epoch": 0.23201,
      "grad_norm": 1.8503743395333405,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 23201
    },
    {
      "epoch": 0.23202,
      "grad_norm": 1.1866492005161333,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 23202
    },
    {
      "epoch": 0.23203,
      "grad_norm": 1.5410260579949557,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 23203
    },
    {
      "epoch": 0.23204,
      "grad_norm": 1.2874196651584062,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 23204
    },
    {
      "epoch": 0.23205,
      "grad_norm": 1.558736019283087,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 23205
    },
    {
      "epoch": 0.23206,
      "grad_norm": 1.1155157635475197,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 23206
    },
    {
      "epoch": 0.23207,
      "grad_norm": 1.6284966721910028,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 23207
    },
    {
      "epoch": 0.23208,
      "grad_norm": 1.4187063451456476,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 23208
    },
    {
      "epoch": 0.23209,
      "grad_norm": 1.428265865797565,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 23209
    },
    {
      "epoch": 0.2321,
      "grad_norm": 1.3748224800991098,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 23210
    },
    {
      "epoch": 0.23211,
      "grad_norm": 1.6895160144380876,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 23211
    },
    {
      "epoch": 0.23212,
      "grad_norm": 1.0415828454277831,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 23212
    },
    {
      "epoch": 0.23213,
      "grad_norm": 1.517687929459321,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 23213
    },
    {
      "epoch": 0.23214,
      "grad_norm": 1.2141206614580975,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 23214
    },
    {
      "epoch": 0.23215,
      "grad_norm": 1.3926596116826988,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 23215
    },
    {
      "epoch": 0.23216,
      "grad_norm": 1.5309294204579795,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 23216
    },
    {
      "epoch": 0.23217,
      "grad_norm": 1.4013125958772263,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 23217
    },
    {
      "epoch": 0.23218,
      "grad_norm": 1.5455137081320858,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 23218
    },
    {
      "epoch": 0.23219,
      "grad_norm": 1.3460428580661958,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 23219
    },
    {
      "epoch": 0.2322,
      "grad_norm": 1.5581252782994168,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 23220
    },
    {
      "epoch": 0.23221,
      "grad_norm": 1.2972087312487386,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 23221
    },
    {
      "epoch": 0.23222,
      "grad_norm": 1.3393994009577652,
      "learning_rate": 0.003,
      "loss": 4.0307,
      "step": 23222
    },
    {
      "epoch": 0.23223,
      "grad_norm": 1.4743440652844046,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 23223
    },
    {
      "epoch": 0.23224,
      "grad_norm": 1.2399382638443128,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 23224
    },
    {
      "epoch": 0.23225,
      "grad_norm": 1.4274691304657934,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 23225
    },
    {
      "epoch": 0.23226,
      "grad_norm": 1.2153235809358585,
      "learning_rate": 0.003,
      "loss": 4.0322,
      "step": 23226
    },
    {
      "epoch": 0.23227,
      "grad_norm": 1.4742385055043994,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 23227
    },
    {
      "epoch": 0.23228,
      "grad_norm": 1.201506525492165,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 23228
    },
    {
      "epoch": 0.23229,
      "grad_norm": 1.55977523256643,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 23229
    },
    {
      "epoch": 0.2323,
      "grad_norm": 1.3237454623368943,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 23230
    },
    {
      "epoch": 0.23231,
      "grad_norm": 1.4198197981554923,
      "learning_rate": 0.003,
      "loss": 4.0234,
      "step": 23231
    },
    {
      "epoch": 0.23232,
      "grad_norm": 1.200003748296258,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 23232
    },
    {
      "epoch": 0.23233,
      "grad_norm": 1.364562004030165,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 23233
    },
    {
      "epoch": 0.23234,
      "grad_norm": 1.2354970282473423,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 23234
    },
    {
      "epoch": 0.23235,
      "grad_norm": 1.6048801376951722,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 23235
    },
    {
      "epoch": 0.23236,
      "grad_norm": 1.144768589358892,
      "learning_rate": 0.003,
      "loss": 4.0215,
      "step": 23236
    },
    {
      "epoch": 0.23237,
      "grad_norm": 1.6146400549623448,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 23237
    },
    {
      "epoch": 0.23238,
      "grad_norm": 1.2845332033949788,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 23238
    },
    {
      "epoch": 0.23239,
      "grad_norm": 1.5198619928502966,
      "learning_rate": 0.003,
      "loss": 4.0918,
      "step": 23239
    },
    {
      "epoch": 0.2324,
      "grad_norm": 1.4796043315387597,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 23240
    },
    {
      "epoch": 0.23241,
      "grad_norm": 1.3232047527976964,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 23241
    },
    {
      "epoch": 0.23242,
      "grad_norm": 1.5909351606870596,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 23242
    },
    {
      "epoch": 0.23243,
      "grad_norm": 1.295351887818341,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 23243
    },
    {
      "epoch": 0.23244,
      "grad_norm": 1.7195299870472625,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 23244
    },
    {
      "epoch": 0.23245,
      "grad_norm": 1.1073968338402673,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 23245
    },
    {
      "epoch": 0.23246,
      "grad_norm": 1.3603625036297036,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 23246
    },
    {
      "epoch": 0.23247,
      "grad_norm": 1.648006095868824,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 23247
    },
    {
      "epoch": 0.23248,
      "grad_norm": 1.0838339786857158,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 23248
    },
    {
      "epoch": 0.23249,
      "grad_norm": 1.8525512974473193,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 23249
    },
    {
      "epoch": 0.2325,
      "grad_norm": 1.0995067287459384,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 23250
    },
    {
      "epoch": 0.23251,
      "grad_norm": 1.501016569416679,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 23251
    },
    {
      "epoch": 0.23252,
      "grad_norm": 1.1455643535743112,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 23252
    },
    {
      "epoch": 0.23253,
      "grad_norm": 1.3518525655773297,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 23253
    },
    {
      "epoch": 0.23254,
      "grad_norm": 1.4105817535103777,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 23254
    },
    {
      "epoch": 0.23255,
      "grad_norm": 1.169917404981231,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 23255
    },
    {
      "epoch": 0.23256,
      "grad_norm": 1.501594932765973,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 23256
    },
    {
      "epoch": 0.23257,
      "grad_norm": 1.295753874583137,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 23257
    },
    {
      "epoch": 0.23258,
      "grad_norm": 2.039609126896833,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 23258
    },
    {
      "epoch": 0.23259,
      "grad_norm": 1.1959699585543309,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 23259
    },
    {
      "epoch": 0.2326,
      "grad_norm": 1.4181179411384424,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 23260
    },
    {
      "epoch": 0.23261,
      "grad_norm": 1.143132251049764,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 23261
    },
    {
      "epoch": 0.23262,
      "grad_norm": 1.2022803882982567,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 23262
    },
    {
      "epoch": 0.23263,
      "grad_norm": 1.4850903264816182,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 23263
    },
    {
      "epoch": 0.23264,
      "grad_norm": 1.1552796642446281,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 23264
    },
    {
      "epoch": 0.23265,
      "grad_norm": 1.4577981987212594,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 23265
    },
    {
      "epoch": 0.23266,
      "grad_norm": 1.5465790001414537,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 23266
    },
    {
      "epoch": 0.23267,
      "grad_norm": 1.2356899901345653,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 23267
    },
    {
      "epoch": 0.23268,
      "grad_norm": 1.4966204405392933,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 23268
    },
    {
      "epoch": 0.23269,
      "grad_norm": 1.1490991958453594,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 23269
    },
    {
      "epoch": 0.2327,
      "grad_norm": 1.632207849955671,
      "learning_rate": 0.003,
      "loss": 4.082,
      "step": 23270
    },
    {
      "epoch": 0.23271,
      "grad_norm": 1.1523017348795916,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 23271
    },
    {
      "epoch": 0.23272,
      "grad_norm": 1.4942461998769034,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 23272
    },
    {
      "epoch": 0.23273,
      "grad_norm": 1.524916558472294,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 23273
    },
    {
      "epoch": 0.23274,
      "grad_norm": 1.3375224337440415,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 23274
    },
    {
      "epoch": 0.23275,
      "grad_norm": 1.640461322521533,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 23275
    },
    {
      "epoch": 0.23276,
      "grad_norm": 1.376257675894278,
      "learning_rate": 0.003,
      "loss": 4.0886,
      "step": 23276
    },
    {
      "epoch": 0.23277,
      "grad_norm": 1.3720448538424719,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 23277
    },
    {
      "epoch": 0.23278,
      "grad_norm": 1.330648971532328,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 23278
    },
    {
      "epoch": 0.23279,
      "grad_norm": 1.4403782791511024,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 23279
    },
    {
      "epoch": 0.2328,
      "grad_norm": 1.3486738279518276,
      "learning_rate": 0.003,
      "loss": 4.0858,
      "step": 23280
    },
    {
      "epoch": 0.23281,
      "grad_norm": 1.5742117054538272,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 23281
    },
    {
      "epoch": 0.23282,
      "grad_norm": 1.360354340322476,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 23282
    },
    {
      "epoch": 0.23283,
      "grad_norm": 1.573803399401895,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 23283
    },
    {
      "epoch": 0.23284,
      "grad_norm": 1.380470787810996,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 23284
    },
    {
      "epoch": 0.23285,
      "grad_norm": 1.1481081913957012,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 23285
    },
    {
      "epoch": 0.23286,
      "grad_norm": 1.378426484420779,
      "learning_rate": 0.003,
      "loss": 4.0234,
      "step": 23286
    },
    {
      "epoch": 0.23287,
      "grad_norm": 1.4891976572396248,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 23287
    },
    {
      "epoch": 0.23288,
      "grad_norm": 1.274367116949293,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 23288
    },
    {
      "epoch": 0.23289,
      "grad_norm": 1.2995741354141095,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 23289
    },
    {
      "epoch": 0.2329,
      "grad_norm": 1.3751707069043315,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 23290
    },
    {
      "epoch": 0.23291,
      "grad_norm": 1.1318703341623686,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 23291
    },
    {
      "epoch": 0.23292,
      "grad_norm": 1.3848642530821986,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 23292
    },
    {
      "epoch": 0.23293,
      "grad_norm": 1.3580898429055006,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 23293
    },
    {
      "epoch": 0.23294,
      "grad_norm": 1.3435476391696566,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 23294
    },
    {
      "epoch": 0.23295,
      "grad_norm": 1.2607504054385046,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 23295
    },
    {
      "epoch": 0.23296,
      "grad_norm": 1.456243137673101,
      "learning_rate": 0.003,
      "loss": 4.0329,
      "step": 23296
    },
    {
      "epoch": 0.23297,
      "grad_norm": 1.3895092586474425,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 23297
    },
    {
      "epoch": 0.23298,
      "grad_norm": 1.3805235974651768,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 23298
    },
    {
      "epoch": 0.23299,
      "grad_norm": 1.4239808245437895,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 23299
    },
    {
      "epoch": 0.233,
      "grad_norm": 1.530693199442528,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 23300
    },
    {
      "epoch": 0.23301,
      "grad_norm": 1.2764472475073199,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 23301
    },
    {
      "epoch": 0.23302,
      "grad_norm": 1.4247669730558707,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 23302
    },
    {
      "epoch": 0.23303,
      "grad_norm": 1.2172673406242815,
      "learning_rate": 0.003,
      "loss": 4.0042,
      "step": 23303
    },
    {
      "epoch": 0.23304,
      "grad_norm": 1.4074013609359952,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 23304
    },
    {
      "epoch": 0.23305,
      "grad_norm": 1.789281302975788,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 23305
    },
    {
      "epoch": 0.23306,
      "grad_norm": 1.1198034725501445,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 23306
    },
    {
      "epoch": 0.23307,
      "grad_norm": 1.5216483930759517,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 23307
    },
    {
      "epoch": 0.23308,
      "grad_norm": 1.3337214740061536,
      "learning_rate": 0.003,
      "loss": 4.0315,
      "step": 23308
    },
    {
      "epoch": 0.23309,
      "grad_norm": 1.379853589673503,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 23309
    },
    {
      "epoch": 0.2331,
      "grad_norm": 1.3763096443011047,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 23310
    },
    {
      "epoch": 0.23311,
      "grad_norm": 1.324091611104481,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 23311
    },
    {
      "epoch": 0.23312,
      "grad_norm": 1.469904166008016,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 23312
    },
    {
      "epoch": 0.23313,
      "grad_norm": 1.3533478110113275,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 23313
    },
    {
      "epoch": 0.23314,
      "grad_norm": 1.3417430157453967,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 23314
    },
    {
      "epoch": 0.23315,
      "grad_norm": 1.4765814465557936,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 23315
    },
    {
      "epoch": 0.23316,
      "grad_norm": 1.5731600437764057,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 23316
    },
    {
      "epoch": 0.23317,
      "grad_norm": 1.4499194957120853,
      "learning_rate": 0.003,
      "loss": 4.0912,
      "step": 23317
    },
    {
      "epoch": 0.23318,
      "grad_norm": 1.461271136541248,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 23318
    },
    {
      "epoch": 0.23319,
      "grad_norm": 1.4191884278936935,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 23319
    },
    {
      "epoch": 0.2332,
      "grad_norm": 1.3395844738642793,
      "learning_rate": 0.003,
      "loss": 4.0588,
      "step": 23320
    },
    {
      "epoch": 0.23321,
      "grad_norm": 1.2927872666501299,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 23321
    },
    {
      "epoch": 0.23322,
      "grad_norm": 1.3284300302078813,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 23322
    },
    {
      "epoch": 0.23323,
      "grad_norm": 1.388065516587412,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 23323
    },
    {
      "epoch": 0.23324,
      "grad_norm": 1.2730160726961077,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 23324
    },
    {
      "epoch": 0.23325,
      "grad_norm": 1.6958348103450889,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 23325
    },
    {
      "epoch": 0.23326,
      "grad_norm": 1.218806575339292,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 23326
    },
    {
      "epoch": 0.23327,
      "grad_norm": 1.4049694173549079,
      "learning_rate": 0.003,
      "loss": 4.0095,
      "step": 23327
    },
    {
      "epoch": 0.23328,
      "grad_norm": 1.3725084742900742,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 23328
    },
    {
      "epoch": 0.23329,
      "grad_norm": 1.4354762724536505,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 23329
    },
    {
      "epoch": 0.2333,
      "grad_norm": 1.2729372688233613,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 23330
    },
    {
      "epoch": 0.23331,
      "grad_norm": 1.3739779736425317,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 23331
    },
    {
      "epoch": 0.23332,
      "grad_norm": 1.2797412439650528,
      "learning_rate": 0.003,
      "loss": 4.0235,
      "step": 23332
    },
    {
      "epoch": 0.23333,
      "grad_norm": 1.1359805215254921,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 23333
    },
    {
      "epoch": 0.23334,
      "grad_norm": 1.4859752571815634,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 23334
    },
    {
      "epoch": 0.23335,
      "grad_norm": 1.2644116192581694,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 23335
    },
    {
      "epoch": 0.23336,
      "grad_norm": 1.7288192148564303,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 23336
    },
    {
      "epoch": 0.23337,
      "grad_norm": 1.4522526863132863,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 23337
    },
    {
      "epoch": 0.23338,
      "grad_norm": 1.3746559438134074,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 23338
    },
    {
      "epoch": 0.23339,
      "grad_norm": 1.4896974277707722,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 23339
    },
    {
      "epoch": 0.2334,
      "grad_norm": 1.115931811769656,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 23340
    },
    {
      "epoch": 0.23341,
      "grad_norm": 1.4310574265050977,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 23341
    },
    {
      "epoch": 0.23342,
      "grad_norm": 1.242085521227107,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 23342
    },
    {
      "epoch": 0.23343,
      "grad_norm": 1.487393175276797,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 23343
    },
    {
      "epoch": 0.23344,
      "grad_norm": 1.1575642661862828,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 23344
    },
    {
      "epoch": 0.23345,
      "grad_norm": 1.399116039326976,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 23345
    },
    {
      "epoch": 0.23346,
      "grad_norm": 1.2296764806064457,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 23346
    },
    {
      "epoch": 0.23347,
      "grad_norm": 1.4347408253791656,
      "learning_rate": 0.003,
      "loss": 4.0179,
      "step": 23347
    },
    {
      "epoch": 0.23348,
      "grad_norm": 1.7674929643590882,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 23348
    },
    {
      "epoch": 0.23349,
      "grad_norm": 1.3096582728277626,
      "learning_rate": 0.003,
      "loss": 4.018,
      "step": 23349
    },
    {
      "epoch": 0.2335,
      "grad_norm": 1.5564933789167448,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 23350
    },
    {
      "epoch": 0.23351,
      "grad_norm": 1.3566410676533962,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 23351
    },
    {
      "epoch": 0.23352,
      "grad_norm": 1.4891266643637302,
      "learning_rate": 0.003,
      "loss": 4.0804,
      "step": 23352
    },
    {
      "epoch": 0.23353,
      "grad_norm": 1.2880685163878165,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 23353
    },
    {
      "epoch": 0.23354,
      "grad_norm": 1.4053611881721693,
      "learning_rate": 0.003,
      "loss": 4.0724,
      "step": 23354
    },
    {
      "epoch": 0.23355,
      "grad_norm": 1.4909643929490488,
      "learning_rate": 0.003,
      "loss": 4.0238,
      "step": 23355
    },
    {
      "epoch": 0.23356,
      "grad_norm": 1.305293227109484,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 23356
    },
    {
      "epoch": 0.23357,
      "grad_norm": 1.8078193423309594,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 23357
    },
    {
      "epoch": 0.23358,
      "grad_norm": 1.0891309342760909,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 23358
    },
    {
      "epoch": 0.23359,
      "grad_norm": 1.6236321420258064,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 23359
    },
    {
      "epoch": 0.2336,
      "grad_norm": 1.3116676798911029,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 23360
    },
    {
      "epoch": 0.23361,
      "grad_norm": 1.3478344689475437,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 23361
    },
    {
      "epoch": 0.23362,
      "grad_norm": 1.5128040823850337,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 23362
    },
    {
      "epoch": 0.23363,
      "grad_norm": 1.3541693429615973,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 23363
    },
    {
      "epoch": 0.23364,
      "grad_norm": 1.205625666698049,
      "learning_rate": 0.003,
      "loss": 4.0787,
      "step": 23364
    },
    {
      "epoch": 0.23365,
      "grad_norm": 1.3524023350765948,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 23365
    },
    {
      "epoch": 0.23366,
      "grad_norm": 1.3072262571821724,
      "learning_rate": 0.003,
      "loss": 4.0293,
      "step": 23366
    },
    {
      "epoch": 0.23367,
      "grad_norm": 1.222400294982097,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 23367
    },
    {
      "epoch": 0.23368,
      "grad_norm": 1.3857796257043968,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 23368
    },
    {
      "epoch": 0.23369,
      "grad_norm": 1.5039811217841277,
      "learning_rate": 0.003,
      "loss": 4.0327,
      "step": 23369
    },
    {
      "epoch": 0.2337,
      "grad_norm": 1.592357064418626,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 23370
    },
    {
      "epoch": 0.23371,
      "grad_norm": 1.3307194504706745,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 23371
    },
    {
      "epoch": 0.23372,
      "grad_norm": 1.7265129605754046,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 23372
    },
    {
      "epoch": 0.23373,
      "grad_norm": 1.2791146988045174,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 23373
    },
    {
      "epoch": 0.23374,
      "grad_norm": 1.2658211887182347,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 23374
    },
    {
      "epoch": 0.23375,
      "grad_norm": 1.615940418471002,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 23375
    },
    {
      "epoch": 0.23376,
      "grad_norm": 1.2488341062337922,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 23376
    },
    {
      "epoch": 0.23377,
      "grad_norm": 1.64317541007616,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 23377
    },
    {
      "epoch": 0.23378,
      "grad_norm": 1.1794176936796756,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 23378
    },
    {
      "epoch": 0.23379,
      "grad_norm": 1.5511462295838057,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 23379
    },
    {
      "epoch": 0.2338,
      "grad_norm": 1.2101383495455327,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 23380
    },
    {
      "epoch": 0.23381,
      "grad_norm": 1.4288709848721513,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 23381
    },
    {
      "epoch": 0.23382,
      "grad_norm": 1.145074381396933,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 23382
    },
    {
      "epoch": 0.23383,
      "grad_norm": 1.3238362213026706,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 23383
    },
    {
      "epoch": 0.23384,
      "grad_norm": 1.5058887305002904,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 23384
    },
    {
      "epoch": 0.23385,
      "grad_norm": 1.2497972129327182,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 23385
    },
    {
      "epoch": 0.23386,
      "grad_norm": 1.5232675410756014,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 23386
    },
    {
      "epoch": 0.23387,
      "grad_norm": 1.3182771221034846,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 23387
    },
    {
      "epoch": 0.23388,
      "grad_norm": 1.6502823950348506,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 23388
    },
    {
      "epoch": 0.23389,
      "grad_norm": 1.513712969471974,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 23389
    },
    {
      "epoch": 0.2339,
      "grad_norm": 1.123901985534363,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 23390
    },
    {
      "epoch": 0.23391,
      "grad_norm": 1.7476835044221801,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 23391
    },
    {
      "epoch": 0.23392,
      "grad_norm": 1.1481292324836674,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 23392
    },
    {
      "epoch": 0.23393,
      "grad_norm": 1.572492849609787,
      "learning_rate": 0.003,
      "loss": 4.0925,
      "step": 23393
    },
    {
      "epoch": 0.23394,
      "grad_norm": 1.3732563583360473,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 23394
    },
    {
      "epoch": 0.23395,
      "grad_norm": 1.275735982851037,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 23395
    },
    {
      "epoch": 0.23396,
      "grad_norm": 1.380916624018907,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 23396
    },
    {
      "epoch": 0.23397,
      "grad_norm": 1.3210506383537612,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 23397
    },
    {
      "epoch": 0.23398,
      "grad_norm": 1.3783554506449633,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 23398
    },
    {
      "epoch": 0.23399,
      "grad_norm": 1.209440185167752,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 23399
    },
    {
      "epoch": 0.234,
      "grad_norm": 1.4810129691853908,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 23400
    },
    {
      "epoch": 0.23401,
      "grad_norm": 1.3207923745865922,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 23401
    },
    {
      "epoch": 0.23402,
      "grad_norm": 1.4871117893603158,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 23402
    },
    {
      "epoch": 0.23403,
      "grad_norm": 1.2725559733965675,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 23403
    },
    {
      "epoch": 0.23404,
      "grad_norm": 1.6976241417923523,
      "learning_rate": 0.003,
      "loss": 4.0253,
      "step": 23404
    },
    {
      "epoch": 0.23405,
      "grad_norm": 1.2316079912852995,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 23405
    },
    {
      "epoch": 0.23406,
      "grad_norm": 1.379303060405435,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 23406
    },
    {
      "epoch": 0.23407,
      "grad_norm": 1.5705799116220056,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 23407
    },
    {
      "epoch": 0.23408,
      "grad_norm": 1.3894820157771752,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 23408
    },
    {
      "epoch": 0.23409,
      "grad_norm": 1.3482000288335323,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 23409
    },
    {
      "epoch": 0.2341,
      "grad_norm": 1.290661024844589,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 23410
    },
    {
      "epoch": 0.23411,
      "grad_norm": 1.3901306049252855,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 23411
    },
    {
      "epoch": 0.23412,
      "grad_norm": 1.2377484876085936,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 23412
    },
    {
      "epoch": 0.23413,
      "grad_norm": 1.3366616591804248,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 23413
    },
    {
      "epoch": 0.23414,
      "grad_norm": 1.4331132534925277,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 23414
    },
    {
      "epoch": 0.23415,
      "grad_norm": 1.3126353689905936,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 23415
    },
    {
      "epoch": 0.23416,
      "grad_norm": 1.4781456314539463,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 23416
    },
    {
      "epoch": 0.23417,
      "grad_norm": 1.4561657212061085,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 23417
    },
    {
      "epoch": 0.23418,
      "grad_norm": 1.4285506376814623,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 23418
    },
    {
      "epoch": 0.23419,
      "grad_norm": 1.370247952989015,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 23419
    },
    {
      "epoch": 0.2342,
      "grad_norm": 1.2749684833331565,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 23420
    },
    {
      "epoch": 0.23421,
      "grad_norm": 1.5250548904588213,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 23421
    },
    {
      "epoch": 0.23422,
      "grad_norm": 1.187391022985111,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 23422
    },
    {
      "epoch": 0.23423,
      "grad_norm": 1.6147469186980699,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 23423
    },
    {
      "epoch": 0.23424,
      "grad_norm": 1.0474255437095539,
      "learning_rate": 0.003,
      "loss": 4.0681,
      "step": 23424
    },
    {
      "epoch": 0.23425,
      "grad_norm": 1.7465891540813754,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 23425
    },
    {
      "epoch": 0.23426,
      "grad_norm": 1.2244151716829588,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 23426
    },
    {
      "epoch": 0.23427,
      "grad_norm": 1.3679252555123618,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 23427
    },
    {
      "epoch": 0.23428,
      "grad_norm": 1.4915060486582232,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 23428
    },
    {
      "epoch": 0.23429,
      "grad_norm": 1.4721209458612163,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 23429
    },
    {
      "epoch": 0.2343,
      "grad_norm": 1.771403773607354,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 23430
    },
    {
      "epoch": 0.23431,
      "grad_norm": 1.2437319053095972,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 23431
    },
    {
      "epoch": 0.23432,
      "grad_norm": 1.4602353551740295,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 23432
    },
    {
      "epoch": 0.23433,
      "grad_norm": 1.212846807146865,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 23433
    },
    {
      "epoch": 0.23434,
      "grad_norm": 1.4052700980998136,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 23434
    },
    {
      "epoch": 0.23435,
      "grad_norm": 1.4041306961096618,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 23435
    },
    {
      "epoch": 0.23436,
      "grad_norm": 1.374945230547344,
      "learning_rate": 0.003,
      "loss": 4.0292,
      "step": 23436
    },
    {
      "epoch": 0.23437,
      "grad_norm": 1.6102002724792641,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 23437
    },
    {
      "epoch": 0.23438,
      "grad_norm": 1.3578097499182225,
      "learning_rate": 0.003,
      "loss": 4.1029,
      "step": 23438
    },
    {
      "epoch": 0.23439,
      "grad_norm": 1.2544931946268074,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 23439
    },
    {
      "epoch": 0.2344,
      "grad_norm": 1.435947825713841,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 23440
    },
    {
      "epoch": 0.23441,
      "grad_norm": 1.1505188288965753,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 23441
    },
    {
      "epoch": 0.23442,
      "grad_norm": 1.5148787047929175,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 23442
    },
    {
      "epoch": 0.23443,
      "grad_norm": 1.3109857199586687,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 23443
    },
    {
      "epoch": 0.23444,
      "grad_norm": 1.6302184377208593,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 23444
    },
    {
      "epoch": 0.23445,
      "grad_norm": 1.0958710629638364,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 23445
    },
    {
      "epoch": 0.23446,
      "grad_norm": 1.6449759672702116,
      "learning_rate": 0.003,
      "loss": 4.0932,
      "step": 23446
    },
    {
      "epoch": 0.23447,
      "grad_norm": 1.1804608987453538,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 23447
    },
    {
      "epoch": 0.23448,
      "grad_norm": 1.5954880830901266,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 23448
    },
    {
      "epoch": 0.23449,
      "grad_norm": 1.4107087451702547,
      "learning_rate": 0.003,
      "loss": 4.0889,
      "step": 23449
    },
    {
      "epoch": 0.2345,
      "grad_norm": 1.6456755065393895,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 23450
    },
    {
      "epoch": 0.23451,
      "grad_norm": 1.4246644443506047,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 23451
    },
    {
      "epoch": 0.23452,
      "grad_norm": 1.2505246247554676,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 23452
    },
    {
      "epoch": 0.23453,
      "grad_norm": 1.3988344508462989,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 23453
    },
    {
      "epoch": 0.23454,
      "grad_norm": 1.4150773598301845,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 23454
    },
    {
      "epoch": 0.23455,
      "grad_norm": 1.0643776084089818,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 23455
    },
    {
      "epoch": 0.23456,
      "grad_norm": 1.4902059310863867,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 23456
    },
    {
      "epoch": 0.23457,
      "grad_norm": 1.2956298542669884,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 23457
    },
    {
      "epoch": 0.23458,
      "grad_norm": 1.5402630597607896,
      "learning_rate": 0.003,
      "loss": 4.0181,
      "step": 23458
    },
    {
      "epoch": 0.23459,
      "grad_norm": 1.379775772744756,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 23459
    },
    {
      "epoch": 0.2346,
      "grad_norm": 1.4297407268821414,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 23460
    },
    {
      "epoch": 0.23461,
      "grad_norm": 1.167866307132083,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 23461
    },
    {
      "epoch": 0.23462,
      "grad_norm": 1.5211096038407923,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 23462
    },
    {
      "epoch": 0.23463,
      "grad_norm": 1.0167766892965715,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 23463
    },
    {
      "epoch": 0.23464,
      "grad_norm": 1.7284721847452755,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 23464
    },
    {
      "epoch": 0.23465,
      "grad_norm": 1.2114230935605876,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 23465
    },
    {
      "epoch": 0.23466,
      "grad_norm": 1.5320697409612032,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 23466
    },
    {
      "epoch": 0.23467,
      "grad_norm": 1.4745310006514158,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 23467
    },
    {
      "epoch": 0.23468,
      "grad_norm": 1.5868292231327064,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 23468
    },
    {
      "epoch": 0.23469,
      "grad_norm": 1.4583457832566906,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 23469
    },
    {
      "epoch": 0.2347,
      "grad_norm": 1.2634409313940942,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 23470
    },
    {
      "epoch": 0.23471,
      "grad_norm": 1.7875095908216263,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 23471
    },
    {
      "epoch": 0.23472,
      "grad_norm": 1.331252310075803,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 23472
    },
    {
      "epoch": 0.23473,
      "grad_norm": 1.4729126453494954,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 23473
    },
    {
      "epoch": 0.23474,
      "grad_norm": 1.6809016283735423,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 23474
    },
    {
      "epoch": 0.23475,
      "grad_norm": 1.1312357787690661,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 23475
    },
    {
      "epoch": 0.23476,
      "grad_norm": 1.216519874183779,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 23476
    },
    {
      "epoch": 0.23477,
      "grad_norm": 1.4284770336089858,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 23477
    },
    {
      "epoch": 0.23478,
      "grad_norm": 1.5610670537451417,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 23478
    },
    {
      "epoch": 0.23479,
      "grad_norm": 1.3721097626999792,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 23479
    },
    {
      "epoch": 0.2348,
      "grad_norm": 1.2099749031404357,
      "learning_rate": 0.003,
      "loss": 4.1087,
      "step": 23480
    },
    {
      "epoch": 0.23481,
      "grad_norm": 1.4651803028437265,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 23481
    },
    {
      "epoch": 0.23482,
      "grad_norm": 1.0910292321437485,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 23482
    },
    {
      "epoch": 0.23483,
      "grad_norm": 1.4626230500530066,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 23483
    },
    {
      "epoch": 0.23484,
      "grad_norm": 1.2148026937237315,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 23484
    },
    {
      "epoch": 0.23485,
      "grad_norm": 1.4405634035580306,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 23485
    },
    {
      "epoch": 0.23486,
      "grad_norm": 1.5469730217530646,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 23486
    },
    {
      "epoch": 0.23487,
      "grad_norm": 1.3719537071087946,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 23487
    },
    {
      "epoch": 0.23488,
      "grad_norm": 1.8411077853873343,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 23488
    },
    {
      "epoch": 0.23489,
      "grad_norm": 1.1661254921174309,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 23489
    },
    {
      "epoch": 0.2349,
      "grad_norm": 1.317792002689811,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 23490
    },
    {
      "epoch": 0.23491,
      "grad_norm": 1.4869573300670418,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 23491
    },
    {
      "epoch": 0.23492,
      "grad_norm": 1.290033490913107,
      "learning_rate": 0.003,
      "loss": 4.0162,
      "step": 23492
    },
    {
      "epoch": 0.23493,
      "grad_norm": 1.7556965055797964,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 23493
    },
    {
      "epoch": 0.23494,
      "grad_norm": 1.2639642209153696,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 23494
    },
    {
      "epoch": 0.23495,
      "grad_norm": 1.5050470251308685,
      "learning_rate": 0.003,
      "loss": 4.0838,
      "step": 23495
    },
    {
      "epoch": 0.23496,
      "grad_norm": 1.269189098340193,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 23496
    },
    {
      "epoch": 0.23497,
      "grad_norm": 1.3678242319170064,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 23497
    },
    {
      "epoch": 0.23498,
      "grad_norm": 1.2846911244072163,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 23498
    },
    {
      "epoch": 0.23499,
      "grad_norm": 1.4047687043640837,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 23499
    },
    {
      "epoch": 0.235,
      "grad_norm": 1.4032908436053146,
      "learning_rate": 0.003,
      "loss": 4.0841,
      "step": 23500
    },
    {
      "epoch": 0.23501,
      "grad_norm": 1.4658861420478264,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 23501
    },
    {
      "epoch": 0.23502,
      "grad_norm": 1.4790123184827366,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 23502
    },
    {
      "epoch": 0.23503,
      "grad_norm": 1.3415461361994145,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 23503
    },
    {
      "epoch": 0.23504,
      "grad_norm": 1.6424081294110777,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 23504
    },
    {
      "epoch": 0.23505,
      "grad_norm": 1.2811723199997707,
      "learning_rate": 0.003,
      "loss": 4.0933,
      "step": 23505
    },
    {
      "epoch": 0.23506,
      "grad_norm": 1.304736790261805,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 23506
    },
    {
      "epoch": 0.23507,
      "grad_norm": 1.2371494547114326,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 23507
    },
    {
      "epoch": 0.23508,
      "grad_norm": 1.536752203983268,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 23508
    },
    {
      "epoch": 0.23509,
      "grad_norm": 1.1185890320027325,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 23509
    },
    {
      "epoch": 0.2351,
      "grad_norm": 1.577830003306467,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 23510
    },
    {
      "epoch": 0.23511,
      "grad_norm": 1.2613423722497226,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 23511
    },
    {
      "epoch": 0.23512,
      "grad_norm": 1.399843933996001,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 23512
    },
    {
      "epoch": 0.23513,
      "grad_norm": 1.6404305390784972,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 23513
    },
    {
      "epoch": 0.23514,
      "grad_norm": 1.069714512392399,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 23514
    },
    {
      "epoch": 0.23515,
      "grad_norm": 1.5734915129907068,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 23515
    },
    {
      "epoch": 0.23516,
      "grad_norm": 1.465511969395423,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 23516
    },
    {
      "epoch": 0.23517,
      "grad_norm": 1.3599987691119655,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 23517
    },
    {
      "epoch": 0.23518,
      "grad_norm": 1.7162863697047077,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 23518
    },
    {
      "epoch": 0.23519,
      "grad_norm": 1.0657951760927105,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 23519
    },
    {
      "epoch": 0.2352,
      "grad_norm": 1.4633888179031667,
      "learning_rate": 0.003,
      "loss": 4.0942,
      "step": 23520
    },
    {
      "epoch": 0.23521,
      "grad_norm": 1.1581660133998444,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 23521
    },
    {
      "epoch": 0.23522,
      "grad_norm": 1.245044795349148,
      "learning_rate": 0.003,
      "loss": 4.033,
      "step": 23522
    },
    {
      "epoch": 0.23523,
      "grad_norm": 1.4553285563825764,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 23523
    },
    {
      "epoch": 0.23524,
      "grad_norm": 1.2864694674159676,
      "learning_rate": 0.003,
      "loss": 4.1077,
      "step": 23524
    },
    {
      "epoch": 0.23525,
      "grad_norm": 1.367469797784762,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 23525
    },
    {
      "epoch": 0.23526,
      "grad_norm": 1.5847550660382819,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 23526
    },
    {
      "epoch": 0.23527,
      "grad_norm": 1.1267107355595851,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 23527
    },
    {
      "epoch": 0.23528,
      "grad_norm": 1.1254508854791694,
      "learning_rate": 0.003,
      "loss": 4.0227,
      "step": 23528
    },
    {
      "epoch": 0.23529,
      "grad_norm": 1.499019604704434,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 23529
    },
    {
      "epoch": 0.2353,
      "grad_norm": 1.3221042060715784,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 23530
    },
    {
      "epoch": 0.23531,
      "grad_norm": 1.500259870860935,
      "learning_rate": 0.003,
      "loss": 4.0031,
      "step": 23531
    },
    {
      "epoch": 0.23532,
      "grad_norm": 1.3774606507063807,
      "learning_rate": 0.003,
      "loss": 4.0891,
      "step": 23532
    },
    {
      "epoch": 0.23533,
      "grad_norm": 1.488951266188294,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 23533
    },
    {
      "epoch": 0.23534,
      "grad_norm": 1.2468561355909225,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 23534
    },
    {
      "epoch": 0.23535,
      "grad_norm": 1.4126856395387974,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 23535
    },
    {
      "epoch": 0.23536,
      "grad_norm": 1.3393107042206556,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 23536
    },
    {
      "epoch": 0.23537,
      "grad_norm": 1.4675866749726905,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 23537
    },
    {
      "epoch": 0.23538,
      "grad_norm": 1.6632827300737214,
      "learning_rate": 0.003,
      "loss": 4.0247,
      "step": 23538
    },
    {
      "epoch": 0.23539,
      "grad_norm": 1.2609099121791207,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 23539
    },
    {
      "epoch": 0.2354,
      "grad_norm": 1.2448403795611518,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 23540
    },
    {
      "epoch": 0.23541,
      "grad_norm": 1.644576973005076,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 23541
    },
    {
      "epoch": 0.23542,
      "grad_norm": 1.1833374314746115,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 23542
    },
    {
      "epoch": 0.23543,
      "grad_norm": 1.5165897407239832,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 23543
    },
    {
      "epoch": 0.23544,
      "grad_norm": 1.3440478216017815,
      "learning_rate": 0.003,
      "loss": 4.0374,
      "step": 23544
    },
    {
      "epoch": 0.23545,
      "grad_norm": 1.1487889574704027,
      "learning_rate": 0.003,
      "loss": 4.0301,
      "step": 23545
    },
    {
      "epoch": 0.23546,
      "grad_norm": 1.551788326118772,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 23546
    },
    {
      "epoch": 0.23547,
      "grad_norm": 1.1445624709477686,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 23547
    },
    {
      "epoch": 0.23548,
      "grad_norm": 1.4151480868609965,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 23548
    },
    {
      "epoch": 0.23549,
      "grad_norm": 1.3238955882498833,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 23549
    },
    {
      "epoch": 0.2355,
      "grad_norm": 1.3394488361646577,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 23550
    },
    {
      "epoch": 0.23551,
      "grad_norm": 1.1350649434168805,
      "learning_rate": 0.003,
      "loss": 4.0266,
      "step": 23551
    },
    {
      "epoch": 0.23552,
      "grad_norm": 1.6686796802772081,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 23552
    },
    {
      "epoch": 0.23553,
      "grad_norm": 1.2501725999871147,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 23553
    },
    {
      "epoch": 0.23554,
      "grad_norm": 1.4502452262016516,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 23554
    },
    {
      "epoch": 0.23555,
      "grad_norm": 1.2627182999775564,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 23555
    },
    {
      "epoch": 0.23556,
      "grad_norm": 1.5865820054319144,
      "learning_rate": 0.003,
      "loss": 4.0309,
      "step": 23556
    },
    {
      "epoch": 0.23557,
      "grad_norm": 1.212166402140409,
      "learning_rate": 0.003,
      "loss": 4.0217,
      "step": 23557
    },
    {
      "epoch": 0.23558,
      "grad_norm": 1.3941679528516313,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 23558
    },
    {
      "epoch": 0.23559,
      "grad_norm": 1.1675756947254337,
      "learning_rate": 0.003,
      "loss": 4.0218,
      "step": 23559
    },
    {
      "epoch": 0.2356,
      "grad_norm": 1.769189889288519,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 23560
    },
    {
      "epoch": 0.23561,
      "grad_norm": 1.2395302226234761,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 23561
    },
    {
      "epoch": 0.23562,
      "grad_norm": 1.7037482148406256,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 23562
    },
    {
      "epoch": 0.23563,
      "grad_norm": 1.3006614121391928,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 23563
    },
    {
      "epoch": 0.23564,
      "grad_norm": 1.7178535780951354,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 23564
    },
    {
      "epoch": 0.23565,
      "grad_norm": 1.2511621782698048,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 23565
    },
    {
      "epoch": 0.23566,
      "grad_norm": 1.5401991286903127,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 23566
    },
    {
      "epoch": 0.23567,
      "grad_norm": 1.127857362854384,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 23567
    },
    {
      "epoch": 0.23568,
      "grad_norm": 1.4358199888136134,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 23568
    },
    {
      "epoch": 0.23569,
      "grad_norm": 1.4887109954764102,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 23569
    },
    {
      "epoch": 0.2357,
      "grad_norm": 1.0503727518447026,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 23570
    },
    {
      "epoch": 0.23571,
      "grad_norm": 1.646107473638904,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 23571
    },
    {
      "epoch": 0.23572,
      "grad_norm": 1.1858287766157665,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 23572
    },
    {
      "epoch": 0.23573,
      "grad_norm": 1.6227116841989064,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 23573
    },
    {
      "epoch": 0.23574,
      "grad_norm": 1.2894256998778117,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 23574
    },
    {
      "epoch": 0.23575,
      "grad_norm": 1.6329288467353305,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 23575
    },
    {
      "epoch": 0.23576,
      "grad_norm": 1.0797845743008918,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 23576
    },
    {
      "epoch": 0.23577,
      "grad_norm": 1.4520211993906205,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 23577
    },
    {
      "epoch": 0.23578,
      "grad_norm": 1.3284805516276323,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 23578
    },
    {
      "epoch": 0.23579,
      "grad_norm": 1.3618990962596662,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 23579
    },
    {
      "epoch": 0.2358,
      "grad_norm": 1.145652833990343,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 23580
    },
    {
      "epoch": 0.23581,
      "grad_norm": 1.514144047022743,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 23581
    },
    {
      "epoch": 0.23582,
      "grad_norm": 1.3285574107066616,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 23582
    },
    {
      "epoch": 0.23583,
      "grad_norm": 1.664632172552976,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 23583
    },
    {
      "epoch": 0.23584,
      "grad_norm": 1.5167512332361512,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 23584
    },
    {
      "epoch": 0.23585,
      "grad_norm": 1.7125282195167086,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 23585
    },
    {
      "epoch": 0.23586,
      "grad_norm": 1.130931265910693,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 23586
    },
    {
      "epoch": 0.23587,
      "grad_norm": 1.4755149683551616,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 23587
    },
    {
      "epoch": 0.23588,
      "grad_norm": 1.5338596236600828,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 23588
    },
    {
      "epoch": 0.23589,
      "grad_norm": 1.6684215415562105,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 23589
    },
    {
      "epoch": 0.2359,
      "grad_norm": 1.2147513914037273,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 23590
    },
    {
      "epoch": 0.23591,
      "grad_norm": 1.7053040574485951,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 23591
    },
    {
      "epoch": 0.23592,
      "grad_norm": 1.2224905964458432,
      "learning_rate": 0.003,
      "loss": 4.0832,
      "step": 23592
    },
    {
      "epoch": 0.23593,
      "grad_norm": 1.3395516006038173,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 23593
    },
    {
      "epoch": 0.23594,
      "grad_norm": 1.0564481744672038,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 23594
    },
    {
      "epoch": 0.23595,
      "grad_norm": 1.5427023138415519,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 23595
    },
    {
      "epoch": 0.23596,
      "grad_norm": 1.1457494099338927,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 23596
    },
    {
      "epoch": 0.23597,
      "grad_norm": 1.4923339434817735,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 23597
    },
    {
      "epoch": 0.23598,
      "grad_norm": 1.3421229492117712,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 23598
    },
    {
      "epoch": 0.23599,
      "grad_norm": 1.3173561641282392,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 23599
    },
    {
      "epoch": 0.236,
      "grad_norm": 1.5246088725331406,
      "learning_rate": 0.003,
      "loss": 4.0899,
      "step": 23600
    },
    {
      "epoch": 0.23601,
      "grad_norm": 1.3983035419160177,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 23601
    },
    {
      "epoch": 0.23602,
      "grad_norm": 1.7450572159300748,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 23602
    },
    {
      "epoch": 0.23603,
      "grad_norm": 1.3108541616789215,
      "learning_rate": 0.003,
      "loss": 4.0904,
      "step": 23603
    },
    {
      "epoch": 0.23604,
      "grad_norm": 1.496498747979966,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 23604
    },
    {
      "epoch": 0.23605,
      "grad_norm": 1.335927751276949,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 23605
    },
    {
      "epoch": 0.23606,
      "grad_norm": 1.376043104839073,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 23606
    },
    {
      "epoch": 0.23607,
      "grad_norm": 1.1983800794163468,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 23607
    },
    {
      "epoch": 0.23608,
      "grad_norm": 1.4116727332495218,
      "learning_rate": 0.003,
      "loss": 4.0329,
      "step": 23608
    },
    {
      "epoch": 0.23609,
      "grad_norm": 1.327947045088272,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 23609
    },
    {
      "epoch": 0.2361,
      "grad_norm": 1.3483372317783033,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 23610
    },
    {
      "epoch": 0.23611,
      "grad_norm": 1.1669563261202986,
      "learning_rate": 0.003,
      "loss": 4.0322,
      "step": 23611
    },
    {
      "epoch": 0.23612,
      "grad_norm": 1.6665118008124027,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 23612
    },
    {
      "epoch": 0.23613,
      "grad_norm": 1.3739468002730715,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 23613
    },
    {
      "epoch": 0.23614,
      "grad_norm": 1.233809584077891,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 23614
    },
    {
      "epoch": 0.23615,
      "grad_norm": 1.572931478480147,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 23615
    },
    {
      "epoch": 0.23616,
      "grad_norm": 1.1738648294276521,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 23616
    },
    {
      "epoch": 0.23617,
      "grad_norm": 1.4122694266819866,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 23617
    },
    {
      "epoch": 0.23618,
      "grad_norm": 1.5351955831772994,
      "learning_rate": 0.003,
      "loss": 4.0209,
      "step": 23618
    },
    {
      "epoch": 0.23619,
      "grad_norm": 1.1480549871649792,
      "learning_rate": 0.003,
      "loss": 4.0234,
      "step": 23619
    },
    {
      "epoch": 0.2362,
      "grad_norm": 1.5886702088393145,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 23620
    },
    {
      "epoch": 0.23621,
      "grad_norm": 1.192709069786014,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 23621
    },
    {
      "epoch": 0.23622,
      "grad_norm": 1.3516921026147195,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 23622
    },
    {
      "epoch": 0.23623,
      "grad_norm": 1.658909331920533,
      "learning_rate": 0.003,
      "loss": 4.0183,
      "step": 23623
    },
    {
      "epoch": 0.23624,
      "grad_norm": 1.2287851205103206,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 23624
    },
    {
      "epoch": 0.23625,
      "grad_norm": 1.556326240733628,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 23625
    },
    {
      "epoch": 0.23626,
      "grad_norm": 1.1704765854460446,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 23626
    },
    {
      "epoch": 0.23627,
      "grad_norm": 1.5200472734273278,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 23627
    },
    {
      "epoch": 0.23628,
      "grad_norm": 1.319051527043223,
      "learning_rate": 0.003,
      "loss": 4.0292,
      "step": 23628
    },
    {
      "epoch": 0.23629,
      "grad_norm": 1.416119703817279,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 23629
    },
    {
      "epoch": 0.2363,
      "grad_norm": 1.3133879921714262,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 23630
    },
    {
      "epoch": 0.23631,
      "grad_norm": 1.437023945086885,
      "learning_rate": 0.003,
      "loss": 4.0231,
      "step": 23631
    },
    {
      "epoch": 0.23632,
      "grad_norm": 1.1513591766520368,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 23632
    },
    {
      "epoch": 0.23633,
      "grad_norm": 1.5193434001319606,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 23633
    },
    {
      "epoch": 0.23634,
      "grad_norm": 1.1623495610484178,
      "learning_rate": 0.003,
      "loss": 4.0811,
      "step": 23634
    },
    {
      "epoch": 0.23635,
      "grad_norm": 1.5844147563868947,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 23635
    },
    {
      "epoch": 0.23636,
      "grad_norm": 1.2318230489096145,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 23636
    },
    {
      "epoch": 0.23637,
      "grad_norm": 1.432070004872038,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 23637
    },
    {
      "epoch": 0.23638,
      "grad_norm": 1.4564620691728094,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 23638
    },
    {
      "epoch": 0.23639,
      "grad_norm": 1.3721671151108517,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 23639
    },
    {
      "epoch": 0.2364,
      "grad_norm": 1.5991308480593305,
      "learning_rate": 0.003,
      "loss": 4.0304,
      "step": 23640
    },
    {
      "epoch": 0.23641,
      "grad_norm": 1.4340117617888781,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 23641
    },
    {
      "epoch": 0.23642,
      "grad_norm": 1.4102524067431634,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 23642
    },
    {
      "epoch": 0.23643,
      "grad_norm": 1.4785654806317532,
      "learning_rate": 0.003,
      "loss": 4.0884,
      "step": 23643
    },
    {
      "epoch": 0.23644,
      "grad_norm": 1.3197359125305035,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 23644
    },
    {
      "epoch": 0.23645,
      "grad_norm": 1.225739894839928,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 23645
    },
    {
      "epoch": 0.23646,
      "grad_norm": 1.549236165081848,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 23646
    },
    {
      "epoch": 0.23647,
      "grad_norm": 1.3554814343574306,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 23647
    },
    {
      "epoch": 0.23648,
      "grad_norm": 1.4858102492073069,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 23648
    },
    {
      "epoch": 0.23649,
      "grad_norm": 1.1460061042209861,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 23649
    },
    {
      "epoch": 0.2365,
      "grad_norm": 1.4003691538584704,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 23650
    },
    {
      "epoch": 0.23651,
      "grad_norm": 1.3420310754803368,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 23651
    },
    {
      "epoch": 0.23652,
      "grad_norm": 1.4427500744279627,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 23652
    },
    {
      "epoch": 0.23653,
      "grad_norm": 0.9981497956929709,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 23653
    },
    {
      "epoch": 0.23654,
      "grad_norm": 1.5434548242262909,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 23654
    },
    {
      "epoch": 0.23655,
      "grad_norm": 1.282829108353015,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 23655
    },
    {
      "epoch": 0.23656,
      "grad_norm": 1.4051043368170837,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 23656
    },
    {
      "epoch": 0.23657,
      "grad_norm": 1.696644316180369,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 23657
    },
    {
      "epoch": 0.23658,
      "grad_norm": 1.3853789495479925,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 23658
    },
    {
      "epoch": 0.23659,
      "grad_norm": 1.465777658921139,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 23659
    },
    {
      "epoch": 0.2366,
      "grad_norm": 1.535637065863811,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 23660
    },
    {
      "epoch": 0.23661,
      "grad_norm": 1.2407916991708687,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 23661
    },
    {
      "epoch": 0.23662,
      "grad_norm": 1.3224541634369102,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 23662
    },
    {
      "epoch": 0.23663,
      "grad_norm": 1.2735361590109953,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 23663
    },
    {
      "epoch": 0.23664,
      "grad_norm": 1.1599813143934437,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 23664
    },
    {
      "epoch": 0.23665,
      "grad_norm": 1.5231624381414275,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 23665
    },
    {
      "epoch": 0.23666,
      "grad_norm": 1.3112522071404034,
      "learning_rate": 0.003,
      "loss": 4.115,
      "step": 23666
    },
    {
      "epoch": 0.23667,
      "grad_norm": 1.6521600127322806,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 23667
    },
    {
      "epoch": 0.23668,
      "grad_norm": 1.424585037913398,
      "learning_rate": 0.003,
      "loss": 4.0328,
      "step": 23668
    },
    {
      "epoch": 0.23669,
      "grad_norm": 1.3715206623300467,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 23669
    },
    {
      "epoch": 0.2367,
      "grad_norm": 1.4647596590040015,
      "learning_rate": 0.003,
      "loss": 4.0238,
      "step": 23670
    },
    {
      "epoch": 0.23671,
      "grad_norm": 1.2028551548756445,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 23671
    },
    {
      "epoch": 0.23672,
      "grad_norm": 1.4944334500334975,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 23672
    },
    {
      "epoch": 0.23673,
      "grad_norm": 1.5016964241815314,
      "learning_rate": 0.003,
      "loss": 4.0212,
      "step": 23673
    },
    {
      "epoch": 0.23674,
      "grad_norm": 1.3039651288047591,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 23674
    },
    {
      "epoch": 0.23675,
      "grad_norm": 1.3626413074604664,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 23675
    },
    {
      "epoch": 0.23676,
      "grad_norm": 1.4496866940216593,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 23676
    },
    {
      "epoch": 0.23677,
      "grad_norm": 1.183866697465119,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 23677
    },
    {
      "epoch": 0.23678,
      "grad_norm": 1.2890338921266171,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 23678
    },
    {
      "epoch": 0.23679,
      "grad_norm": 1.1721886344381682,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 23679
    },
    {
      "epoch": 0.2368,
      "grad_norm": 1.6278739187063578,
      "learning_rate": 0.003,
      "loss": 4.0189,
      "step": 23680
    },
    {
      "epoch": 0.23681,
      "grad_norm": 1.2920736306502882,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 23681
    },
    {
      "epoch": 0.23682,
      "grad_norm": 1.5685630119267018,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 23682
    },
    {
      "epoch": 0.23683,
      "grad_norm": 1.3732226614011787,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 23683
    },
    {
      "epoch": 0.23684,
      "grad_norm": 1.5887688049102116,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 23684
    },
    {
      "epoch": 0.23685,
      "grad_norm": 1.3730346857562286,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 23685
    },
    {
      "epoch": 0.23686,
      "grad_norm": 1.4689113599286796,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 23686
    },
    {
      "epoch": 0.23687,
      "grad_norm": 1.4095915160540724,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 23687
    },
    {
      "epoch": 0.23688,
      "grad_norm": 1.257061519749002,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 23688
    },
    {
      "epoch": 0.23689,
      "grad_norm": 1.5853550954893474,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 23689
    },
    {
      "epoch": 0.2369,
      "grad_norm": 1.1632440511212632,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 23690
    },
    {
      "epoch": 0.23691,
      "grad_norm": 1.4970128095142774,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 23691
    },
    {
      "epoch": 0.23692,
      "grad_norm": 1.3638489336068609,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 23692
    },
    {
      "epoch": 0.23693,
      "grad_norm": 1.6696265294868635,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 23693
    },
    {
      "epoch": 0.23694,
      "grad_norm": 1.2733102986013913,
      "learning_rate": 0.003,
      "loss": 4.0162,
      "step": 23694
    },
    {
      "epoch": 0.23695,
      "grad_norm": 1.2757334531279756,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 23695
    },
    {
      "epoch": 0.23696,
      "grad_norm": 1.592273124766524,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 23696
    },
    {
      "epoch": 0.23697,
      "grad_norm": 1.2613170095067048,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 23697
    },
    {
      "epoch": 0.23698,
      "grad_norm": 1.4239489887722652,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 23698
    },
    {
      "epoch": 0.23699,
      "grad_norm": 1.5151470075544053,
      "learning_rate": 0.003,
      "loss": 4.0229,
      "step": 23699
    },
    {
      "epoch": 0.237,
      "grad_norm": 1.3190693526624828,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 23700
    },
    {
      "epoch": 0.23701,
      "grad_norm": 1.2625702194076969,
      "learning_rate": 0.003,
      "loss": 4.0185,
      "step": 23701
    },
    {
      "epoch": 0.23702,
      "grad_norm": 1.3163639093166508,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 23702
    },
    {
      "epoch": 0.23703,
      "grad_norm": 1.4585642788834488,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 23703
    },
    {
      "epoch": 0.23704,
      "grad_norm": 1.2463201430820265,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 23704
    },
    {
      "epoch": 0.23705,
      "grad_norm": 1.399710281982181,
      "learning_rate": 0.003,
      "loss": 4.0247,
      "step": 23705
    },
    {
      "epoch": 0.23706,
      "grad_norm": 1.2254111778445447,
      "learning_rate": 0.003,
      "loss": 4.0209,
      "step": 23706
    },
    {
      "epoch": 0.23707,
      "grad_norm": 1.4938337992820137,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 23707
    },
    {
      "epoch": 0.23708,
      "grad_norm": 1.471848255121138,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 23708
    },
    {
      "epoch": 0.23709,
      "grad_norm": 1.6508759526546322,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 23709
    },
    {
      "epoch": 0.2371,
      "grad_norm": 1.3579753007102955,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 23710
    },
    {
      "epoch": 0.23711,
      "grad_norm": 1.2823517492476701,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 23711
    },
    {
      "epoch": 0.23712,
      "grad_norm": 1.4418318137878419,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 23712
    },
    {
      "epoch": 0.23713,
      "grad_norm": 1.3365225311086284,
      "learning_rate": 0.003,
      "loss": 4.0436,
      "step": 23713
    },
    {
      "epoch": 0.23714,
      "grad_norm": 1.4961142610676437,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 23714
    },
    {
      "epoch": 0.23715,
      "grad_norm": 1.2835543909532312,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 23715
    },
    {
      "epoch": 0.23716,
      "grad_norm": 1.4022699343194178,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 23716
    },
    {
      "epoch": 0.23717,
      "grad_norm": 1.2939316952306115,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 23717
    },
    {
      "epoch": 0.23718,
      "grad_norm": 1.4151173691120322,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 23718
    },
    {
      "epoch": 0.23719,
      "grad_norm": 1.460839796525741,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 23719
    },
    {
      "epoch": 0.2372,
      "grad_norm": 1.5745424089000837,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 23720
    },
    {
      "epoch": 0.23721,
      "grad_norm": 1.3016430484884807,
      "learning_rate": 0.003,
      "loss": 4.0185,
      "step": 23721
    },
    {
      "epoch": 0.23722,
      "grad_norm": 1.4548967565753528,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 23722
    },
    {
      "epoch": 0.23723,
      "grad_norm": 1.308716035128893,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 23723
    },
    {
      "epoch": 0.23724,
      "grad_norm": 1.5032293477234924,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 23724
    },
    {
      "epoch": 0.23725,
      "grad_norm": 1.0581030438208996,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 23725
    },
    {
      "epoch": 0.23726,
      "grad_norm": 1.4633059635079007,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 23726
    },
    {
      "epoch": 0.23727,
      "grad_norm": 1.2753225451283543,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 23727
    },
    {
      "epoch": 0.23728,
      "grad_norm": 1.3293575075976654,
      "learning_rate": 0.003,
      "loss": 4.0044,
      "step": 23728
    },
    {
      "epoch": 0.23729,
      "grad_norm": 1.4792940776092705,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 23729
    },
    {
      "epoch": 0.2373,
      "grad_norm": 1.2234705801377,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 23730
    },
    {
      "epoch": 0.23731,
      "grad_norm": 1.3607529940793255,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 23731
    },
    {
      "epoch": 0.23732,
      "grad_norm": 1.4660074124153655,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 23732
    },
    {
      "epoch": 0.23733,
      "grad_norm": 1.3950089055417283,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 23733
    },
    {
      "epoch": 0.23734,
      "grad_norm": 1.4855922809082651,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 23734
    },
    {
      "epoch": 0.23735,
      "grad_norm": 1.1619249957598732,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 23735
    },
    {
      "epoch": 0.23736,
      "grad_norm": 1.515395474163773,
      "learning_rate": 0.003,
      "loss": 4.026,
      "step": 23736
    },
    {
      "epoch": 0.23737,
      "grad_norm": 1.2710030388482008,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 23737
    },
    {
      "epoch": 0.23738,
      "grad_norm": 1.5125834984056037,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 23738
    },
    {
      "epoch": 0.23739,
      "grad_norm": 1.4517012803959244,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 23739
    },
    {
      "epoch": 0.2374,
      "grad_norm": 1.67274048647102,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 23740
    },
    {
      "epoch": 0.23741,
      "grad_norm": 1.292461838734173,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 23741
    },
    {
      "epoch": 0.23742,
      "grad_norm": 1.2820159233178188,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 23742
    },
    {
      "epoch": 0.23743,
      "grad_norm": 1.7082296540956725,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 23743
    },
    {
      "epoch": 0.23744,
      "grad_norm": 1.4325084137542587,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 23744
    },
    {
      "epoch": 0.23745,
      "grad_norm": 1.4595301237648413,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 23745
    },
    {
      "epoch": 0.23746,
      "grad_norm": 1.4245005043264383,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 23746
    },
    {
      "epoch": 0.23747,
      "grad_norm": 1.304128097745544,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 23747
    },
    {
      "epoch": 0.23748,
      "grad_norm": 1.2330748303873393,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 23748
    },
    {
      "epoch": 0.23749,
      "grad_norm": 1.2902122167464298,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 23749
    },
    {
      "epoch": 0.2375,
      "grad_norm": 1.357336213341285,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 23750
    },
    {
      "epoch": 0.23751,
      "grad_norm": 1.4101977731690332,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 23751
    },
    {
      "epoch": 0.23752,
      "grad_norm": 1.5671052386243283,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 23752
    },
    {
      "epoch": 0.23753,
      "grad_norm": 1.4561874458523836,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 23753
    },
    {
      "epoch": 0.23754,
      "grad_norm": 1.6496280310214197,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 23754
    },
    {
      "epoch": 0.23755,
      "grad_norm": 1.6085701026443695,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 23755
    },
    {
      "epoch": 0.23756,
      "grad_norm": 1.1582347693150215,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 23756
    },
    {
      "epoch": 0.23757,
      "grad_norm": 1.562853208179962,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 23757
    },
    {
      "epoch": 0.23758,
      "grad_norm": 1.2483496470144355,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 23758
    },
    {
      "epoch": 0.23759,
      "grad_norm": 1.3270835572321285,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 23759
    },
    {
      "epoch": 0.2376,
      "grad_norm": 1.3381027466485975,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 23760
    },
    {
      "epoch": 0.23761,
      "grad_norm": 1.3795053871405227,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 23761
    },
    {
      "epoch": 0.23762,
      "grad_norm": 1.4713292606496422,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 23762
    },
    {
      "epoch": 0.23763,
      "grad_norm": 1.13310181388237,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 23763
    },
    {
      "epoch": 0.23764,
      "grad_norm": 1.4365549016004904,
      "learning_rate": 0.003,
      "loss": 4.0202,
      "step": 23764
    },
    {
      "epoch": 0.23765,
      "grad_norm": 1.1297188680876304,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 23765
    },
    {
      "epoch": 0.23766,
      "grad_norm": 1.4116402948496631,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 23766
    },
    {
      "epoch": 0.23767,
      "grad_norm": 1.3164830502265807,
      "learning_rate": 0.003,
      "loss": 4.0267,
      "step": 23767
    },
    {
      "epoch": 0.23768,
      "grad_norm": 1.2918661745473947,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 23768
    },
    {
      "epoch": 0.23769,
      "grad_norm": 1.5637865132804893,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 23769
    },
    {
      "epoch": 0.2377,
      "grad_norm": 1.3124835634283885,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 23770
    },
    {
      "epoch": 0.23771,
      "grad_norm": 1.8116991522849801,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 23771
    },
    {
      "epoch": 0.23772,
      "grad_norm": 1.350272520718708,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 23772
    },
    {
      "epoch": 0.23773,
      "grad_norm": 1.4044574815881063,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 23773
    },
    {
      "epoch": 0.23774,
      "grad_norm": 1.4913650240195102,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 23774
    },
    {
      "epoch": 0.23775,
      "grad_norm": 1.3945788119693396,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 23775
    },
    {
      "epoch": 0.23776,
      "grad_norm": 1.4266427732762825,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 23776
    },
    {
      "epoch": 0.23777,
      "grad_norm": 1.3372979525506867,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 23777
    },
    {
      "epoch": 0.23778,
      "grad_norm": 1.3489631019606934,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 23778
    },
    {
      "epoch": 0.23779,
      "grad_norm": 1.3421415503862049,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 23779
    },
    {
      "epoch": 0.2378,
      "grad_norm": 1.675774883772357,
      "learning_rate": 0.003,
      "loss": 4.0086,
      "step": 23780
    },
    {
      "epoch": 0.23781,
      "grad_norm": 1.1704258782292785,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 23781
    },
    {
      "epoch": 0.23782,
      "grad_norm": 1.4356232365793093,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 23782
    },
    {
      "epoch": 0.23783,
      "grad_norm": 1.364742331801125,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 23783
    },
    {
      "epoch": 0.23784,
      "grad_norm": 1.275911914822051,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 23784
    },
    {
      "epoch": 0.23785,
      "grad_norm": 1.4887763898630668,
      "learning_rate": 0.003,
      "loss": 4.0308,
      "step": 23785
    },
    {
      "epoch": 0.23786,
      "grad_norm": 1.331803732350777,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 23786
    },
    {
      "epoch": 0.23787,
      "grad_norm": 1.3698619919514319,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 23787
    },
    {
      "epoch": 0.23788,
      "grad_norm": 1.5433500274899132,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 23788
    },
    {
      "epoch": 0.23789,
      "grad_norm": 1.20440428110954,
      "learning_rate": 0.003,
      "loss": 4.0333,
      "step": 23789
    },
    {
      "epoch": 0.2379,
      "grad_norm": 1.3825456419984896,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 23790
    },
    {
      "epoch": 0.23791,
      "grad_norm": 1.5089479161473718,
      "learning_rate": 0.003,
      "loss": 4.027,
      "step": 23791
    },
    {
      "epoch": 0.23792,
      "grad_norm": 1.4206593693592973,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 23792
    },
    {
      "epoch": 0.23793,
      "grad_norm": 1.5144565745928913,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 23793
    },
    {
      "epoch": 0.23794,
      "grad_norm": 1.3032704970872415,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 23794
    },
    {
      "epoch": 0.23795,
      "grad_norm": 1.3063518106157503,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 23795
    },
    {
      "epoch": 0.23796,
      "grad_norm": 1.4414575007722408,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 23796
    },
    {
      "epoch": 0.23797,
      "grad_norm": 1.446387987669235,
      "learning_rate": 0.003,
      "loss": 4.0271,
      "step": 23797
    },
    {
      "epoch": 0.23798,
      "grad_norm": 1.54667755487038,
      "learning_rate": 0.003,
      "loss": 4.0289,
      "step": 23798
    },
    {
      "epoch": 0.23799,
      "grad_norm": 1.266682287593711,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 23799
    },
    {
      "epoch": 0.238,
      "grad_norm": 1.312003475435821,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 23800
    },
    {
      "epoch": 0.23801,
      "grad_norm": 1.4088559393432345,
      "learning_rate": 0.003,
      "loss": 4.092,
      "step": 23801
    },
    {
      "epoch": 0.23802,
      "grad_norm": 1.6838138418644581,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 23802
    },
    {
      "epoch": 0.23803,
      "grad_norm": 1.215442220818897,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 23803
    },
    {
      "epoch": 0.23804,
      "grad_norm": 1.6589090290364392,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 23804
    },
    {
      "epoch": 0.23805,
      "grad_norm": 0.9436150220405386,
      "learning_rate": 0.003,
      "loss": 4.0306,
      "step": 23805
    },
    {
      "epoch": 0.23806,
      "grad_norm": 1.7691316974642126,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 23806
    },
    {
      "epoch": 0.23807,
      "grad_norm": 1.477498995331464,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 23807
    },
    {
      "epoch": 0.23808,
      "grad_norm": 1.304295383588438,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 23808
    },
    {
      "epoch": 0.23809,
      "grad_norm": 1.614707971805161,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 23809
    },
    {
      "epoch": 0.2381,
      "grad_norm": 1.0162396579929045,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 23810
    },
    {
      "epoch": 0.23811,
      "grad_norm": 1.5559334686374409,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 23811
    },
    {
      "epoch": 0.23812,
      "grad_norm": 1.5982991422660973,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 23812
    },
    {
      "epoch": 0.23813,
      "grad_norm": 1.276807995696549,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 23813
    },
    {
      "epoch": 0.23814,
      "grad_norm": 1.2772523958571014,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 23814
    },
    {
      "epoch": 0.23815,
      "grad_norm": 1.3132961813792767,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 23815
    },
    {
      "epoch": 0.23816,
      "grad_norm": 1.309818868243345,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 23816
    },
    {
      "epoch": 0.23817,
      "grad_norm": 1.4249962150156523,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 23817
    },
    {
      "epoch": 0.23818,
      "grad_norm": 1.2297845460360541,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 23818
    },
    {
      "epoch": 0.23819,
      "grad_norm": 1.4499472388015047,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 23819
    },
    {
      "epoch": 0.2382,
      "grad_norm": 1.6833199053099153,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 23820
    },
    {
      "epoch": 0.23821,
      "grad_norm": 1.432908495373751,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 23821
    },
    {
      "epoch": 0.23822,
      "grad_norm": 1.0842052591519995,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 23822
    },
    {
      "epoch": 0.23823,
      "grad_norm": 1.4007313293813313,
      "learning_rate": 0.003,
      "loss": 4.0628,
      "step": 23823
    },
    {
      "epoch": 0.23824,
      "grad_norm": 1.6372371721978412,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 23824
    },
    {
      "epoch": 0.23825,
      "grad_norm": 1.1039519693304536,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 23825
    },
    {
      "epoch": 0.23826,
      "grad_norm": 1.7034974107340861,
      "learning_rate": 0.003,
      "loss": 4.0333,
      "step": 23826
    },
    {
      "epoch": 0.23827,
      "grad_norm": 1.1050668562005355,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 23827
    },
    {
      "epoch": 0.23828,
      "grad_norm": 1.5089155098277798,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 23828
    },
    {
      "epoch": 0.23829,
      "grad_norm": 1.2680754184912342,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 23829
    },
    {
      "epoch": 0.2383,
      "grad_norm": 1.3294701242887348,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 23830
    },
    {
      "epoch": 0.23831,
      "grad_norm": 1.4721807512161498,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 23831
    },
    {
      "epoch": 0.23832,
      "grad_norm": 1.2474083008461583,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 23832
    },
    {
      "epoch": 0.23833,
      "grad_norm": 1.5024746972000815,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 23833
    },
    {
      "epoch": 0.23834,
      "grad_norm": 1.511506487520023,
      "learning_rate": 0.003,
      "loss": 4.0247,
      "step": 23834
    },
    {
      "epoch": 0.23835,
      "grad_norm": 1.6908269131540616,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 23835
    },
    {
      "epoch": 0.23836,
      "grad_norm": 1.3790388819882633,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 23836
    },
    {
      "epoch": 0.23837,
      "grad_norm": 1.3285309395252207,
      "learning_rate": 0.003,
      "loss": 4.0953,
      "step": 23837
    },
    {
      "epoch": 0.23838,
      "grad_norm": 1.2743391450563046,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 23838
    },
    {
      "epoch": 0.23839,
      "grad_norm": 1.6217943225615725,
      "learning_rate": 0.003,
      "loss": 4.063,
      "step": 23839
    },
    {
      "epoch": 0.2384,
      "grad_norm": 1.1152541200610773,
      "learning_rate": 0.003,
      "loss": 4.0246,
      "step": 23840
    },
    {
      "epoch": 0.23841,
      "grad_norm": 1.6173808582178677,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 23841
    },
    {
      "epoch": 0.23842,
      "grad_norm": 1.1833309294393404,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 23842
    },
    {
      "epoch": 0.23843,
      "grad_norm": 1.7304148497630552,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 23843
    },
    {
      "epoch": 0.23844,
      "grad_norm": 1.2326838908494742,
      "learning_rate": 0.003,
      "loss": 4.0652,
      "step": 23844
    },
    {
      "epoch": 0.23845,
      "grad_norm": 1.47916800059893,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 23845
    },
    {
      "epoch": 0.23846,
      "grad_norm": 1.1302090760844912,
      "learning_rate": 0.003,
      "loss": 4.0202,
      "step": 23846
    },
    {
      "epoch": 0.23847,
      "grad_norm": 1.5664279076266978,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 23847
    },
    {
      "epoch": 0.23848,
      "grad_norm": 1.3737071177931959,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 23848
    },
    {
      "epoch": 0.23849,
      "grad_norm": 1.3654776330550957,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 23849
    },
    {
      "epoch": 0.2385,
      "grad_norm": 1.5070315706528226,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 23850
    },
    {
      "epoch": 0.23851,
      "grad_norm": 1.3873273151489298,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 23851
    },
    {
      "epoch": 0.23852,
      "grad_norm": 1.33552502928097,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 23852
    },
    {
      "epoch": 0.23853,
      "grad_norm": 1.4264683902740416,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 23853
    },
    {
      "epoch": 0.23854,
      "grad_norm": 1.1861298323305576,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 23854
    },
    {
      "epoch": 0.23855,
      "grad_norm": 1.4632123550935487,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 23855
    },
    {
      "epoch": 0.23856,
      "grad_norm": 1.064430085926257,
      "learning_rate": 0.003,
      "loss": 4.0245,
      "step": 23856
    },
    {
      "epoch": 0.23857,
      "grad_norm": 1.9269232944162906,
      "learning_rate": 0.003,
      "loss": 4.0966,
      "step": 23857
    },
    {
      "epoch": 0.23858,
      "grad_norm": 1.204356903464043,
      "learning_rate": 0.003,
      "loss": 4.1017,
      "step": 23858
    },
    {
      "epoch": 0.23859,
      "grad_norm": 1.9279509446903418,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 23859
    },
    {
      "epoch": 0.2386,
      "grad_norm": 1.3213780425828707,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 23860
    },
    {
      "epoch": 0.23861,
      "grad_norm": 1.2273116828630017,
      "learning_rate": 0.003,
      "loss": 4.0232,
      "step": 23861
    },
    {
      "epoch": 0.23862,
      "grad_norm": 1.4482013010905208,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 23862
    },
    {
      "epoch": 0.23863,
      "grad_norm": 1.5255913592993389,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 23863
    },
    {
      "epoch": 0.23864,
      "grad_norm": 1.2667835745332046,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 23864
    },
    {
      "epoch": 0.23865,
      "grad_norm": 1.4638919124615997,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 23865
    },
    {
      "epoch": 0.23866,
      "grad_norm": 1.6537109980718634,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 23866
    },
    {
      "epoch": 0.23867,
      "grad_norm": 1.144264157861541,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 23867
    },
    {
      "epoch": 0.23868,
      "grad_norm": 1.45679501662993,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 23868
    },
    {
      "epoch": 0.23869,
      "grad_norm": 1.2851293505655565,
      "learning_rate": 0.003,
      "loss": 4.0759,
      "step": 23869
    },
    {
      "epoch": 0.2387,
      "grad_norm": 1.53453233126814,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 23870
    },
    {
      "epoch": 0.23871,
      "grad_norm": 1.2324819540586525,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 23871
    },
    {
      "epoch": 0.23872,
      "grad_norm": 1.4576342550602204,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 23872
    },
    {
      "epoch": 0.23873,
      "grad_norm": 1.3095921949579354,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 23873
    },
    {
      "epoch": 0.23874,
      "grad_norm": 1.6303465975553029,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 23874
    },
    {
      "epoch": 0.23875,
      "grad_norm": 1.5271946277937312,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 23875
    },
    {
      "epoch": 0.23876,
      "grad_norm": 1.3545128670544826,
      "learning_rate": 0.003,
      "loss": 4.033,
      "step": 23876
    },
    {
      "epoch": 0.23877,
      "grad_norm": 1.4882755565741435,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 23877
    },
    {
      "epoch": 0.23878,
      "grad_norm": 1.4555933551625828,
      "learning_rate": 0.003,
      "loss": 4.0261,
      "step": 23878
    },
    {
      "epoch": 0.23879,
      "grad_norm": 1.4442452418839198,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 23879
    },
    {
      "epoch": 0.2388,
      "grad_norm": 1.3838006327606363,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 23880
    },
    {
      "epoch": 0.23881,
      "grad_norm": 1.353045979434275,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 23881
    },
    {
      "epoch": 0.23882,
      "grad_norm": 1.2035868230847475,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 23882
    },
    {
      "epoch": 0.23883,
      "grad_norm": 1.3151637884857132,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 23883
    },
    {
      "epoch": 0.23884,
      "grad_norm": 1.3469918439785171,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 23884
    },
    {
      "epoch": 0.23885,
      "grad_norm": 1.6276977312150183,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 23885
    },
    {
      "epoch": 0.23886,
      "grad_norm": 0.9870363651716452,
      "learning_rate": 0.003,
      "loss": 4.0166,
      "step": 23886
    },
    {
      "epoch": 0.23887,
      "grad_norm": 1.687312232772964,
      "learning_rate": 0.003,
      "loss": 4.0547,
      "step": 23887
    },
    {
      "epoch": 0.23888,
      "grad_norm": 1.2719392440031376,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 23888
    },
    {
      "epoch": 0.23889,
      "grad_norm": 1.7781724738599882,
      "learning_rate": 0.003,
      "loss": 4.0274,
      "step": 23889
    },
    {
      "epoch": 0.2389,
      "grad_norm": 1.429271397424128,
      "learning_rate": 0.003,
      "loss": 4.0741,
      "step": 23890
    },
    {
      "epoch": 0.23891,
      "grad_norm": 1.2064605716494519,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 23891
    },
    {
      "epoch": 0.23892,
      "grad_norm": 1.574314808234865,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 23892
    },
    {
      "epoch": 0.23893,
      "grad_norm": 1.2306207363246748,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 23893
    },
    {
      "epoch": 0.23894,
      "grad_norm": 1.4800015072971937,
      "learning_rate": 0.003,
      "loss": 4.0868,
      "step": 23894
    },
    {
      "epoch": 0.23895,
      "grad_norm": 1.1766164342089749,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 23895
    },
    {
      "epoch": 0.23896,
      "grad_norm": 2.038374218099904,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 23896
    },
    {
      "epoch": 0.23897,
      "grad_norm": 1.209997615477808,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 23897
    },
    {
      "epoch": 0.23898,
      "grad_norm": 1.527994793198077,
      "learning_rate": 0.003,
      "loss": 4.073,
      "step": 23898
    },
    {
      "epoch": 0.23899,
      "grad_norm": 1.5598152877419624,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 23899
    },
    {
      "epoch": 0.239,
      "grad_norm": 1.0747574357175296,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 23900
    },
    {
      "epoch": 0.23901,
      "grad_norm": 1.5540337504735566,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 23901
    },
    {
      "epoch": 0.23902,
      "grad_norm": 1.2864827706828081,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 23902
    },
    {
      "epoch": 0.23903,
      "grad_norm": 1.5679024715751169,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 23903
    },
    {
      "epoch": 0.23904,
      "grad_norm": 1.2940647261642273,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 23904
    },
    {
      "epoch": 0.23905,
      "grad_norm": 1.4599982353565195,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 23905
    },
    {
      "epoch": 0.23906,
      "grad_norm": 1.1975209818326515,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 23906
    },
    {
      "epoch": 0.23907,
      "grad_norm": 1.4749190018515188,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 23907
    },
    {
      "epoch": 0.23908,
      "grad_norm": 1.3128009532242484,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 23908
    },
    {
      "epoch": 0.23909,
      "grad_norm": 1.3401151975136683,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 23909
    },
    {
      "epoch": 0.2391,
      "grad_norm": 1.759963670211569,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 23910
    },
    {
      "epoch": 0.23911,
      "grad_norm": 1.1500155017694818,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 23911
    },
    {
      "epoch": 0.23912,
      "grad_norm": 1.3889662282538109,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 23912
    },
    {
      "epoch": 0.23913,
      "grad_norm": 1.5685165044324842,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 23913
    },
    {
      "epoch": 0.23914,
      "grad_norm": 1.35687960144389,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 23914
    },
    {
      "epoch": 0.23915,
      "grad_norm": 1.5036410374791425,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 23915
    },
    {
      "epoch": 0.23916,
      "grad_norm": 1.3774942534493801,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 23916
    },
    {
      "epoch": 0.23917,
      "grad_norm": 1.3622200146525703,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 23917
    },
    {
      "epoch": 0.23918,
      "grad_norm": 1.4785813270006374,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 23918
    },
    {
      "epoch": 0.23919,
      "grad_norm": 1.4166758330222942,
      "learning_rate": 0.003,
      "loss": 4.0951,
      "step": 23919
    },
    {
      "epoch": 0.2392,
      "grad_norm": 1.137203057813097,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 23920
    },
    {
      "epoch": 0.23921,
      "grad_norm": 1.4881849967255394,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 23921
    },
    {
      "epoch": 0.23922,
      "grad_norm": 1.2638340477372327,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 23922
    },
    {
      "epoch": 0.23923,
      "grad_norm": 1.232610847895491,
      "learning_rate": 0.003,
      "loss": 4.0218,
      "step": 23923
    },
    {
      "epoch": 0.23924,
      "grad_norm": 1.428442090677148,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 23924
    },
    {
      "epoch": 0.23925,
      "grad_norm": 1.229763622838009,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 23925
    },
    {
      "epoch": 0.23926,
      "grad_norm": 1.5704113702178863,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 23926
    },
    {
      "epoch": 0.23927,
      "grad_norm": 1.3759363736172487,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 23927
    },
    {
      "epoch": 0.23928,
      "grad_norm": 1.479622646250979,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 23928
    },
    {
      "epoch": 0.23929,
      "grad_norm": 1.5193335088531157,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 23929
    },
    {
      "epoch": 0.2393,
      "grad_norm": 1.2385853878345658,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 23930
    },
    {
      "epoch": 0.23931,
      "grad_norm": 1.3301591585359105,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 23931
    },
    {
      "epoch": 0.23932,
      "grad_norm": 1.4967629933181466,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 23932
    },
    {
      "epoch": 0.23933,
      "grad_norm": 1.239061230910692,
      "learning_rate": 0.003,
      "loss": 4.033,
      "step": 23933
    },
    {
      "epoch": 0.23934,
      "grad_norm": 1.5911761601168937,
      "learning_rate": 0.003,
      "loss": 4.028,
      "step": 23934
    },
    {
      "epoch": 0.23935,
      "grad_norm": 1.426816165529343,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 23935
    },
    {
      "epoch": 0.23936,
      "grad_norm": 1.2027913915117059,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 23936
    },
    {
      "epoch": 0.23937,
      "grad_norm": 1.4237114034771414,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 23937
    },
    {
      "epoch": 0.23938,
      "grad_norm": 1.3361100789464557,
      "learning_rate": 0.003,
      "loss": 4.0735,
      "step": 23938
    },
    {
      "epoch": 0.23939,
      "grad_norm": 1.7015890090381045,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 23939
    },
    {
      "epoch": 0.2394,
      "grad_norm": 1.1599771038186024,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 23940
    },
    {
      "epoch": 0.23941,
      "grad_norm": 1.389903527458184,
      "learning_rate": 0.003,
      "loss": 4.0127,
      "step": 23941
    },
    {
      "epoch": 0.23942,
      "grad_norm": 1.2259445035966947,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 23942
    },
    {
      "epoch": 0.23943,
      "grad_norm": 1.3392741013180798,
      "learning_rate": 0.003,
      "loss": 4.0213,
      "step": 23943
    },
    {
      "epoch": 0.23944,
      "grad_norm": 1.2351941856468855,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 23944
    },
    {
      "epoch": 0.23945,
      "grad_norm": 1.5485213693209419,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 23945
    },
    {
      "epoch": 0.23946,
      "grad_norm": 1.2991648162126856,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 23946
    },
    {
      "epoch": 0.23947,
      "grad_norm": 1.4474743768636082,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 23947
    },
    {
      "epoch": 0.23948,
      "grad_norm": 1.3183307411077938,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 23948
    },
    {
      "epoch": 0.23949,
      "grad_norm": 1.3685403069930697,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 23949
    },
    {
      "epoch": 0.2395,
      "grad_norm": 1.3755717188009384,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 23950
    },
    {
      "epoch": 0.23951,
      "grad_norm": 1.6721210950286103,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 23951
    },
    {
      "epoch": 0.23952,
      "grad_norm": 1.3083439455606527,
      "learning_rate": 0.003,
      "loss": 4.0288,
      "step": 23952
    },
    {
      "epoch": 0.23953,
      "grad_norm": 1.4422213495504526,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 23953
    },
    {
      "epoch": 0.23954,
      "grad_norm": 1.387026715996028,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 23954
    },
    {
      "epoch": 0.23955,
      "grad_norm": 1.440346263084508,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 23955
    },
    {
      "epoch": 0.23956,
      "grad_norm": 1.590818189868967,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 23956
    },
    {
      "epoch": 0.23957,
      "grad_norm": 1.3889756185660593,
      "learning_rate": 0.003,
      "loss": 4.0213,
      "step": 23957
    },
    {
      "epoch": 0.23958,
      "grad_norm": 1.0926375524677692,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 23958
    },
    {
      "epoch": 0.23959,
      "grad_norm": 1.558191918429594,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 23959
    },
    {
      "epoch": 0.2396,
      "grad_norm": 1.5299441214630756,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 23960
    },
    {
      "epoch": 0.23961,
      "grad_norm": 1.239989674867137,
      "learning_rate": 0.003,
      "loss": 4.0188,
      "step": 23961
    },
    {
      "epoch": 0.23962,
      "grad_norm": 1.6305351077706236,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 23962
    },
    {
      "epoch": 0.23963,
      "grad_norm": 1.1062324901786194,
      "learning_rate": 0.003,
      "loss": 4.0092,
      "step": 23963
    },
    {
      "epoch": 0.23964,
      "grad_norm": 1.4694025967846787,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 23964
    },
    {
      "epoch": 0.23965,
      "grad_norm": 1.333277997412035,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 23965
    },
    {
      "epoch": 0.23966,
      "grad_norm": 1.5332268470613066,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 23966
    },
    {
      "epoch": 0.23967,
      "grad_norm": 1.2938365569303085,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 23967
    },
    {
      "epoch": 0.23968,
      "grad_norm": 1.1061559486835393,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 23968
    },
    {
      "epoch": 0.23969,
      "grad_norm": 1.4417066760279533,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 23969
    },
    {
      "epoch": 0.2397,
      "grad_norm": 1.0927795175523582,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 23970
    },
    {
      "epoch": 0.23971,
      "grad_norm": 1.5746775711428718,
      "learning_rate": 0.003,
      "loss": 4.0821,
      "step": 23971
    },
    {
      "epoch": 0.23972,
      "grad_norm": 1.4299606579525728,
      "learning_rate": 0.003,
      "loss": 4.0418,
      "step": 23972
    },
    {
      "epoch": 0.23973,
      "grad_norm": 1.6227309934797456,
      "learning_rate": 0.003,
      "loss": 4.0751,
      "step": 23973
    },
    {
      "epoch": 0.23974,
      "grad_norm": 1.517868044457317,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 23974
    },
    {
      "epoch": 0.23975,
      "grad_norm": 1.5062625468166382,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 23975
    },
    {
      "epoch": 0.23976,
      "grad_norm": 1.6528882048371647,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 23976
    },
    {
      "epoch": 0.23977,
      "grad_norm": 1.2776247453207008,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 23977
    },
    {
      "epoch": 0.23978,
      "grad_norm": 1.4202344502240514,
      "learning_rate": 0.003,
      "loss": 4.0271,
      "step": 23978
    },
    {
      "epoch": 0.23979,
      "grad_norm": 1.168376450636879,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 23979
    },
    {
      "epoch": 0.2398,
      "grad_norm": 1.3600901694620695,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 23980
    },
    {
      "epoch": 0.23981,
      "grad_norm": 1.3406922849924334,
      "learning_rate": 0.003,
      "loss": 4.0842,
      "step": 23981
    },
    {
      "epoch": 0.23982,
      "grad_norm": 1.4085494383855803,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 23982
    },
    {
      "epoch": 0.23983,
      "grad_norm": 1.1953433239102236,
      "learning_rate": 0.003,
      "loss": 4.0359,
      "step": 23983
    },
    {
      "epoch": 0.23984,
      "grad_norm": 1.7073756513123728,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 23984
    },
    {
      "epoch": 0.23985,
      "grad_norm": 1.1673486075400417,
      "learning_rate": 0.003,
      "loss": 4.0241,
      "step": 23985
    },
    {
      "epoch": 0.23986,
      "grad_norm": 1.6603190878239664,
      "learning_rate": 0.003,
      "loss": 4.0351,
      "step": 23986
    },
    {
      "epoch": 0.23987,
      "grad_norm": 1.2459795657565698,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 23987
    },
    {
      "epoch": 0.23988,
      "grad_norm": 1.2623862909418935,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 23988
    },
    {
      "epoch": 0.23989,
      "grad_norm": 1.4939180244219257,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 23989
    },
    {
      "epoch": 0.2399,
      "grad_norm": 1.4221846092754766,
      "learning_rate": 0.003,
      "loss": 4.0826,
      "step": 23990
    },
    {
      "epoch": 0.23991,
      "grad_norm": 1.3174146944497975,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 23991
    },
    {
      "epoch": 0.23992,
      "grad_norm": 1.231965197781564,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 23992
    },
    {
      "epoch": 0.23993,
      "grad_norm": 1.4611481267350466,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 23993
    },
    {
      "epoch": 0.23994,
      "grad_norm": 1.4764328218480265,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 23994
    },
    {
      "epoch": 0.23995,
      "grad_norm": 1.2652128922069787,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 23995
    },
    {
      "epoch": 0.23996,
      "grad_norm": 1.6664184416192886,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 23996
    },
    {
      "epoch": 0.23997,
      "grad_norm": 1.2647859807629176,
      "learning_rate": 0.003,
      "loss": 4.0721,
      "step": 23997
    },
    {
      "epoch": 0.23998,
      "grad_norm": 1.6670294215699575,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 23998
    },
    {
      "epoch": 0.23999,
      "grad_norm": 1.3493151760352913,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 23999
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5097572940765318,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 24000
    },
    {
      "epoch": 0.24001,
      "grad_norm": 1.3208655695161489,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 24001
    },
    {
      "epoch": 0.24002,
      "grad_norm": 1.369927948056816,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 24002
    },
    {
      "epoch": 0.24003,
      "grad_norm": 1.5871583498951725,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 24003
    },
    {
      "epoch": 0.24004,
      "grad_norm": 1.6102392093801234,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 24004
    },
    {
      "epoch": 0.24005,
      "grad_norm": 1.34507653505968,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 24005
    },
    {
      "epoch": 0.24006,
      "grad_norm": 1.5630318864677974,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 24006
    },
    {
      "epoch": 0.24007,
      "grad_norm": 1.0674417926438253,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 24007
    },
    {
      "epoch": 0.24008,
      "grad_norm": 1.6015581748821677,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 24008
    },
    {
      "epoch": 0.24009,
      "grad_norm": 1.2303992524679244,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 24009
    },
    {
      "epoch": 0.2401,
      "grad_norm": 1.4950705278510021,
      "learning_rate": 0.003,
      "loss": 4.0727,
      "step": 24010
    },
    {
      "epoch": 0.24011,
      "grad_norm": 1.133061847566615,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 24011
    },
    {
      "epoch": 0.24012,
      "grad_norm": 1.6713085072539149,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 24012
    },
    {
      "epoch": 0.24013,
      "grad_norm": 1.6075171742907977,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 24013
    },
    {
      "epoch": 0.24014,
      "grad_norm": 1.2231569309114756,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 24014
    },
    {
      "epoch": 0.24015,
      "grad_norm": 1.4613631005558372,
      "learning_rate": 0.003,
      "loss": 4.0337,
      "step": 24015
    },
    {
      "epoch": 0.24016,
      "grad_norm": 1.574652594837066,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24016
    },
    {
      "epoch": 0.24017,
      "grad_norm": 1.4814976148306211,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 24017
    },
    {
      "epoch": 0.24018,
      "grad_norm": 1.2243178435485185,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24018
    },
    {
      "epoch": 0.24019,
      "grad_norm": 1.5514268268011124,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 24019
    },
    {
      "epoch": 0.2402,
      "grad_norm": 1.2261703220940252,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 24020
    },
    {
      "epoch": 0.24021,
      "grad_norm": 1.253688213915527,
      "learning_rate": 0.003,
      "loss": 4.0322,
      "step": 24021
    },
    {
      "epoch": 0.24022,
      "grad_norm": 1.3775097145449553,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 24022
    },
    {
      "epoch": 0.24023,
      "grad_norm": 1.321556310405547,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 24023
    },
    {
      "epoch": 0.24024,
      "grad_norm": 1.607243932832859,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 24024
    },
    {
      "epoch": 0.24025,
      "grad_norm": 1.341042583470396,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 24025
    },
    {
      "epoch": 0.24026,
      "grad_norm": 1.4578221252401997,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 24026
    },
    {
      "epoch": 0.24027,
      "grad_norm": 1.3098491479749361,
      "learning_rate": 0.003,
      "loss": 4.0855,
      "step": 24027
    },
    {
      "epoch": 0.24028,
      "grad_norm": 1.4545823917202343,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 24028
    },
    {
      "epoch": 0.24029,
      "grad_norm": 1.2893726150280804,
      "learning_rate": 0.003,
      "loss": 4.0301,
      "step": 24029
    },
    {
      "epoch": 0.2403,
      "grad_norm": 1.3363865057682953,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 24030
    },
    {
      "epoch": 0.24031,
      "grad_norm": 1.6045281258319553,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 24031
    },
    {
      "epoch": 0.24032,
      "grad_norm": 1.3380129367690536,
      "learning_rate": 0.003,
      "loss": 4.0059,
      "step": 24032
    },
    {
      "epoch": 0.24033,
      "grad_norm": 1.5289132545255149,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24033
    },
    {
      "epoch": 0.24034,
      "grad_norm": 1.2731089616102533,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 24034
    },
    {
      "epoch": 0.24035,
      "grad_norm": 1.3720113652692365,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24035
    },
    {
      "epoch": 0.24036,
      "grad_norm": 1.1645157447080348,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 24036
    },
    {
      "epoch": 0.24037,
      "grad_norm": 1.4602016963125448,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 24037
    },
    {
      "epoch": 0.24038,
      "grad_norm": 1.17121783826053,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 24038
    },
    {
      "epoch": 0.24039,
      "grad_norm": 1.565316292029105,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 24039
    },
    {
      "epoch": 0.2404,
      "grad_norm": 1.4431977852581739,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 24040
    },
    {
      "epoch": 0.24041,
      "grad_norm": 1.2971511719510866,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 24041
    },
    {
      "epoch": 0.24042,
      "grad_norm": 1.5378137044571127,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 24042
    },
    {
      "epoch": 0.24043,
      "grad_norm": 1.538340658409097,
      "learning_rate": 0.003,
      "loss": 4.0851,
      "step": 24043
    },
    {
      "epoch": 0.24044,
      "grad_norm": 1.7107100056471156,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 24044
    },
    {
      "epoch": 0.24045,
      "grad_norm": 1.1631157122618356,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 24045
    },
    {
      "epoch": 0.24046,
      "grad_norm": 1.3447005065554585,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 24046
    },
    {
      "epoch": 0.24047,
      "grad_norm": 1.6684331329788673,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 24047
    },
    {
      "epoch": 0.24048,
      "grad_norm": 1.199564465199791,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 24048
    },
    {
      "epoch": 0.24049,
      "grad_norm": 1.4851422530365161,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 24049
    },
    {
      "epoch": 0.2405,
      "grad_norm": 1.562186527883096,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 24050
    },
    {
      "epoch": 0.24051,
      "grad_norm": 1.2951359561812292,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 24051
    },
    {
      "epoch": 0.24052,
      "grad_norm": 1.4070257444670804,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 24052
    },
    {
      "epoch": 0.24053,
      "grad_norm": 1.2996402174222985,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 24053
    },
    {
      "epoch": 0.24054,
      "grad_norm": 1.4405012769108854,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 24054
    },
    {
      "epoch": 0.24055,
      "grad_norm": 1.3124257817813214,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 24055
    },
    {
      "epoch": 0.24056,
      "grad_norm": 1.5220651917606063,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 24056
    },
    {
      "epoch": 0.24057,
      "grad_norm": 1.8254056278058537,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 24057
    },
    {
      "epoch": 0.24058,
      "grad_norm": 1.4577066872880862,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 24058
    },
    {
      "epoch": 0.24059,
      "grad_norm": 1.3197777921445313,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 24059
    },
    {
      "epoch": 0.2406,
      "grad_norm": 1.3289676010536744,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 24060
    },
    {
      "epoch": 0.24061,
      "grad_norm": 1.363120612686246,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 24061
    },
    {
      "epoch": 0.24062,
      "grad_norm": 1.251584936235425,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 24062
    },
    {
      "epoch": 0.24063,
      "grad_norm": 1.629686642487354,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 24063
    },
    {
      "epoch": 0.24064,
      "grad_norm": 1.1974476306693564,
      "learning_rate": 0.003,
      "loss": 4.0155,
      "step": 24064
    },
    {
      "epoch": 0.24065,
      "grad_norm": 1.5052029223205199,
      "learning_rate": 0.003,
      "loss": 4.0958,
      "step": 24065
    },
    {
      "epoch": 0.24066,
      "grad_norm": 1.3714932148527221,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 24066
    },
    {
      "epoch": 0.24067,
      "grad_norm": 1.5123863749372173,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 24067
    },
    {
      "epoch": 0.24068,
      "grad_norm": 1.427258807086581,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 24068
    },
    {
      "epoch": 0.24069,
      "grad_norm": 1.3667381992617014,
      "learning_rate": 0.003,
      "loss": 4.0814,
      "step": 24069
    },
    {
      "epoch": 0.2407,
      "grad_norm": 1.336350845274785,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24070
    },
    {
      "epoch": 0.24071,
      "grad_norm": 1.3208790369639019,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 24071
    },
    {
      "epoch": 0.24072,
      "grad_norm": 1.1108388793306836,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 24072
    },
    {
      "epoch": 0.24073,
      "grad_norm": 1.4814179727466374,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 24073
    },
    {
      "epoch": 0.24074,
      "grad_norm": 1.305090190150312,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 24074
    },
    {
      "epoch": 0.24075,
      "grad_norm": 1.553580445107228,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 24075
    },
    {
      "epoch": 0.24076,
      "grad_norm": 1.3514565595428543,
      "learning_rate": 0.003,
      "loss": 4.1108,
      "step": 24076
    },
    {
      "epoch": 0.24077,
      "grad_norm": 1.8428150937722338,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 24077
    },
    {
      "epoch": 0.24078,
      "grad_norm": 1.15136528152379,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 24078
    },
    {
      "epoch": 0.24079,
      "grad_norm": 1.2394692787565729,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 24079
    },
    {
      "epoch": 0.2408,
      "grad_norm": 1.5167698228484925,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 24080
    },
    {
      "epoch": 0.24081,
      "grad_norm": 1.4595376744030748,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 24081
    },
    {
      "epoch": 0.24082,
      "grad_norm": 1.2838236761478816,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 24082
    },
    {
      "epoch": 0.24083,
      "grad_norm": 1.4465599909295825,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 24083
    },
    {
      "epoch": 0.24084,
      "grad_norm": 1.4072058842213704,
      "learning_rate": 0.003,
      "loss": 4.095,
      "step": 24084
    },
    {
      "epoch": 0.24085,
      "grad_norm": 1.2014843558169472,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 24085
    },
    {
      "epoch": 0.24086,
      "grad_norm": 1.7230270251193622,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 24086
    },
    {
      "epoch": 0.24087,
      "grad_norm": 1.1460395686823723,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 24087
    },
    {
      "epoch": 0.24088,
      "grad_norm": 1.9497605446726536,
      "learning_rate": 0.003,
      "loss": 4.0695,
      "step": 24088
    },
    {
      "epoch": 0.24089,
      "grad_norm": 1.0301781552385405,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 24089
    },
    {
      "epoch": 0.2409,
      "grad_norm": 1.6469408464810509,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 24090
    },
    {
      "epoch": 0.24091,
      "grad_norm": 1.273926825135361,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 24091
    },
    {
      "epoch": 0.24092,
      "grad_norm": 1.4330414027792062,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 24092
    },
    {
      "epoch": 0.24093,
      "grad_norm": 1.4755580105387949,
      "learning_rate": 0.003,
      "loss": 4.0268,
      "step": 24093
    },
    {
      "epoch": 0.24094,
      "grad_norm": 1.3552857859565144,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 24094
    },
    {
      "epoch": 0.24095,
      "grad_norm": 1.2449461527733876,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 24095
    },
    {
      "epoch": 0.24096,
      "grad_norm": 1.3337860286244896,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 24096
    },
    {
      "epoch": 0.24097,
      "grad_norm": 1.3142564338112297,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 24097
    },
    {
      "epoch": 0.24098,
      "grad_norm": 1.649099599663682,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 24098
    },
    {
      "epoch": 0.24099,
      "grad_norm": 1.3280762681798943,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 24099
    },
    {
      "epoch": 0.241,
      "grad_norm": 1.654781855622568,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 24100
    },
    {
      "epoch": 0.24101,
      "grad_norm": 1.2119325482687349,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 24101
    },
    {
      "epoch": 0.24102,
      "grad_norm": 1.5160643133768692,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 24102
    },
    {
      "epoch": 0.24103,
      "grad_norm": 1.3073909264012171,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 24103
    },
    {
      "epoch": 0.24104,
      "grad_norm": 1.2471836493913577,
      "learning_rate": 0.003,
      "loss": 4.0093,
      "step": 24104
    },
    {
      "epoch": 0.24105,
      "grad_norm": 1.2986683022158774,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 24105
    },
    {
      "epoch": 0.24106,
      "grad_norm": 1.4753876400658665,
      "learning_rate": 0.003,
      "loss": 4.0837,
      "step": 24106
    },
    {
      "epoch": 0.24107,
      "grad_norm": 1.2841130899235613,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 24107
    },
    {
      "epoch": 0.24108,
      "grad_norm": 1.298390054165471,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 24108
    },
    {
      "epoch": 0.24109,
      "grad_norm": 1.458470898975051,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 24109
    },
    {
      "epoch": 0.2411,
      "grad_norm": 1.0671846941033294,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 24110
    },
    {
      "epoch": 0.24111,
      "grad_norm": 1.5408516818685556,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 24111
    },
    {
      "epoch": 0.24112,
      "grad_norm": 0.9886222221704883,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 24112
    },
    {
      "epoch": 0.24113,
      "grad_norm": 1.6609339760141486,
      "learning_rate": 0.003,
      "loss": 4.0896,
      "step": 24113
    },
    {
      "epoch": 0.24114,
      "grad_norm": 1.1177725002355285,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 24114
    },
    {
      "epoch": 0.24115,
      "grad_norm": 1.6618753087966223,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 24115
    },
    {
      "epoch": 0.24116,
      "grad_norm": 1.593161191206839,
      "learning_rate": 0.003,
      "loss": 4.0614,
      "step": 24116
    },
    {
      "epoch": 0.24117,
      "grad_norm": 1.2873678293225177,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 24117
    },
    {
      "epoch": 0.24118,
      "grad_norm": 1.7329763851569069,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 24118
    },
    {
      "epoch": 0.24119,
      "grad_norm": 1.2714595493373622,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 24119
    },
    {
      "epoch": 0.2412,
      "grad_norm": 1.5502666064407715,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 24120
    },
    {
      "epoch": 0.24121,
      "grad_norm": 1.2214682016660192,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 24121
    },
    {
      "epoch": 0.24122,
      "grad_norm": 1.4618454169690436,
      "learning_rate": 0.003,
      "loss": 4.0175,
      "step": 24122
    },
    {
      "epoch": 0.24123,
      "grad_norm": 1.2804479803066813,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 24123
    },
    {
      "epoch": 0.24124,
      "grad_norm": 1.5313227284482278,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 24124
    },
    {
      "epoch": 0.24125,
      "grad_norm": 1.5613304791889409,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 24125
    },
    {
      "epoch": 0.24126,
      "grad_norm": 1.0656993644403323,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 24126
    },
    {
      "epoch": 0.24127,
      "grad_norm": 1.539046537649022,
      "learning_rate": 0.003,
      "loss": 4.0286,
      "step": 24127
    },
    {
      "epoch": 0.24128,
      "grad_norm": 1.1492346054391356,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 24128
    },
    {
      "epoch": 0.24129,
      "grad_norm": 1.8960752256071003,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 24129
    },
    {
      "epoch": 0.2413,
      "grad_norm": 1.2089011267547258,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 24130
    },
    {
      "epoch": 0.24131,
      "grad_norm": 1.3396534965253424,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 24131
    },
    {
      "epoch": 0.24132,
      "grad_norm": 1.4842610594009207,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 24132
    },
    {
      "epoch": 0.24133,
      "grad_norm": 1.3976309450236981,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 24133
    },
    {
      "epoch": 0.24134,
      "grad_norm": 1.5902169018715517,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 24134
    },
    {
      "epoch": 0.24135,
      "grad_norm": 1.4054456342875623,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 24135
    },
    {
      "epoch": 0.24136,
      "grad_norm": 1.1633319820102586,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 24136
    },
    {
      "epoch": 0.24137,
      "grad_norm": 1.6028449933793758,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 24137
    },
    {
      "epoch": 0.24138,
      "grad_norm": 1.7416615377617044,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 24138
    },
    {
      "epoch": 0.24139,
      "grad_norm": 1.1040978857034935,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 24139
    },
    {
      "epoch": 0.2414,
      "grad_norm": 1.4326342747758016,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 24140
    },
    {
      "epoch": 0.24141,
      "grad_norm": 1.3173080638906012,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 24141
    },
    {
      "epoch": 0.24142,
      "grad_norm": 1.199641285163167,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 24142
    },
    {
      "epoch": 0.24143,
      "grad_norm": 1.3731662312621917,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 24143
    },
    {
      "epoch": 0.24144,
      "grad_norm": 1.4187046283538451,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 24144
    },
    {
      "epoch": 0.24145,
      "grad_norm": 1.2963124746011216,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 24145
    },
    {
      "epoch": 0.24146,
      "grad_norm": 1.2583703473186578,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 24146
    },
    {
      "epoch": 0.24147,
      "grad_norm": 1.378754668740218,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 24147
    },
    {
      "epoch": 0.24148,
      "grad_norm": 1.3044351423436642,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 24148
    },
    {
      "epoch": 0.24149,
      "grad_norm": 1.3123845621937755,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 24149
    },
    {
      "epoch": 0.2415,
      "grad_norm": 1.3925292135149083,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 24150
    },
    {
      "epoch": 0.24151,
      "grad_norm": 1.229044455895039,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 24151
    },
    {
      "epoch": 0.24152,
      "grad_norm": 1.5548229529941022,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 24152
    },
    {
      "epoch": 0.24153,
      "grad_norm": 1.226909830678718,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 24153
    },
    {
      "epoch": 0.24154,
      "grad_norm": 1.7686656047696907,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 24154
    },
    {
      "epoch": 0.24155,
      "grad_norm": 1.2456571851748708,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 24155
    },
    {
      "epoch": 0.24156,
      "grad_norm": 1.3966131277600515,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 24156
    },
    {
      "epoch": 0.24157,
      "grad_norm": 1.2597456621338394,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 24157
    },
    {
      "epoch": 0.24158,
      "grad_norm": 1.5881093924879088,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 24158
    },
    {
      "epoch": 0.24159,
      "grad_norm": 1.3635991818037387,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 24159
    },
    {
      "epoch": 0.2416,
      "grad_norm": 1.9670775413511101,
      "learning_rate": 0.003,
      "loss": 4.0574,
      "step": 24160
    },
    {
      "epoch": 0.24161,
      "grad_norm": 1.3054958691614253,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 24161
    },
    {
      "epoch": 0.24162,
      "grad_norm": 1.0894464236970764,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 24162
    },
    {
      "epoch": 0.24163,
      "grad_norm": 1.6673970714260016,
      "learning_rate": 0.003,
      "loss": 4.0717,
      "step": 24163
    },
    {
      "epoch": 0.24164,
      "grad_norm": 1.3907948894269935,
      "learning_rate": 0.003,
      "loss": 4.022,
      "step": 24164
    },
    {
      "epoch": 0.24165,
      "grad_norm": 1.303472901631941,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 24165
    },
    {
      "epoch": 0.24166,
      "grad_norm": 1.5688314266097538,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 24166
    },
    {
      "epoch": 0.24167,
      "grad_norm": 1.4828007101737333,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 24167
    },
    {
      "epoch": 0.24168,
      "grad_norm": 1.153020032594496,
      "learning_rate": 0.003,
      "loss": 4.0921,
      "step": 24168
    },
    {
      "epoch": 0.24169,
      "grad_norm": 1.6807951751398211,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 24169
    },
    {
      "epoch": 0.2417,
      "grad_norm": 1.1395687023333403,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 24170
    },
    {
      "epoch": 0.24171,
      "grad_norm": 1.710114409140257,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 24171
    },
    {
      "epoch": 0.24172,
      "grad_norm": 1.2453411558166612,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 24172
    },
    {
      "epoch": 0.24173,
      "grad_norm": 1.4640730242339421,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 24173
    },
    {
      "epoch": 0.24174,
      "grad_norm": 1.2053949287522918,
      "learning_rate": 0.003,
      "loss": 4.021,
      "step": 24174
    },
    {
      "epoch": 0.24175,
      "grad_norm": 1.674565837253066,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 24175
    },
    {
      "epoch": 0.24176,
      "grad_norm": 1.3836774832370602,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 24176
    },
    {
      "epoch": 0.24177,
      "grad_norm": 1.3490456470272745,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 24177
    },
    {
      "epoch": 0.24178,
      "grad_norm": 1.369329690781937,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 24178
    },
    {
      "epoch": 0.24179,
      "grad_norm": 1.1655394574473443,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 24179
    },
    {
      "epoch": 0.2418,
      "grad_norm": 1.637493211341957,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 24180
    },
    {
      "epoch": 0.24181,
      "grad_norm": 1.1968948824780248,
      "learning_rate": 0.003,
      "loss": 4.0264,
      "step": 24181
    },
    {
      "epoch": 0.24182,
      "grad_norm": 1.5602879931997047,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 24182
    },
    {
      "epoch": 0.24183,
      "grad_norm": 1.0345435895912405,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 24183
    },
    {
      "epoch": 0.24184,
      "grad_norm": 1.6200732593921277,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 24184
    },
    {
      "epoch": 0.24185,
      "grad_norm": 1.2023060083441939,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 24185
    },
    {
      "epoch": 0.24186,
      "grad_norm": 1.4663230827324905,
      "learning_rate": 0.003,
      "loss": 4.08,
      "step": 24186
    },
    {
      "epoch": 0.24187,
      "grad_norm": 1.5532612101258794,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 24187
    },
    {
      "epoch": 0.24188,
      "grad_norm": 1.4739841593131926,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 24188
    },
    {
      "epoch": 0.24189,
      "grad_norm": 1.3376844071359788,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24189
    },
    {
      "epoch": 0.2419,
      "grad_norm": 1.3448150107076327,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 24190
    },
    {
      "epoch": 0.24191,
      "grad_norm": 1.3102973802513778,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 24191
    },
    {
      "epoch": 0.24192,
      "grad_norm": 1.2912464168049576,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 24192
    },
    {
      "epoch": 0.24193,
      "grad_norm": 1.3284029491857814,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 24193
    },
    {
      "epoch": 0.24194,
      "grad_norm": 1.419056951370366,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 24194
    },
    {
      "epoch": 0.24195,
      "grad_norm": 1.4665238512294811,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 24195
    },
    {
      "epoch": 0.24196,
      "grad_norm": 1.4440262793436895,
      "learning_rate": 0.003,
      "loss": 4.0965,
      "step": 24196
    },
    {
      "epoch": 0.24197,
      "grad_norm": 1.3368358354618906,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 24197
    },
    {
      "epoch": 0.24198,
      "grad_norm": 1.294532460452846,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 24198
    },
    {
      "epoch": 0.24199,
      "grad_norm": 1.155168883050232,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 24199
    },
    {
      "epoch": 0.242,
      "grad_norm": 1.5366001992696707,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 24200
    },
    {
      "epoch": 0.24201,
      "grad_norm": 1.186991280202697,
      "learning_rate": 0.003,
      "loss": 4.0718,
      "step": 24201
    },
    {
      "epoch": 0.24202,
      "grad_norm": 1.5046129820830885,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 24202
    },
    {
      "epoch": 0.24203,
      "grad_norm": 1.5490891294740057,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 24203
    },
    {
      "epoch": 0.24204,
      "grad_norm": 1.1992394383361793,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 24204
    },
    {
      "epoch": 0.24205,
      "grad_norm": 1.540735703971023,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 24205
    },
    {
      "epoch": 0.24206,
      "grad_norm": 1.1360236479870494,
      "learning_rate": 0.003,
      "loss": 4.0191,
      "step": 24206
    },
    {
      "epoch": 0.24207,
      "grad_norm": 1.5499068140833971,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 24207
    },
    {
      "epoch": 0.24208,
      "grad_norm": 1.3230316006938978,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 24208
    },
    {
      "epoch": 0.24209,
      "grad_norm": 1.2977193847469652,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 24209
    },
    {
      "epoch": 0.2421,
      "grad_norm": 1.8675181796009408,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 24210
    },
    {
      "epoch": 0.24211,
      "grad_norm": 1.2728049259065852,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 24211
    },
    {
      "epoch": 0.24212,
      "grad_norm": 1.3735599001217964,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 24212
    },
    {
      "epoch": 0.24213,
      "grad_norm": 1.361356516951775,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 24213
    },
    {
      "epoch": 0.24214,
      "grad_norm": 1.4157889372005399,
      "learning_rate": 0.003,
      "loss": 4.0712,
      "step": 24214
    },
    {
      "epoch": 0.24215,
      "grad_norm": 1.2223427196954062,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 24215
    },
    {
      "epoch": 0.24216,
      "grad_norm": 1.441977318289308,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 24216
    },
    {
      "epoch": 0.24217,
      "grad_norm": 1.2370739232957682,
      "learning_rate": 0.003,
      "loss": 4.0562,
      "step": 24217
    },
    {
      "epoch": 0.24218,
      "grad_norm": 1.2690633201751553,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 24218
    },
    {
      "epoch": 0.24219,
      "grad_norm": 1.6734778404795343,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 24219
    },
    {
      "epoch": 0.2422,
      "grad_norm": 1.2219899883140208,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 24220
    },
    {
      "epoch": 0.24221,
      "grad_norm": 1.4097330247284796,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 24221
    },
    {
      "epoch": 0.24222,
      "grad_norm": 1.4938325003799242,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 24222
    },
    {
      "epoch": 0.24223,
      "grad_norm": 1.3487625356047497,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 24223
    },
    {
      "epoch": 0.24224,
      "grad_norm": 1.4262319837233057,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 24224
    },
    {
      "epoch": 0.24225,
      "grad_norm": 1.5446225618111487,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 24225
    },
    {
      "epoch": 0.24226,
      "grad_norm": 1.3232341728510078,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 24226
    },
    {
      "epoch": 0.24227,
      "grad_norm": 1.683012094414617,
      "learning_rate": 0.003,
      "loss": 4.0829,
      "step": 24227
    },
    {
      "epoch": 0.24228,
      "grad_norm": 1.295498521942109,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 24228
    },
    {
      "epoch": 0.24229,
      "grad_norm": 1.3345779702149414,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 24229
    },
    {
      "epoch": 0.2423,
      "grad_norm": 1.220376470555496,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24230
    },
    {
      "epoch": 0.24231,
      "grad_norm": 1.4614746663436433,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 24231
    },
    {
      "epoch": 0.24232,
      "grad_norm": 1.105429062896927,
      "learning_rate": 0.003,
      "loss": 4.0197,
      "step": 24232
    },
    {
      "epoch": 0.24233,
      "grad_norm": 1.40791526406954,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 24233
    },
    {
      "epoch": 0.24234,
      "grad_norm": 1.5286186035685638,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 24234
    },
    {
      "epoch": 0.24235,
      "grad_norm": 1.2760029744853594,
      "learning_rate": 0.003,
      "loss": 4.0099,
      "step": 24235
    },
    {
      "epoch": 0.24236,
      "grad_norm": 1.6618425248039033,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 24236
    },
    {
      "epoch": 0.24237,
      "grad_norm": 1.124526517736685,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 24237
    },
    {
      "epoch": 0.24238,
      "grad_norm": 1.416640233410491,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 24238
    },
    {
      "epoch": 0.24239,
      "grad_norm": 1.347448856826729,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 24239
    },
    {
      "epoch": 0.2424,
      "grad_norm": 1.4961608163368172,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 24240
    },
    {
      "epoch": 0.24241,
      "grad_norm": 1.3471709137260668,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 24241
    },
    {
      "epoch": 0.24242,
      "grad_norm": 1.3148583875188227,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 24242
    },
    {
      "epoch": 0.24243,
      "grad_norm": 1.1937711995588174,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 24243
    },
    {
      "epoch": 0.24244,
      "grad_norm": 1.4749043608917174,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 24244
    },
    {
      "epoch": 0.24245,
      "grad_norm": 1.498681909173045,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 24245
    },
    {
      "epoch": 0.24246,
      "grad_norm": 1.578179945223917,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 24246
    },
    {
      "epoch": 0.24247,
      "grad_norm": 1.44493706097333,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 24247
    },
    {
      "epoch": 0.24248,
      "grad_norm": 1.565795094573324,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 24248
    },
    {
      "epoch": 0.24249,
      "grad_norm": 1.3342083156197733,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 24249
    },
    {
      "epoch": 0.2425,
      "grad_norm": 1.2513184429047661,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 24250
    },
    {
      "epoch": 0.24251,
      "grad_norm": 1.2281229389318682,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 24251
    },
    {
      "epoch": 0.24252,
      "grad_norm": 1.0860596610360937,
      "learning_rate": 0.003,
      "loss": 4.0299,
      "step": 24252
    },
    {
      "epoch": 0.24253,
      "grad_norm": 1.5039491917515584,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 24253
    },
    {
      "epoch": 0.24254,
      "grad_norm": 1.4064376154439924,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 24254
    },
    {
      "epoch": 0.24255,
      "grad_norm": 1.6241444988948979,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 24255
    },
    {
      "epoch": 0.24256,
      "grad_norm": 1.2205957992114362,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 24256
    },
    {
      "epoch": 0.24257,
      "grad_norm": 1.5260255640938487,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 24257
    },
    {
      "epoch": 0.24258,
      "grad_norm": 1.5989556429015173,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 24258
    },
    {
      "epoch": 0.24259,
      "grad_norm": 1.0751015815729699,
      "learning_rate": 0.003,
      "loss": 4.0033,
      "step": 24259
    },
    {
      "epoch": 0.2426,
      "grad_norm": 1.702472906850061,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 24260
    },
    {
      "epoch": 0.24261,
      "grad_norm": 1.5207047352202259,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 24261
    },
    {
      "epoch": 0.24262,
      "grad_norm": 1.2409427212254576,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 24262
    },
    {
      "epoch": 0.24263,
      "grad_norm": 1.4261002465171797,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 24263
    },
    {
      "epoch": 0.24264,
      "grad_norm": 1.3528463451310437,
      "learning_rate": 0.003,
      "loss": 4.0974,
      "step": 24264
    },
    {
      "epoch": 0.24265,
      "grad_norm": 1.4199721692575715,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 24265
    },
    {
      "epoch": 0.24266,
      "grad_norm": 1.1801013207495208,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 24266
    },
    {
      "epoch": 0.24267,
      "grad_norm": 1.4976461476618232,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 24267
    },
    {
      "epoch": 0.24268,
      "grad_norm": 1.450681518453227,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 24268
    },
    {
      "epoch": 0.24269,
      "grad_norm": 1.5218552775333807,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 24269
    },
    {
      "epoch": 0.2427,
      "grad_norm": 1.3646692833450003,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 24270
    },
    {
      "epoch": 0.24271,
      "grad_norm": 1.372212462846825,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 24271
    },
    {
      "epoch": 0.24272,
      "grad_norm": 1.476927481122054,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 24272
    },
    {
      "epoch": 0.24273,
      "grad_norm": 1.6055376498452612,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 24273
    },
    {
      "epoch": 0.24274,
      "grad_norm": 1.1388426209900409,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 24274
    },
    {
      "epoch": 0.24275,
      "grad_norm": 1.6332333310164349,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 24275
    },
    {
      "epoch": 0.24276,
      "grad_norm": 1.1006605511883762,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 24276
    },
    {
      "epoch": 0.24277,
      "grad_norm": 1.5254223487723864,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 24277
    },
    {
      "epoch": 0.24278,
      "grad_norm": 1.0211464488351767,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 24278
    },
    {
      "epoch": 0.24279,
      "grad_norm": 1.6620715130789216,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 24279
    },
    {
      "epoch": 0.2428,
      "grad_norm": 1.2899780131352225,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 24280
    },
    {
      "epoch": 0.24281,
      "grad_norm": 1.6287916717460589,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 24281
    },
    {
      "epoch": 0.24282,
      "grad_norm": 1.3662707506483824,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 24282
    },
    {
      "epoch": 0.24283,
      "grad_norm": 1.26474848782294,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 24283
    },
    {
      "epoch": 0.24284,
      "grad_norm": 1.3738238600431016,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 24284
    },
    {
      "epoch": 0.24285,
      "grad_norm": 1.4668232371604881,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 24285
    },
    {
      "epoch": 0.24286,
      "grad_norm": 1.3705409294532989,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 24286
    },
    {
      "epoch": 0.24287,
      "grad_norm": 1.4100367607106197,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 24287
    },
    {
      "epoch": 0.24288,
      "grad_norm": 1.3674569502200074,
      "learning_rate": 0.003,
      "loss": 4.0293,
      "step": 24288
    },
    {
      "epoch": 0.24289,
      "grad_norm": 1.5153245137137263,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 24289
    },
    {
      "epoch": 0.2429,
      "grad_norm": 1.2030234602807022,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 24290
    },
    {
      "epoch": 0.24291,
      "grad_norm": 1.453601677945708,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 24291
    },
    {
      "epoch": 0.24292,
      "grad_norm": 1.4259857001108331,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 24292
    },
    {
      "epoch": 0.24293,
      "grad_norm": 1.3952410742227412,
      "learning_rate": 0.003,
      "loss": 4.0074,
      "step": 24293
    },
    {
      "epoch": 0.24294,
      "grad_norm": 1.294020014575058,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 24294
    },
    {
      "epoch": 0.24295,
      "grad_norm": 1.4638737452182335,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 24295
    },
    {
      "epoch": 0.24296,
      "grad_norm": 1.363088574657565,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 24296
    },
    {
      "epoch": 0.24297,
      "grad_norm": 1.5362387615785396,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 24297
    },
    {
      "epoch": 0.24298,
      "grad_norm": 1.4637221422097164,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 24298
    },
    {
      "epoch": 0.24299,
      "grad_norm": 1.4573293428199658,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 24299
    },
    {
      "epoch": 0.243,
      "grad_norm": 1.3336338228792428,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 24300
    },
    {
      "epoch": 0.24301,
      "grad_norm": 1.4459234739305402,
      "learning_rate": 0.003,
      "loss": 4.0203,
      "step": 24301
    },
    {
      "epoch": 0.24302,
      "grad_norm": 1.128787365746679,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 24302
    },
    {
      "epoch": 0.24303,
      "grad_norm": 1.530955486995232,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 24303
    },
    {
      "epoch": 0.24304,
      "grad_norm": 1.1271872822650284,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 24304
    },
    {
      "epoch": 0.24305,
      "grad_norm": 1.2707499654126093,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 24305
    },
    {
      "epoch": 0.24306,
      "grad_norm": 1.4113113896743796,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 24306
    },
    {
      "epoch": 0.24307,
      "grad_norm": 1.522721069518732,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 24307
    },
    {
      "epoch": 0.24308,
      "grad_norm": 1.3787396396927958,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 24308
    },
    {
      "epoch": 0.24309,
      "grad_norm": 1.456770842668709,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 24309
    },
    {
      "epoch": 0.2431,
      "grad_norm": 1.4023116403124465,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 24310
    },
    {
      "epoch": 0.24311,
      "grad_norm": 1.4478873108427872,
      "learning_rate": 0.003,
      "loss": 4.0427,
      "step": 24311
    },
    {
      "epoch": 0.24312,
      "grad_norm": 1.2028607088997458,
      "learning_rate": 0.003,
      "loss": 4.0197,
      "step": 24312
    },
    {
      "epoch": 0.24313,
      "grad_norm": 1.3631454952837638,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 24313
    },
    {
      "epoch": 0.24314,
      "grad_norm": 1.1105654090559485,
      "learning_rate": 0.003,
      "loss": 4.0241,
      "step": 24314
    },
    {
      "epoch": 0.24315,
      "grad_norm": 1.5136857873274912,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 24315
    },
    {
      "epoch": 0.24316,
      "grad_norm": 1.3313712629597232,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 24316
    },
    {
      "epoch": 0.24317,
      "grad_norm": 1.5155104946271194,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 24317
    },
    {
      "epoch": 0.24318,
      "grad_norm": 1.522851616703085,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 24318
    },
    {
      "epoch": 0.24319,
      "grad_norm": 1.3383762316729688,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 24319
    },
    {
      "epoch": 0.2432,
      "grad_norm": 1.3100848345872873,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 24320
    },
    {
      "epoch": 0.24321,
      "grad_norm": 1.3435673846156737,
      "learning_rate": 0.003,
      "loss": 4.0309,
      "step": 24321
    },
    {
      "epoch": 0.24322,
      "grad_norm": 1.2516632175398126,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 24322
    },
    {
      "epoch": 0.24323,
      "grad_norm": 1.5232425443425288,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 24323
    },
    {
      "epoch": 0.24324,
      "grad_norm": 1.4367358145086015,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 24324
    },
    {
      "epoch": 0.24325,
      "grad_norm": 1.4684628974815295,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 24325
    },
    {
      "epoch": 0.24326,
      "grad_norm": 1.3624352636286248,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 24326
    },
    {
      "epoch": 0.24327,
      "grad_norm": 1.5706965201661032,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 24327
    },
    {
      "epoch": 0.24328,
      "grad_norm": 1.5638520719360105,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 24328
    },
    {
      "epoch": 0.24329,
      "grad_norm": 1.2353108973359153,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 24329
    },
    {
      "epoch": 0.2433,
      "grad_norm": 1.3871393259640707,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 24330
    },
    {
      "epoch": 0.24331,
      "grad_norm": 1.2881701036380848,
      "learning_rate": 0.003,
      "loss": 4.0257,
      "step": 24331
    },
    {
      "epoch": 0.24332,
      "grad_norm": 1.3615269539507815,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 24332
    },
    {
      "epoch": 0.24333,
      "grad_norm": 1.5615727208485255,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 24333
    },
    {
      "epoch": 0.24334,
      "grad_norm": 1.2817409757654774,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 24334
    },
    {
      "epoch": 0.24335,
      "grad_norm": 1.4257980343726475,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 24335
    },
    {
      "epoch": 0.24336,
      "grad_norm": 1.2940428962366473,
      "learning_rate": 0.003,
      "loss": 4.0273,
      "step": 24336
    },
    {
      "epoch": 0.24337,
      "grad_norm": 1.2305778335972597,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 24337
    },
    {
      "epoch": 0.24338,
      "grad_norm": 1.4471247474330844,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 24338
    },
    {
      "epoch": 0.24339,
      "grad_norm": 1.385055081545386,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 24339
    },
    {
      "epoch": 0.2434,
      "grad_norm": 1.3640820062649563,
      "learning_rate": 0.003,
      "loss": 4.0338,
      "step": 24340
    },
    {
      "epoch": 0.24341,
      "grad_norm": 1.483300219861865,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 24341
    },
    {
      "epoch": 0.24342,
      "grad_norm": 1.270825362616498,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 24342
    },
    {
      "epoch": 0.24343,
      "grad_norm": 1.3257500639714206,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 24343
    },
    {
      "epoch": 0.24344,
      "grad_norm": 1.3964866735472208,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 24344
    },
    {
      "epoch": 0.24345,
      "grad_norm": 1.5052621243505104,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 24345
    },
    {
      "epoch": 0.24346,
      "grad_norm": 1.4123858896856982,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 24346
    },
    {
      "epoch": 0.24347,
      "grad_norm": 1.4832504684088665,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 24347
    },
    {
      "epoch": 0.24348,
      "grad_norm": 1.4298142844095485,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 24348
    },
    {
      "epoch": 0.24349,
      "grad_norm": 1.5737336294592716,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 24349
    },
    {
      "epoch": 0.2435,
      "grad_norm": 1.4858935974430687,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 24350
    },
    {
      "epoch": 0.24351,
      "grad_norm": 1.4123014696818081,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 24351
    },
    {
      "epoch": 0.24352,
      "grad_norm": 1.521035894910265,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 24352
    },
    {
      "epoch": 0.24353,
      "grad_norm": 1.2365922099599098,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 24353
    },
    {
      "epoch": 0.24354,
      "grad_norm": 1.429933876519162,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 24354
    },
    {
      "epoch": 0.24355,
      "grad_norm": 1.19182519647841,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 24355
    },
    {
      "epoch": 0.24356,
      "grad_norm": 1.3457778123474984,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 24356
    },
    {
      "epoch": 0.24357,
      "grad_norm": 1.3159769497093734,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 24357
    },
    {
      "epoch": 0.24358,
      "grad_norm": 1.592912543197476,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 24358
    },
    {
      "epoch": 0.24359,
      "grad_norm": 1.0522976764991934,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 24359
    },
    {
      "epoch": 0.2436,
      "grad_norm": 1.8177060944721388,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 24360
    },
    {
      "epoch": 0.24361,
      "grad_norm": 1.3556473724145826,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 24361
    },
    {
      "epoch": 0.24362,
      "grad_norm": 1.3194578770130425,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 24362
    },
    {
      "epoch": 0.24363,
      "grad_norm": 1.352932434724565,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 24363
    },
    {
      "epoch": 0.24364,
      "grad_norm": 1.3340297145132445,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 24364
    },
    {
      "epoch": 0.24365,
      "grad_norm": 1.4323645307855688,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 24365
    },
    {
      "epoch": 0.24366,
      "grad_norm": 1.0952847900510025,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 24366
    },
    {
      "epoch": 0.24367,
      "grad_norm": 1.9483655270056768,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 24367
    },
    {
      "epoch": 0.24368,
      "grad_norm": 1.184671412937592,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 24368
    },
    {
      "epoch": 0.24369,
      "grad_norm": 1.323735956311098,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 24369
    },
    {
      "epoch": 0.2437,
      "grad_norm": 1.2287185524637567,
      "learning_rate": 0.003,
      "loss": 4.0243,
      "step": 24370
    },
    {
      "epoch": 0.24371,
      "grad_norm": 1.5477768258982456,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 24371
    },
    {
      "epoch": 0.24372,
      "grad_norm": 1.5782593451961935,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 24372
    },
    {
      "epoch": 0.24373,
      "grad_norm": 1.4108284316744677,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 24373
    },
    {
      "epoch": 0.24374,
      "grad_norm": 1.2284705573126384,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 24374
    },
    {
      "epoch": 0.24375,
      "grad_norm": 1.37467768781896,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 24375
    },
    {
      "epoch": 0.24376,
      "grad_norm": 1.471349874890904,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 24376
    },
    {
      "epoch": 0.24377,
      "grad_norm": 1.3749388319854385,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 24377
    },
    {
      "epoch": 0.24378,
      "grad_norm": 1.274363233056151,
      "learning_rate": 0.003,
      "loss": 4.0191,
      "step": 24378
    },
    {
      "epoch": 0.24379,
      "grad_norm": 1.6912229841315816,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 24379
    },
    {
      "epoch": 0.2438,
      "grad_norm": 1.400093247837791,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 24380
    },
    {
      "epoch": 0.24381,
      "grad_norm": 1.2895066337843406,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 24381
    },
    {
      "epoch": 0.24382,
      "grad_norm": 1.4194758712654583,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 24382
    },
    {
      "epoch": 0.24383,
      "grad_norm": 1.2433499914638473,
      "learning_rate": 0.003,
      "loss": 4.0329,
      "step": 24383
    },
    {
      "epoch": 0.24384,
      "grad_norm": 1.4821397422065814,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 24384
    },
    {
      "epoch": 0.24385,
      "grad_norm": 1.295262408381157,
      "learning_rate": 0.003,
      "loss": 4.0377,
      "step": 24385
    },
    {
      "epoch": 0.24386,
      "grad_norm": 1.2954067866727659,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 24386
    },
    {
      "epoch": 0.24387,
      "grad_norm": 1.5892041704504634,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 24387
    },
    {
      "epoch": 0.24388,
      "grad_norm": 1.430285687375732,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 24388
    },
    {
      "epoch": 0.24389,
      "grad_norm": 1.712218677082825,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 24389
    },
    {
      "epoch": 0.2439,
      "grad_norm": 1.4151262536645157,
      "learning_rate": 0.003,
      "loss": 4.0304,
      "step": 24390
    },
    {
      "epoch": 0.24391,
      "grad_norm": 1.3112013316467788,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 24391
    },
    {
      "epoch": 0.24392,
      "grad_norm": 1.7240463615981416,
      "learning_rate": 0.003,
      "loss": 4.0779,
      "step": 24392
    },
    {
      "epoch": 0.24393,
      "grad_norm": 0.8858424824078097,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 24393
    },
    {
      "epoch": 0.24394,
      "grad_norm": 1.4116201647707651,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 24394
    },
    {
      "epoch": 0.24395,
      "grad_norm": 1.3334782424316771,
      "learning_rate": 0.003,
      "loss": 4.0205,
      "step": 24395
    },
    {
      "epoch": 0.24396,
      "grad_norm": 1.4632406948876087,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 24396
    },
    {
      "epoch": 0.24397,
      "grad_norm": 1.2817814009624928,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 24397
    },
    {
      "epoch": 0.24398,
      "grad_norm": 1.487348617246001,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 24398
    },
    {
      "epoch": 0.24399,
      "grad_norm": 1.5460169285214382,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 24399
    },
    {
      "epoch": 0.244,
      "grad_norm": 1.4556184245199653,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 24400
    },
    {
      "epoch": 0.24401,
      "grad_norm": 1.3171795632224563,
      "learning_rate": 0.003,
      "loss": 4.019,
      "step": 24401
    },
    {
      "epoch": 0.24402,
      "grad_norm": 1.185485189642518,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 24402
    },
    {
      "epoch": 0.24403,
      "grad_norm": 1.3650193870209284,
      "learning_rate": 0.003,
      "loss": 4.0737,
      "step": 24403
    },
    {
      "epoch": 0.24404,
      "grad_norm": 1.3693011354662386,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 24404
    },
    {
      "epoch": 0.24405,
      "grad_norm": 1.3979817367995035,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 24405
    },
    {
      "epoch": 0.24406,
      "grad_norm": 1.4139725650330204,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 24406
    },
    {
      "epoch": 0.24407,
      "grad_norm": 1.5673551822882998,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 24407
    },
    {
      "epoch": 0.24408,
      "grad_norm": 1.4348223488692182,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24408
    },
    {
      "epoch": 0.24409,
      "grad_norm": 1.2870111952470822,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 24409
    },
    {
      "epoch": 0.2441,
      "grad_norm": 1.5607075385910172,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 24410
    },
    {
      "epoch": 0.24411,
      "grad_norm": 1.465916095422506,
      "learning_rate": 0.003,
      "loss": 4.0357,
      "step": 24411
    },
    {
      "epoch": 0.24412,
      "grad_norm": 1.3937935936403099,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 24412
    },
    {
      "epoch": 0.24413,
      "grad_norm": 1.2495468899332896,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 24413
    },
    {
      "epoch": 0.24414,
      "grad_norm": 1.4251258114352572,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 24414
    },
    {
      "epoch": 0.24415,
      "grad_norm": 1.516055056087047,
      "learning_rate": 0.003,
      "loss": 4.0212,
      "step": 24415
    },
    {
      "epoch": 0.24416,
      "grad_norm": 1.1966696834797699,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 24416
    },
    {
      "epoch": 0.24417,
      "grad_norm": 1.5138749495250303,
      "learning_rate": 0.003,
      "loss": 4.0385,
      "step": 24417
    },
    {
      "epoch": 0.24418,
      "grad_norm": 1.5030877361485186,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 24418
    },
    {
      "epoch": 0.24419,
      "grad_norm": 1.4994207623423623,
      "learning_rate": 0.003,
      "loss": 4.0457,
      "step": 24419
    },
    {
      "epoch": 0.2442,
      "grad_norm": 1.5317987422931272,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 24420
    },
    {
      "epoch": 0.24421,
      "grad_norm": 1.358310738936398,
      "learning_rate": 0.003,
      "loss": 4.0389,
      "step": 24421
    },
    {
      "epoch": 0.24422,
      "grad_norm": 1.4750672972351306,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 24422
    },
    {
      "epoch": 0.24423,
      "grad_norm": 1.353864942876393,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 24423
    },
    {
      "epoch": 0.24424,
      "grad_norm": 1.3545312285939408,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 24424
    },
    {
      "epoch": 0.24425,
      "grad_norm": 1.3341578590560708,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 24425
    },
    {
      "epoch": 0.24426,
      "grad_norm": 1.3757113987183092,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24426
    },
    {
      "epoch": 0.24427,
      "grad_norm": 1.5416574636024183,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24427
    },
    {
      "epoch": 0.24428,
      "grad_norm": 1.570906714197422,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 24428
    },
    {
      "epoch": 0.24429,
      "grad_norm": 1.4222210092104266,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 24429
    },
    {
      "epoch": 0.2443,
      "grad_norm": 1.0247992703851363,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 24430
    },
    {
      "epoch": 0.24431,
      "grad_norm": 1.810736259743895,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 24431
    },
    {
      "epoch": 0.24432,
      "grad_norm": 1.2365585636560195,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 24432
    },
    {
      "epoch": 0.24433,
      "grad_norm": 1.674257260914585,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 24433
    },
    {
      "epoch": 0.24434,
      "grad_norm": 1.3286671090444853,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 24434
    },
    {
      "epoch": 0.24435,
      "grad_norm": 1.0904345767145893,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 24435
    },
    {
      "epoch": 0.24436,
      "grad_norm": 1.5762453636013694,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 24436
    },
    {
      "epoch": 0.24437,
      "grad_norm": 1.158325997917941,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 24437
    },
    {
      "epoch": 0.24438,
      "grad_norm": 1.680406328144425,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 24438
    },
    {
      "epoch": 0.24439,
      "grad_norm": 1.311732222890648,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 24439
    },
    {
      "epoch": 0.2444,
      "grad_norm": 1.2699507423947751,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 24440
    },
    {
      "epoch": 0.24441,
      "grad_norm": 1.3243951161664762,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 24441
    },
    {
      "epoch": 0.24442,
      "grad_norm": 1.5229048088252097,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 24442
    },
    {
      "epoch": 0.24443,
      "grad_norm": 1.4887025761334038,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 24443
    },
    {
      "epoch": 0.24444,
      "grad_norm": 1.586094301592182,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 24444
    },
    {
      "epoch": 0.24445,
      "grad_norm": 1.302344126469741,
      "learning_rate": 0.003,
      "loss": 4.085,
      "step": 24445
    },
    {
      "epoch": 0.24446,
      "grad_norm": 1.186296232078591,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 24446
    },
    {
      "epoch": 0.24447,
      "grad_norm": 1.6377893578773677,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 24447
    },
    {
      "epoch": 0.24448,
      "grad_norm": 1.3549837039624482,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 24448
    },
    {
      "epoch": 0.24449,
      "grad_norm": 1.4502058738900307,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 24449
    },
    {
      "epoch": 0.2445,
      "grad_norm": 1.3386576209257963,
      "learning_rate": 0.003,
      "loss": 4.0264,
      "step": 24450
    },
    {
      "epoch": 0.24451,
      "grad_norm": 1.1957847782425934,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 24451
    },
    {
      "epoch": 0.24452,
      "grad_norm": 1.412730486244832,
      "learning_rate": 0.003,
      "loss": 4.0554,
      "step": 24452
    },
    {
      "epoch": 0.24453,
      "grad_norm": 1.2428989416283058,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 24453
    },
    {
      "epoch": 0.24454,
      "grad_norm": 1.517494911075393,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 24454
    },
    {
      "epoch": 0.24455,
      "grad_norm": 1.271593906889155,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 24455
    },
    {
      "epoch": 0.24456,
      "grad_norm": 1.7310280058005147,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 24456
    },
    {
      "epoch": 0.24457,
      "grad_norm": 1.387073538895397,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 24457
    },
    {
      "epoch": 0.24458,
      "grad_norm": 1.782320903754428,
      "learning_rate": 0.003,
      "loss": 4.079,
      "step": 24458
    },
    {
      "epoch": 0.24459,
      "grad_norm": 1.1960676787134221,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 24459
    },
    {
      "epoch": 0.2446,
      "grad_norm": 1.5858316416307017,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 24460
    },
    {
      "epoch": 0.24461,
      "grad_norm": 1.709659418447422,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 24461
    },
    {
      "epoch": 0.24462,
      "grad_norm": 1.2395379160011137,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 24462
    },
    {
      "epoch": 0.24463,
      "grad_norm": 1.4529047209427233,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 24463
    },
    {
      "epoch": 0.24464,
      "grad_norm": 1.3186117408301783,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 24464
    },
    {
      "epoch": 0.24465,
      "grad_norm": 1.2927826551197905,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 24465
    },
    {
      "epoch": 0.24466,
      "grad_norm": 1.6199789647127618,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 24466
    },
    {
      "epoch": 0.24467,
      "grad_norm": 1.0021142676261792,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 24467
    },
    {
      "epoch": 0.24468,
      "grad_norm": 1.4961842732304016,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 24468
    },
    {
      "epoch": 0.24469,
      "grad_norm": 1.30394600125921,
      "learning_rate": 0.003,
      "loss": 4.1019,
      "step": 24469
    },
    {
      "epoch": 0.2447,
      "grad_norm": 1.5561301183312808,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 24470
    },
    {
      "epoch": 0.24471,
      "grad_norm": 1.2900322797382657,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 24471
    },
    {
      "epoch": 0.24472,
      "grad_norm": 1.3229581521811908,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 24472
    },
    {
      "epoch": 0.24473,
      "grad_norm": 1.3858105877881572,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 24473
    },
    {
      "epoch": 0.24474,
      "grad_norm": 1.496441007578832,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 24474
    },
    {
      "epoch": 0.24475,
      "grad_norm": 1.4317586032144547,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 24475
    },
    {
      "epoch": 0.24476,
      "grad_norm": 1.6698655719421154,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 24476
    },
    {
      "epoch": 0.24477,
      "grad_norm": 1.6093669618707407,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 24477
    },
    {
      "epoch": 0.24478,
      "grad_norm": 1.2589770699717462,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 24478
    },
    {
      "epoch": 0.24479,
      "grad_norm": 1.4768781549212349,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 24479
    },
    {
      "epoch": 0.2448,
      "grad_norm": 1.6449614413144684,
      "learning_rate": 0.003,
      "loss": 4.0282,
      "step": 24480
    },
    {
      "epoch": 0.24481,
      "grad_norm": 1.2132655621083248,
      "learning_rate": 0.003,
      "loss": 4.0283,
      "step": 24481
    },
    {
      "epoch": 0.24482,
      "grad_norm": 1.3357216242531542,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 24482
    },
    {
      "epoch": 0.24483,
      "grad_norm": 1.582000784823852,
      "learning_rate": 0.003,
      "loss": 4.019,
      "step": 24483
    },
    {
      "epoch": 0.24484,
      "grad_norm": 1.2914068837901693,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 24484
    },
    {
      "epoch": 0.24485,
      "grad_norm": 1.2532583593279065,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 24485
    },
    {
      "epoch": 0.24486,
      "grad_norm": 1.4148623858568774,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 24486
    },
    {
      "epoch": 0.24487,
      "grad_norm": 1.3746186948771522,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 24487
    },
    {
      "epoch": 0.24488,
      "grad_norm": 1.335901034283889,
      "learning_rate": 0.003,
      "loss": 4.0768,
      "step": 24488
    },
    {
      "epoch": 0.24489,
      "grad_norm": 1.3127438995233012,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 24489
    },
    {
      "epoch": 0.2449,
      "grad_norm": 1.2952384595072532,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 24490
    },
    {
      "epoch": 0.24491,
      "grad_norm": 1.3252811911960327,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 24491
    },
    {
      "epoch": 0.24492,
      "grad_norm": 1.416293001547004,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 24492
    },
    {
      "epoch": 0.24493,
      "grad_norm": 1.6077015621828852,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 24493
    },
    {
      "epoch": 0.24494,
      "grad_norm": 1.653231955767635,
      "learning_rate": 0.003,
      "loss": 4.046,
      "step": 24494
    },
    {
      "epoch": 0.24495,
      "grad_norm": 1.3480367016954977,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 24495
    },
    {
      "epoch": 0.24496,
      "grad_norm": 1.3594085884946376,
      "learning_rate": 0.003,
      "loss": 4.0563,
      "step": 24496
    },
    {
      "epoch": 0.24497,
      "grad_norm": 1.1981812674244252,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 24497
    },
    {
      "epoch": 0.24498,
      "grad_norm": 1.498331640084471,
      "learning_rate": 0.003,
      "loss": 4.0603,
      "step": 24498
    },
    {
      "epoch": 0.24499,
      "grad_norm": 1.3094852746117118,
      "learning_rate": 0.003,
      "loss": 4.0191,
      "step": 24499
    },
    {
      "epoch": 0.245,
      "grad_norm": 1.472095960571987,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 24500
    },
    {
      "epoch": 0.24501,
      "grad_norm": 1.3808112648803745,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 24501
    },
    {
      "epoch": 0.24502,
      "grad_norm": 1.2129734417411704,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 24502
    },
    {
      "epoch": 0.24503,
      "grad_norm": 1.5611790014750202,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 24503
    },
    {
      "epoch": 0.24504,
      "grad_norm": 1.1533309815977,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 24504
    },
    {
      "epoch": 0.24505,
      "grad_norm": 1.6488448674654328,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 24505
    },
    {
      "epoch": 0.24506,
      "grad_norm": 1.294658197306135,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 24506
    },
    {
      "epoch": 0.24507,
      "grad_norm": 1.4197784989873545,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 24507
    },
    {
      "epoch": 0.24508,
      "grad_norm": 1.2750191872365566,
      "learning_rate": 0.003,
      "loss": 4.0256,
      "step": 24508
    },
    {
      "epoch": 0.24509,
      "grad_norm": 1.4359010091783646,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 24509
    },
    {
      "epoch": 0.2451,
      "grad_norm": 1.3417831220305074,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 24510
    },
    {
      "epoch": 0.24511,
      "grad_norm": 1.4232112183549983,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 24511
    },
    {
      "epoch": 0.24512,
      "grad_norm": 1.2173747227926879,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 24512
    },
    {
      "epoch": 0.24513,
      "grad_norm": 1.2648743666521953,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 24513
    },
    {
      "epoch": 0.24514,
      "grad_norm": 1.3260981005183057,
      "learning_rate": 0.003,
      "loss": 4.0887,
      "step": 24514
    },
    {
      "epoch": 0.24515,
      "grad_norm": 1.862719067652409,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 24515
    },
    {
      "epoch": 0.24516,
      "grad_norm": 1.202946518556536,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 24516
    },
    {
      "epoch": 0.24517,
      "grad_norm": 2.04262011083375,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 24517
    },
    {
      "epoch": 0.24518,
      "grad_norm": 1.0726008388250743,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 24518
    },
    {
      "epoch": 0.24519,
      "grad_norm": 1.6750839788348069,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 24519
    },
    {
      "epoch": 0.2452,
      "grad_norm": 1.1668555665196068,
      "learning_rate": 0.003,
      "loss": 4.0771,
      "step": 24520
    },
    {
      "epoch": 0.24521,
      "grad_norm": 1.5162086637467458,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 24521
    },
    {
      "epoch": 0.24522,
      "grad_norm": 1.5733418179620733,
      "learning_rate": 0.003,
      "loss": 4.1011,
      "step": 24522
    },
    {
      "epoch": 0.24523,
      "grad_norm": 1.4948107770526557,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 24523
    },
    {
      "epoch": 0.24524,
      "grad_norm": 1.6909722365205078,
      "learning_rate": 0.003,
      "loss": 4.0813,
      "step": 24524
    },
    {
      "epoch": 0.24525,
      "grad_norm": 1.6588137622291679,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 24525
    },
    {
      "epoch": 0.24526,
      "grad_norm": 1.40047849015704,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 24526
    },
    {
      "epoch": 0.24527,
      "grad_norm": 1.6987966478577776,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 24527
    },
    {
      "epoch": 0.24528,
      "grad_norm": 1.1681302398291613,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 24528
    },
    {
      "epoch": 0.24529,
      "grad_norm": 1.4112970283358313,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 24529
    },
    {
      "epoch": 0.2453,
      "grad_norm": 1.2115141575097237,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 24530
    },
    {
      "epoch": 0.24531,
      "grad_norm": 1.3956583977814705,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 24531
    },
    {
      "epoch": 0.24532,
      "grad_norm": 1.2277481321412032,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 24532
    },
    {
      "epoch": 0.24533,
      "grad_norm": 1.1711346197082155,
      "learning_rate": 0.003,
      "loss": 4.0376,
      "step": 24533
    },
    {
      "epoch": 0.24534,
      "grad_norm": 1.361342262928631,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 24534
    },
    {
      "epoch": 0.24535,
      "grad_norm": 1.465196438698652,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 24535
    },
    {
      "epoch": 0.24536,
      "grad_norm": 1.3227436156788686,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 24536
    },
    {
      "epoch": 0.24537,
      "grad_norm": 1.7352730851433897,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 24537
    },
    {
      "epoch": 0.24538,
      "grad_norm": 1.4420467752851898,
      "learning_rate": 0.003,
      "loss": 4.0182,
      "step": 24538
    },
    {
      "epoch": 0.24539,
      "grad_norm": 1.8879252281652532,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 24539
    },
    {
      "epoch": 0.2454,
      "grad_norm": 1.0139930085724325,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 24540
    },
    {
      "epoch": 0.24541,
      "grad_norm": 1.4078807796121886,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 24541
    },
    {
      "epoch": 0.24542,
      "grad_norm": 1.5171695189987984,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 24542
    },
    {
      "epoch": 0.24543,
      "grad_norm": 1.2287151872288782,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 24543
    },
    {
      "epoch": 0.24544,
      "grad_norm": 1.827233745827479,
      "learning_rate": 0.003,
      "loss": 4.0865,
      "step": 24544
    },
    {
      "epoch": 0.24545,
      "grad_norm": 1.5425358275520964,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 24545
    },
    {
      "epoch": 0.24546,
      "grad_norm": 1.1074978733238126,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 24546
    },
    {
      "epoch": 0.24547,
      "grad_norm": 1.41572862405677,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 24547
    },
    {
      "epoch": 0.24548,
      "grad_norm": 1.4618538064043483,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 24548
    },
    {
      "epoch": 0.24549,
      "grad_norm": 1.5358867638614377,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 24549
    },
    {
      "epoch": 0.2455,
      "grad_norm": 1.4764067874278128,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 24550
    },
    {
      "epoch": 0.24551,
      "grad_norm": 1.2656017796107097,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 24551
    },
    {
      "epoch": 0.24552,
      "grad_norm": 1.422792356176076,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 24552
    },
    {
      "epoch": 0.24553,
      "grad_norm": 1.1951801034404221,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 24553
    },
    {
      "epoch": 0.24554,
      "grad_norm": 1.3188185892123274,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 24554
    },
    {
      "epoch": 0.24555,
      "grad_norm": 1.3115026082608847,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 24555
    },
    {
      "epoch": 0.24556,
      "grad_norm": 1.3329702336810645,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 24556
    },
    {
      "epoch": 0.24557,
      "grad_norm": 1.3978328960150062,
      "learning_rate": 0.003,
      "loss": 4.0197,
      "step": 24557
    },
    {
      "epoch": 0.24558,
      "grad_norm": 1.1175116108372913,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 24558
    },
    {
      "epoch": 0.24559,
      "grad_norm": 1.773148591028893,
      "learning_rate": 0.003,
      "loss": 4.0683,
      "step": 24559
    },
    {
      "epoch": 0.2456,
      "grad_norm": 1.2411366591922672,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 24560
    },
    {
      "epoch": 0.24561,
      "grad_norm": 1.3765841855025884,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 24561
    },
    {
      "epoch": 0.24562,
      "grad_norm": 1.2549496069463526,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 24562
    },
    {
      "epoch": 0.24563,
      "grad_norm": 1.346449356926344,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 24563
    },
    {
      "epoch": 0.24564,
      "grad_norm": 1.413674698156181,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 24564
    },
    {
      "epoch": 0.24565,
      "grad_norm": 1.2789473351579235,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 24565
    },
    {
      "epoch": 0.24566,
      "grad_norm": 1.339566785711176,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 24566
    },
    {
      "epoch": 0.24567,
      "grad_norm": 1.3650356101008385,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 24567
    },
    {
      "epoch": 0.24568,
      "grad_norm": 1.3876420647866352,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 24568
    },
    {
      "epoch": 0.24569,
      "grad_norm": 1.4338418726194089,
      "learning_rate": 0.003,
      "loss": 4.0231,
      "step": 24569
    },
    {
      "epoch": 0.2457,
      "grad_norm": 1.753807450464722,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 24570
    },
    {
      "epoch": 0.24571,
      "grad_norm": 1.5146053908835122,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 24571
    },
    {
      "epoch": 0.24572,
      "grad_norm": 1.4814182439570196,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 24572
    },
    {
      "epoch": 0.24573,
      "grad_norm": 1.3029561351089674,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 24573
    },
    {
      "epoch": 0.24574,
      "grad_norm": 1.5578404170827305,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 24574
    },
    {
      "epoch": 0.24575,
      "grad_norm": 1.1894949968217112,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 24575
    },
    {
      "epoch": 0.24576,
      "grad_norm": 1.5470838354769207,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 24576
    },
    {
      "epoch": 0.24577,
      "grad_norm": 1.3500795966059036,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 24577
    },
    {
      "epoch": 0.24578,
      "grad_norm": 1.4092926111890176,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 24578
    },
    {
      "epoch": 0.24579,
      "grad_norm": 1.52709199684596,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 24579
    },
    {
      "epoch": 0.2458,
      "grad_norm": 1.1814001688213163,
      "learning_rate": 0.003,
      "loss": 4.0305,
      "step": 24580
    },
    {
      "epoch": 0.24581,
      "grad_norm": 1.4261722649888424,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 24581
    },
    {
      "epoch": 0.24582,
      "grad_norm": 1.1442515720367656,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 24582
    },
    {
      "epoch": 0.24583,
      "grad_norm": 1.487270336446602,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 24583
    },
    {
      "epoch": 0.24584,
      "grad_norm": 1.0401735167002224,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 24584
    },
    {
      "epoch": 0.24585,
      "grad_norm": 1.4650482110188905,
      "learning_rate": 0.003,
      "loss": 4.0938,
      "step": 24585
    },
    {
      "epoch": 0.24586,
      "grad_norm": 1.242945969827525,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 24586
    },
    {
      "epoch": 0.24587,
      "grad_norm": 1.6521344087701795,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 24587
    },
    {
      "epoch": 0.24588,
      "grad_norm": 1.5645848951245152,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 24588
    },
    {
      "epoch": 0.24589,
      "grad_norm": 1.4555479561093512,
      "learning_rate": 0.003,
      "loss": 4.0267,
      "step": 24589
    },
    {
      "epoch": 0.2459,
      "grad_norm": 1.694459714484806,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 24590
    },
    {
      "epoch": 0.24591,
      "grad_norm": 1.4258825484612274,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 24591
    },
    {
      "epoch": 0.24592,
      "grad_norm": 1.3288767041200285,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 24592
    },
    {
      "epoch": 0.24593,
      "grad_norm": 1.393988006884798,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 24593
    },
    {
      "epoch": 0.24594,
      "grad_norm": 1.2835427756796285,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 24594
    },
    {
      "epoch": 0.24595,
      "grad_norm": 1.4058980884817271,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 24595
    },
    {
      "epoch": 0.24596,
      "grad_norm": 1.3398051499422285,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 24596
    },
    {
      "epoch": 0.24597,
      "grad_norm": 1.6835506575380206,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 24597
    },
    {
      "epoch": 0.24598,
      "grad_norm": 1.3022608969836875,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 24598
    },
    {
      "epoch": 0.24599,
      "grad_norm": 1.866922719544958,
      "learning_rate": 0.003,
      "loss": 4.0313,
      "step": 24599
    },
    {
      "epoch": 0.246,
      "grad_norm": 1.1664846654114946,
      "learning_rate": 0.003,
      "loss": 4.0098,
      "step": 24600
    },
    {
      "epoch": 0.24601,
      "grad_norm": 1.5861335434926875,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 24601
    },
    {
      "epoch": 0.24602,
      "grad_norm": 1.2087072445416438,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 24602
    },
    {
      "epoch": 0.24603,
      "grad_norm": 1.6703543958167697,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 24603
    },
    {
      "epoch": 0.24604,
      "grad_norm": 1.2758211439689626,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 24604
    },
    {
      "epoch": 0.24605,
      "grad_norm": 1.4340962875835834,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 24605
    },
    {
      "epoch": 0.24606,
      "grad_norm": 1.1954652855240864,
      "learning_rate": 0.003,
      "loss": 4.0294,
      "step": 24606
    },
    {
      "epoch": 0.24607,
      "grad_norm": 1.4462666793842442,
      "learning_rate": 0.003,
      "loss": 4.0719,
      "step": 24607
    },
    {
      "epoch": 0.24608,
      "grad_norm": 1.4156766202597262,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 24608
    },
    {
      "epoch": 0.24609,
      "grad_norm": 1.323020574912954,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 24609
    },
    {
      "epoch": 0.2461,
      "grad_norm": 1.387088035102824,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 24610
    },
    {
      "epoch": 0.24611,
      "grad_norm": 1.3002607368516486,
      "learning_rate": 0.003,
      "loss": 4.0226,
      "step": 24611
    },
    {
      "epoch": 0.24612,
      "grad_norm": 1.5218306101556822,
      "learning_rate": 0.003,
      "loss": 4.0266,
      "step": 24612
    },
    {
      "epoch": 0.24613,
      "grad_norm": 1.1749153785406177,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24613
    },
    {
      "epoch": 0.24614,
      "grad_norm": 1.5835294723077846,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 24614
    },
    {
      "epoch": 0.24615,
      "grad_norm": 1.3003192097520786,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 24615
    },
    {
      "epoch": 0.24616,
      "grad_norm": 1.373273185063886,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 24616
    },
    {
      "epoch": 0.24617,
      "grad_norm": 1.3320281377586736,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 24617
    },
    {
      "epoch": 0.24618,
      "grad_norm": 1.611281347154727,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 24618
    },
    {
      "epoch": 0.24619,
      "grad_norm": 1.3408690334115998,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 24619
    },
    {
      "epoch": 0.2462,
      "grad_norm": 1.6464263087993103,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 24620
    },
    {
      "epoch": 0.24621,
      "grad_norm": 1.3693516194600759,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 24621
    },
    {
      "epoch": 0.24622,
      "grad_norm": 1.458470456610755,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 24622
    },
    {
      "epoch": 0.24623,
      "grad_norm": 1.399964577553382,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 24623
    },
    {
      "epoch": 0.24624,
      "grad_norm": 1.3420808827799742,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 24624
    },
    {
      "epoch": 0.24625,
      "grad_norm": 1.3963823107514195,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 24625
    },
    {
      "epoch": 0.24626,
      "grad_norm": 1.2297394037438159,
      "learning_rate": 0.003,
      "loss": 4.018,
      "step": 24626
    },
    {
      "epoch": 0.24627,
      "grad_norm": 1.1268844287216007,
      "learning_rate": 0.003,
      "loss": 4.0385,
      "step": 24627
    },
    {
      "epoch": 0.24628,
      "grad_norm": 1.3406638305777105,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 24628
    },
    {
      "epoch": 0.24629,
      "grad_norm": 1.3450601486899039,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 24629
    },
    {
      "epoch": 0.2463,
      "grad_norm": 1.5864310279963865,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 24630
    },
    {
      "epoch": 0.24631,
      "grad_norm": 1.133502531197477,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 24631
    },
    {
      "epoch": 0.24632,
      "grad_norm": 1.6448342809364231,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 24632
    },
    {
      "epoch": 0.24633,
      "grad_norm": 1.4034757528785844,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 24633
    },
    {
      "epoch": 0.24634,
      "grad_norm": 1.556867051293386,
      "learning_rate": 0.003,
      "loss": 4.0266,
      "step": 24634
    },
    {
      "epoch": 0.24635,
      "grad_norm": 1.5121569647113713,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 24635
    },
    {
      "epoch": 0.24636,
      "grad_norm": 1.1961739620650575,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 24636
    },
    {
      "epoch": 0.24637,
      "grad_norm": 1.4239970723771453,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 24637
    },
    {
      "epoch": 0.24638,
      "grad_norm": 1.102409428379748,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 24638
    },
    {
      "epoch": 0.24639,
      "grad_norm": 1.7719140279355743,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 24639
    },
    {
      "epoch": 0.2464,
      "grad_norm": 1.3712803644169622,
      "learning_rate": 0.003,
      "loss": 4.0368,
      "step": 24640
    },
    {
      "epoch": 0.24641,
      "grad_norm": 1.5045699212811086,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 24641
    },
    {
      "epoch": 0.24642,
      "grad_norm": 1.3761270797693337,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 24642
    },
    {
      "epoch": 0.24643,
      "grad_norm": 1.4943490813640457,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 24643
    },
    {
      "epoch": 0.24644,
      "grad_norm": 1.5344596961688393,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 24644
    },
    {
      "epoch": 0.24645,
      "grad_norm": 1.22811239361762,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24645
    },
    {
      "epoch": 0.24646,
      "grad_norm": 1.6091391728534283,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 24646
    },
    {
      "epoch": 0.24647,
      "grad_norm": 1.0788366757862968,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 24647
    },
    {
      "epoch": 0.24648,
      "grad_norm": 1.7664175405023235,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 24648
    },
    {
      "epoch": 0.24649,
      "grad_norm": 1.3672920496192789,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 24649
    },
    {
      "epoch": 0.2465,
      "grad_norm": 1.4143340175968577,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 24650
    },
    {
      "epoch": 0.24651,
      "grad_norm": 1.5061316213091294,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 24651
    },
    {
      "epoch": 0.24652,
      "grad_norm": 1.3424203535012875,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 24652
    },
    {
      "epoch": 0.24653,
      "grad_norm": 1.5086800307564445,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 24653
    },
    {
      "epoch": 0.24654,
      "grad_norm": 1.415792946661826,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 24654
    },
    {
      "epoch": 0.24655,
      "grad_norm": 1.33784359813273,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 24655
    },
    {
      "epoch": 0.24656,
      "grad_norm": 1.2431502701811876,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 24656
    },
    {
      "epoch": 0.24657,
      "grad_norm": 1.3791456290761512,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 24657
    },
    {
      "epoch": 0.24658,
      "grad_norm": 1.234715877137626,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 24658
    },
    {
      "epoch": 0.24659,
      "grad_norm": 1.5439065876747495,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 24659
    },
    {
      "epoch": 0.2466,
      "grad_norm": 1.0873679317458538,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 24660
    },
    {
      "epoch": 0.24661,
      "grad_norm": 1.6673409970089326,
      "learning_rate": 0.003,
      "loss": 4.044,
      "step": 24661
    },
    {
      "epoch": 0.24662,
      "grad_norm": 1.0444852034726346,
      "learning_rate": 0.003,
      "loss": 4.0196,
      "step": 24662
    },
    {
      "epoch": 0.24663,
      "grad_norm": 1.4755195416697107,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 24663
    },
    {
      "epoch": 0.24664,
      "grad_norm": 1.5490630187821017,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 24664
    },
    {
      "epoch": 0.24665,
      "grad_norm": 1.4389212193285406,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 24665
    },
    {
      "epoch": 0.24666,
      "grad_norm": 1.5898485789007988,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 24666
    },
    {
      "epoch": 0.24667,
      "grad_norm": 1.3414327554826224,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 24667
    },
    {
      "epoch": 0.24668,
      "grad_norm": 1.5083138034996968,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 24668
    },
    {
      "epoch": 0.24669,
      "grad_norm": 1.4375052700800133,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 24669
    },
    {
      "epoch": 0.2467,
      "grad_norm": 1.5605833834304783,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 24670
    },
    {
      "epoch": 0.24671,
      "grad_norm": 1.3886377108529842,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 24671
    },
    {
      "epoch": 0.24672,
      "grad_norm": 1.32058004497019,
      "learning_rate": 0.003,
      "loss": 4.0885,
      "step": 24672
    },
    {
      "epoch": 0.24673,
      "grad_norm": 1.4354588066798009,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 24673
    },
    {
      "epoch": 0.24674,
      "grad_norm": 1.1846883296170725,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 24674
    },
    {
      "epoch": 0.24675,
      "grad_norm": 1.6357411637273922,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 24675
    },
    {
      "epoch": 0.24676,
      "grad_norm": 1.6304879496370692,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 24676
    },
    {
      "epoch": 0.24677,
      "grad_norm": 1.1608645715542714,
      "learning_rate": 0.003,
      "loss": 4.0705,
      "step": 24677
    },
    {
      "epoch": 0.24678,
      "grad_norm": 1.426252735297271,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 24678
    },
    {
      "epoch": 0.24679,
      "grad_norm": 1.3048441360426093,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 24679
    },
    {
      "epoch": 0.2468,
      "grad_norm": 1.9295635246293055,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 24680
    },
    {
      "epoch": 0.24681,
      "grad_norm": 1.1741759928736533,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 24681
    },
    {
      "epoch": 0.24682,
      "grad_norm": 1.5398171618549163,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 24682
    },
    {
      "epoch": 0.24683,
      "grad_norm": 1.516394595940718,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 24683
    },
    {
      "epoch": 0.24684,
      "grad_norm": 1.1237933799901774,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 24684
    },
    {
      "epoch": 0.24685,
      "grad_norm": 1.6189441512891052,
      "learning_rate": 0.003,
      "loss": 4.0303,
      "step": 24685
    },
    {
      "epoch": 0.24686,
      "grad_norm": 1.3418548017049141,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 24686
    },
    {
      "epoch": 0.24687,
      "grad_norm": 1.2101976318774845,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 24687
    },
    {
      "epoch": 0.24688,
      "grad_norm": 1.5088613613462525,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 24688
    },
    {
      "epoch": 0.24689,
      "grad_norm": 1.396871327334466,
      "learning_rate": 0.003,
      "loss": 4.019,
      "step": 24689
    },
    {
      "epoch": 0.2469,
      "grad_norm": 1.432637212109617,
      "learning_rate": 0.003,
      "loss": 4.0781,
      "step": 24690
    },
    {
      "epoch": 0.24691,
      "grad_norm": 1.4407538030132299,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 24691
    },
    {
      "epoch": 0.24692,
      "grad_norm": 1.3150026238012944,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 24692
    },
    {
      "epoch": 0.24693,
      "grad_norm": 1.454035732610563,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 24693
    },
    {
      "epoch": 0.24694,
      "grad_norm": 1.4155672294803454,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 24694
    },
    {
      "epoch": 0.24695,
      "grad_norm": 1.4857061136107959,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 24695
    },
    {
      "epoch": 0.24696,
      "grad_norm": 1.2884400314255022,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 24696
    },
    {
      "epoch": 0.24697,
      "grad_norm": 1.2763712061446344,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 24697
    },
    {
      "epoch": 0.24698,
      "grad_norm": 1.404145056826194,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 24698
    },
    {
      "epoch": 0.24699,
      "grad_norm": 1.4530050644988275,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 24699
    },
    {
      "epoch": 0.247,
      "grad_norm": 1.2924308540326528,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 24700
    },
    {
      "epoch": 0.24701,
      "grad_norm": 1.437510979020967,
      "learning_rate": 0.003,
      "loss": 4.03,
      "step": 24701
    },
    {
      "epoch": 0.24702,
      "grad_norm": 1.4454755117993687,
      "learning_rate": 0.003,
      "loss": 4.0442,
      "step": 24702
    },
    {
      "epoch": 0.24703,
      "grad_norm": 1.3855527548886926,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 24703
    },
    {
      "epoch": 0.24704,
      "grad_norm": 1.2122761754217306,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 24704
    },
    {
      "epoch": 0.24705,
      "grad_norm": 1.5411351433529628,
      "learning_rate": 0.003,
      "loss": 4.0675,
      "step": 24705
    },
    {
      "epoch": 0.24706,
      "grad_norm": 1.36596179369511,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 24706
    },
    {
      "epoch": 0.24707,
      "grad_norm": 1.5830397978622857,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 24707
    },
    {
      "epoch": 0.24708,
      "grad_norm": 1.5662877541642863,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 24708
    },
    {
      "epoch": 0.24709,
      "grad_norm": 1.2411658668010157,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 24709
    },
    {
      "epoch": 0.2471,
      "grad_norm": 1.372268665719959,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 24710
    },
    {
      "epoch": 0.24711,
      "grad_norm": 1.2982392326994294,
      "learning_rate": 0.003,
      "loss": 4.0252,
      "step": 24711
    },
    {
      "epoch": 0.24712,
      "grad_norm": 1.317207009060153,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 24712
    },
    {
      "epoch": 0.24713,
      "grad_norm": 1.295956405394906,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 24713
    },
    {
      "epoch": 0.24714,
      "grad_norm": 1.47415354394144,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 24714
    },
    {
      "epoch": 0.24715,
      "grad_norm": 1.1830980536372186,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 24715
    },
    {
      "epoch": 0.24716,
      "grad_norm": 1.4562620130790478,
      "learning_rate": 0.003,
      "loss": 4.0303,
      "step": 24716
    },
    {
      "epoch": 0.24717,
      "grad_norm": 1.4242449287789543,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 24717
    },
    {
      "epoch": 0.24718,
      "grad_norm": 1.3875396746958109,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 24718
    },
    {
      "epoch": 0.24719,
      "grad_norm": 1.2596669274496077,
      "learning_rate": 0.003,
      "loss": 4.0355,
      "step": 24719
    },
    {
      "epoch": 0.2472,
      "grad_norm": 1.3599444745526024,
      "learning_rate": 0.003,
      "loss": 4.0367,
      "step": 24720
    },
    {
      "epoch": 0.24721,
      "grad_norm": 1.6367321665569092,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 24721
    },
    {
      "epoch": 0.24722,
      "grad_norm": 1.3975867186835018,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 24722
    },
    {
      "epoch": 0.24723,
      "grad_norm": 1.492204042599613,
      "learning_rate": 0.003,
      "loss": 4.0185,
      "step": 24723
    },
    {
      "epoch": 0.24724,
      "grad_norm": 1.276593404764828,
      "learning_rate": 0.003,
      "loss": 4.0583,
      "step": 24724
    },
    {
      "epoch": 0.24725,
      "grad_norm": 1.4397073350127796,
      "learning_rate": 0.003,
      "loss": 4.0315,
      "step": 24725
    },
    {
      "epoch": 0.24726,
      "grad_norm": 1.2502827618745087,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 24726
    },
    {
      "epoch": 0.24727,
      "grad_norm": 1.5028846024404838,
      "learning_rate": 0.003,
      "loss": 4.027,
      "step": 24727
    },
    {
      "epoch": 0.24728,
      "grad_norm": 1.367116680061631,
      "learning_rate": 0.003,
      "loss": 4.0224,
      "step": 24728
    },
    {
      "epoch": 0.24729,
      "grad_norm": 1.3241220634655897,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 24729
    },
    {
      "epoch": 0.2473,
      "grad_norm": 1.494587052328523,
      "learning_rate": 0.003,
      "loss": 4.026,
      "step": 24730
    },
    {
      "epoch": 0.24731,
      "grad_norm": 1.463582765087985,
      "learning_rate": 0.003,
      "loss": 4.0754,
      "step": 24731
    },
    {
      "epoch": 0.24732,
      "grad_norm": 1.5693980068020192,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 24732
    },
    {
      "epoch": 0.24733,
      "grad_norm": 1.1560774062930037,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 24733
    },
    {
      "epoch": 0.24734,
      "grad_norm": 1.2579879998812238,
      "learning_rate": 0.003,
      "loss": 4.0104,
      "step": 24734
    },
    {
      "epoch": 0.24735,
      "grad_norm": 1.6737570852668278,
      "learning_rate": 0.003,
      "loss": 4.0745,
      "step": 24735
    },
    {
      "epoch": 0.24736,
      "grad_norm": 1.422753624810444,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 24736
    },
    {
      "epoch": 0.24737,
      "grad_norm": 1.3081567642689587,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 24737
    },
    {
      "epoch": 0.24738,
      "grad_norm": 1.4681337506181662,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 24738
    },
    {
      "epoch": 0.24739,
      "grad_norm": 1.353013517803222,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 24739
    },
    {
      "epoch": 0.2474,
      "grad_norm": 1.5044503178582012,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 24740
    },
    {
      "epoch": 0.24741,
      "grad_norm": 1.2147016615036763,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 24741
    },
    {
      "epoch": 0.24742,
      "grad_norm": 1.3548029930576675,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 24742
    },
    {
      "epoch": 0.24743,
      "grad_norm": 1.1736661393469732,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 24743
    },
    {
      "epoch": 0.24744,
      "grad_norm": 1.523797189567375,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 24744
    },
    {
      "epoch": 0.24745,
      "grad_norm": 1.4938887946099033,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 24745
    },
    {
      "epoch": 0.24746,
      "grad_norm": 1.6129495765344535,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 24746
    },
    {
      "epoch": 0.24747,
      "grad_norm": 1.5465153082862477,
      "learning_rate": 0.003,
      "loss": 4.0749,
      "step": 24747
    },
    {
      "epoch": 0.24748,
      "grad_norm": 1.3470657073245074,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 24748
    },
    {
      "epoch": 0.24749,
      "grad_norm": 1.4144434361944425,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 24749
    },
    {
      "epoch": 0.2475,
      "grad_norm": 1.6091144458303441,
      "learning_rate": 0.003,
      "loss": 4.026,
      "step": 24750
    },
    {
      "epoch": 0.24751,
      "grad_norm": 1.151243467795758,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 24751
    },
    {
      "epoch": 0.24752,
      "grad_norm": 1.487955722009135,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 24752
    },
    {
      "epoch": 0.24753,
      "grad_norm": 1.3489445381704375,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 24753
    },
    {
      "epoch": 0.24754,
      "grad_norm": 1.4814548584561162,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 24754
    },
    {
      "epoch": 0.24755,
      "grad_norm": 1.3338594890886353,
      "learning_rate": 0.003,
      "loss": 4.0039,
      "step": 24755
    },
    {
      "epoch": 0.24756,
      "grad_norm": 1.3463998506375616,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 24756
    },
    {
      "epoch": 0.24757,
      "grad_norm": 1.4057127181307358,
      "learning_rate": 0.003,
      "loss": 4.0278,
      "step": 24757
    },
    {
      "epoch": 0.24758,
      "grad_norm": 1.3194816805840566,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 24758
    },
    {
      "epoch": 0.24759,
      "grad_norm": 1.3604347150425697,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 24759
    },
    {
      "epoch": 0.2476,
      "grad_norm": 1.4327687092121575,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 24760
    },
    {
      "epoch": 0.24761,
      "grad_norm": 1.2454851239598583,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 24761
    },
    {
      "epoch": 0.24762,
      "grad_norm": 1.304836654064837,
      "learning_rate": 0.003,
      "loss": 4.0792,
      "step": 24762
    },
    {
      "epoch": 0.24763,
      "grad_norm": 1.4698849853851503,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 24763
    },
    {
      "epoch": 0.24764,
      "grad_norm": 1.331294383622434,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 24764
    },
    {
      "epoch": 0.24765,
      "grad_norm": 1.541256627652232,
      "learning_rate": 0.003,
      "loss": 4.0503,
      "step": 24765
    },
    {
      "epoch": 0.24766,
      "grad_norm": 1.577679557555772,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 24766
    },
    {
      "epoch": 0.24767,
      "grad_norm": 1.1012168018809079,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 24767
    },
    {
      "epoch": 0.24768,
      "grad_norm": 1.5592017317726659,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 24768
    },
    {
      "epoch": 0.24769,
      "grad_norm": 1.2909868546716334,
      "learning_rate": 0.003,
      "loss": 4.028,
      "step": 24769
    },
    {
      "epoch": 0.2477,
      "grad_norm": 1.4914432464569773,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 24770
    },
    {
      "epoch": 0.24771,
      "grad_norm": 1.3316743496743026,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 24771
    },
    {
      "epoch": 0.24772,
      "grad_norm": 1.4832628542594846,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 24772
    },
    {
      "epoch": 0.24773,
      "grad_norm": 1.3558204825540894,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 24773
    },
    {
      "epoch": 0.24774,
      "grad_norm": 1.3845284802193545,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 24774
    },
    {
      "epoch": 0.24775,
      "grad_norm": 1.6918078209206067,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 24775
    },
    {
      "epoch": 0.24776,
      "grad_norm": 1.218607779578139,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 24776
    },
    {
      "epoch": 0.24777,
      "grad_norm": 1.5667096008125976,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 24777
    },
    {
      "epoch": 0.24778,
      "grad_norm": 1.2915817404671388,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 24778
    },
    {
      "epoch": 0.24779,
      "grad_norm": 1.35905369335716,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 24779
    },
    {
      "epoch": 0.2478,
      "grad_norm": 1.403991505646697,
      "learning_rate": 0.003,
      "loss": 4.065,
      "step": 24780
    },
    {
      "epoch": 0.24781,
      "grad_norm": 1.3525827146462097,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 24781
    },
    {
      "epoch": 0.24782,
      "grad_norm": 1.3993274299502594,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 24782
    },
    {
      "epoch": 0.24783,
      "grad_norm": 1.3927380904293887,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 24783
    },
    {
      "epoch": 0.24784,
      "grad_norm": 1.4019995753607755,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 24784
    },
    {
      "epoch": 0.24785,
      "grad_norm": 1.456500031497397,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 24785
    },
    {
      "epoch": 0.24786,
      "grad_norm": 1.3937884265844593,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 24786
    },
    {
      "epoch": 0.24787,
      "grad_norm": 1.499569326172104,
      "learning_rate": 0.003,
      "loss": 4.068,
      "step": 24787
    },
    {
      "epoch": 0.24788,
      "grad_norm": 1.2575954064825454,
      "learning_rate": 0.003,
      "loss": 4.0434,
      "step": 24788
    },
    {
      "epoch": 0.24789,
      "grad_norm": 1.49778299411736,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 24789
    },
    {
      "epoch": 0.2479,
      "grad_norm": 1.3586828791251255,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 24790
    },
    {
      "epoch": 0.24791,
      "grad_norm": 1.3173905183874424,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 24791
    },
    {
      "epoch": 0.24792,
      "grad_norm": 1.5420085687784748,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 24792
    },
    {
      "epoch": 0.24793,
      "grad_norm": 1.959099866201587,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 24793
    },
    {
      "epoch": 0.24794,
      "grad_norm": 1.2520886598351242,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 24794
    },
    {
      "epoch": 0.24795,
      "grad_norm": 1.370389659314976,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 24795
    },
    {
      "epoch": 0.24796,
      "grad_norm": 1.3983090798660438,
      "learning_rate": 0.003,
      "loss": 4.066,
      "step": 24796
    },
    {
      "epoch": 0.24797,
      "grad_norm": 1.646825361741049,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 24797
    },
    {
      "epoch": 0.24798,
      "grad_norm": 1.2272973826331968,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 24798
    },
    {
      "epoch": 0.24799,
      "grad_norm": 1.466392956421392,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 24799
    },
    {
      "epoch": 0.248,
      "grad_norm": 1.2716286009691005,
      "learning_rate": 0.003,
      "loss": 4.0249,
      "step": 24800
    },
    {
      "epoch": 0.24801,
      "grad_norm": 1.3834781240407417,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 24801
    },
    {
      "epoch": 0.24802,
      "grad_norm": 1.3723594513286763,
      "learning_rate": 0.003,
      "loss": 4.0279,
      "step": 24802
    },
    {
      "epoch": 0.24803,
      "grad_norm": 1.2643603060613744,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 24803
    },
    {
      "epoch": 0.24804,
      "grad_norm": 1.319402698735918,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 24804
    },
    {
      "epoch": 0.24805,
      "grad_norm": 1.5612727928705017,
      "learning_rate": 0.003,
      "loss": 4.0261,
      "step": 24805
    },
    {
      "epoch": 0.24806,
      "grad_norm": 1.2740989483540295,
      "learning_rate": 0.003,
      "loss": 4.015,
      "step": 24806
    },
    {
      "epoch": 0.24807,
      "grad_norm": 1.40594330473953,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 24807
    },
    {
      "epoch": 0.24808,
      "grad_norm": 1.3658506105532042,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 24808
    },
    {
      "epoch": 0.24809,
      "grad_norm": 1.5299280710476353,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 24809
    },
    {
      "epoch": 0.2481,
      "grad_norm": 1.43084151160238,
      "learning_rate": 0.003,
      "loss": 4.0247,
      "step": 24810
    },
    {
      "epoch": 0.24811,
      "grad_norm": 1.41360182111721,
      "learning_rate": 0.003,
      "loss": 4.0396,
      "step": 24811
    },
    {
      "epoch": 0.24812,
      "grad_norm": 1.5230535965569425,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 24812
    },
    {
      "epoch": 0.24813,
      "grad_norm": 1.1047908473832206,
      "learning_rate": 0.003,
      "loss": 4.0309,
      "step": 24813
    },
    {
      "epoch": 0.24814,
      "grad_norm": 1.2984868528247686,
      "learning_rate": 0.003,
      "loss": 4.053,
      "step": 24814
    },
    {
      "epoch": 0.24815,
      "grad_norm": 1.3578828905489941,
      "learning_rate": 0.003,
      "loss": 4.0283,
      "step": 24815
    },
    {
      "epoch": 0.24816,
      "grad_norm": 1.5434596633809294,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 24816
    },
    {
      "epoch": 0.24817,
      "grad_norm": 1.123125374566577,
      "learning_rate": 0.003,
      "loss": 4.0501,
      "step": 24817
    },
    {
      "epoch": 0.24818,
      "grad_norm": 1.7423886550183643,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 24818
    },
    {
      "epoch": 0.24819,
      "grad_norm": 1.2024239741592904,
      "learning_rate": 0.003,
      "loss": 4.0294,
      "step": 24819
    },
    {
      "epoch": 0.2482,
      "grad_norm": 1.4649236260931853,
      "learning_rate": 0.003,
      "loss": 4.0888,
      "step": 24820
    },
    {
      "epoch": 0.24821,
      "grad_norm": 1.2166068174318763,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 24821
    },
    {
      "epoch": 0.24822,
      "grad_norm": 1.4652583334148943,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 24822
    },
    {
      "epoch": 0.24823,
      "grad_norm": 1.585850347497253,
      "learning_rate": 0.003,
      "loss": 4.0708,
      "step": 24823
    },
    {
      "epoch": 0.24824,
      "grad_norm": 1.3916526111881091,
      "learning_rate": 0.003,
      "loss": 4.021,
      "step": 24824
    },
    {
      "epoch": 0.24825,
      "grad_norm": 1.6536254524236857,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 24825
    },
    {
      "epoch": 0.24826,
      "grad_norm": 1.4018207123308375,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 24826
    },
    {
      "epoch": 0.24827,
      "grad_norm": 1.7208651229403833,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 24827
    },
    {
      "epoch": 0.24828,
      "grad_norm": 1.2741941037851814,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 24828
    },
    {
      "epoch": 0.24829,
      "grad_norm": 1.486304567488449,
      "learning_rate": 0.003,
      "loss": 4.0878,
      "step": 24829
    },
    {
      "epoch": 0.2483,
      "grad_norm": 1.4911574450730498,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 24830
    },
    {
      "epoch": 0.24831,
      "grad_norm": 1.2176633831592192,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 24831
    },
    {
      "epoch": 0.24832,
      "grad_norm": 1.3517968058011713,
      "learning_rate": 0.003,
      "loss": 4.0429,
      "step": 24832
    },
    {
      "epoch": 0.24833,
      "grad_norm": 1.301456914131885,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 24833
    },
    {
      "epoch": 0.24834,
      "grad_norm": 1.2259073306915889,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 24834
    },
    {
      "epoch": 0.24835,
      "grad_norm": 1.4886572650448724,
      "learning_rate": 0.003,
      "loss": 4.096,
      "step": 24835
    },
    {
      "epoch": 0.24836,
      "grad_norm": 1.5991193231226448,
      "learning_rate": 0.003,
      "loss": 4.0734,
      "step": 24836
    },
    {
      "epoch": 0.24837,
      "grad_norm": 1.172590946791768,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 24837
    },
    {
      "epoch": 0.24838,
      "grad_norm": 1.6931173646800393,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 24838
    },
    {
      "epoch": 0.24839,
      "grad_norm": 1.251729650022794,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 24839
    },
    {
      "epoch": 0.2484,
      "grad_norm": 1.8671057562896713,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 24840
    },
    {
      "epoch": 0.24841,
      "grad_norm": 1.1413728903592155,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 24841
    },
    {
      "epoch": 0.24842,
      "grad_norm": 1.4479764225510083,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 24842
    },
    {
      "epoch": 0.24843,
      "grad_norm": 1.473154459160328,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 24843
    },
    {
      "epoch": 0.24844,
      "grad_norm": 1.3335312756841913,
      "learning_rate": 0.003,
      "loss": 4.0653,
      "step": 24844
    },
    {
      "epoch": 0.24845,
      "grad_norm": 1.3417832457179788,
      "learning_rate": 0.003,
      "loss": 4.0674,
      "step": 24845
    },
    {
      "epoch": 0.24846,
      "grad_norm": 1.5881214403623038,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 24846
    },
    {
      "epoch": 0.24847,
      "grad_norm": 0.9745628327870453,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 24847
    },
    {
      "epoch": 0.24848,
      "grad_norm": 1.5852238210899334,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 24848
    },
    {
      "epoch": 0.24849,
      "grad_norm": 1.3723568450317898,
      "learning_rate": 0.003,
      "loss": 4.0831,
      "step": 24849
    },
    {
      "epoch": 0.2485,
      "grad_norm": 1.2900442154879015,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 24850
    },
    {
      "epoch": 0.24851,
      "grad_norm": 1.3980421100565765,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 24851
    },
    {
      "epoch": 0.24852,
      "grad_norm": 1.3663843712744006,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 24852
    },
    {
      "epoch": 0.24853,
      "grad_norm": 1.282759220882982,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 24853
    },
    {
      "epoch": 0.24854,
      "grad_norm": 1.4978189660133514,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 24854
    },
    {
      "epoch": 0.24855,
      "grad_norm": 1.2896105125080048,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 24855
    },
    {
      "epoch": 0.24856,
      "grad_norm": 1.402952601496086,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 24856
    },
    {
      "epoch": 0.24857,
      "grad_norm": 1.2786785021417044,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 24857
    },
    {
      "epoch": 0.24858,
      "grad_norm": 1.5481757204827589,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 24858
    },
    {
      "epoch": 0.24859,
      "grad_norm": 1.351570389855419,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 24859
    },
    {
      "epoch": 0.2486,
      "grad_norm": 1.397430048986852,
      "learning_rate": 0.003,
      "loss": 4.0636,
      "step": 24860
    },
    {
      "epoch": 0.24861,
      "grad_norm": 1.3551863557561121,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 24861
    },
    {
      "epoch": 0.24862,
      "grad_norm": 1.4185473871724321,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 24862
    },
    {
      "epoch": 0.24863,
      "grad_norm": 1.0862842495971277,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 24863
    },
    {
      "epoch": 0.24864,
      "grad_norm": 1.6818731455903855,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 24864
    },
    {
      "epoch": 0.24865,
      "grad_norm": 1.2511071361237098,
      "learning_rate": 0.003,
      "loss": 4.0491,
      "step": 24865
    },
    {
      "epoch": 0.24866,
      "grad_norm": 1.4338302577905506,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 24866
    },
    {
      "epoch": 0.24867,
      "grad_norm": 1.4837660041330187,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 24867
    },
    {
      "epoch": 0.24868,
      "grad_norm": 1.3639552229035876,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 24868
    },
    {
      "epoch": 0.24869,
      "grad_norm": 1.6062538649364084,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 24869
    },
    {
      "epoch": 0.2487,
      "grad_norm": 1.306422248173135,
      "learning_rate": 0.003,
      "loss": 4.075,
      "step": 24870
    },
    {
      "epoch": 0.24871,
      "grad_norm": 1.2408179046414147,
      "learning_rate": 0.003,
      "loss": 4.0548,
      "step": 24871
    },
    {
      "epoch": 0.24872,
      "grad_norm": 1.3788783268952154,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 24872
    },
    {
      "epoch": 0.24873,
      "grad_norm": 1.1628844349626586,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 24873
    },
    {
      "epoch": 0.24874,
      "grad_norm": 1.4655332628857765,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 24874
    },
    {
      "epoch": 0.24875,
      "grad_norm": 1.384914918385266,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 24875
    },
    {
      "epoch": 0.24876,
      "grad_norm": 1.4130782634140597,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 24876
    },
    {
      "epoch": 0.24877,
      "grad_norm": 1.355909140788544,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 24877
    },
    {
      "epoch": 0.24878,
      "grad_norm": 1.1676441454980664,
      "learning_rate": 0.003,
      "loss": 4.0271,
      "step": 24878
    },
    {
      "epoch": 0.24879,
      "grad_norm": 1.5883908599015018,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 24879
    },
    {
      "epoch": 0.2488,
      "grad_norm": 1.3787898491179174,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 24880
    },
    {
      "epoch": 0.24881,
      "grad_norm": 1.5499210880341256,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 24881
    },
    {
      "epoch": 0.24882,
      "grad_norm": 1.3268620001276943,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 24882
    },
    {
      "epoch": 0.24883,
      "grad_norm": 1.3964274047136032,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 24883
    },
    {
      "epoch": 0.24884,
      "grad_norm": 1.6682214680093097,
      "learning_rate": 0.003,
      "loss": 4.0795,
      "step": 24884
    },
    {
      "epoch": 0.24885,
      "grad_norm": 1.1521562131146341,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 24885
    },
    {
      "epoch": 0.24886,
      "grad_norm": 1.6896948860704655,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 24886
    },
    {
      "epoch": 0.24887,
      "grad_norm": 1.1755749713396866,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 24887
    },
    {
      "epoch": 0.24888,
      "grad_norm": 1.3413675390733668,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 24888
    },
    {
      "epoch": 0.24889,
      "grad_norm": 1.416136348657915,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 24889
    },
    {
      "epoch": 0.2489,
      "grad_norm": 1.225528924865507,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 24890
    },
    {
      "epoch": 0.24891,
      "grad_norm": 2.094430239331432,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 24891
    },
    {
      "epoch": 0.24892,
      "grad_norm": 1.0868170182591497,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 24892
    },
    {
      "epoch": 0.24893,
      "grad_norm": 1.5105464776711766,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 24893
    },
    {
      "epoch": 0.24894,
      "grad_norm": 1.2482318290860162,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 24894
    },
    {
      "epoch": 0.24895,
      "grad_norm": 1.5186923811710138,
      "learning_rate": 0.003,
      "loss": 4.0733,
      "step": 24895
    },
    {
      "epoch": 0.24896,
      "grad_norm": 1.309405107142853,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 24896
    },
    {
      "epoch": 0.24897,
      "grad_norm": 1.5169604372824392,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 24897
    },
    {
      "epoch": 0.24898,
      "grad_norm": 1.5921853381062598,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 24898
    },
    {
      "epoch": 0.24899,
      "grad_norm": 1.4648493377128722,
      "learning_rate": 0.003,
      "loss": 4.0453,
      "step": 24899
    },
    {
      "epoch": 0.249,
      "grad_norm": 1.2439150950204947,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 24900
    },
    {
      "epoch": 0.24901,
      "grad_norm": 1.74117714099156,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 24901
    },
    {
      "epoch": 0.24902,
      "grad_norm": 1.3645948388309788,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 24902
    },
    {
      "epoch": 0.24903,
      "grad_norm": 1.227528824834206,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 24903
    },
    {
      "epoch": 0.24904,
      "grad_norm": 1.4694408437430944,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 24904
    },
    {
      "epoch": 0.24905,
      "grad_norm": 1.6293815404863183,
      "learning_rate": 0.003,
      "loss": 4.0797,
      "step": 24905
    },
    {
      "epoch": 0.24906,
      "grad_norm": 1.229020343389747,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 24906
    },
    {
      "epoch": 0.24907,
      "grad_norm": 1.3833909364053913,
      "learning_rate": 0.003,
      "loss": 4.0439,
      "step": 24907
    },
    {
      "epoch": 0.24908,
      "grad_norm": 1.5717682911050141,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 24908
    },
    {
      "epoch": 0.24909,
      "grad_norm": 1.3164742775962954,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 24909
    },
    {
      "epoch": 0.2491,
      "grad_norm": 1.5423291639592938,
      "learning_rate": 0.003,
      "loss": 4.0428,
      "step": 24910
    },
    {
      "epoch": 0.24911,
      "grad_norm": 1.286871212691636,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 24911
    },
    {
      "epoch": 0.24912,
      "grad_norm": 1.5492135276404744,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 24912
    },
    {
      "epoch": 0.24913,
      "grad_norm": 1.3480728891768714,
      "learning_rate": 0.003,
      "loss": 4.0596,
      "step": 24913
    },
    {
      "epoch": 0.24914,
      "grad_norm": 1.4481688106902018,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 24914
    },
    {
      "epoch": 0.24915,
      "grad_norm": 1.3349488120589184,
      "learning_rate": 0.003,
      "loss": 4.0835,
      "step": 24915
    },
    {
      "epoch": 0.24916,
      "grad_norm": 1.3652534876243616,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 24916
    },
    {
      "epoch": 0.24917,
      "grad_norm": 1.5501035773819616,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 24917
    },
    {
      "epoch": 0.24918,
      "grad_norm": 1.1708953688393553,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 24918
    },
    {
      "epoch": 0.24919,
      "grad_norm": 1.4982117647734141,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 24919
    },
    {
      "epoch": 0.2492,
      "grad_norm": 0.9986467472437912,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 24920
    },
    {
      "epoch": 0.24921,
      "grad_norm": 1.6723162577163804,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 24921
    },
    {
      "epoch": 0.24922,
      "grad_norm": 1.1854858068943663,
      "learning_rate": 0.003,
      "loss": 4.0511,
      "step": 24922
    },
    {
      "epoch": 0.24923,
      "grad_norm": 1.5399502472417759,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 24923
    },
    {
      "epoch": 0.24924,
      "grad_norm": 1.3444110408533056,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 24924
    },
    {
      "epoch": 0.24925,
      "grad_norm": 1.370379849328035,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 24925
    },
    {
      "epoch": 0.24926,
      "grad_norm": 1.1605349139968573,
      "learning_rate": 0.003,
      "loss": 4.051,
      "step": 24926
    },
    {
      "epoch": 0.24927,
      "grad_norm": 1.3521096311721168,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 24927
    },
    {
      "epoch": 0.24928,
      "grad_norm": 1.2241155366659324,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 24928
    },
    {
      "epoch": 0.24929,
      "grad_norm": 1.674015221783975,
      "learning_rate": 0.003,
      "loss": 4.0844,
      "step": 24929
    },
    {
      "epoch": 0.2493,
      "grad_norm": 1.3273645716043994,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 24930
    },
    {
      "epoch": 0.24931,
      "grad_norm": 1.42044499953458,
      "learning_rate": 0.003,
      "loss": 4.0879,
      "step": 24931
    },
    {
      "epoch": 0.24932,
      "grad_norm": 1.4143382443837755,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 24932
    },
    {
      "epoch": 0.24933,
      "grad_norm": 1.7790182925022033,
      "learning_rate": 0.003,
      "loss": 4.0796,
      "step": 24933
    },
    {
      "epoch": 0.24934,
      "grad_norm": 1.2015171015364932,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 24934
    },
    {
      "epoch": 0.24935,
      "grad_norm": 1.6905300146912643,
      "learning_rate": 0.003,
      "loss": 4.0809,
      "step": 24935
    },
    {
      "epoch": 0.24936,
      "grad_norm": 1.5317808660507717,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 24936
    },
    {
      "epoch": 0.24937,
      "grad_norm": 1.510811048729096,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 24937
    },
    {
      "epoch": 0.24938,
      "grad_norm": 1.5739336242025765,
      "learning_rate": 0.003,
      "loss": 4.0688,
      "step": 24938
    },
    {
      "epoch": 0.24939,
      "grad_norm": 1.3278074605778212,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 24939
    },
    {
      "epoch": 0.2494,
      "grad_norm": 1.461548147720003,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 24940
    },
    {
      "epoch": 0.24941,
      "grad_norm": 1.0999535839830032,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 24941
    },
    {
      "epoch": 0.24942,
      "grad_norm": 1.4332565291133144,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 24942
    },
    {
      "epoch": 0.24943,
      "grad_norm": 1.515141909449798,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 24943
    },
    {
      "epoch": 0.24944,
      "grad_norm": 1.5380607304798846,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 24944
    },
    {
      "epoch": 0.24945,
      "grad_norm": 1.3992941597860196,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 24945
    },
    {
      "epoch": 0.24946,
      "grad_norm": 1.4545334976515218,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 24946
    },
    {
      "epoch": 0.24947,
      "grad_norm": 1.2777979900092966,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 24947
    },
    {
      "epoch": 0.24948,
      "grad_norm": 1.6710438025633039,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 24948
    },
    {
      "epoch": 0.24949,
      "grad_norm": 1.3390451864209725,
      "learning_rate": 0.003,
      "loss": 4.0298,
      "step": 24949
    },
    {
      "epoch": 0.2495,
      "grad_norm": 1.4898233288977503,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 24950
    },
    {
      "epoch": 0.24951,
      "grad_norm": 1.2959355284493708,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 24951
    },
    {
      "epoch": 0.24952,
      "grad_norm": 1.413154750131716,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 24952
    },
    {
      "epoch": 0.24953,
      "grad_norm": 1.3960260011173238,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 24953
    },
    {
      "epoch": 0.24954,
      "grad_norm": 1.1679144432834494,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 24954
    },
    {
      "epoch": 0.24955,
      "grad_norm": 1.557044717174055,
      "learning_rate": 0.003,
      "loss": 4.0744,
      "step": 24955
    },
    {
      "epoch": 0.24956,
      "grad_norm": 1.3195799815444853,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 24956
    },
    {
      "epoch": 0.24957,
      "grad_norm": 1.2553260762254992,
      "learning_rate": 0.003,
      "loss": 4.0805,
      "step": 24957
    },
    {
      "epoch": 0.24958,
      "grad_norm": 1.3083190563309135,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 24958
    },
    {
      "epoch": 0.24959,
      "grad_norm": 1.7430249716467296,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 24959
    },
    {
      "epoch": 0.2496,
      "grad_norm": 1.2486684097059337,
      "learning_rate": 0.003,
      "loss": 4.059,
      "step": 24960
    },
    {
      "epoch": 0.24961,
      "grad_norm": 1.462288326704534,
      "learning_rate": 0.003,
      "loss": 4.0315,
      "step": 24961
    },
    {
      "epoch": 0.24962,
      "grad_norm": 1.6236630141571975,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 24962
    },
    {
      "epoch": 0.24963,
      "grad_norm": 1.1187082060403732,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 24963
    },
    {
      "epoch": 0.24964,
      "grad_norm": 1.5318702321218611,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 24964
    },
    {
      "epoch": 0.24965,
      "grad_norm": 1.3332413918668171,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 24965
    },
    {
      "epoch": 0.24966,
      "grad_norm": 1.1564175191270192,
      "learning_rate": 0.003,
      "loss": 4.0685,
      "step": 24966
    },
    {
      "epoch": 0.24967,
      "grad_norm": 1.428895740734984,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 24967
    },
    {
      "epoch": 0.24968,
      "grad_norm": 1.2742530340491622,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 24968
    },
    {
      "epoch": 0.24969,
      "grad_norm": 1.541860631267734,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 24969
    },
    {
      "epoch": 0.2497,
      "grad_norm": 1.2837649030426541,
      "learning_rate": 0.003,
      "loss": 4.0306,
      "step": 24970
    },
    {
      "epoch": 0.24971,
      "grad_norm": 1.4671433508690153,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 24971
    },
    {
      "epoch": 0.24972,
      "grad_norm": 1.4063292914102563,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 24972
    },
    {
      "epoch": 0.24973,
      "grad_norm": 1.5678528858891951,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 24973
    },
    {
      "epoch": 0.24974,
      "grad_norm": 1.6838405654390767,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 24974
    },
    {
      "epoch": 0.24975,
      "grad_norm": 1.4335037590487514,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 24975
    },
    {
      "epoch": 0.24976,
      "grad_norm": 1.294616511862058,
      "learning_rate": 0.003,
      "loss": 4.0689,
      "step": 24976
    },
    {
      "epoch": 0.24977,
      "grad_norm": 1.477863145139648,
      "learning_rate": 0.003,
      "loss": 4.0567,
      "step": 24977
    },
    {
      "epoch": 0.24978,
      "grad_norm": 1.1322766215325784,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 24978
    },
    {
      "epoch": 0.24979,
      "grad_norm": 1.397481628428317,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 24979
    },
    {
      "epoch": 0.2498,
      "grad_norm": 1.2790383466875175,
      "learning_rate": 0.003,
      "loss": 4.0273,
      "step": 24980
    },
    {
      "epoch": 0.24981,
      "grad_norm": 1.4565413286850377,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 24981
    },
    {
      "epoch": 0.24982,
      "grad_norm": 1.3285689190026917,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 24982
    },
    {
      "epoch": 0.24983,
      "grad_norm": 1.5237320103316905,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 24983
    },
    {
      "epoch": 0.24984,
      "grad_norm": 1.4263999868936332,
      "learning_rate": 0.003,
      "loss": 4.0021,
      "step": 24984
    },
    {
      "epoch": 0.24985,
      "grad_norm": 1.5723829723682632,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 24985
    },
    {
      "epoch": 0.24986,
      "grad_norm": 1.3118673213997734,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 24986
    },
    {
      "epoch": 0.24987,
      "grad_norm": 1.291787280514421,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 24987
    },
    {
      "epoch": 0.24988,
      "grad_norm": 1.7411938909897147,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 24988
    },
    {
      "epoch": 0.24989,
      "grad_norm": 1.1754841604041872,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 24989
    },
    {
      "epoch": 0.2499,
      "grad_norm": 1.5195210592200663,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 24990
    },
    {
      "epoch": 0.24991,
      "grad_norm": 1.3669007011543888,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 24991
    },
    {
      "epoch": 0.24992,
      "grad_norm": 1.221543626014081,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 24992
    },
    {
      "epoch": 0.24993,
      "grad_norm": 1.228415036087146,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 24993
    },
    {
      "epoch": 0.24994,
      "grad_norm": 1.4075007937712496,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 24994
    },
    {
      "epoch": 0.24995,
      "grad_norm": 1.1173673761004927,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 24995
    },
    {
      "epoch": 0.24996,
      "grad_norm": 1.5485276760896818,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 24996
    },
    {
      "epoch": 0.24997,
      "grad_norm": 1.1573506280325772,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 24997
    },
    {
      "epoch": 0.24998,
      "grad_norm": 1.448005405494121,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 24998
    },
    {
      "epoch": 0.24999,
      "grad_norm": 1.4040346319688222,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 24999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8838509897983278,
      "learning_rate": 0.003,
      "loss": 4.0714,
      "step": 25000
    },
    {
      "epoch": 0.25001,
      "grad_norm": 1.2748017475324664,
      "learning_rate": 0.003,
      "loss": 4.0288,
      "step": 25001
    },
    {
      "epoch": 0.25002,
      "grad_norm": 1.4801517152524883,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 25002
    },
    {
      "epoch": 0.25003,
      "grad_norm": 1.3330920461551459,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 25003
    },
    {
      "epoch": 0.25004,
      "grad_norm": 1.135724062116385,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 25004
    },
    {
      "epoch": 0.25005,
      "grad_norm": 1.7225988356124133,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 25005
    },
    {
      "epoch": 0.25006,
      "grad_norm": 1.4821539595897881,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 25006
    },
    {
      "epoch": 0.25007,
      "grad_norm": 1.344104848645791,
      "learning_rate": 0.003,
      "loss": 4.0197,
      "step": 25007
    },
    {
      "epoch": 0.25008,
      "grad_norm": 1.2711149140336329,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 25008
    },
    {
      "epoch": 0.25009,
      "grad_norm": 1.8915259742696044,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 25009
    },
    {
      "epoch": 0.2501,
      "grad_norm": 1.1617670539810043,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 25010
    },
    {
      "epoch": 0.25011,
      "grad_norm": 1.4106537485260828,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 25011
    },
    {
      "epoch": 0.25012,
      "grad_norm": 1.6012213409493314,
      "learning_rate": 0.003,
      "loss": 4.0626,
      "step": 25012
    },
    {
      "epoch": 0.25013,
      "grad_norm": 1.1638164507160436,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 25013
    },
    {
      "epoch": 0.25014,
      "grad_norm": 1.4016524508438235,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 25014
    },
    {
      "epoch": 0.25015,
      "grad_norm": 1.2380151210316184,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 25015
    },
    {
      "epoch": 0.25016,
      "grad_norm": 1.5357464557476528,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 25016
    },
    {
      "epoch": 0.25017,
      "grad_norm": 1.538934875163108,
      "learning_rate": 0.003,
      "loss": 4.0785,
      "step": 25017
    },
    {
      "epoch": 0.25018,
      "grad_norm": 1.3479982683605563,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 25018
    },
    {
      "epoch": 0.25019,
      "grad_norm": 1.46247496258233,
      "learning_rate": 0.003,
      "loss": 4.086,
      "step": 25019
    },
    {
      "epoch": 0.2502,
      "grad_norm": 1.1007571816329893,
      "learning_rate": 0.003,
      "loss": 4.0667,
      "step": 25020
    },
    {
      "epoch": 0.25021,
      "grad_norm": 1.4440317923634345,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 25021
    },
    {
      "epoch": 0.25022,
      "grad_norm": 1.6107645955759808,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 25022
    },
    {
      "epoch": 0.25023,
      "grad_norm": 1.2881296338059511,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 25023
    },
    {
      "epoch": 0.25024,
      "grad_norm": 1.4144028382262763,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 25024
    },
    {
      "epoch": 0.25025,
      "grad_norm": 1.5649497075553052,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 25025
    },
    {
      "epoch": 0.25026,
      "grad_norm": 1.1499792536365803,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 25026
    },
    {
      "epoch": 0.25027,
      "grad_norm": 1.436633311360459,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 25027
    },
    {
      "epoch": 0.25028,
      "grad_norm": 1.2893043720731796,
      "learning_rate": 0.003,
      "loss": 4.0462,
      "step": 25028
    },
    {
      "epoch": 0.25029,
      "grad_norm": 1.3135017011424721,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 25029
    },
    {
      "epoch": 0.2503,
      "grad_norm": 1.3907877043148116,
      "learning_rate": 0.003,
      "loss": 4.0483,
      "step": 25030
    },
    {
      "epoch": 0.25031,
      "grad_norm": 1.529935768172165,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 25031
    },
    {
      "epoch": 0.25032,
      "grad_norm": 1.5475090834382632,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 25032
    },
    {
      "epoch": 0.25033,
      "grad_norm": 1.3376815131287971,
      "learning_rate": 0.003,
      "loss": 4.0693,
      "step": 25033
    },
    {
      "epoch": 0.25034,
      "grad_norm": 1.3254297699267132,
      "learning_rate": 0.003,
      "loss": 4.0394,
      "step": 25034
    },
    {
      "epoch": 0.25035,
      "grad_norm": 1.359602359603589,
      "learning_rate": 0.003,
      "loss": 4.0387,
      "step": 25035
    },
    {
      "epoch": 0.25036,
      "grad_norm": 1.1715099649607268,
      "learning_rate": 0.003,
      "loss": 4.0502,
      "step": 25036
    },
    {
      "epoch": 0.25037,
      "grad_norm": 1.5000779841111187,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25037
    },
    {
      "epoch": 0.25038,
      "grad_norm": 1.2164945876141462,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 25038
    },
    {
      "epoch": 0.25039,
      "grad_norm": 1.3229696305792358,
      "learning_rate": 0.003,
      "loss": 4.0235,
      "step": 25039
    },
    {
      "epoch": 0.2504,
      "grad_norm": 1.4793524610030588,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 25040
    },
    {
      "epoch": 0.25041,
      "grad_norm": 1.550371363016915,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 25041
    },
    {
      "epoch": 0.25042,
      "grad_norm": 1.5546323928883719,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 25042
    },
    {
      "epoch": 0.25043,
      "grad_norm": 1.27409114915906,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 25043
    },
    {
      "epoch": 0.25044,
      "grad_norm": 1.6421988949703035,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 25044
    },
    {
      "epoch": 0.25045,
      "grad_norm": 1.207263186027553,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 25045
    },
    {
      "epoch": 0.25046,
      "grad_norm": 1.628015587652915,
      "learning_rate": 0.003,
      "loss": 4.012,
      "step": 25046
    },
    {
      "epoch": 0.25047,
      "grad_norm": 1.2303938396009948,
      "learning_rate": 0.003,
      "loss": 4.0621,
      "step": 25047
    },
    {
      "epoch": 0.25048,
      "grad_norm": 1.2675196435956482,
      "learning_rate": 0.003,
      "loss": 4.0038,
      "step": 25048
    },
    {
      "epoch": 0.25049,
      "grad_norm": 1.449009113313399,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 25049
    },
    {
      "epoch": 0.2505,
      "grad_norm": 1.3391571994921696,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 25050
    },
    {
      "epoch": 0.25051,
      "grad_norm": 1.5504349144109788,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 25051
    },
    {
      "epoch": 0.25052,
      "grad_norm": 1.273381778899984,
      "learning_rate": 0.003,
      "loss": 4.0183,
      "step": 25052
    },
    {
      "epoch": 0.25053,
      "grad_norm": 1.2570499703351385,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 25053
    },
    {
      "epoch": 0.25054,
      "grad_norm": 1.3450530257740156,
      "learning_rate": 0.003,
      "loss": 4.0242,
      "step": 25054
    },
    {
      "epoch": 0.25055,
      "grad_norm": 1.3723494824881228,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 25055
    },
    {
      "epoch": 0.25056,
      "grad_norm": 1.5026395755157202,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 25056
    },
    {
      "epoch": 0.25057,
      "grad_norm": 1.3738937049339626,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 25057
    },
    {
      "epoch": 0.25058,
      "grad_norm": 1.6493501919757914,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 25058
    },
    {
      "epoch": 0.25059,
      "grad_norm": 2.067010137092943,
      "learning_rate": 0.003,
      "loss": 4.0272,
      "step": 25059
    },
    {
      "epoch": 0.2506,
      "grad_norm": 1.332216595669157,
      "learning_rate": 0.003,
      "loss": 4.0402,
      "step": 25060
    },
    {
      "epoch": 0.25061,
      "grad_norm": 1.2092073858482,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 25061
    },
    {
      "epoch": 0.25062,
      "grad_norm": 1.3714865804427254,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 25062
    },
    {
      "epoch": 0.25063,
      "grad_norm": 1.3298674353430395,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 25063
    },
    {
      "epoch": 0.25064,
      "grad_norm": 1.4359353365071301,
      "learning_rate": 0.003,
      "loss": 4.0662,
      "step": 25064
    },
    {
      "epoch": 0.25065,
      "grad_norm": 1.471839901228785,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 25065
    },
    {
      "epoch": 0.25066,
      "grad_norm": 1.406811254447624,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 25066
    },
    {
      "epoch": 0.25067,
      "grad_norm": 1.384814010059198,
      "learning_rate": 0.003,
      "loss": 4.0677,
      "step": 25067
    },
    {
      "epoch": 0.25068,
      "grad_norm": 1.2837779920048389,
      "learning_rate": 0.003,
      "loss": 4.0101,
      "step": 25068
    },
    {
      "epoch": 0.25069,
      "grad_norm": 1.3865271012867195,
      "learning_rate": 0.003,
      "loss": 4.0615,
      "step": 25069
    },
    {
      "epoch": 0.2507,
      "grad_norm": 1.5067677462514886,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 25070
    },
    {
      "epoch": 0.25071,
      "grad_norm": 1.2625014372525856,
      "learning_rate": 0.003,
      "loss": 4.0277,
      "step": 25071
    },
    {
      "epoch": 0.25072,
      "grad_norm": 1.6130587445811615,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 25072
    },
    {
      "epoch": 0.25073,
      "grad_norm": 1.2906004924780379,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 25073
    },
    {
      "epoch": 0.25074,
      "grad_norm": 1.3010535286702674,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 25074
    },
    {
      "epoch": 0.25075,
      "grad_norm": 1.3461636181166996,
      "learning_rate": 0.003,
      "loss": 4.0041,
      "step": 25075
    },
    {
      "epoch": 0.25076,
      "grad_norm": 1.5663004864494208,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 25076
    },
    {
      "epoch": 0.25077,
      "grad_norm": 1.2892589301736825,
      "learning_rate": 0.003,
      "loss": 4.0333,
      "step": 25077
    },
    {
      "epoch": 0.25078,
      "grad_norm": 1.2352698688091392,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 25078
    },
    {
      "epoch": 0.25079,
      "grad_norm": 1.3442533519893591,
      "learning_rate": 0.003,
      "loss": 4.0224,
      "step": 25079
    },
    {
      "epoch": 0.2508,
      "grad_norm": 1.4542872670051366,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 25080
    },
    {
      "epoch": 0.25081,
      "grad_norm": 1.2801046803551044,
      "learning_rate": 0.003,
      "loss": 4.0243,
      "step": 25081
    },
    {
      "epoch": 0.25082,
      "grad_norm": 1.420217542975841,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 25082
    },
    {
      "epoch": 0.25083,
      "grad_norm": 1.7141721616532908,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 25083
    },
    {
      "epoch": 0.25084,
      "grad_norm": 1.331859307732692,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 25084
    },
    {
      "epoch": 0.25085,
      "grad_norm": 1.2380851877688794,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 25085
    },
    {
      "epoch": 0.25086,
      "grad_norm": 1.9382257193389576,
      "learning_rate": 0.003,
      "loss": 4.0556,
      "step": 25086
    },
    {
      "epoch": 0.25087,
      "grad_norm": 0.9514925501712712,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 25087
    },
    {
      "epoch": 0.25088,
      "grad_norm": 1.5746573300504183,
      "learning_rate": 0.003,
      "loss": 4.0789,
      "step": 25088
    },
    {
      "epoch": 0.25089,
      "grad_norm": 1.1497653308679063,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 25089
    },
    {
      "epoch": 0.2509,
      "grad_norm": 1.4877931758857417,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25090
    },
    {
      "epoch": 0.25091,
      "grad_norm": 1.5172605757609134,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 25091
    },
    {
      "epoch": 0.25092,
      "grad_norm": 1.2571439915881732,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 25092
    },
    {
      "epoch": 0.25093,
      "grad_norm": 1.6019225827112673,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 25093
    },
    {
      "epoch": 0.25094,
      "grad_norm": 1.3508186049678066,
      "learning_rate": 0.003,
      "loss": 4.0319,
      "step": 25094
    },
    {
      "epoch": 0.25095,
      "grad_norm": 1.2758583196980002,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 25095
    },
    {
      "epoch": 0.25096,
      "grad_norm": 1.4366718068242452,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 25096
    },
    {
      "epoch": 0.25097,
      "grad_norm": 1.6128743785652717,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 25097
    },
    {
      "epoch": 0.25098,
      "grad_norm": 1.421883497722055,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 25098
    },
    {
      "epoch": 0.25099,
      "grad_norm": 1.5359042444124875,
      "learning_rate": 0.003,
      "loss": 4.0124,
      "step": 25099
    },
    {
      "epoch": 0.251,
      "grad_norm": 1.5560587269101052,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 25100
    },
    {
      "epoch": 0.25101,
      "grad_norm": 1.1870444886318026,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 25101
    },
    {
      "epoch": 0.25102,
      "grad_norm": 1.5022136824037668,
      "learning_rate": 0.003,
      "loss": 4.012,
      "step": 25102
    },
    {
      "epoch": 0.25103,
      "grad_norm": 1.2523542449419982,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 25103
    },
    {
      "epoch": 0.25104,
      "grad_norm": 1.6053678501454782,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 25104
    },
    {
      "epoch": 0.25105,
      "grad_norm": 1.1830984434372966,
      "learning_rate": 0.003,
      "loss": 4.0641,
      "step": 25105
    },
    {
      "epoch": 0.25106,
      "grad_norm": 1.7028503508424542,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 25106
    },
    {
      "epoch": 0.25107,
      "grad_norm": 1.276465655350452,
      "learning_rate": 0.003,
      "loss": 4.0153,
      "step": 25107
    },
    {
      "epoch": 0.25108,
      "grad_norm": 1.642978517199037,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 25108
    },
    {
      "epoch": 0.25109,
      "grad_norm": 1.3285579061365733,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 25109
    },
    {
      "epoch": 0.2511,
      "grad_norm": 1.5164706875427652,
      "learning_rate": 0.003,
      "loss": 4.0722,
      "step": 25110
    },
    {
      "epoch": 0.25111,
      "grad_norm": 1.2564825677002043,
      "learning_rate": 0.003,
      "loss": 4.0352,
      "step": 25111
    },
    {
      "epoch": 0.25112,
      "grad_norm": 1.3576272139447902,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 25112
    },
    {
      "epoch": 0.25113,
      "grad_norm": 1.4842960772650207,
      "learning_rate": 0.003,
      "loss": 4.0856,
      "step": 25113
    },
    {
      "epoch": 0.25114,
      "grad_norm": 1.5729813113545792,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 25114
    },
    {
      "epoch": 0.25115,
      "grad_norm": 1.201038438127107,
      "learning_rate": 0.003,
      "loss": 4.025,
      "step": 25115
    },
    {
      "epoch": 0.25116,
      "grad_norm": 1.3821215330755574,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 25116
    },
    {
      "epoch": 0.25117,
      "grad_norm": 1.361043474531906,
      "learning_rate": 0.003,
      "loss": 4.0657,
      "step": 25117
    },
    {
      "epoch": 0.25118,
      "grad_norm": 1.429792382444159,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 25118
    },
    {
      "epoch": 0.25119,
      "grad_norm": 1.2984450847415157,
      "learning_rate": 0.003,
      "loss": 4.0052,
      "step": 25119
    },
    {
      "epoch": 0.2512,
      "grad_norm": 1.251387180293601,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25120
    },
    {
      "epoch": 0.25121,
      "grad_norm": 1.6212657568444315,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 25121
    },
    {
      "epoch": 0.25122,
      "grad_norm": 1.101721795296369,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 25122
    },
    {
      "epoch": 0.25123,
      "grad_norm": 1.7175279157835124,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 25123
    },
    {
      "epoch": 0.25124,
      "grad_norm": 1.0755713648383642,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 25124
    },
    {
      "epoch": 0.25125,
      "grad_norm": 1.7882614117403746,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 25125
    },
    {
      "epoch": 0.25126,
      "grad_norm": 1.436601426219007,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 25126
    },
    {
      "epoch": 0.25127,
      "grad_norm": 1.4731193208094815,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 25127
    },
    {
      "epoch": 0.25128,
      "grad_norm": 1.8291166228073243,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 25128
    },
    {
      "epoch": 0.25129,
      "grad_norm": 1.2940235491169902,
      "learning_rate": 0.003,
      "loss": 4.0515,
      "step": 25129
    },
    {
      "epoch": 0.2513,
      "grad_norm": 1.401892583160338,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 25130
    },
    {
      "epoch": 0.25131,
      "grad_norm": 1.5127252425077868,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 25131
    },
    {
      "epoch": 0.25132,
      "grad_norm": 1.3940964706079149,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 25132
    },
    {
      "epoch": 0.25133,
      "grad_norm": 1.357102229005622,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 25133
    },
    {
      "epoch": 0.25134,
      "grad_norm": 1.4684367662800373,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25134
    },
    {
      "epoch": 0.25135,
      "grad_norm": 1.4236044972482442,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 25135
    },
    {
      "epoch": 0.25136,
      "grad_norm": 1.2394641907887927,
      "learning_rate": 0.003,
      "loss": 4.012,
      "step": 25136
    },
    {
      "epoch": 0.25137,
      "grad_norm": 1.2520251277594996,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 25137
    },
    {
      "epoch": 0.25138,
      "grad_norm": 1.3601293011990994,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 25138
    },
    {
      "epoch": 0.25139,
      "grad_norm": 1.3660228536740826,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 25139
    },
    {
      "epoch": 0.2514,
      "grad_norm": 1.4537555643947053,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 25140
    },
    {
      "epoch": 0.25141,
      "grad_norm": 1.5232967625090827,
      "learning_rate": 0.003,
      "loss": 4.0464,
      "step": 25141
    },
    {
      "epoch": 0.25142,
      "grad_norm": 1.4963666099124064,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 25142
    },
    {
      "epoch": 0.25143,
      "grad_norm": 1.4013324726574117,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 25143
    },
    {
      "epoch": 0.25144,
      "grad_norm": 1.3046829409451521,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 25144
    },
    {
      "epoch": 0.25145,
      "grad_norm": 1.537355149465918,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 25145
    },
    {
      "epoch": 0.25146,
      "grad_norm": 1.0734506048543841,
      "learning_rate": 0.003,
      "loss": 4.0241,
      "step": 25146
    },
    {
      "epoch": 0.25147,
      "grad_norm": 1.6936472664094977,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 25147
    },
    {
      "epoch": 0.25148,
      "grad_norm": 1.184668133259422,
      "learning_rate": 0.003,
      "loss": 4.0682,
      "step": 25148
    },
    {
      "epoch": 0.25149,
      "grad_norm": 1.6557792673181408,
      "learning_rate": 0.003,
      "loss": 4.0245,
      "step": 25149
    },
    {
      "epoch": 0.2515,
      "grad_norm": 1.2444052042044118,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 25150
    },
    {
      "epoch": 0.25151,
      "grad_norm": 1.7276337200454241,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25151
    },
    {
      "epoch": 0.25152,
      "grad_norm": 1.134267436836086,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 25152
    },
    {
      "epoch": 0.25153,
      "grad_norm": 1.4210589908143996,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 25153
    },
    {
      "epoch": 0.25154,
      "grad_norm": 1.3573451675136259,
      "learning_rate": 0.003,
      "loss": 4.0866,
      "step": 25154
    },
    {
      "epoch": 0.25155,
      "grad_norm": 1.5440684094347121,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 25155
    },
    {
      "epoch": 0.25156,
      "grad_norm": 1.4149666272875177,
      "learning_rate": 0.003,
      "loss": 4.0295,
      "step": 25156
    },
    {
      "epoch": 0.25157,
      "grad_norm": 1.516900080469773,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 25157
    },
    {
      "epoch": 0.25158,
      "grad_norm": 1.439199693827732,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 25158
    },
    {
      "epoch": 0.25159,
      "grad_norm": 1.1999588454352765,
      "learning_rate": 0.003,
      "loss": 4.0456,
      "step": 25159
    },
    {
      "epoch": 0.2516,
      "grad_norm": 1.637985824999422,
      "learning_rate": 0.003,
      "loss": 4.0369,
      "step": 25160
    },
    {
      "epoch": 0.25161,
      "grad_norm": 1.2078896254049998,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 25161
    },
    {
      "epoch": 0.25162,
      "grad_norm": 1.5540013680699354,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 25162
    },
    {
      "epoch": 0.25163,
      "grad_norm": 1.1133042461942622,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 25163
    },
    {
      "epoch": 0.25164,
      "grad_norm": 1.8615331735865226,
      "learning_rate": 0.003,
      "loss": 4.0613,
      "step": 25164
    },
    {
      "epoch": 0.25165,
      "grad_norm": 1.1747796455899546,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 25165
    },
    {
      "epoch": 0.25166,
      "grad_norm": 1.4022974315076093,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 25166
    },
    {
      "epoch": 0.25167,
      "grad_norm": 1.6004397743087728,
      "learning_rate": 0.003,
      "loss": 4.077,
      "step": 25167
    },
    {
      "epoch": 0.25168,
      "grad_norm": 1.2728967979811856,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 25168
    },
    {
      "epoch": 0.25169,
      "grad_norm": 1.560009772990896,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 25169
    },
    {
      "epoch": 0.2517,
      "grad_norm": 1.202206915278775,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 25170
    },
    {
      "epoch": 0.25171,
      "grad_norm": 1.4360783404367445,
      "learning_rate": 0.003,
      "loss": 4.029,
      "step": 25171
    },
    {
      "epoch": 0.25172,
      "grad_norm": 1.3178475774984317,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 25172
    },
    {
      "epoch": 0.25173,
      "grad_norm": 1.3286607421612373,
      "learning_rate": 0.003,
      "loss": 3.9995,
      "step": 25173
    },
    {
      "epoch": 0.25174,
      "grad_norm": 1.7317664582148886,
      "learning_rate": 0.003,
      "loss": 4.0769,
      "step": 25174
    },
    {
      "epoch": 0.25175,
      "grad_norm": 1.1587860139328252,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 25175
    },
    {
      "epoch": 0.25176,
      "grad_norm": 1.6043646579839161,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 25176
    },
    {
      "epoch": 0.25177,
      "grad_norm": 1.5277343890698323,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 25177
    },
    {
      "epoch": 0.25178,
      "grad_norm": 1.272206851210595,
      "learning_rate": 0.003,
      "loss": 4.0817,
      "step": 25178
    },
    {
      "epoch": 0.25179,
      "grad_norm": 1.643293505702538,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 25179
    },
    {
      "epoch": 0.2518,
      "grad_norm": 0.9575810638322739,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 25180
    },
    {
      "epoch": 0.25181,
      "grad_norm": 1.5991334240527573,
      "learning_rate": 0.003,
      "loss": 4.0373,
      "step": 25181
    },
    {
      "epoch": 0.25182,
      "grad_norm": 1.192800793906798,
      "learning_rate": 0.003,
      "loss": 4.0944,
      "step": 25182
    },
    {
      "epoch": 0.25183,
      "grad_norm": 1.8500253568772047,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 25183
    },
    {
      "epoch": 0.25184,
      "grad_norm": 1.3156410732671433,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 25184
    },
    {
      "epoch": 0.25185,
      "grad_norm": 1.3647590890373258,
      "learning_rate": 0.003,
      "loss": 4.0322,
      "step": 25185
    },
    {
      "epoch": 0.25186,
      "grad_norm": 1.3464265359380858,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 25186
    },
    {
      "epoch": 0.25187,
      "grad_norm": 1.4572536800134575,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 25187
    },
    {
      "epoch": 0.25188,
      "grad_norm": 1.3175917856223025,
      "learning_rate": 0.003,
      "loss": 4.0404,
      "step": 25188
    },
    {
      "epoch": 0.25189,
      "grad_norm": 1.288567220097874,
      "learning_rate": 0.003,
      "loss": 4.0927,
      "step": 25189
    },
    {
      "epoch": 0.2519,
      "grad_norm": 1.484782114060467,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 25190
    },
    {
      "epoch": 0.25191,
      "grad_norm": 1.2869256707428494,
      "learning_rate": 0.003,
      "loss": 4.043,
      "step": 25191
    },
    {
      "epoch": 0.25192,
      "grad_norm": 1.3245246568168316,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 25192
    },
    {
      "epoch": 0.25193,
      "grad_norm": 1.5690624476461152,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 25193
    },
    {
      "epoch": 0.25194,
      "grad_norm": 1.2659452680962804,
      "learning_rate": 0.003,
      "loss": 4.0184,
      "step": 25194
    },
    {
      "epoch": 0.25195,
      "grad_norm": 1.4583903814389454,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 25195
    },
    {
      "epoch": 0.25196,
      "grad_norm": 1.2397058449658471,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 25196
    },
    {
      "epoch": 0.25197,
      "grad_norm": 1.339395723429851,
      "learning_rate": 0.003,
      "loss": 4.0353,
      "step": 25197
    },
    {
      "epoch": 0.25198,
      "grad_norm": 1.328676936692069,
      "learning_rate": 0.003,
      "loss": 4.0395,
      "step": 25198
    },
    {
      "epoch": 0.25199,
      "grad_norm": 1.6238119822845876,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 25199
    },
    {
      "epoch": 0.252,
      "grad_norm": 1.269888360143885,
      "learning_rate": 0.003,
      "loss": 4.0579,
      "step": 25200
    },
    {
      "epoch": 0.25201,
      "grad_norm": 1.4267044277087728,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 25201
    },
    {
      "epoch": 0.25202,
      "grad_norm": 1.4218801507011976,
      "learning_rate": 0.003,
      "loss": 4.033,
      "step": 25202
    },
    {
      "epoch": 0.25203,
      "grad_norm": 1.3688132860757418,
      "learning_rate": 0.003,
      "loss": 4.0153,
      "step": 25203
    },
    {
      "epoch": 0.25204,
      "grad_norm": 1.609201944142336,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 25204
    },
    {
      "epoch": 0.25205,
      "grad_norm": 1.3787918809355866,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 25205
    },
    {
      "epoch": 0.25206,
      "grad_norm": 1.6419170881232383,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 25206
    },
    {
      "epoch": 0.25207,
      "grad_norm": 1.3773995047148953,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 25207
    },
    {
      "epoch": 0.25208,
      "grad_norm": 1.467466751074815,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 25208
    },
    {
      "epoch": 0.25209,
      "grad_norm": 1.2283268077267135,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 25209
    },
    {
      "epoch": 0.2521,
      "grad_norm": 1.4422807603201062,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 25210
    },
    {
      "epoch": 0.25211,
      "grad_norm": 1.2376673380059546,
      "learning_rate": 0.003,
      "loss": 4.0635,
      "step": 25211
    },
    {
      "epoch": 0.25212,
      "grad_norm": 1.6167020652642925,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25212
    },
    {
      "epoch": 0.25213,
      "grad_norm": 1.1562475019273848,
      "learning_rate": 0.003,
      "loss": 4.0403,
      "step": 25213
    },
    {
      "epoch": 0.25214,
      "grad_norm": 1.6284489472696146,
      "learning_rate": 0.003,
      "loss": 4.0329,
      "step": 25214
    },
    {
      "epoch": 0.25215,
      "grad_norm": 1.4914284717135833,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 25215
    },
    {
      "epoch": 0.25216,
      "grad_norm": 1.5671720529987918,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 25216
    },
    {
      "epoch": 0.25217,
      "grad_norm": 1.279759773672114,
      "learning_rate": 0.003,
      "loss": 4.0212,
      "step": 25217
    },
    {
      "epoch": 0.25218,
      "grad_norm": 1.3908419171119077,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 25218
    },
    {
      "epoch": 0.25219,
      "grad_norm": 1.4238686154367661,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 25219
    },
    {
      "epoch": 0.2522,
      "grad_norm": 1.1857520037593197,
      "learning_rate": 0.003,
      "loss": 4.0521,
      "step": 25220
    },
    {
      "epoch": 0.25221,
      "grad_norm": 1.3730914307492301,
      "learning_rate": 0.003,
      "loss": 4.0928,
      "step": 25221
    },
    {
      "epoch": 0.25222,
      "grad_norm": 1.2651121726250283,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 25222
    },
    {
      "epoch": 0.25223,
      "grad_norm": 1.4139279888094562,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 25223
    },
    {
      "epoch": 0.25224,
      "grad_norm": 1.1753096524842168,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 25224
    },
    {
      "epoch": 0.25225,
      "grad_norm": 1.4300409723200138,
      "learning_rate": 0.003,
      "loss": 4.0316,
      "step": 25225
    },
    {
      "epoch": 0.25226,
      "grad_norm": 1.1391466203413618,
      "learning_rate": 0.003,
      "loss": 4.0833,
      "step": 25226
    },
    {
      "epoch": 0.25227,
      "grad_norm": 1.6894102544410714,
      "learning_rate": 0.003,
      "loss": 4.0692,
      "step": 25227
    },
    {
      "epoch": 0.25228,
      "grad_norm": 1.1246710077672635,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 25228
    },
    {
      "epoch": 0.25229,
      "grad_norm": 1.5565339525433832,
      "learning_rate": 0.003,
      "loss": 4.0513,
      "step": 25229
    },
    {
      "epoch": 0.2523,
      "grad_norm": 1.2863038928226818,
      "learning_rate": 0.003,
      "loss": 4.0496,
      "step": 25230
    },
    {
      "epoch": 0.25231,
      "grad_norm": 1.676789127526627,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 25231
    },
    {
      "epoch": 0.25232,
      "grad_norm": 1.2318633956853078,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25232
    },
    {
      "epoch": 0.25233,
      "grad_norm": 1.7270268075070208,
      "learning_rate": 0.003,
      "loss": 4.084,
      "step": 25233
    },
    {
      "epoch": 0.25234,
      "grad_norm": 1.8843912582486868,
      "learning_rate": 0.003,
      "loss": 4.0794,
      "step": 25234
    },
    {
      "epoch": 0.25235,
      "grad_norm": 1.2539211454200057,
      "learning_rate": 0.003,
      "loss": 4.0437,
      "step": 25235
    },
    {
      "epoch": 0.25236,
      "grad_norm": 1.0934542740270976,
      "learning_rate": 0.003,
      "loss": 4.0807,
      "step": 25236
    },
    {
      "epoch": 0.25237,
      "grad_norm": 1.664552691951597,
      "learning_rate": 0.003,
      "loss": 4.0452,
      "step": 25237
    },
    {
      "epoch": 0.25238,
      "grad_norm": 1.3523323860632475,
      "learning_rate": 0.003,
      "loss": 4.0301,
      "step": 25238
    },
    {
      "epoch": 0.25239,
      "grad_norm": 1.7030880227832823,
      "learning_rate": 0.003,
      "loss": 4.0425,
      "step": 25239
    },
    {
      "epoch": 0.2524,
      "grad_norm": 1.4292686011186586,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 25240
    },
    {
      "epoch": 0.25241,
      "grad_norm": 1.1257082469293838,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25241
    },
    {
      "epoch": 0.25242,
      "grad_norm": 1.3709438714996827,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 25242
    },
    {
      "epoch": 0.25243,
      "grad_norm": 1.363709261568603,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 25243
    },
    {
      "epoch": 0.25244,
      "grad_norm": 1.5986290113544008,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 25244
    },
    {
      "epoch": 0.25245,
      "grad_norm": 1.5283901507907833,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 25245
    },
    {
      "epoch": 0.25246,
      "grad_norm": 1.3552740125178642,
      "learning_rate": 0.003,
      "loss": 4.0237,
      "step": 25246
    },
    {
      "epoch": 0.25247,
      "grad_norm": 1.4763989174990026,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25247
    },
    {
      "epoch": 0.25248,
      "grad_norm": 1.3853022269493331,
      "learning_rate": 0.003,
      "loss": 4.0254,
      "step": 25248
    },
    {
      "epoch": 0.25249,
      "grad_norm": 1.380827051311956,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 25249
    },
    {
      "epoch": 0.2525,
      "grad_norm": 1.4990658995688257,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 25250
    },
    {
      "epoch": 0.25251,
      "grad_norm": 1.3669744332486737,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 25251
    },
    {
      "epoch": 0.25252,
      "grad_norm": 1.3500189881507845,
      "learning_rate": 0.003,
      "loss": 4.0192,
      "step": 25252
    },
    {
      "epoch": 0.25253,
      "grad_norm": 1.245215618245871,
      "learning_rate": 0.003,
      "loss": 4.0872,
      "step": 25253
    },
    {
      "epoch": 0.25254,
      "grad_norm": 1.3994491652014052,
      "learning_rate": 0.003,
      "loss": 4.0268,
      "step": 25254
    },
    {
      "epoch": 0.25255,
      "grad_norm": 1.12317261140991,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 25255
    },
    {
      "epoch": 0.25256,
      "grad_norm": 1.636785252583222,
      "learning_rate": 0.003,
      "loss": 4.0447,
      "step": 25256
    },
    {
      "epoch": 0.25257,
      "grad_norm": 1.1551172065144573,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 25257
    },
    {
      "epoch": 0.25258,
      "grad_norm": 1.4391707178789181,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 25258
    },
    {
      "epoch": 0.25259,
      "grad_norm": 1.518728799871394,
      "learning_rate": 0.003,
      "loss": 4.0799,
      "step": 25259
    },
    {
      "epoch": 0.2526,
      "grad_norm": 1.291050329160415,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 25260
    },
    {
      "epoch": 0.25261,
      "grad_norm": 1.3886281933205307,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 25261
    },
    {
      "epoch": 0.25262,
      "grad_norm": 1.2639215308699423,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 25262
    },
    {
      "epoch": 0.25263,
      "grad_norm": 1.2449024687391204,
      "learning_rate": 0.003,
      "loss": 4.0431,
      "step": 25263
    },
    {
      "epoch": 0.25264,
      "grad_norm": 1.6348950672560394,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 25264
    },
    {
      "epoch": 0.25265,
      "grad_norm": 1.7705397155507583,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 25265
    },
    {
      "epoch": 0.25266,
      "grad_norm": 1.5916546516380479,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 25266
    },
    {
      "epoch": 0.25267,
      "grad_norm": 1.4357868472063537,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 25267
    },
    {
      "epoch": 0.25268,
      "grad_norm": 1.2137546219048938,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 25268
    },
    {
      "epoch": 0.25269,
      "grad_norm": 1.5021698413914775,
      "learning_rate": 0.003,
      "loss": 4.0179,
      "step": 25269
    },
    {
      "epoch": 0.2527,
      "grad_norm": 1.3569008962632036,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 25270
    },
    {
      "epoch": 0.25271,
      "grad_norm": 1.6240817817906592,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 25271
    },
    {
      "epoch": 0.25272,
      "grad_norm": 1.1948694351999662,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 25272
    },
    {
      "epoch": 0.25273,
      "grad_norm": 1.493489422202077,
      "learning_rate": 0.003,
      "loss": 4.0468,
      "step": 25273
    },
    {
      "epoch": 0.25274,
      "grad_norm": 1.3185568053747854,
      "learning_rate": 0.003,
      "loss": 4.029,
      "step": 25274
    },
    {
      "epoch": 0.25275,
      "grad_norm": 1.447258551626343,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 25275
    },
    {
      "epoch": 0.25276,
      "grad_norm": 1.3908947661523872,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 25276
    },
    {
      "epoch": 0.25277,
      "grad_norm": 1.4495112107734367,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 25277
    },
    {
      "epoch": 0.25278,
      "grad_norm": 1.5572027865147067,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 25278
    },
    {
      "epoch": 0.25279,
      "grad_norm": 1.2244474680635473,
      "learning_rate": 0.003,
      "loss": 4.0311,
      "step": 25279
    },
    {
      "epoch": 0.2528,
      "grad_norm": 1.4015021802087044,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 25280
    },
    {
      "epoch": 0.25281,
      "grad_norm": 1.5014720273624802,
      "learning_rate": 0.003,
      "loss": 4.0716,
      "step": 25281
    },
    {
      "epoch": 0.25282,
      "grad_norm": 1.4039071490433639,
      "learning_rate": 0.003,
      "loss": 4.0406,
      "step": 25282
    },
    {
      "epoch": 0.25283,
      "grad_norm": 1.5923833113653523,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 25283
    },
    {
      "epoch": 0.25284,
      "grad_norm": 1.4991002940986895,
      "learning_rate": 0.003,
      "loss": 4.0516,
      "step": 25284
    },
    {
      "epoch": 0.25285,
      "grad_norm": 1.846203411630082,
      "learning_rate": 0.003,
      "loss": 4.071,
      "step": 25285
    },
    {
      "epoch": 0.25286,
      "grad_norm": 1.1134305180409663,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 25286
    },
    {
      "epoch": 0.25287,
      "grad_norm": 1.6651144200303902,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 25287
    },
    {
      "epoch": 0.25288,
      "grad_norm": 1.3111997370810007,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 25288
    },
    {
      "epoch": 0.25289,
      "grad_norm": 1.5443862247101616,
      "learning_rate": 0.003,
      "loss": 4.0446,
      "step": 25289
    },
    {
      "epoch": 0.2529,
      "grad_norm": 1.5555542843265413,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 25290
    },
    {
      "epoch": 0.25291,
      "grad_norm": 1.222826730149652,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 25291
    },
    {
      "epoch": 0.25292,
      "grad_norm": 1.2559938765362613,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 25292
    },
    {
      "epoch": 0.25293,
      "grad_norm": 1.5355502562598609,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 25293
    },
    {
      "epoch": 0.25294,
      "grad_norm": 1.3170346789557994,
      "learning_rate": 0.003,
      "loss": 4.0238,
      "step": 25294
    },
    {
      "epoch": 0.25295,
      "grad_norm": 1.237767382911788,
      "learning_rate": 0.003,
      "loss": 4.0218,
      "step": 25295
    },
    {
      "epoch": 0.25296,
      "grad_norm": 1.413900295177227,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 25296
    },
    {
      "epoch": 0.25297,
      "grad_norm": 1.257466286911203,
      "learning_rate": 0.003,
      "loss": 4.0828,
      "step": 25297
    },
    {
      "epoch": 0.25298,
      "grad_norm": 1.4962081789133097,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 25298
    },
    {
      "epoch": 0.25299,
      "grad_norm": 1.2628159445568492,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 25299
    },
    {
      "epoch": 0.253,
      "grad_norm": 1.8174857011302987,
      "learning_rate": 0.003,
      "loss": 4.0686,
      "step": 25300
    },
    {
      "epoch": 0.25301,
      "grad_norm": 1.2035343529271079,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 25301
    },
    {
      "epoch": 0.25302,
      "grad_norm": 1.6560577664185787,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 25302
    },
    {
      "epoch": 0.25303,
      "grad_norm": 1.3482229308444569,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 25303
    },
    {
      "epoch": 0.25304,
      "grad_norm": 1.688565516809209,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 25304
    },
    {
      "epoch": 0.25305,
      "grad_norm": 1.1769296822558077,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 25305
    },
    {
      "epoch": 0.25306,
      "grad_norm": 1.6848923174320518,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 25306
    },
    {
      "epoch": 0.25307,
      "grad_norm": 1.376158276833462,
      "learning_rate": 0.003,
      "loss": 4.0254,
      "step": 25307
    },
    {
      "epoch": 0.25308,
      "grad_norm": 1.3371662072263286,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 25308
    },
    {
      "epoch": 0.25309,
      "grad_norm": 1.2956318033021297,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 25309
    },
    {
      "epoch": 0.2531,
      "grad_norm": 1.6719974740738792,
      "learning_rate": 0.003,
      "loss": 4.0599,
      "step": 25310
    },
    {
      "epoch": 0.25311,
      "grad_norm": 1.2701102834006786,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 25311
    },
    {
      "epoch": 0.25312,
      "grad_norm": 1.155714883626628,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 25312
    },
    {
      "epoch": 0.25313,
      "grad_norm": 1.6508745898760928,
      "learning_rate": 0.003,
      "loss": 4.0774,
      "step": 25313
    },
    {
      "epoch": 0.25314,
      "grad_norm": 1.1854889941185416,
      "learning_rate": 0.003,
      "loss": 4.0187,
      "step": 25314
    },
    {
      "epoch": 0.25315,
      "grad_norm": 1.751531085829464,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 25315
    },
    {
      "epoch": 0.25316,
      "grad_norm": 1.15988291149224,
      "learning_rate": 0.003,
      "loss": 4.0223,
      "step": 25316
    },
    {
      "epoch": 0.25317,
      "grad_norm": 1.545473273781796,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 25317
    },
    {
      "epoch": 0.25318,
      "grad_norm": 1.1814362412392383,
      "learning_rate": 0.003,
      "loss": 4.031,
      "step": 25318
    },
    {
      "epoch": 0.25319,
      "grad_norm": 1.417648847282533,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 25319
    },
    {
      "epoch": 0.2532,
      "grad_norm": 1.3642856434639747,
      "learning_rate": 0.003,
      "loss": 4.0726,
      "step": 25320
    },
    {
      "epoch": 0.25321,
      "grad_norm": 1.3247494612287611,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 25321
    },
    {
      "epoch": 0.25322,
      "grad_norm": 1.72306792822049,
      "learning_rate": 0.003,
      "loss": 4.074,
      "step": 25322
    },
    {
      "epoch": 0.25323,
      "grad_norm": 1.470529794566376,
      "learning_rate": 0.003,
      "loss": 4.0767,
      "step": 25323
    },
    {
      "epoch": 0.25324,
      "grad_norm": 1.2170346363395002,
      "learning_rate": 0.003,
      "loss": 4.0591,
      "step": 25324
    },
    {
      "epoch": 0.25325,
      "grad_norm": 1.568263916775638,
      "learning_rate": 0.003,
      "loss": 4.0424,
      "step": 25325
    },
    {
      "epoch": 0.25326,
      "grad_norm": 1.5491882537504937,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 25326
    },
    {
      "epoch": 0.25327,
      "grad_norm": 1.0427265548153881,
      "learning_rate": 0.003,
      "loss": 4.035,
      "step": 25327
    },
    {
      "epoch": 0.25328,
      "grad_norm": 1.6310818781840875,
      "learning_rate": 0.003,
      "loss": 4.062,
      "step": 25328
    },
    {
      "epoch": 0.25329,
      "grad_norm": 1.0730968676960242,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 25329
    },
    {
      "epoch": 0.2533,
      "grad_norm": 1.5533450401337339,
      "learning_rate": 0.003,
      "loss": 4.06,
      "step": 25330
    },
    {
      "epoch": 0.25331,
      "grad_norm": 1.091160367129258,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 25331
    },
    {
      "epoch": 0.25332,
      "grad_norm": 1.4582832354874897,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 25332
    },
    {
      "epoch": 0.25333,
      "grad_norm": 1.1904747000840705,
      "learning_rate": 0.003,
      "loss": 4.0846,
      "step": 25333
    },
    {
      "epoch": 0.25334,
      "grad_norm": 1.6746989650572792,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 25334
    },
    {
      "epoch": 0.25335,
      "grad_norm": 1.3943335118458742,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 25335
    },
    {
      "epoch": 0.25336,
      "grad_norm": 1.3007989107596187,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 25336
    },
    {
      "epoch": 0.25337,
      "grad_norm": 1.2758247238218996,
      "learning_rate": 0.003,
      "loss": 4.0704,
      "step": 25337
    },
    {
      "epoch": 0.25338,
      "grad_norm": 1.9076240102285584,
      "learning_rate": 0.003,
      "loss": 4.1027,
      "step": 25338
    },
    {
      "epoch": 0.25339,
      "grad_norm": 1.1404905957926628,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 25339
    },
    {
      "epoch": 0.2534,
      "grad_norm": 1.8523073011824807,
      "learning_rate": 0.003,
      "loss": 4.069,
      "step": 25340
    },
    {
      "epoch": 0.25341,
      "grad_norm": 1.1194040882344323,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 25341
    },
    {
      "epoch": 0.25342,
      "grad_norm": 1.549142310175357,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 25342
    },
    {
      "epoch": 0.25343,
      "grad_norm": 1.2006496823268282,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 25343
    },
    {
      "epoch": 0.25344,
      "grad_norm": 1.6631956676763877,
      "learning_rate": 0.003,
      "loss": 4.0742,
      "step": 25344
    },
    {
      "epoch": 0.25345,
      "grad_norm": 1.2454918394718975,
      "learning_rate": 0.003,
      "loss": 4.0673,
      "step": 25345
    },
    {
      "epoch": 0.25346,
      "grad_norm": 1.7132471770695137,
      "learning_rate": 0.003,
      "loss": 4.0729,
      "step": 25346
    },
    {
      "epoch": 0.25347,
      "grad_norm": 1.306744519301823,
      "learning_rate": 0.003,
      "loss": 4.0895,
      "step": 25347
    },
    {
      "epoch": 0.25348,
      "grad_norm": 1.8369378981985613,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 25348
    },
    {
      "epoch": 0.25349,
      "grad_norm": 1.2057702365716583,
      "learning_rate": 0.003,
      "loss": 4.0584,
      "step": 25349
    },
    {
      "epoch": 0.2535,
      "grad_norm": 1.3978518034395102,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 25350
    },
    {
      "epoch": 0.25351,
      "grad_norm": 1.5167059645036451,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 25351
    },
    {
      "epoch": 0.25352,
      "grad_norm": 1.1736513542821312,
      "learning_rate": 0.003,
      "loss": 4.0658,
      "step": 25352
    },
    {
      "epoch": 0.25353,
      "grad_norm": 1.366786288754727,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 25353
    },
    {
      "epoch": 0.25354,
      "grad_norm": 1.5529538620022543,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 25354
    },
    {
      "epoch": 0.25355,
      "grad_norm": 1.4534921537422647,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 25355
    },
    {
      "epoch": 0.25356,
      "grad_norm": 1.1231021103701653,
      "learning_rate": 0.003,
      "loss": 4.0776,
      "step": 25356
    },
    {
      "epoch": 0.25357,
      "grad_norm": 1.398818329809975,
      "learning_rate": 0.003,
      "loss": 4.0409,
      "step": 25357
    },
    {
      "epoch": 0.25358,
      "grad_norm": 1.3066952279349477,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 25358
    },
    {
      "epoch": 0.25359,
      "grad_norm": 1.5085211200614557,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25359
    },
    {
      "epoch": 0.2536,
      "grad_norm": 1.3792822156418256,
      "learning_rate": 0.003,
      "loss": 4.0274,
      "step": 25360
    },
    {
      "epoch": 0.25361,
      "grad_norm": 1.437914146825342,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 25361
    },
    {
      "epoch": 0.25362,
      "grad_norm": 1.3442472799101133,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25362
    },
    {
      "epoch": 0.25363,
      "grad_norm": 1.2775616597390271,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 25363
    },
    {
      "epoch": 0.25364,
      "grad_norm": 1.4903277099078978,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 25364
    },
    {
      "epoch": 0.25365,
      "grad_norm": 1.0867282015038917,
      "learning_rate": 0.003,
      "loss": 4.0911,
      "step": 25365
    },
    {
      "epoch": 0.25366,
      "grad_norm": 1.5969745409458787,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 25366
    },
    {
      "epoch": 0.25367,
      "grad_norm": 1.354055787199915,
      "learning_rate": 0.003,
      "loss": 4.0935,
      "step": 25367
    },
    {
      "epoch": 0.25368,
      "grad_norm": 1.4261475991246289,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 25368
    },
    {
      "epoch": 0.25369,
      "grad_norm": 1.4494309302630086,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25369
    },
    {
      "epoch": 0.2537,
      "grad_norm": 1.6908373777218801,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 25370
    },
    {
      "epoch": 0.25371,
      "grad_norm": 1.2693320946214979,
      "learning_rate": 0.003,
      "loss": 4.036,
      "step": 25371
    },
    {
      "epoch": 0.25372,
      "grad_norm": 1.5912014458095827,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 25372
    },
    {
      "epoch": 0.25373,
      "grad_norm": 1.429442354639141,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 25373
    },
    {
      "epoch": 0.25374,
      "grad_norm": 1.2106835598717447,
      "learning_rate": 0.003,
      "loss": 4.0385,
      "step": 25374
    },
    {
      "epoch": 0.25375,
      "grad_norm": 1.2291352644114955,
      "learning_rate": 0.003,
      "loss": 4.0481,
      "step": 25375
    },
    {
      "epoch": 0.25376,
      "grad_norm": 1.5491030380805002,
      "learning_rate": 0.003,
      "loss": 4.0471,
      "step": 25376
    },
    {
      "epoch": 0.25377,
      "grad_norm": 1.349354695469733,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 25377
    },
    {
      "epoch": 0.25378,
      "grad_norm": 1.2550556685135807,
      "learning_rate": 0.003,
      "loss": 4.0039,
      "step": 25378
    },
    {
      "epoch": 0.25379,
      "grad_norm": 1.70959528564685,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 25379
    },
    {
      "epoch": 0.2538,
      "grad_norm": 1.071350400387467,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 25380
    },
    {
      "epoch": 0.25381,
      "grad_norm": 1.4934836952308728,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 25381
    },
    {
      "epoch": 0.25382,
      "grad_norm": 1.253833647497224,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 25382
    },
    {
      "epoch": 0.25383,
      "grad_norm": 1.4168675618424733,
      "learning_rate": 0.003,
      "loss": 4.0314,
      "step": 25383
    },
    {
      "epoch": 0.25384,
      "grad_norm": 1.4055443780686308,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 25384
    },
    {
      "epoch": 0.25385,
      "grad_norm": 1.3802201908647092,
      "learning_rate": 0.003,
      "loss": 4.0478,
      "step": 25385
    },
    {
      "epoch": 0.25386,
      "grad_norm": 1.61344484365093,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 25386
    },
    {
      "epoch": 0.25387,
      "grad_norm": 1.4473526834834542,
      "learning_rate": 0.003,
      "loss": 4.0604,
      "step": 25387
    },
    {
      "epoch": 0.25388,
      "grad_norm": 1.3572345545708926,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 25388
    },
    {
      "epoch": 0.25389,
      "grad_norm": 1.4380307115348705,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 25389
    },
    {
      "epoch": 0.2539,
      "grad_norm": 1.5985142399858883,
      "learning_rate": 0.003,
      "loss": 4.0753,
      "step": 25390
    },
    {
      "epoch": 0.25391,
      "grad_norm": 1.1681266889310944,
      "learning_rate": 0.003,
      "loss": 4.0301,
      "step": 25391
    },
    {
      "epoch": 0.25392,
      "grad_norm": 1.6881797846404536,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 25392
    },
    {
      "epoch": 0.25393,
      "grad_norm": 0.9404537895376099,
      "learning_rate": 0.003,
      "loss": 4.0508,
      "step": 25393
    },
    {
      "epoch": 0.25394,
      "grad_norm": 1.5971862911816268,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 25394
    },
    {
      "epoch": 0.25395,
      "grad_norm": 1.3150112109260046,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 25395
    },
    {
      "epoch": 0.25396,
      "grad_norm": 1.5841399325369836,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 25396
    },
    {
      "epoch": 0.25397,
      "grad_norm": 1.2896557300443863,
      "learning_rate": 0.003,
      "loss": 4.0672,
      "step": 25397
    },
    {
      "epoch": 0.25398,
      "grad_norm": 1.4597374381102795,
      "learning_rate": 0.003,
      "loss": 4.0229,
      "step": 25398
    },
    {
      "epoch": 0.25399,
      "grad_norm": 1.2407152853191694,
      "learning_rate": 0.003,
      "loss": 4.0381,
      "step": 25399
    },
    {
      "epoch": 0.254,
      "grad_norm": 1.7234638239605042,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 25400
    },
    {
      "epoch": 0.25401,
      "grad_norm": 1.1062642598266486,
      "learning_rate": 0.003,
      "loss": 4.0332,
      "step": 25401
    },
    {
      "epoch": 0.25402,
      "grad_norm": 1.5092134498788248,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 25402
    },
    {
      "epoch": 0.25403,
      "grad_norm": 1.3190106082444915,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 25403
    },
    {
      "epoch": 0.25404,
      "grad_norm": 1.3631759003200474,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 25404
    },
    {
      "epoch": 0.25405,
      "grad_norm": 1.2454031135005021,
      "learning_rate": 0.003,
      "loss": 4.0707,
      "step": 25405
    },
    {
      "epoch": 0.25406,
      "grad_norm": 1.3316368945211745,
      "learning_rate": 0.003,
      "loss": 4.0288,
      "step": 25406
    },
    {
      "epoch": 0.25407,
      "grad_norm": 1.216654701130413,
      "learning_rate": 0.003,
      "loss": 4.0448,
      "step": 25407
    },
    {
      "epoch": 0.25408,
      "grad_norm": 1.3947282269274737,
      "learning_rate": 0.003,
      "loss": 4.0479,
      "step": 25408
    },
    {
      "epoch": 0.25409,
      "grad_norm": 1.045474083004841,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 25409
    },
    {
      "epoch": 0.2541,
      "grad_norm": 1.3733577526266227,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 25410
    },
    {
      "epoch": 0.25411,
      "grad_norm": 1.356634390863771,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 25411
    },
    {
      "epoch": 0.25412,
      "grad_norm": 1.4267529153713783,
      "learning_rate": 0.003,
      "loss": 4.0178,
      "step": 25412
    },
    {
      "epoch": 0.25413,
      "grad_norm": 1.3964855542087644,
      "learning_rate": 0.003,
      "loss": 4.0343,
      "step": 25413
    },
    {
      "epoch": 0.25414,
      "grad_norm": 1.7791704413924154,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 25414
    },
    {
      "epoch": 0.25415,
      "grad_norm": 1.7759494880818907,
      "learning_rate": 0.003,
      "loss": 4.0422,
      "step": 25415
    },
    {
      "epoch": 0.25416,
      "grad_norm": 1.0945053015374115,
      "learning_rate": 0.003,
      "loss": 4.0713,
      "step": 25416
    },
    {
      "epoch": 0.25417,
      "grad_norm": 1.5732961289276437,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 25417
    },
    {
      "epoch": 0.25418,
      "grad_norm": 1.7129251975752757,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 25418
    },
    {
      "epoch": 0.25419,
      "grad_norm": 1.213056209907701,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 25419
    },
    {
      "epoch": 0.2542,
      "grad_norm": 1.3752080791935162,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 25420
    },
    {
      "epoch": 0.25421,
      "grad_norm": 1.6566408869926255,
      "learning_rate": 0.003,
      "loss": 4.0238,
      "step": 25421
    },
    {
      "epoch": 0.25422,
      "grad_norm": 1.1857458251141733,
      "learning_rate": 0.003,
      "loss": 4.0477,
      "step": 25422
    },
    {
      "epoch": 0.25423,
      "grad_norm": 1.5654506277682456,
      "learning_rate": 0.003,
      "loss": 4.0324,
      "step": 25423
    },
    {
      "epoch": 0.25424,
      "grad_norm": 1.3534971447060011,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 25424
    },
    {
      "epoch": 0.25425,
      "grad_norm": 1.3945769427685644,
      "learning_rate": 0.003,
      "loss": 4.0509,
      "step": 25425
    },
    {
      "epoch": 0.25426,
      "grad_norm": 1.417318171662554,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 25426
    },
    {
      "epoch": 0.25427,
      "grad_norm": 1.1261987609957194,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 25427
    },
    {
      "epoch": 0.25428,
      "grad_norm": 1.3967926642107085,
      "learning_rate": 0.003,
      "loss": 4.0233,
      "step": 25428
    },
    {
      "epoch": 0.25429,
      "grad_norm": 1.3087608074691588,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 25429
    },
    {
      "epoch": 0.2543,
      "grad_norm": 1.515608708485061,
      "learning_rate": 0.003,
      "loss": 4.048,
      "step": 25430
    },
    {
      "epoch": 0.25431,
      "grad_norm": 1.6166523837265636,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 25431
    },
    {
      "epoch": 0.25432,
      "grad_norm": 1.290464585157251,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 25432
    },
    {
      "epoch": 0.25433,
      "grad_norm": 1.4610107350983912,
      "learning_rate": 0.003,
      "loss": 4.0273,
      "step": 25433
    },
    {
      "epoch": 0.25434,
      "grad_norm": 1.3292558713759466,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 25434
    },
    {
      "epoch": 0.25435,
      "grad_norm": 1.4190979527478365,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 25435
    },
    {
      "epoch": 0.25436,
      "grad_norm": 1.4469213125140579,
      "learning_rate": 0.003,
      "loss": 4.0169,
      "step": 25436
    },
    {
      "epoch": 0.25437,
      "grad_norm": 1.5297520558522897,
      "learning_rate": 0.003,
      "loss": 4.025,
      "step": 25437
    },
    {
      "epoch": 0.25438,
      "grad_norm": 1.631054984413253,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 25438
    },
    {
      "epoch": 0.25439,
      "grad_norm": 1.4467444768516275,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25439
    },
    {
      "epoch": 0.2544,
      "grad_norm": 1.4579086730313882,
      "learning_rate": 0.003,
      "loss": 4.0839,
      "step": 25440
    },
    {
      "epoch": 0.25441,
      "grad_norm": 1.5767669818177323,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 25441
    },
    {
      "epoch": 0.25442,
      "grad_norm": 1.0968711455614821,
      "learning_rate": 0.003,
      "loss": 4.0438,
      "step": 25442
    },
    {
      "epoch": 0.25443,
      "grad_norm": 1.419885756274376,
      "learning_rate": 0.003,
      "loss": 4.0251,
      "step": 25443
    },
    {
      "epoch": 0.25444,
      "grad_norm": 1.3317378902711083,
      "learning_rate": 0.003,
      "loss": 4.0276,
      "step": 25444
    },
    {
      "epoch": 0.25445,
      "grad_norm": 1.540388880840489,
      "learning_rate": 0.003,
      "loss": 4.0339,
      "step": 25445
    },
    {
      "epoch": 0.25446,
      "grad_norm": 1.2031396547544115,
      "learning_rate": 0.003,
      "loss": 4.0638,
      "step": 25446
    },
    {
      "epoch": 0.25447,
      "grad_norm": 1.2472168293952166,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 25447
    },
    {
      "epoch": 0.25448,
      "grad_norm": 1.4286423213250237,
      "learning_rate": 0.003,
      "loss": 4.0284,
      "step": 25448
    },
    {
      "epoch": 0.25449,
      "grad_norm": 1.3767987734319982,
      "learning_rate": 0.003,
      "loss": 4.0459,
      "step": 25449
    },
    {
      "epoch": 0.2545,
      "grad_norm": 1.2072036752512636,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 25450
    },
    {
      "epoch": 0.25451,
      "grad_norm": 1.3885889860302425,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 25451
    },
    {
      "epoch": 0.25452,
      "grad_norm": 1.2807127650982557,
      "learning_rate": 0.003,
      "loss": 4.0258,
      "step": 25452
    },
    {
      "epoch": 0.25453,
      "grad_norm": 1.4622716565462128,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 25453
    },
    {
      "epoch": 0.25454,
      "grad_norm": 1.3984585133403915,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 25454
    },
    {
      "epoch": 0.25455,
      "grad_norm": 1.722589543330776,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25455
    },
    {
      "epoch": 0.25456,
      "grad_norm": 1.049234477652921,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 25456
    },
    {
      "epoch": 0.25457,
      "grad_norm": 1.6585121093231572,
      "learning_rate": 0.003,
      "loss": 4.0461,
      "step": 25457
    },
    {
      "epoch": 0.25458,
      "grad_norm": 1.6568689101169995,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 25458
    },
    {
      "epoch": 0.25459,
      "grad_norm": 1.3065026531125563,
      "learning_rate": 0.003,
      "loss": 4.0354,
      "step": 25459
    },
    {
      "epoch": 0.2546,
      "grad_norm": 1.6461219953234143,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 25460
    },
    {
      "epoch": 0.25461,
      "grad_norm": 1.4595907393033671,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 25461
    },
    {
      "epoch": 0.25462,
      "grad_norm": 1.4673569175305712,
      "learning_rate": 0.003,
      "loss": 4.0557,
      "step": 25462
    },
    {
      "epoch": 0.25463,
      "grad_norm": 1.343289518619863,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 25463
    },
    {
      "epoch": 0.25464,
      "grad_norm": 1.5496183489886368,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 25464
    },
    {
      "epoch": 0.25465,
      "grad_norm": 1.245319571820456,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 25465
    },
    {
      "epoch": 0.25466,
      "grad_norm": 1.5463524494688146,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25466
    },
    {
      "epoch": 0.25467,
      "grad_norm": 1.4283336557576143,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 25467
    },
    {
      "epoch": 0.25468,
      "grad_norm": 1.3001518310217621,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 25468
    },
    {
      "epoch": 0.25469,
      "grad_norm": 1.2773717276405157,
      "learning_rate": 0.003,
      "loss": 4.0523,
      "step": 25469
    },
    {
      "epoch": 0.2547,
      "grad_norm": 1.38340031269305,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 25470
    },
    {
      "epoch": 0.25471,
      "grad_norm": 1.4828503726044868,
      "learning_rate": 0.003,
      "loss": 4.05,
      "step": 25471
    },
    {
      "epoch": 0.25472,
      "grad_norm": 1.1984016203073116,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25472
    },
    {
      "epoch": 0.25473,
      "grad_norm": 1.5418996007598897,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 25473
    },
    {
      "epoch": 0.25474,
      "grad_norm": 1.2062283675408971,
      "learning_rate": 0.003,
      "loss": 4.0455,
      "step": 25474
    },
    {
      "epoch": 0.25475,
      "grad_norm": 1.5981634597954275,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 25475
    },
    {
      "epoch": 0.25476,
      "grad_norm": 1.24141003092722,
      "learning_rate": 0.003,
      "loss": 4.0488,
      "step": 25476
    },
    {
      "epoch": 0.25477,
      "grad_norm": 1.4308010302219056,
      "learning_rate": 0.003,
      "loss": 4.0631,
      "step": 25477
    },
    {
      "epoch": 0.25478,
      "grad_norm": 1.3492380519691354,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 25478
    },
    {
      "epoch": 0.25479,
      "grad_norm": 1.6521006888376721,
      "learning_rate": 0.003,
      "loss": 4.0618,
      "step": 25479
    },
    {
      "epoch": 0.2548,
      "grad_norm": 1.2105233944318274,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 25480
    },
    {
      "epoch": 0.25481,
      "grad_norm": 1.8273802969537531,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 25481
    },
    {
      "epoch": 0.25482,
      "grad_norm": 1.4317374315117672,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 25482
    },
    {
      "epoch": 0.25483,
      "grad_norm": 1.6709654773685005,
      "learning_rate": 0.003,
      "loss": 4.0388,
      "step": 25483
    },
    {
      "epoch": 0.25484,
      "grad_norm": 1.122015636225175,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 25484
    },
    {
      "epoch": 0.25485,
      "grad_norm": 1.8720606423002597,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 25485
    },
    {
      "epoch": 0.25486,
      "grad_norm": 1.2414872521489775,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 25486
    },
    {
      "epoch": 0.25487,
      "grad_norm": 1.3501391275763572,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 25487
    },
    {
      "epoch": 0.25488,
      "grad_norm": 1.2657488936108148,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 25488
    },
    {
      "epoch": 0.25489,
      "grad_norm": 1.2866693287512825,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 25489
    },
    {
      "epoch": 0.2549,
      "grad_norm": 1.1507462943989544,
      "learning_rate": 0.003,
      "loss": 4.0908,
      "step": 25490
    },
    {
      "epoch": 0.25491,
      "grad_norm": 1.6505210383597093,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 25491
    },
    {
      "epoch": 0.25492,
      "grad_norm": 1.655999320337114,
      "learning_rate": 0.003,
      "loss": 4.0854,
      "step": 25492
    },
    {
      "epoch": 0.25493,
      "grad_norm": 1.489003942159722,
      "learning_rate": 0.003,
      "loss": 4.0272,
      "step": 25493
    },
    {
      "epoch": 0.25494,
      "grad_norm": 1.4469608474854367,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 25494
    },
    {
      "epoch": 0.25495,
      "grad_norm": 1.5283449341925222,
      "learning_rate": 0.003,
      "loss": 4.0783,
      "step": 25495
    },
    {
      "epoch": 0.25496,
      "grad_norm": 1.3635115107183724,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 25496
    },
    {
      "epoch": 0.25497,
      "grad_norm": 1.1995007646145033,
      "learning_rate": 0.003,
      "loss": 4.0058,
      "step": 25497
    },
    {
      "epoch": 0.25498,
      "grad_norm": 1.5854153604104129,
      "learning_rate": 0.003,
      "loss": 4.055,
      "step": 25498
    },
    {
      "epoch": 0.25499,
      "grad_norm": 1.3648984887424485,
      "learning_rate": 0.003,
      "loss": 4.0433,
      "step": 25499
    },
    {
      "epoch": 0.255,
      "grad_norm": 1.250932743135752,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 25500
    },
    {
      "epoch": 0.25501,
      "grad_norm": 1.287839439068214,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 25501
    },
    {
      "epoch": 0.25502,
      "grad_norm": 1.3817175731409814,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 25502
    },
    {
      "epoch": 0.25503,
      "grad_norm": 1.3792964016289855,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 25503
    },
    {
      "epoch": 0.25504,
      "grad_norm": 1.204147996878464,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 25504
    },
    {
      "epoch": 0.25505,
      "grad_norm": 1.5397284941873677,
      "learning_rate": 0.003,
      "loss": 4.0356,
      "step": 25505
    },
    {
      "epoch": 0.25506,
      "grad_norm": 1.0146692603730831,
      "learning_rate": 0.003,
      "loss": 4.0384,
      "step": 25506
    },
    {
      "epoch": 0.25507,
      "grad_norm": 1.8067172621579008,
      "learning_rate": 0.003,
      "loss": 4.0849,
      "step": 25507
    },
    {
      "epoch": 0.25508,
      "grad_norm": 1.3891274380035143,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 25508
    },
    {
      "epoch": 0.25509,
      "grad_norm": 1.3754320305357979,
      "learning_rate": 0.003,
      "loss": 4.0444,
      "step": 25509
    },
    {
      "epoch": 0.2551,
      "grad_norm": 1.8460865750632554,
      "learning_rate": 0.003,
      "loss": 4.0412,
      "step": 25510
    },
    {
      "epoch": 0.25511,
      "grad_norm": 1.4117599576402156,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 25511
    },
    {
      "epoch": 0.25512,
      "grad_norm": 1.2375998029830446,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 25512
    },
    {
      "epoch": 0.25513,
      "grad_norm": 1.7249260792517664,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 25513
    },
    {
      "epoch": 0.25514,
      "grad_norm": 1.3667986530658038,
      "learning_rate": 0.003,
      "loss": 4.0537,
      "step": 25514
    },
    {
      "epoch": 0.25515,
      "grad_norm": 1.2065183453708739,
      "learning_rate": 0.003,
      "loss": 4.0246,
      "step": 25515
    },
    {
      "epoch": 0.25516,
      "grad_norm": 1.499885123957226,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25516
    },
    {
      "epoch": 0.25517,
      "grad_norm": 1.6053158692127931,
      "learning_rate": 0.003,
      "loss": 4.0568,
      "step": 25517
    },
    {
      "epoch": 0.25518,
      "grad_norm": 1.1784143439425059,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 25518
    },
    {
      "epoch": 0.25519,
      "grad_norm": 1.663841417551379,
      "learning_rate": 0.003,
      "loss": 4.0788,
      "step": 25519
    },
    {
      "epoch": 0.2552,
      "grad_norm": 1.2454679350478837,
      "learning_rate": 0.003,
      "loss": 4.0364,
      "step": 25520
    },
    {
      "epoch": 0.25521,
      "grad_norm": 1.4526655998402291,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 25521
    },
    {
      "epoch": 0.25522,
      "grad_norm": 1.3503624720970302,
      "learning_rate": 0.003,
      "loss": 4.0772,
      "step": 25522
    },
    {
      "epoch": 0.25523,
      "grad_norm": 1.5321714143195468,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 25523
    },
    {
      "epoch": 0.25524,
      "grad_norm": 1.2074238680178804,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 25524
    },
    {
      "epoch": 0.25525,
      "grad_norm": 1.3414394790024693,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 25525
    },
    {
      "epoch": 0.25526,
      "grad_norm": 1.5064976773081298,
      "learning_rate": 0.003,
      "loss": 4.0336,
      "step": 25526
    },
    {
      "epoch": 0.25527,
      "grad_norm": 1.574143445729999,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 25527
    },
    {
      "epoch": 0.25528,
      "grad_norm": 1.3103603861903474,
      "learning_rate": 0.003,
      "loss": 4.0321,
      "step": 25528
    },
    {
      "epoch": 0.25529,
      "grad_norm": 1.4242531305185333,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 25529
    },
    {
      "epoch": 0.2553,
      "grad_norm": 1.7002054477903017,
      "learning_rate": 0.003,
      "loss": 4.0278,
      "step": 25530
    },
    {
      "epoch": 0.25531,
      "grad_norm": 1.305686598220321,
      "learning_rate": 0.003,
      "loss": 4.0575,
      "step": 25531
    },
    {
      "epoch": 0.25532,
      "grad_norm": 1.6031756885553046,
      "learning_rate": 0.003,
      "loss": 4.0619,
      "step": 25532
    },
    {
      "epoch": 0.25533,
      "grad_norm": 1.6535578698077482,
      "learning_rate": 0.003,
      "loss": 4.0414,
      "step": 25533
    },
    {
      "epoch": 0.25534,
      "grad_norm": 1.4011224076301971,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 25534
    },
    {
      "epoch": 0.25535,
      "grad_norm": 1.3194927226083388,
      "learning_rate": 0.003,
      "loss": 4.0242,
      "step": 25535
    },
    {
      "epoch": 0.25536,
      "grad_norm": 1.4719984387637652,
      "learning_rate": 0.003,
      "loss": 4.0752,
      "step": 25536
    },
    {
      "epoch": 0.25537,
      "grad_norm": 1.413405276629304,
      "learning_rate": 0.003,
      "loss": 4.0445,
      "step": 25537
    },
    {
      "epoch": 0.25538,
      "grad_norm": 1.2029370892731581,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 25538
    },
    {
      "epoch": 0.25539,
      "grad_norm": 1.4745098513515444,
      "learning_rate": 0.003,
      "loss": 4.0518,
      "step": 25539
    },
    {
      "epoch": 0.2554,
      "grad_norm": 1.4683139081476744,
      "learning_rate": 0.003,
      "loss": 4.0344,
      "step": 25540
    },
    {
      "epoch": 0.25541,
      "grad_norm": 1.4207176040469283,
      "learning_rate": 0.003,
      "loss": 4.0593,
      "step": 25541
    },
    {
      "epoch": 0.25542,
      "grad_norm": 1.345149981613017,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 25542
    },
    {
      "epoch": 0.25543,
      "grad_norm": 1.2976518051036532,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 25543
    },
    {
      "epoch": 0.25544,
      "grad_norm": 1.302915675300831,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 25544
    },
    {
      "epoch": 0.25545,
      "grad_norm": 1.3851046737962214,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 25545
    },
    {
      "epoch": 0.25546,
      "grad_norm": 1.1542478170592156,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 25546
    },
    {
      "epoch": 0.25547,
      "grad_norm": 1.2668581097510914,
      "learning_rate": 0.003,
      "loss": 4.0823,
      "step": 25547
    },
    {
      "epoch": 0.25548,
      "grad_norm": 1.408565277981055,
      "learning_rate": 0.003,
      "loss": 4.0263,
      "step": 25548
    },
    {
      "epoch": 0.25549,
      "grad_norm": 1.379448097975971,
      "learning_rate": 0.003,
      "loss": 4.0845,
      "step": 25549
    },
    {
      "epoch": 0.2555,
      "grad_norm": 1.47746390899516,
      "learning_rate": 0.003,
      "loss": 4.0806,
      "step": 25550
    },
    {
      "epoch": 0.25551,
      "grad_norm": 1.4060110220394637,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 25551
    },
    {
      "epoch": 0.25552,
      "grad_norm": 1.5521657059476335,
      "learning_rate": 0.003,
      "loss": 4.032,
      "step": 25552
    },
    {
      "epoch": 0.25553,
      "grad_norm": 1.3460182335681414,
      "learning_rate": 0.003,
      "loss": 4.0489,
      "step": 25553
    },
    {
      "epoch": 0.25554,
      "grad_norm": 1.3951274717807889,
      "learning_rate": 0.003,
      "loss": 4.024,
      "step": 25554
    },
    {
      "epoch": 0.25555,
      "grad_norm": 1.2605631411112124,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 25555
    },
    {
      "epoch": 0.25556,
      "grad_norm": 1.387365403509073,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 25556
    },
    {
      "epoch": 0.25557,
      "grad_norm": 1.1365117688393942,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 25557
    },
    {
      "epoch": 0.25558,
      "grad_norm": 1.5523552618741925,
      "learning_rate": 0.003,
      "loss": 4.0196,
      "step": 25558
    },
    {
      "epoch": 0.25559,
      "grad_norm": 1.2852773056090736,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 25559
    },
    {
      "epoch": 0.2556,
      "grad_norm": 1.4867682850994497,
      "learning_rate": 0.003,
      "loss": 4.0281,
      "step": 25560
    },
    {
      "epoch": 0.25561,
      "grad_norm": 1.4433288723808106,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 25561
    },
    {
      "epoch": 0.25562,
      "grad_norm": 1.484708219144454,
      "learning_rate": 0.003,
      "loss": 4.0803,
      "step": 25562
    },
    {
      "epoch": 0.25563,
      "grad_norm": 1.2217009187906631,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 25563
    },
    {
      "epoch": 0.25564,
      "grad_norm": 1.6273487398124253,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 25564
    },
    {
      "epoch": 0.25565,
      "grad_norm": 1.1418039625357443,
      "learning_rate": 0.003,
      "loss": 4.041,
      "step": 25565
    },
    {
      "epoch": 0.25566,
      "grad_norm": 1.9081119274794787,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 25566
    },
    {
      "epoch": 0.25567,
      "grad_norm": 1.1309008990510911,
      "learning_rate": 0.003,
      "loss": 4.0764,
      "step": 25567
    },
    {
      "epoch": 0.25568,
      "grad_norm": 1.4415724374870398,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 25568
    },
    {
      "epoch": 0.25569,
      "grad_norm": 1.5751933103145532,
      "learning_rate": 0.003,
      "loss": 4.1016,
      "step": 25569
    },
    {
      "epoch": 0.2557,
      "grad_norm": 1.4707785212510334,
      "learning_rate": 0.003,
      "loss": 4.0663,
      "step": 25570
    },
    {
      "epoch": 0.25571,
      "grad_norm": 1.4149629158922243,
      "learning_rate": 0.003,
      "loss": 4.0808,
      "step": 25571
    },
    {
      "epoch": 0.25572,
      "grad_norm": 1.374790990623414,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 25572
    },
    {
      "epoch": 0.25573,
      "grad_norm": 1.210199376353704,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 25573
    },
    {
      "epoch": 0.25574,
      "grad_norm": 1.5109900352152457,
      "learning_rate": 0.003,
      "loss": 4.0485,
      "step": 25574
    },
    {
      "epoch": 0.25575,
      "grad_norm": 1.3628665564438698,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 25575
    },
    {
      "epoch": 0.25576,
      "grad_norm": 1.3713555492479177,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 25576
    },
    {
      "epoch": 0.25577,
      "grad_norm": 1.456544716391132,
      "learning_rate": 0.003,
      "loss": 3.9986,
      "step": 25577
    },
    {
      "epoch": 0.25578,
      "grad_norm": 1.424732368650631,
      "learning_rate": 0.003,
      "loss": 4.0415,
      "step": 25578
    },
    {
      "epoch": 0.25579,
      "grad_norm": 1.4844816880175222,
      "learning_rate": 0.003,
      "loss": 4.0731,
      "step": 25579
    },
    {
      "epoch": 0.2558,
      "grad_norm": 1.5947452454680795,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 25580
    },
    {
      "epoch": 0.25581,
      "grad_norm": 1.277538620835759,
      "learning_rate": 0.003,
      "loss": 4.0378,
      "step": 25581
    },
    {
      "epoch": 0.25582,
      "grad_norm": 1.441032111193491,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 25582
    },
    {
      "epoch": 0.25583,
      "grad_norm": 1.5593337561585785,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 25583
    },
    {
      "epoch": 0.25584,
      "grad_norm": 1.2656265666580455,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 25584
    },
    {
      "epoch": 0.25585,
      "grad_norm": 1.572161089478733,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 25585
    },
    {
      "epoch": 0.25586,
      "grad_norm": 1.3725402641669586,
      "learning_rate": 0.003,
      "loss": 4.0212,
      "step": 25586
    },
    {
      "epoch": 0.25587,
      "grad_norm": 1.2928108511865175,
      "learning_rate": 0.003,
      "loss": 4.0482,
      "step": 25587
    },
    {
      "epoch": 0.25588,
      "grad_norm": 1.7000502042071954,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 25588
    },
    {
      "epoch": 0.25589,
      "grad_norm": 1.0391278695442063,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 25589
    },
    {
      "epoch": 0.2559,
      "grad_norm": 1.670549943871537,
      "learning_rate": 0.003,
      "loss": 4.0801,
      "step": 25590
    },
    {
      "epoch": 0.25591,
      "grad_norm": 1.3502080895288906,
      "learning_rate": 0.003,
      "loss": 4.0697,
      "step": 25591
    },
    {
      "epoch": 0.25592,
      "grad_norm": 1.5998127949138552,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 25592
    },
    {
      "epoch": 0.25593,
      "grad_norm": 1.4927779283903124,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 25593
    },
    {
      "epoch": 0.25594,
      "grad_norm": 1.3512325946577588,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 25594
    },
    {
      "epoch": 0.25595,
      "grad_norm": 1.545728011810376,
      "learning_rate": 0.003,
      "loss": 4.054,
      "step": 25595
    },
    {
      "epoch": 0.25596,
      "grad_norm": 1.143924173770256,
      "learning_rate": 0.003,
      "loss": 4.0125,
      "step": 25596
    },
    {
      "epoch": 0.25597,
      "grad_norm": 1.4719395316672117,
      "learning_rate": 0.003,
      "loss": 4.0151,
      "step": 25597
    },
    {
      "epoch": 0.25598,
      "grad_norm": 1.2823896317577903,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 25598
    },
    {
      "epoch": 0.25599,
      "grad_norm": 1.4977329146274463,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 25599
    },
    {
      "epoch": 0.256,
      "grad_norm": 1.2146057423019825,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 25600
    },
    {
      "epoch": 0.25601,
      "grad_norm": 1.5364502447600348,
      "learning_rate": 0.003,
      "loss": 4.0386,
      "step": 25601
    },
    {
      "epoch": 0.25602,
      "grad_norm": 1.2953270765922653,
      "learning_rate": 0.003,
      "loss": 4.0534,
      "step": 25602
    },
    {
      "epoch": 0.25603,
      "grad_norm": 1.5422478467612095,
      "learning_rate": 0.003,
      "loss": 4.0507,
      "step": 25603
    },
    {
      "epoch": 0.25604,
      "grad_norm": 1.504897084769102,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 25604
    },
    {
      "epoch": 0.25605,
      "grad_norm": 1.227914010864788,
      "learning_rate": 0.003,
      "loss": 4.0545,
      "step": 25605
    },
    {
      "epoch": 0.25606,
      "grad_norm": 1.3701605516147022,
      "learning_rate": 0.003,
      "loss": 4.0867,
      "step": 25606
    },
    {
      "epoch": 0.25607,
      "grad_norm": 1.4373366965215018,
      "learning_rate": 0.003,
      "loss": 4.0549,
      "step": 25607
    },
    {
      "epoch": 0.25608,
      "grad_norm": 1.3863671750769027,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25608
    },
    {
      "epoch": 0.25609,
      "grad_norm": 1.3586256980547762,
      "learning_rate": 0.003,
      "loss": 4.0291,
      "step": 25609
    },
    {
      "epoch": 0.2561,
      "grad_norm": 1.4221874924687905,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 25610
    },
    {
      "epoch": 0.25611,
      "grad_norm": 1.2933945943071143,
      "learning_rate": 0.003,
      "loss": 4.0555,
      "step": 25611
    },
    {
      "epoch": 0.25612,
      "grad_norm": 1.8888523979511467,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 25612
    },
    {
      "epoch": 0.25613,
      "grad_norm": 1.1553774649680226,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 25613
    },
    {
      "epoch": 0.25614,
      "grad_norm": 1.382682284795602,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 25614
    },
    {
      "epoch": 0.25615,
      "grad_norm": 1.5200154382208988,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 25615
    },
    {
      "epoch": 0.25616,
      "grad_norm": 1.3677227848629085,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 25616
    },
    {
      "epoch": 0.25617,
      "grad_norm": 1.3587337029225195,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 25617
    },
    {
      "epoch": 0.25618,
      "grad_norm": 1.3218479755322596,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 25618
    },
    {
      "epoch": 0.25619,
      "grad_norm": 1.6378284934670104,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 25619
    },
    {
      "epoch": 0.2562,
      "grad_norm": 1.1891262651888395,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 25620
    },
    {
      "epoch": 0.25621,
      "grad_norm": 1.8150342195290958,
      "learning_rate": 0.003,
      "loss": 4.0765,
      "step": 25621
    },
    {
      "epoch": 0.25622,
      "grad_norm": 1.2266477034103502,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 25622
    },
    {
      "epoch": 0.25623,
      "grad_norm": 1.4550482710301471,
      "learning_rate": 0.003,
      "loss": 4.0363,
      "step": 25623
    },
    {
      "epoch": 0.25624,
      "grad_norm": 1.6517687336703877,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 25624
    },
    {
      "epoch": 0.25625,
      "grad_norm": 1.2874298181925097,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 25625
    },
    {
      "epoch": 0.25626,
      "grad_norm": 1.5608521150812815,
      "learning_rate": 0.003,
      "loss": 4.026,
      "step": 25626
    },
    {
      "epoch": 0.25627,
      "grad_norm": 1.2844718717936994,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 25627
    },
    {
      "epoch": 0.25628,
      "grad_norm": 1.5219868861506365,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 25628
    },
    {
      "epoch": 0.25629,
      "grad_norm": 1.1072690521254682,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 25629
    },
    {
      "epoch": 0.2563,
      "grad_norm": 1.6770526752316803,
      "learning_rate": 0.003,
      "loss": 4.0366,
      "step": 25630
    },
    {
      "epoch": 0.25631,
      "grad_norm": 1.1924272460918408,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 25631
    },
    {
      "epoch": 0.25632,
      "grad_norm": 1.6524297052037216,
      "learning_rate": 0.003,
      "loss": 4.0648,
      "step": 25632
    },
    {
      "epoch": 0.25633,
      "grad_norm": 1.3786302641111716,
      "learning_rate": 0.003,
      "loss": 4.0375,
      "step": 25633
    },
    {
      "epoch": 0.25634,
      "grad_norm": 1.4738184669781345,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 25634
    },
    {
      "epoch": 0.25635,
      "grad_norm": 1.4271402770465842,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 25635
    },
    {
      "epoch": 0.25636,
      "grad_norm": 1.5840391634191457,
      "learning_rate": 0.003,
      "loss": 4.0467,
      "step": 25636
    },
    {
      "epoch": 0.25637,
      "grad_norm": 1.3924577605942232,
      "learning_rate": 0.003,
      "loss": 4.0775,
      "step": 25637
    },
    {
      "epoch": 0.25638,
      "grad_norm": 1.5127636962257702,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 25638
    },
    {
      "epoch": 0.25639,
      "grad_norm": 1.125998535681846,
      "learning_rate": 0.003,
      "loss": 4.0272,
      "step": 25639
    },
    {
      "epoch": 0.2564,
      "grad_norm": 1.4315293157848545,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 25640
    },
    {
      "epoch": 0.25641,
      "grad_norm": 1.166717423192097,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 25641
    },
    {
      "epoch": 0.25642,
      "grad_norm": 1.4611707778140268,
      "learning_rate": 0.003,
      "loss": 4.0589,
      "step": 25642
    },
    {
      "epoch": 0.25643,
      "grad_norm": 1.3541098870906236,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 25643
    },
    {
      "epoch": 0.25644,
      "grad_norm": 1.439131328468955,
      "learning_rate": 0.003,
      "loss": 4.0317,
      "step": 25644
    },
    {
      "epoch": 0.25645,
      "grad_norm": 1.5520526523202323,
      "learning_rate": 0.003,
      "loss": 4.072,
      "step": 25645
    },
    {
      "epoch": 0.25646,
      "grad_norm": 1.3276712651446125,
      "learning_rate": 0.003,
      "loss": 4.0443,
      "step": 25646
    },
    {
      "epoch": 0.25647,
      "grad_norm": 1.3719425216373475,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 25647
    },
    {
      "epoch": 0.25648,
      "grad_norm": 1.3012783655688513,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25648
    },
    {
      "epoch": 0.25649,
      "grad_norm": 1.3732277507595656,
      "learning_rate": 0.003,
      "loss": 4.0572,
      "step": 25649
    },
    {
      "epoch": 0.2565,
      "grad_norm": 1.2922094344270323,
      "learning_rate": 0.003,
      "loss": 4.047,
      "step": 25650
    },
    {
      "epoch": 0.25651,
      "grad_norm": 1.4127963947497395,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 25651
    },
    {
      "epoch": 0.25652,
      "grad_norm": 1.6479457181059827,
      "learning_rate": 0.003,
      "loss": 4.0214,
      "step": 25652
    },
    {
      "epoch": 0.25653,
      "grad_norm": 1.5008279493265089,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25653
    },
    {
      "epoch": 0.25654,
      "grad_norm": 1.3590110056169176,
      "learning_rate": 0.003,
      "loss": 4.0542,
      "step": 25654
    },
    {
      "epoch": 0.25655,
      "grad_norm": 1.6594833834902327,
      "learning_rate": 0.003,
      "loss": 4.0175,
      "step": 25655
    },
    {
      "epoch": 0.25656,
      "grad_norm": 1.0363940422854796,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 25656
    },
    {
      "epoch": 0.25657,
      "grad_norm": 1.9050085010301476,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 25657
    },
    {
      "epoch": 0.25658,
      "grad_norm": 1.0781457885298993,
      "learning_rate": 0.003,
      "loss": 4.0627,
      "step": 25658
    },
    {
      "epoch": 0.25659,
      "grad_norm": 1.5519917422087668,
      "learning_rate": 0.003,
      "loss": 4.0825,
      "step": 25659
    },
    {
      "epoch": 0.2566,
      "grad_norm": 1.432387047987245,
      "learning_rate": 0.003,
      "loss": 4.0756,
      "step": 25660
    },
    {
      "epoch": 0.25661,
      "grad_norm": 1.4014516490127664,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 25661
    },
    {
      "epoch": 0.25662,
      "grad_norm": 1.5124493225855538,
      "learning_rate": 0.003,
      "loss": 4.0393,
      "step": 25662
    },
    {
      "epoch": 0.25663,
      "grad_norm": 1.3104131918819921,
      "learning_rate": 0.003,
      "loss": 4.076,
      "step": 25663
    },
    {
      "epoch": 0.25664,
      "grad_norm": 1.6319781966993143,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 25664
    },
    {
      "epoch": 0.25665,
      "grad_norm": 1.40101755375911,
      "learning_rate": 0.003,
      "loss": 4.0816,
      "step": 25665
    },
    {
      "epoch": 0.25666,
      "grad_norm": 1.2432996400599061,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 25666
    },
    {
      "epoch": 0.25667,
      "grad_norm": 1.3739814362869518,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 25667
    },
    {
      "epoch": 0.25668,
      "grad_norm": 1.3465405971500355,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 25668
    },
    {
      "epoch": 0.25669,
      "grad_norm": 1.0850199712691657,
      "learning_rate": 0.003,
      "loss": 4.0709,
      "step": 25669
    },
    {
      "epoch": 0.2567,
      "grad_norm": 1.5972457607952169,
      "learning_rate": 0.003,
      "loss": 4.0859,
      "step": 25670
    },
    {
      "epoch": 0.25671,
      "grad_norm": 1.3516584668589968,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25671
    },
    {
      "epoch": 0.25672,
      "grad_norm": 1.1213905471187569,
      "learning_rate": 0.003,
      "loss": 4.0215,
      "step": 25672
    },
    {
      "epoch": 0.25673,
      "grad_norm": 1.6695261828303127,
      "learning_rate": 0.003,
      "loss": 4.0786,
      "step": 25673
    },
    {
      "epoch": 0.25674,
      "grad_norm": 1.6851991452851967,
      "learning_rate": 0.003,
      "loss": 4.057,
      "step": 25674
    },
    {
      "epoch": 0.25675,
      "grad_norm": 1.4957603568627118,
      "learning_rate": 0.003,
      "loss": 4.0524,
      "step": 25675
    },
    {
      "epoch": 0.25676,
      "grad_norm": 1.4393328321796806,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 25676
    },
    {
      "epoch": 0.25677,
      "grad_norm": 1.318697217303443,
      "learning_rate": 0.003,
      "loss": 4.0937,
      "step": 25677
    },
    {
      "epoch": 0.25678,
      "grad_norm": 1.6135854417095907,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 25678
    },
    {
      "epoch": 0.25679,
      "grad_norm": 1.500621109228186,
      "learning_rate": 0.003,
      "loss": 4.0901,
      "step": 25679
    },
    {
      "epoch": 0.2568,
      "grad_norm": 1.2092819112978743,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 25680
    },
    {
      "epoch": 0.25681,
      "grad_norm": 1.5647121698206703,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 25681
    },
    {
      "epoch": 0.25682,
      "grad_norm": 1.2488500341440882,
      "learning_rate": 0.003,
      "loss": 4.078,
      "step": 25682
    },
    {
      "epoch": 0.25683,
      "grad_norm": 1.3307927660995165,
      "learning_rate": 0.003,
      "loss": 4.0486,
      "step": 25683
    },
    {
      "epoch": 0.25684,
      "grad_norm": 1.2020345417632035,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 25684
    },
    {
      "epoch": 0.25685,
      "grad_norm": 1.3690097645369093,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 25685
    },
    {
      "epoch": 0.25686,
      "grad_norm": 1.3572202738675507,
      "learning_rate": 0.003,
      "loss": 4.0607,
      "step": 25686
    },
    {
      "epoch": 0.25687,
      "grad_norm": 1.674455321942972,
      "learning_rate": 0.003,
      "loss": 4.0365,
      "step": 25687
    },
    {
      "epoch": 0.25688,
      "grad_norm": 1.5836338673174564,
      "learning_rate": 0.003,
      "loss": 4.0408,
      "step": 25688
    },
    {
      "epoch": 0.25689,
      "grad_norm": 1.3288305723084326,
      "learning_rate": 0.003,
      "loss": 4.0576,
      "step": 25689
    },
    {
      "epoch": 0.2569,
      "grad_norm": 1.2606732885831478,
      "learning_rate": 0.003,
      "loss": 4.0668,
      "step": 25690
    },
    {
      "epoch": 0.25691,
      "grad_norm": 1.6622957764526318,
      "learning_rate": 0.003,
      "loss": 4.0701,
      "step": 25691
    },
    {
      "epoch": 0.25692,
      "grad_norm": 1.4238687590002466,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 25692
    },
    {
      "epoch": 0.25693,
      "grad_norm": 1.2600048246169704,
      "learning_rate": 0.003,
      "loss": 3.9995,
      "step": 25693
    },
    {
      "epoch": 0.25694,
      "grad_norm": 1.6758560804396716,
      "learning_rate": 0.003,
      "loss": 4.038,
      "step": 25694
    },
    {
      "epoch": 0.25695,
      "grad_norm": 1.3004927686085215,
      "learning_rate": 0.003,
      "loss": 4.0382,
      "step": 25695
    },
    {
      "epoch": 0.25696,
      "grad_norm": 1.3234999045842284,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 25696
    },
    {
      "epoch": 0.25697,
      "grad_norm": 1.3130852781360365,
      "learning_rate": 0.003,
      "loss": 4.0318,
      "step": 25697
    },
    {
      "epoch": 0.25698,
      "grad_norm": 1.409466515261072,
      "learning_rate": 0.003,
      "loss": 4.0463,
      "step": 25698
    },
    {
      "epoch": 0.25699,
      "grad_norm": 1.953456834564288,
      "learning_rate": 0.003,
      "loss": 4.0597,
      "step": 25699
    },
    {
      "epoch": 0.257,
      "grad_norm": 1.1811700948959156,
      "learning_rate": 0.003,
      "loss": 4.0331,
      "step": 25700
    },
    {
      "epoch": 0.25701,
      "grad_norm": 1.3793797791911686,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 25701
    },
    {
      "epoch": 0.25702,
      "grad_norm": 1.4917660759265097,
      "learning_rate": 0.003,
      "loss": 4.0586,
      "step": 25702
    },
    {
      "epoch": 0.25703,
      "grad_norm": 1.1470374359095978,
      "learning_rate": 0.003,
      "loss": 4.0265,
      "step": 25703
    },
    {
      "epoch": 0.25704,
      "grad_norm": 1.5744649941338507,
      "learning_rate": 0.003,
      "loss": 4.0497,
      "step": 25704
    },
    {
      "epoch": 0.25705,
      "grad_norm": 1.239173001518195,
      "learning_rate": 0.003,
      "loss": 4.042,
      "step": 25705
    },
    {
      "epoch": 0.25706,
      "grad_norm": 1.459386466978485,
      "learning_rate": 0.003,
      "loss": 4.0676,
      "step": 25706
    },
    {
      "epoch": 0.25707,
      "grad_norm": 1.2680497930950785,
      "learning_rate": 0.003,
      "loss": 4.0582,
      "step": 25707
    },
    {
      "epoch": 0.25708,
      "grad_norm": 1.4292873426686685,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 25708
    },
    {
      "epoch": 0.25709,
      "grad_norm": 1.4518833240958424,
      "learning_rate": 0.003,
      "loss": 4.0827,
      "step": 25709
    },
    {
      "epoch": 0.2571,
      "grad_norm": 1.4613610877265584,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 25710
    },
    {
      "epoch": 0.25711,
      "grad_norm": 1.3905270100159586,
      "learning_rate": 0.003,
      "loss": 4.0535,
      "step": 25711
    },
    {
      "epoch": 0.25712,
      "grad_norm": 1.5639175758899766,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 25712
    },
    {
      "epoch": 0.25713,
      "grad_norm": 1.4409150133627313,
      "learning_rate": 0.003,
      "loss": 4.0569,
      "step": 25713
    },
    {
      "epoch": 0.25714,
      "grad_norm": 1.252541035045259,
      "learning_rate": 0.003,
      "loss": 4.0818,
      "step": 25714
    },
    {
      "epoch": 0.25715,
      "grad_norm": 1.6163380686679378,
      "learning_rate": 0.003,
      "loss": 4.0706,
      "step": 25715
    },
    {
      "epoch": 0.25716,
      "grad_norm": 1.1820498209170367,
      "learning_rate": 0.003,
      "loss": 4.0458,
      "step": 25716
    },
    {
      "epoch": 0.25717,
      "grad_norm": 1.301598833424317,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 25717
    },
    {
      "epoch": 0.25718,
      "grad_norm": 1.2792018926899467,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 25718
    },
    {
      "epoch": 0.25719,
      "grad_norm": 1.3501303322912859,
      "learning_rate": 0.003,
      "loss": 4.0441,
      "step": 25719
    },
    {
      "epoch": 0.2572,
      "grad_norm": 1.3177568924249792,
      "learning_rate": 0.003,
      "loss": 4.0698,
      "step": 25720
    },
    {
      "epoch": 0.25721,
      "grad_norm": 1.4874371828159054,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 25721
    },
    {
      "epoch": 0.25722,
      "grad_norm": 1.4517723696697118,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 25722
    },
    {
      "epoch": 0.25723,
      "grad_norm": 1.7467544869978353,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 25723
    },
    {
      "epoch": 0.25724,
      "grad_norm": 1.1058695475593083,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 25724
    },
    {
      "epoch": 0.25725,
      "grad_norm": 1.6426173155731811,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 25725
    },
    {
      "epoch": 0.25726,
      "grad_norm": 1.6369307065972698,
      "learning_rate": 0.003,
      "loss": 4.0766,
      "step": 25726
    },
    {
      "epoch": 0.25727,
      "grad_norm": 1.2751965527246372,
      "learning_rate": 0.003,
      "loss": 4.0632,
      "step": 25727
    },
    {
      "epoch": 0.25728,
      "grad_norm": 1.4786255229428193,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 25728
    },
    {
      "epoch": 0.25729,
      "grad_norm": 1.456060724550526,
      "learning_rate": 0.003,
      "loss": 4.0432,
      "step": 25729
    },
    {
      "epoch": 0.2573,
      "grad_norm": 1.33075720510049,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 25730
    },
    {
      "epoch": 0.25731,
      "grad_norm": 1.366940157978389,
      "learning_rate": 0.003,
      "loss": 4.0566,
      "step": 25731
    },
    {
      "epoch": 0.25732,
      "grad_norm": 1.518974674670828,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 25732
    },
    {
      "epoch": 0.25733,
      "grad_norm": 1.4038927932142937,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 25733
    },
    {
      "epoch": 0.25734,
      "grad_norm": 1.3190757994773379,
      "learning_rate": 0.003,
      "loss": 4.0113,
      "step": 25734
    },
    {
      "epoch": 0.25735,
      "grad_norm": 1.4512224467330683,
      "learning_rate": 0.003,
      "loss": 4.0275,
      "step": 25735
    },
    {
      "epoch": 0.25736,
      "grad_norm": 1.3863419204178877,
      "learning_rate": 0.003,
      "loss": 4.0302,
      "step": 25736
    },
    {
      "epoch": 0.25737,
      "grad_norm": 1.6052178635467413,
      "learning_rate": 0.003,
      "loss": 4.0528,
      "step": 25737
    },
    {
      "epoch": 0.25738,
      "grad_norm": 1.2400021749999985,
      "learning_rate": 0.003,
      "loss": 3.9923,
      "step": 25738
    },
    {
      "epoch": 0.25739,
      "grad_norm": 1.4807450909830673,
      "learning_rate": 0.003,
      "loss": 4.028,
      "step": 25739
    },
    {
      "epoch": 0.2574,
      "grad_norm": 1.1886029489537362,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 25740
    },
    {
      "epoch": 0.25741,
      "grad_norm": 1.4392580724147757,
      "learning_rate": 0.003,
      "loss": 4.0342,
      "step": 25741
    },
    {
      "epoch": 0.25742,
      "grad_norm": 1.2713361839644788,
      "learning_rate": 0.003,
      "loss": 4.0287,
      "step": 25742
    },
    {
      "epoch": 0.25743,
      "grad_norm": 1.4427922784725438,
      "learning_rate": 0.003,
      "loss": 4.0157,
      "step": 25743
    },
    {
      "epoch": 0.25744,
      "grad_norm": 1.5399738532177305,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 25744
    },
    {
      "epoch": 0.25745,
      "grad_norm": 1.1017972537118668,
      "learning_rate": 0.003,
      "loss": 4.0499,
      "step": 25745
    },
    {
      "epoch": 0.25746,
      "grad_norm": 1.4519954814509688,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 25746
    },
    {
      "epoch": 0.25747,
      "grad_norm": 1.3861194903314558,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 25747
    },
    {
      "epoch": 0.25748,
      "grad_norm": 1.4039774307830237,
      "learning_rate": 0.003,
      "loss": 4.0325,
      "step": 25748
    },
    {
      "epoch": 0.25749,
      "grad_norm": 1.4899174733046028,
      "learning_rate": 0.003,
      "loss": 4.052,
      "step": 25749
    },
    {
      "epoch": 0.2575,
      "grad_norm": 1.2485636621023675,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 25750
    },
    {
      "epoch": 0.25751,
      "grad_norm": 1.3900415948773714,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 25751
    },
    {
      "epoch": 0.25752,
      "grad_norm": 1.412189065251773,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 25752
    },
    {
      "epoch": 0.25753,
      "grad_norm": 1.8201969398357412,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25753
    },
    {
      "epoch": 0.25754,
      "grad_norm": 1.2402487925046348,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 25754
    },
    {
      "epoch": 0.25755,
      "grad_norm": 1.512449877552699,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 25755
    },
    {
      "epoch": 0.25756,
      "grad_norm": 1.2832081561586905,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 25756
    },
    {
      "epoch": 0.25757,
      "grad_norm": 1.4919667998007922,
      "learning_rate": 0.003,
      "loss": 4.0494,
      "step": 25757
    },
    {
      "epoch": 0.25758,
      "grad_norm": 1.6293179742207706,
      "learning_rate": 0.003,
      "loss": 4.0379,
      "step": 25758
    },
    {
      "epoch": 0.25759,
      "grad_norm": 1.4613255786136594,
      "learning_rate": 0.003,
      "loss": 4.0493,
      "step": 25759
    },
    {
      "epoch": 0.2576,
      "grad_norm": 1.5924353041415318,
      "learning_rate": 0.003,
      "loss": 4.0743,
      "step": 25760
    },
    {
      "epoch": 0.25761,
      "grad_norm": 1.4793877081838278,
      "learning_rate": 0.003,
      "loss": 4.0495,
      "step": 25761
    },
    {
      "epoch": 0.25762,
      "grad_norm": 1.2943966666867441,
      "learning_rate": 0.003,
      "loss": 4.0323,
      "step": 25762
    },
    {
      "epoch": 0.25763,
      "grad_norm": 1.5622737201337684,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 25763
    },
    {
      "epoch": 0.25764,
      "grad_norm": 1.1941422050212882,
      "learning_rate": 0.003,
      "loss": 4.0594,
      "step": 25764
    },
    {
      "epoch": 0.25765,
      "grad_norm": 1.495325959856617,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 25765
    },
    {
      "epoch": 0.25766,
      "grad_norm": 1.130020433845397,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 25766
    },
    {
      "epoch": 0.25767,
      "grad_norm": 1.708171967378693,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 25767
    },
    {
      "epoch": 0.25768,
      "grad_norm": 0.9975118265844913,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 25768
    },
    {
      "epoch": 0.25769,
      "grad_norm": 1.732516829154019,
      "learning_rate": 0.003,
      "loss": 4.0602,
      "step": 25769
    },
    {
      "epoch": 0.2577,
      "grad_norm": 1.2597843860679456,
      "learning_rate": 0.003,
      "loss": 4.0687,
      "step": 25770
    },
    {
      "epoch": 0.25771,
      "grad_norm": 1.5335957054617897,
      "learning_rate": 0.003,
      "loss": 4.0617,
      "step": 25771
    },
    {
      "epoch": 0.25772,
      "grad_norm": 1.31186682266505,
      "learning_rate": 0.003,
      "loss": 4.0761,
      "step": 25772
    },
    {
      "epoch": 0.25773,
      "grad_norm": 1.6527018481229452,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 25773
    },
    {
      "epoch": 0.25774,
      "grad_norm": 1.2172463731152112,
      "learning_rate": 0.003,
      "loss": 4.0536,
      "step": 25774
    },
    {
      "epoch": 0.25775,
      "grad_norm": 1.6520069739135184,
      "learning_rate": 0.003,
      "loss": 4.0565,
      "step": 25775
    },
    {
      "epoch": 0.25776,
      "grad_norm": 1.4380218738376331,
      "learning_rate": 0.003,
      "loss": 4.0703,
      "step": 25776
    },
    {
      "epoch": 0.25777,
      "grad_norm": 1.3792681041955943,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 25777
    },
    {
      "epoch": 0.25778,
      "grad_norm": 1.3921307262827065,
      "learning_rate": 0.003,
      "loss": 4.0644,
      "step": 25778
    },
    {
      "epoch": 0.25779,
      "grad_norm": 1.3682029684365404,
      "learning_rate": 0.003,
      "loss": 4.0711,
      "step": 25779
    },
    {
      "epoch": 0.2578,
      "grad_norm": 1.233905160603916,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 25780
    },
    {
      "epoch": 0.25781,
      "grad_norm": 1.4994398708565877,
      "learning_rate": 0.003,
      "loss": 4.0187,
      "step": 25781
    },
    {
      "epoch": 0.25782,
      "grad_norm": 1.3428404440572903,
      "learning_rate": 0.003,
      "loss": 4.0435,
      "step": 25782
    },
    {
      "epoch": 0.25783,
      "grad_norm": 1.4965462579487645,
      "learning_rate": 0.003,
      "loss": 4.0517,
      "step": 25783
    },
    {
      "epoch": 0.25784,
      "grad_norm": 1.4961692116748513,
      "learning_rate": 0.003,
      "loss": 4.0728,
      "step": 25784
    },
    {
      "epoch": 0.25785,
      "grad_norm": 1.2021955570297993,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 25785
    },
    {
      "epoch": 0.25786,
      "grad_norm": 1.4632467341558268,
      "learning_rate": 0.003,
      "loss": 4.0487,
      "step": 25786
    },
    {
      "epoch": 0.25787,
      "grad_norm": 1.3477823937379005,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 25787
    },
    {
      "epoch": 0.25788,
      "grad_norm": 1.2806958470443732,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 25788
    },
    {
      "epoch": 0.25789,
      "grad_norm": 1.2430195123888563,
      "learning_rate": 0.003,
      "loss": 4.0211,
      "step": 25789
    },
    {
      "epoch": 0.2579,
      "grad_norm": 1.9115407089694612,
      "learning_rate": 0.003,
      "loss": 4.0634,
      "step": 25790
    },
    {
      "epoch": 0.25791,
      "grad_norm": 1.213164328000405,
      "learning_rate": 0.003,
      "loss": 4.0691,
      "step": 25791
    },
    {
      "epoch": 0.25792,
      "grad_norm": 1.5413253583696145,
      "learning_rate": 0.003,
      "loss": 4.0391,
      "step": 25792
    },
    {
      "epoch": 0.25793,
      "grad_norm": 1.5369040950220545,
      "learning_rate": 0.003,
      "loss": 4.0297,
      "step": 25793
    },
    {
      "epoch": 0.25794,
      "grad_norm": 1.287590219552472,
      "learning_rate": 0.003,
      "loss": 4.0755,
      "step": 25794
    },
    {
      "epoch": 0.25795,
      "grad_norm": 1.3687037765605559,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 25795
    },
    {
      "epoch": 0.25796,
      "grad_norm": 1.282031588292454,
      "learning_rate": 0.003,
      "loss": 4.0221,
      "step": 25796
    },
    {
      "epoch": 0.25797,
      "grad_norm": 1.5006802561169799,
      "learning_rate": 0.003,
      "loss": 4.0506,
      "step": 25797
    },
    {
      "epoch": 0.25798,
      "grad_norm": 1.303402137518414,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 25798
    },
    {
      "epoch": 0.25799,
      "grad_norm": 1.326103615117249,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 25799
    },
    {
      "epoch": 0.258,
      "grad_norm": 1.3655312105769033,
      "learning_rate": 0.003,
      "loss": 4.0527,
      "step": 25800
    },
    {
      "epoch": 0.25801,
      "grad_norm": 1.5553004008696156,
      "learning_rate": 0.003,
      "loss": 4.0417,
      "step": 25801
    },
    {
      "epoch": 0.25802,
      "grad_norm": 1.3217126095448872,
      "learning_rate": 0.003,
      "loss": 4.0629,
      "step": 25802
    },
    {
      "epoch": 0.25803,
      "grad_norm": 1.5199853871559588,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 25803
    },
    {
      "epoch": 0.25804,
      "grad_norm": 1.3853183223047778,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 25804
    },
    {
      "epoch": 0.25805,
      "grad_norm": 1.5399550439739562,
      "learning_rate": 0.003,
      "loss": 4.0702,
      "step": 25805
    },
    {
      "epoch": 0.25806,
      "grad_norm": 1.3326718898696486,
      "learning_rate": 0.003,
      "loss": 4.0525,
      "step": 25806
    },
    {
      "epoch": 0.25807,
      "grad_norm": 1.2571037602440756,
      "learning_rate": 0.003,
      "loss": 4.0773,
      "step": 25807
    },
    {
      "epoch": 0.25808,
      "grad_norm": 1.3764438141405884,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 25808
    },
    {
      "epoch": 0.25809,
      "grad_norm": 1.2389826627729208,
      "learning_rate": 0.003,
      "loss": 4.04,
      "step": 25809
    },
    {
      "epoch": 0.2581,
      "grad_norm": 1.7141548719383164,
      "learning_rate": 0.003,
      "loss": 4.0561,
      "step": 25810
    },
    {
      "epoch": 0.25811,
      "grad_norm": 1.2884270080294644,
      "learning_rate": 0.003,
      "loss": 4.0595,
      "step": 25811
    },
    {
      "epoch": 0.25812,
      "grad_norm": 1.3525926731880507,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 25812
    },
    {
      "epoch": 0.25813,
      "grad_norm": 1.2412140720868727,
      "learning_rate": 0.003,
      "loss": 4.0454,
      "step": 25813
    },
    {
      "epoch": 0.25814,
      "grad_norm": 1.434993971607768,
      "learning_rate": 0.003,
      "loss": 4.081,
      "step": 25814
    },
    {
      "epoch": 0.25815,
      "grad_norm": 1.5584543317064383,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 25815
    },
    {
      "epoch": 0.25816,
      "grad_norm": 1.088982180340461,
      "learning_rate": 0.003,
      "loss": 4.0643,
      "step": 25816
    },
    {
      "epoch": 0.25817,
      "grad_norm": 1.458333364605955,
      "learning_rate": 0.003,
      "loss": 4.0762,
      "step": 25817
    },
    {
      "epoch": 0.25818,
      "grad_norm": 1.2596061817867001,
      "learning_rate": 0.003,
      "loss": 4.0725,
      "step": 25818
    },
    {
      "epoch": 0.25819,
      "grad_norm": 1.3691866607641443,
      "learning_rate": 0.003,
      "loss": 4.0834,
      "step": 25819
    },
    {
      "epoch": 0.2582,
      "grad_norm": 1.4594653395400055,
      "learning_rate": 0.003,
      "loss": 4.0649,
      "step": 25820
    },
    {
      "epoch": 0.25821,
      "grad_norm": 1.633895425150386,
      "learning_rate": 0.003,
      "loss": 4.0095,
      "step": 25821
    },
    {
      "epoch": 0.25822,
      "grad_norm": 1.3673656056026295,
      "learning_rate": 0.003,
      "loss": 4.0647,
      "step": 25822
    },
    {
      "epoch": 0.25823,
      "grad_norm": 1.2212225322647978,
      "learning_rate": 0.003,
      "loss": 4.0423,
      "step": 25823
    },
    {
      "epoch": 0.25824,
      "grad_norm": 1.7094744529305244,
      "learning_rate": 0.003,
      "loss": 4.0529,
      "step": 25824
    },
    {
      "epoch": 0.25825,
      "grad_norm": 0.9633848464898301,
      "learning_rate": 0.003,
      "loss": 4.0472,
      "step": 25825
    },
    {
      "epoch": 0.25826,
      "grad_norm": 1.6949635919488364,
      "learning_rate": 0.003,
      "loss": 4.064,
      "step": 25826
    },
    {
      "epoch": 0.25827,
      "grad_norm": 1.5900793526759451,
      "learning_rate": 0.003,
      "loss": 4.0696,
      "step": 25827
    },
    {
      "epoch": 0.25828,
      "grad_norm": 1.6316238400329914,
      "learning_rate": 0.003,
      "loss": 4.037,
      "step": 25828
    },
    {
      "epoch": 0.25829,
      "grad_norm": 1.4877724100021106,
      "learning_rate": 0.003,
      "loss": 4.083,
      "step": 25829
    },
    {
      "epoch": 0.2583,
      "grad_norm": 1.063959535273306,
      "learning_rate": 0.003,
      "loss": 4.0581,
      "step": 25830
    },
    {
      "epoch": 0.25831,
      "grad_norm": 1.693513770148217,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 25831
    },
    {
      "epoch": 0.25832,
      "grad_norm": 1.1227150917589217,
      "learning_rate": 0.003,
      "loss": 4.0552,
      "step": 25832
    },
    {
      "epoch": 0.25833,
      "grad_norm": 1.555061526083666,
      "learning_rate": 0.003,
      "loss": 4.0739,
      "step": 25833
    },
    {
      "epoch": 0.25834,
      "grad_norm": 1.1771953174490324,
      "learning_rate": 0.003,
      "loss": 4.0514,
      "step": 25834
    },
    {
      "epoch": 0.25835,
      "grad_norm": 1.4463921611936497,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 25835
    },
    {
      "epoch": 0.25836,
      "grad_norm": 1.209374213224575,
      "learning_rate": 0.003,
      "loss": 4.0622,
      "step": 25836
    },
    {
      "epoch": 0.25837,
      "grad_norm": 1.4100623060802875,
      "learning_rate": 0.003,
      "loss": 4.034,
      "step": 25837
    },
    {
      "epoch": 0.25838,
      "grad_norm": 1.4893210326625075,
      "learning_rate": 0.003,
      "loss": 4.0777,
      "step": 25838
    },
    {
      "epoch": 0.25839,
      "grad_norm": 1.3983396255015352,
      "learning_rate": 0.003,
      "loss": 4.0864,
      "step": 25839
    },
    {
      "epoch": 0.2584,
      "grad_norm": 1.4738249081823351,
      "learning_rate": 0.003,
      "loss": 4.049,
      "step": 25840
    },
    {
      "epoch": 0.25841,
      "grad_norm": 1.3411580400428451,
      "learning_rate": 0.003,
      "loss": 4.0616,
      "step": 25841
    },
    {
      "epoch": 0.25842,
      "grad_norm": 1.590027162773852,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 25842
    },
    {
      "epoch": 0.25843,
      "grad_norm": 1.4321730226213878,
      "learning_rate": 0.003,
      "loss": 4.0419,
      "step": 25843
    },
    {
      "epoch": 0.25844,
      "grad_norm": 1.5059054946020332,
      "learning_rate": 0.003,
      "loss": 4.0605,
      "step": 25844
    },
    {
      "epoch": 0.25845,
      "grad_norm": 1.3981638150751274,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 25845
    },
    {
      "epoch": 0.25846,
      "grad_norm": 1.3661536368106877,
      "learning_rate": 0.003,
      "loss": 4.0397,
      "step": 25846
    },
    {
      "epoch": 0.25847,
      "grad_norm": 1.4269252423372771,
      "learning_rate": 0.003,
      "loss": 4.0824,
      "step": 25847
    },
    {
      "epoch": 0.25848,
      "grad_norm": 1.2746697136898264,
      "learning_rate": 0.003,
      "loss": 4.0229,
      "step": 25848
    },
    {
      "epoch": 0.25849,
      "grad_norm": 1.4284738821220324,
      "learning_rate": 0.003,
      "loss": 4.0401,
      "step": 25849
    },
    {
      "epoch": 0.2585,
      "grad_norm": 1.0158981420254636,
      "learning_rate": 0.003,
      "loss": 4.0222,
      "step": 25850
    },
    {
      "epoch": 0.25851,
      "grad_norm": 1.5779896004785696,
      "learning_rate": 0.003,
      "loss": 4.0411,
      "step": 25851
    },
    {
      "epoch": 0.25852,
      "grad_norm": 1.2983258197763992,
      "learning_rate": 0.003,
      "loss": 4.0421,
      "step": 25852
    },
    {
      "epoch": 0.25853,
      "grad_norm": 1.6845078781875717,
      "learning_rate": 0.003,
      "loss": 4.0601,
      "step": 25853
    },
    {
      "epoch": 0.25854,
      "grad_norm": 1.1986544876414655,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 25854
    },
    {
      "epoch": 0.25855,
      "grad_norm": 1.5729150886691499,
      "learning_rate": 0.003,
      "loss": 4.0473,
      "step": 25855
    },
    {
      "epoch": 0.25856,
      "grad_norm": 1.3392915306796007,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25856
    },
    {
      "epoch": 0.25857,
      "grad_norm": 1.3631361177319512,
      "learning_rate": 0.003,
      "loss": 4.0651,
      "step": 25857
    },
    {
      "epoch": 0.25858,
      "grad_norm": 1.7809939238505188,
      "learning_rate": 0.003,
      "loss": 4.067,
      "step": 25858
    },
    {
      "epoch": 0.25859,
      "grad_norm": 1.543515764385466,
      "learning_rate": 0.003,
      "loss": 4.0748,
      "step": 25859
    },
    {
      "epoch": 0.2586,
      "grad_norm": 1.3737959323869435,
      "learning_rate": 0.003,
      "loss": 4.0564,
      "step": 25860
    },
    {
      "epoch": 0.25861,
      "grad_norm": 1.3540438262391012,
      "learning_rate": 0.003,
      "loss": 4.0642,
      "step": 25861
    },
    {
      "epoch": 0.25862,
      "grad_norm": 1.5414773649869888,
      "learning_rate": 0.003,
      "loss": 4.0633,
      "step": 25862
    },
    {
      "epoch": 0.25863,
      "grad_norm": 1.243365671387855,
      "learning_rate": 0.003,
      "loss": 4.0531,
      "step": 25863
    },
    {
      "epoch": 0.25864,
      "grad_norm": 1.3895325389113213,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 25864
    },
    {
      "epoch": 0.25865,
      "grad_norm": 1.151712296482096,
      "learning_rate": 0.003,
      "loss": 4.0285,
      "step": 25865
    },
    {
      "epoch": 0.25866,
      "grad_norm": 1.4360038086544018,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 25866
    },
    {
      "epoch": 0.25867,
      "grad_norm": 1.2929408018680493,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 25867
    },
    {
      "epoch": 0.25868,
      "grad_norm": 1.3149036231710793,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 25868
    },
    {
      "epoch": 0.25869,
      "grad_norm": 1.4299760667538082,
      "learning_rate": 0.003,
      "loss": 4.0639,
      "step": 25869
    },
    {
      "epoch": 0.2587,
      "grad_norm": 1.3049319738197966,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 25870
    },
    {
      "epoch": 0.25871,
      "grad_norm": 1.8497211546857757,
      "learning_rate": 0.003,
      "loss": 4.039,
      "step": 25871
    },
    {
      "epoch": 0.25872,
      "grad_norm": 1.3471229107379368,
      "learning_rate": 0.003,
      "loss": 4.058,
      "step": 25872
    },
    {
      "epoch": 0.25873,
      "grad_norm": 1.5283683520592855,
      "learning_rate": 0.003,
      "loss": 4.0341,
      "step": 25873
    },
    {
      "epoch": 0.25874,
      "grad_norm": 1.3668622164834656,
      "learning_rate": 0.003,
      "loss": 4.0522,
      "step": 25874
    },
    {
      "epoch": 0.25875,
      "grad_norm": 1.4353794924885628,
      "learning_rate": 0.003,
      "loss": 4.0646,
      "step": 25875
    },
    {
      "epoch": 0.25876,
      "grad_norm": 1.3529544417632908,
      "learning_rate": 0.003,
      "loss": 4.0538,
      "step": 25876
    },
    {
      "epoch": 0.25877,
      "grad_norm": 1.4498900800868253,
      "learning_rate": 0.003,
      "loss": 4.0661,
      "step": 25877
    },
    {
      "epoch": 0.25878,
      "grad_norm": 1.4364274602052436,
      "learning_rate": 0.003,
      "loss": 4.0877,
      "step": 25878
    },
    {
      "epoch": 0.25879,
      "grad_norm": 1.29028391936865,
      "learning_rate": 0.003,
      "loss": 4.0288,
      "step": 25879
    },
    {
      "epoch": 0.2588,
      "grad_norm": 1.4779018540409474,
      "learning_rate": 0.003,
      "loss": 4.0684,
      "step": 25880
    },
    {
      "epoch": 0.25881,
      "grad_norm": 1.2336714565414102,
      "learning_rate": 0.003,
      "loss": 4.0573,
      "step": 25881
    },
    {
      "epoch": 0.25882,
      "grad_norm": 1.7462766422344733,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 25882
    },
    {
      "epoch": 0.25883,
      "grad_norm": 1.3169954918792477,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 25883
    },
    {
      "epoch": 0.25884,
      "grad_norm": 1.584913381609622,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 25884
    },
    {
      "epoch": 0.25885,
      "grad_norm": 1.3720890571185853,
      "learning_rate": 0.003,
      "loss": 4.0426,
      "step": 25885
    },
    {
      "epoch": 0.25886,
      "grad_norm": 1.6966985311526994,
      "learning_rate": 0.003,
      "loss": 4.0504,
      "step": 25886
    },
    {
      "epoch": 0.25887,
      "grad_norm": 1.2111718973440997,
      "learning_rate": 0.003,
      "loss": 4.0857,
      "step": 25887
    },
    {
      "epoch": 0.25888,
      "grad_norm": 1.5375861801955801,
      "learning_rate": 0.003,
      "loss": 4.0608,
      "step": 25888
    },
    {
      "epoch": 0.25889,
      "grad_norm": 1.5573593268619124,
      "learning_rate": 0.003,
      "loss": 4.0637,
      "step": 25889
    },
    {
      "epoch": 0.2589,
      "grad_norm": 1.3598938148443485,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 25890
    },
    {
      "epoch": 0.25891,
      "grad_norm": 1.536737696286913,
      "learning_rate": 0.003,
      "loss": 4.0913,
      "step": 25891
    },
    {
      "epoch": 0.25892,
      "grad_norm": 1.2306886748574855,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25892
    },
    {
      "epoch": 0.25893,
      "grad_norm": 1.5911207196467794,
      "learning_rate": 0.003,
      "loss": 4.0758,
      "step": 25893
    },
    {
      "epoch": 0.25894,
      "grad_norm": 1.5960649998147831,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 25894
    },
    {
      "epoch": 0.25895,
      "grad_norm": 1.8014799959070236,
      "learning_rate": 0.003,
      "loss": 4.0526,
      "step": 25895
    },
    {
      "epoch": 0.25896,
      "grad_norm": 1.1943238857273202,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 25896
    },
    {
      "epoch": 0.25897,
      "grad_norm": 1.4254415322533194,
      "learning_rate": 0.003,
      "loss": 4.0334,
      "step": 25897
    },
    {
      "epoch": 0.25898,
      "grad_norm": 1.7471029558675286,
      "learning_rate": 0.003,
      "loss": 4.0512,
      "step": 25898
    },
    {
      "epoch": 0.25899,
      "grad_norm": 1.0702999858831728,
      "learning_rate": 0.003,
      "loss": 4.0897,
      "step": 25899
    },
    {
      "epoch": 0.259,
      "grad_norm": 1.6960684513089446,
      "learning_rate": 0.003,
      "loss": 4.0519,
      "step": 25900
    },
    {
      "epoch": 0.25901,
      "grad_norm": 1.4095364530171197,
      "learning_rate": 0.003,
      "loss": 4.0843,
      "step": 25901
    },
    {
      "epoch": 0.25902,
      "grad_norm": 1.5962430791332929,
      "learning_rate": 0.003,
      "loss": 4.0399,
      "step": 25902
    },
    {
      "epoch": 0.25903,
      "grad_norm": 1.2283704094120844,
      "learning_rate": 0.003,
      "loss": 4.0592,
      "step": 25903
    },
    {
      "epoch": 0.25904,
      "grad_norm": 1.4244908969617915,
      "learning_rate": 0.003,
      "loss": 4.0474,
      "step": 25904
    },
    {
      "epoch": 0.25905,
      "grad_norm": 1.2567023770419468,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 25905
    },
    {
      "epoch": 0.25906,
      "grad_norm": 1.4056920028431878,
      "learning_rate": 0.003,
      "loss": 4.0812,
      "step": 25906
    },
    {
      "epoch": 0.25907,
      "grad_norm": 1.1940805887856745,
      "learning_rate": 0.003,
      "loss": 4.0679,
      "step": 25907
    },
    {
      "epoch": 0.25908,
      "grad_norm": 1.443914232137034,
      "learning_rate": 0.003,
      "loss": 4.0654,
      "step": 25908
    },
    {
      "epoch": 0.25909,
      "grad_norm": 1.2586347841342749,
      "learning_rate": 0.003,
      "loss": 4.0699,
      "step": 25909
    },
    {
      "epoch": 0.2591,
      "grad_norm": 1.4355098144136065,
      "learning_rate": 0.003,
      "loss": 4.0546,
      "step": 25910
    },
    {
      "epoch": 0.25911,
      "grad_norm": 1.9966617665864,
      "learning_rate": 0.003,
      "loss": 4.0664,
      "step": 25911
    },
    {
      "epoch": 0.25912,
      "grad_norm": 1.2705773935075908,
      "learning_rate": 0.003,
      "loss": 4.0669,
      "step": 25912
    },
    {
      "epoch": 0.25913,
      "grad_norm": 1.6154203845070347,
      "learning_rate": 0.003,
      "loss": 4.045,
      "step": 25913
    },
    {
      "epoch": 0.25914,
      "grad_norm": 1.2523853860935699,
      "learning_rate": 0.003,
      "loss": 4.0362,
      "step": 25914
    },
    {
      "epoch": 0.25915,
      "grad_norm": 1.3920245565841947,
      "learning_rate": 0.003,
      "loss": 4.0558,
      "step": 25915
    },
    {
      "epoch": 0.25916,
      "grad_norm": 1.3577179122459062,
      "learning_rate": 0.003,
      "loss": 4.0694,
      "step": 25916
    },
    {
      "epoch": 0.25917,
      "grad_norm": 1.4608612031271264,
      "learning_rate": 0.003,
      "loss": 4.0645,
      "step": 25917
    },
    {
      "epoch": 0.25918,
      "grad_norm": 1.3977284702206003,
      "learning_rate": 0.003,
      "loss": 4.0671,
      "step": 25918
    },
    {
      "epoch": 0.25919,
      "grad_norm": 1.2728415589647335,
      "learning_rate": 0.003,
      "loss": 4.0606,
      "step": 25919
    },
    {
      "epoch": 0.2592,
      "grad_norm": 1.4042861080005589,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 25920
    },
    {
      "epoch": 0.25921,
      "grad_norm": 1.367977769991152,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 25921
    },
    {
      "epoch": 0.25922,
      "grad_norm": 1.4001055246272784,
      "learning_rate": 0.003,
      "loss": 4.0335,
      "step": 25922
    },
    {
      "epoch": 0.25923,
      "grad_norm": 1.5724131637743832,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 25923
    },
    {
      "epoch": 0.25924,
      "grad_norm": 1.7570110282193514,
      "learning_rate": 0.003,
      "loss": 4.0398,
      "step": 25924
    },
    {
      "epoch": 0.25925,
      "grad_norm": 1.2471720108652042,
      "learning_rate": 0.003,
      "loss": 4.0346,
      "step": 25925
    },
    {
      "epoch": 0.25926,
      "grad_norm": 1.7590488070068033,
      "learning_rate": 0.003,
      "loss": 4.061,
      "step": 25926
    },
    {
      "epoch": 0.25927,
      "grad_norm": 1.1566800277862708,
      "learning_rate": 0.003,
      "loss": 4.0383,
      "step": 25927
    },
    {
      "epoch": 0.25928,
      "grad_norm": 1.6654547912856716,
      "learning_rate": 0.003,
      "loss": 4.0543,
      "step": 25928
    },
    {
      "epoch": 0.25929,
      "grad_norm": 1.4548410622588641,
      "learning_rate": 0.003,
      "loss": 4.0585,
      "step": 25929
    },
    {
      "epoch": 0.2593,
      "grad_norm": 1.3240441326667984,
      "learning_rate": 0.003,
      "loss": 4.0371,
      "step": 25930
    },
    {
      "epoch": 0.25931,
      "grad_norm": 1.3468820633157403,
      "learning_rate": 0.003,
      "loss": 4.0611,
      "step": 25931
    },
    {
      "epoch": 0.25932,
      "grad_norm": 1.234752623714711,
      "learning_rate": 0.003,
      "loss": 4.0413,
      "step": 25932
    },
    {
      "epoch": 0.25933,
      "grad_norm": 1.489783748503926,
      "learning_rate": 0.003,
      "loss": 4.0466,
      "step": 25933
    },
    {
      "epoch": 0.25934,
      "grad_norm": 1.5703388537052494,
      "learning_rate": 0.003,
      "loss": 4.0909,
      "step": 25934
    },
    {
      "epoch": 0.25935,
      "grad_norm": 1.359146624253078,
      "learning_rate": 0.003,
      "loss": 4.0182,
      "step": 25935
    },
    {
      "epoch": 0.25936,
      "grad_norm": 1.4311469619932031,
      "learning_rate": 0.003,
      "loss": 4.0625,
      "step": 25936
    },
    {
      "epoch": 0.25937,
      "grad_norm": 1.548093724859345,
      "learning_rate": 0.003,
      "loss": 4.0609,
      "step": 25937
    },
    {
      "epoch": 0.25938,
      "grad_norm": 1.1705270753155963,
      "learning_rate": 0.003,
      "loss": 4.0073,
      "step": 25938
    },
    {
      "epoch": 0.25939,
      "grad_norm": 1.4512579070685991,
      "learning_rate": 0.003,
      "loss": 4.0571,
      "step": 25939
    },
    {
      "epoch": 0.2594,
      "grad_norm": 1.383681518901621,
      "learning_rate": 0.003,
      "loss": 4.0372,
      "step": 25940
    },
    {
      "epoch": 0.25941,
      "grad_norm": 1.4528343122391336,
      "learning_rate": 0.003,
      "loss": 4.0655,
      "step": 25941
    },
    {
      "epoch": 0.25942,
      "grad_norm": 1.2724837522093335,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 25942
    },
    {
      "epoch": 0.25943,
      "grad_norm": 1.335361592598783,
      "learning_rate": 0.003,
      "loss": 4.0666,
      "step": 25943
    },
    {
      "epoch": 0.25944,
      "grad_norm": 1.8521552296619705,
      "learning_rate": 0.003,
      "loss": 4.0784,
      "step": 25944
    },
    {
      "epoch": 0.25945,
      "grad_norm": 1.2096373900506994,
      "learning_rate": 0.003,
      "loss": 4.0598,
      "step": 25945
    },
    {
      "epoch": 0.25946,
      "grad_norm": 1.811153102155675,
      "learning_rate": 0.003,
      "loss": 4.056,
      "step": 25946
    },
    {
      "epoch": 0.25947,
      "grad_norm": 1.318550659821529,
      "learning_rate": 0.003,
      "loss": 4.0793,
      "step": 25947
    },
    {
      "epoch": 0.25948,
      "grad_norm": 1.4040401329610572,
      "learning_rate": 0.003,
      "loss": 4.0757,
      "step": 25948
    },
    {
      "epoch": 0.25949,
      "grad_norm": 1.2728069476774184,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 25949
    },
    {
      "epoch": 0.2595,
      "grad_norm": 1.572538744908175,
      "learning_rate": 0.003,
      "loss": 4.0308,
      "step": 25950
    },
    {
      "epoch": 0.25951,
      "grad_norm": 1.3586307321276074,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 25951
    },
    {
      "epoch": 0.25952,
      "grad_norm": 1.6110430967188807,
      "learning_rate": 0.003,
      "loss": 4.0347,
      "step": 25952
    },
    {
      "epoch": 0.25953,
      "grad_norm": 1.2698854533034047,
      "learning_rate": 0.003,
      "loss": 4.07,
      "step": 25953
    },
    {
      "epoch": 0.25954,
      "grad_norm": 1.375021416491968,
      "learning_rate": 0.003,
      "loss": 4.0553,
      "step": 25954
    },
    {
      "epoch": 0.25955,
      "grad_norm": 1.1818258227142686,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 25955
    },
    {
      "epoch": 0.25956,
      "grad_norm": 1.4977947701076593,
      "learning_rate": 0.003,
      "loss": 4.0492,
      "step": 25956
    },
    {
      "epoch": 0.25957,
      "grad_norm": 1.0816061410918312,
      "learning_rate": 0.003,
      "loss": 4.0416,
      "step": 25957
    },
    {
      "epoch": 0.25958,
      "grad_norm": 1.6780020100966895,
      "learning_rate": 0.003,
      "loss": 4.0778,
      "step": 25958
    },
    {
      "epoch": 0.25959,
      "grad_norm": 1.0487390752286232,
      "learning_rate": 0.003,
      "loss": 4.0541,
      "step": 25959
    },
    {
      "epoch": 0.2596,
      "grad_norm": 1.4932867315397373,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 25960
    },
    {
      "epoch": 0.25961,
      "grad_norm": 1.2606805109056465,
      "learning_rate": 0.003,
      "loss": 4.0451,
      "step": 25961
    },
    {
      "epoch": 0.25962,
      "grad_norm": 1.4666911533275648,
      "learning_rate": 0.003,
      "loss": 4.0659,
      "step": 25962
    },
    {
      "epoch": 0.25963,
      "grad_norm": 1.541871918588217,
      "learning_rate": 0.003,
      "loss": 4.0559,
      "step": 25963
    },
    {
      "epoch": 0.25964,
      "grad_norm": 1.6697087055813573,
      "learning_rate": 0.003,
      "loss": 4.0747,
      "step": 25964
    },
    {
      "epoch": 0.25965,
      "grad_norm": 1.2259853941984973,
      "learning_rate": 0.003,
      "loss": 4.0875,
      "step": 25965
    },
    {
      "epoch": 0.25966,
      "grad_norm": 1.5250675842116186,
      "learning_rate": 0.003,
      "loss": 4.0539,
      "step": 25966
    },
    {
      "epoch": 0.25967,
      "grad_norm": 1.3753817342983226,
      "learning_rate": 0.003,
      "loss": 4.0476,
      "step": 25967
    },
    {
      "epoch": 0.25968,
      "grad_norm": 1.376575143593774,
      "learning_rate": 0.003,
      "loss": 4.0551,
      "step": 25968
    },
    {
      "epoch": 0.25969,
      "grad_norm": 1.2994905161975587,
      "learning_rate": 0.003,
      "loss": 4.0732,
      "step": 25969
    },
    {
      "epoch": 0.2597,
      "grad_norm": 1.5169319110368726,
      "learning_rate": 0.003,
      "loss": 4.0533,
      "step": 25970
    },
    {
      "epoch": 0.25971,
      "grad_norm": 1.2386824205147875,
      "learning_rate": 0.003,
      "loss": 4.0612,
      "step": 25971
    },
    {
      "epoch": 0.25972,
      "grad_norm": 1.7383666529143638,
      "learning_rate": 0.003,
      "loss": 4.0157,
      "step": 25972
    },
    {
      "epoch": 0.25973,
      "grad_norm": 1.2663148660338621,
      "learning_rate": 0.003,
      "loss": 4.0349,
      "step": 25973
    },
    {
      "epoch": 0.25974,
      "grad_norm": 1.410954093833394,
      "learning_rate": 0.003,
      "loss": 4.0179,
      "step": 25974
    },
    {
      "epoch": 0.25975,
      "grad_norm": 1.3959748346323602,
      "learning_rate": 0.003,
      "loss": 4.0345,
      "step": 25975
    },
    {
      "epoch": 0.25976,
      "grad_norm": 1.6033639859116886,
      "learning_rate": 0.003,
      "loss": 4.0746,
      "step": 25976
    },
    {
      "epoch": 0.25977,
      "grad_norm": 1.015590349717408,
      "learning_rate": 0.003,
      "loss": 4.0715,
      "step": 25977
    },
    {
      "epoch": 0.25978,
      "grad_norm": 1.5386007579331171,
      "learning_rate": 0.003,
      "loss": 4.0723,
      "step": 25978
    },
    {
      "epoch": 0.25979,
      "grad_norm": 1.2730667385894219,
      "learning_rate": 0.003,
      "loss": 4.0498,
      "step": 25979
    },
    {
      "epoch": 0.2598,
      "grad_norm": 1.6525263209786583,
      "learning_rate": 0.003,
      "loss": 4.0484,
      "step": 25980
    },
    {
      "epoch": 0.25981,
      "grad_norm": 1.320399834548752,
      "learning_rate": 0.003,
      "loss": 4.0475,
      "step": 25981
    },
    {
      "epoch": 0.25982,
      "grad_norm": 1.4240212741695537,
      "learning_rate": 0.003,
      "loss": 4.0358,
      "step": 25982
    },
    {
      "epoch": 0.25983,
      "grad_norm": 1.2681146865078845,
      "learning_rate": 0.003,
      "loss": 4.0532,
      "step": 25983
    },
    {
      "epoch": 0.25984,
      "grad_norm": 1.350591087072189,
      "learning_rate": 0.003,
      "loss": 4.0465,
      "step": 25984
    },
    {
      "epoch": 0.25985,
      "grad_norm": 1.3509651928041615,
      "learning_rate": 0.003,
      "loss": 4.016,
      "step": 25985
    },
    {
      "epoch": 0.25986,
      "grad_norm": 1.7312014585479307,
      "learning_rate": 0.003,
      "loss": 4.0326,
      "step": 25986
    },
    {
      "epoch": 0.25987,
      "grad_norm": 1.3613277487845337,
      "learning_rate": 0.003,
      "loss": 4.0361,
      "step": 25987
    },
    {
      "epoch": 0.25988,
      "grad_norm": 1.5766365237863522,
      "learning_rate": 0.003,
      "loss": 4.0392,
      "step": 25988
    },
    {
      "epoch": 0.25989,
      "grad_norm": 1.3156267855734007,
      "learning_rate": 0.003,
      "loss": 4.0624,
      "step": 25989
    },
    {
      "epoch": 0.2599,
      "grad_norm": 1.3163374138023691,
      "learning_rate": 0.003,
      "loss": 4.0407,
      "step": 25990
    },
    {
      "epoch": 0.25991,
      "grad_norm": 1.2630937017814599,
      "learning_rate": 0.003,
      "loss": 4.0587,
      "step": 25991
    },
    {
      "epoch": 0.25992,
      "grad_norm": 1.5870309025748812,
      "learning_rate": 0.003,
      "loss": 4.0449,
      "step": 25992
    },
    {
      "epoch": 0.25993,
      "grad_norm": 1.1307578692165547,
      "learning_rate": 0.003,
      "loss": 4.0578,
      "step": 25993
    },
    {
      "epoch": 0.25994,
      "grad_norm": 1.656157416411466,
      "learning_rate": 0.003,
      "loss": 4.0623,
      "step": 25994
    },
    {
      "epoch": 0.25995,
      "grad_norm": 1.0316959825670473,
      "learning_rate": 0.003,
      "loss": 4.0577,
      "step": 25995
    },
    {
      "epoch": 0.25996,
      "grad_norm": 1.6570808197962392,
      "learning_rate": 0.003,
      "loss": 4.0405,
      "step": 25996
    },
    {
      "epoch": 0.25997,
      "grad_norm": 1.2620136405913034,
      "learning_rate": 0.003,
      "loss": 4.0544,
      "step": 25997
    },
    {
      "epoch": 0.25998,
      "grad_norm": 1.550969458116143,
      "learning_rate": 0.003,
      "loss": 4.0678,
      "step": 25998
    },
    {
      "epoch": 0.25999,
      "grad_norm": 1.634554826716528,
      "learning_rate": 0.003,
      "loss": 4.0736,
      "step": 25999
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6771422883617966,
      "learning_rate": 0.003,
      "loss": 4.0469,
      "step": 26000
    }
  ],
  "logging_steps": 1,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.030734714241024e+18,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}