{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.12,
  "eval_steps": 500,
  "global_step": 12000,
  "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
    }
  ],
  "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": 4.75723714265088e+17,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}