{
  "best_metric": 0.03610680252313614,
  "best_model_checkpoint": "./phishing-email_sender-detection/checkpoint-2733",
  "epoch": 2.0,
  "eval_steps": 1,
  "global_step": 5466,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00036589828027808267,
      "grad_norm": 1.3633811473846436,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.6954,
      "step": 1
    },
    {
      "epoch": 0.0007317965605561653,
      "grad_norm": 1.6500754356384277,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.6961,
      "step": 2
    },
    {
      "epoch": 0.0010976948408342481,
      "grad_norm": 1.5891097784042358,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "step": 3
    },
    {
      "epoch": 0.0014635931211123307,
      "grad_norm": 1.4567039012908936,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6587,
      "step": 4
    },
    {
      "epoch": 0.0018294914013904135,
      "grad_norm": 1.9978994131088257,
      "learning_rate": 5e-06,
      "loss": 0.7206,
      "step": 5
    },
    {
      "epoch": 0.0021953896816684962,
      "grad_norm": 4.357173442840576,
      "learning_rate": 6e-06,
      "loss": 0.7146,
      "step": 6
    },
    {
      "epoch": 0.002561287961946579,
      "grad_norm": 1.4282116889953613,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.685,
      "step": 7
    },
    {
      "epoch": 0.0029271862422246614,
      "grad_norm": 1.5508620738983154,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.6985,
      "step": 8
    },
    {
      "epoch": 0.003293084522502744,
      "grad_norm": 1.2599506378173828,
      "learning_rate": 9e-06,
      "loss": 0.6892,
      "step": 9
    },
    {
      "epoch": 0.003658982802780827,
      "grad_norm": 4.196310043334961,
      "learning_rate": 1e-05,
      "loss": 0.6676,
      "step": 10
    },
    {
      "epoch": 0.00402488108305891,
      "grad_norm": 1.1830265522003174,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.6984,
      "step": 11
    },
    {
      "epoch": 0.0043907793633369925,
      "grad_norm": 1.586369276046753,
      "learning_rate": 1.2e-05,
      "loss": 0.6875,
      "step": 12
    },
    {
      "epoch": 0.004756677643615075,
      "grad_norm": 1.5021487474441528,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.7011,
      "step": 13
    },
    {
      "epoch": 0.005122575923893158,
      "grad_norm": 2.3305892944335938,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.7182,
      "step": 14
    },
    {
      "epoch": 0.005488474204171241,
      "grad_norm": 1.8336869478225708,
      "learning_rate": 1.5e-05,
      "loss": 0.6868,
      "step": 15
    },
    {
      "epoch": 0.005854372484449323,
      "grad_norm": 1.3777838945388794,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.6677,
      "step": 16
    },
    {
      "epoch": 0.0062202707647274055,
      "grad_norm": 1.4701541662216187,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.6806,
      "step": 17
    },
    {
      "epoch": 0.006586169045005488,
      "grad_norm": 1.5502036809921265,
      "learning_rate": 1.8e-05,
      "loss": 0.6526,
      "step": 18
    },
    {
      "epoch": 0.006952067325283571,
      "grad_norm": 1.9634182453155518,
      "learning_rate": 1.9e-05,
      "loss": 0.706,
      "step": 19
    },
    {
      "epoch": 0.007317965605561654,
      "grad_norm": 4.49736213684082,
      "learning_rate": 2e-05,
      "loss": 0.7364,
      "step": 20
    },
    {
      "epoch": 0.007683863885839737,
      "grad_norm": 2.065563917160034,
      "learning_rate": 2.1e-05,
      "loss": 0.6919,
      "step": 21
    },
    {
      "epoch": 0.00804976216611782,
      "grad_norm": 1.344577670097351,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.6552,
      "step": 22
    },
    {
      "epoch": 0.008415660446395901,
      "grad_norm": 7.409795761108398,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.7729,
      "step": 23
    },
    {
      "epoch": 0.008781558726673985,
      "grad_norm": 1.2730001211166382,
      "learning_rate": 2.4e-05,
      "loss": 0.6847,
      "step": 24
    },
    {
      "epoch": 0.009147457006952067,
      "grad_norm": 1.9178658723831177,
      "learning_rate": 2.5e-05,
      "loss": 0.7042,
      "step": 25
    },
    {
      "epoch": 0.00951335528723015,
      "grad_norm": 2.6368801593780518,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.6956,
      "step": 26
    },
    {
      "epoch": 0.009879253567508232,
      "grad_norm": 2.288346529006958,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.696,
      "step": 27
    },
    {
      "epoch": 0.010245151847786316,
      "grad_norm": 1.299471378326416,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.6824,
      "step": 28
    },
    {
      "epoch": 0.010611050128064398,
      "grad_norm": 1.5386204719543457,
      "learning_rate": 2.9e-05,
      "loss": 0.6515,
      "step": 29
    },
    {
      "epoch": 0.010976948408342482,
      "grad_norm": 2.7759652137756348,
      "learning_rate": 3e-05,
      "loss": 0.6813,
      "step": 30
    },
    {
      "epoch": 0.011342846688620564,
      "grad_norm": 2.154977560043335,
      "learning_rate": 3.1e-05,
      "loss": 0.6635,
      "step": 31
    },
    {
      "epoch": 0.011708744968898645,
      "grad_norm": 1.8970507383346558,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.6523,
      "step": 32
    },
    {
      "epoch": 0.012074643249176729,
      "grad_norm": 2.192265272140503,
      "learning_rate": 3.3e-05,
      "loss": 0.6733,
      "step": 33
    },
    {
      "epoch": 0.012440541529454811,
      "grad_norm": 4.805243492126465,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.7286,
      "step": 34
    },
    {
      "epoch": 0.012806439809732895,
      "grad_norm": 1.9246649742126465,
      "learning_rate": 3.5e-05,
      "loss": 0.6687,
      "step": 35
    },
    {
      "epoch": 0.013172338090010977,
      "grad_norm": 2.9424471855163574,
      "learning_rate": 3.6e-05,
      "loss": 0.6974,
      "step": 36
    },
    {
      "epoch": 0.01353823637028906,
      "grad_norm": 6.86552619934082,
      "learning_rate": 3.7e-05,
      "loss": 0.7049,
      "step": 37
    },
    {
      "epoch": 0.013904134650567142,
      "grad_norm": 1.7828360795974731,
      "learning_rate": 3.8e-05,
      "loss": 0.6579,
      "step": 38
    },
    {
      "epoch": 0.014270032930845226,
      "grad_norm": 1.6522778272628784,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.6221,
      "step": 39
    },
    {
      "epoch": 0.014635931211123308,
      "grad_norm": 6.218058109283447,
      "learning_rate": 4e-05,
      "loss": 0.7465,
      "step": 40
    },
    {
      "epoch": 0.01500182949140139,
      "grad_norm": 1.788253903388977,
      "learning_rate": 4.1e-05,
      "loss": 0.7007,
      "step": 41
    },
    {
      "epoch": 0.015367727771679473,
      "grad_norm": 5.290188789367676,
      "learning_rate": 4.2e-05,
      "loss": 0.6741,
      "step": 42
    },
    {
      "epoch": 0.015733626051957557,
      "grad_norm": 1.9791131019592285,
      "learning_rate": 4.3e-05,
      "loss": 0.5557,
      "step": 43
    },
    {
      "epoch": 0.01609952433223564,
      "grad_norm": 4.857782363891602,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6061,
      "step": 44
    },
    {
      "epoch": 0.01646542261251372,
      "grad_norm": 6.5399088859558105,
      "learning_rate": 4.5e-05,
      "loss": 0.6755,
      "step": 45
    },
    {
      "epoch": 0.016831320892791803,
      "grad_norm": 5.36041784286499,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5255,
      "step": 46
    },
    {
      "epoch": 0.017197219173069888,
      "grad_norm": 10.285097122192383,
      "learning_rate": 4.7e-05,
      "loss": 0.6438,
      "step": 47
    },
    {
      "epoch": 0.01756311745334797,
      "grad_norm": 4.434288501739502,
      "learning_rate": 4.8e-05,
      "loss": 0.5803,
      "step": 48
    },
    {
      "epoch": 0.017929015733626052,
      "grad_norm": 6.69638204574585,
      "learning_rate": 4.9e-05,
      "loss": 0.6828,
      "step": 49
    },
    {
      "epoch": 0.018294914013904134,
      "grad_norm": 3.003572940826416,
      "learning_rate": 5e-05,
      "loss": 0.6395,
      "step": 50
    },
    {
      "epoch": 0.018660812294182216,
      "grad_norm": 4.034135341644287,
      "learning_rate": 4.999076809453471e-05,
      "loss": 0.4351,
      "step": 51
    },
    {
      "epoch": 0.0190267105744603,
      "grad_norm": 6.919787406921387,
      "learning_rate": 4.998153618906943e-05,
      "loss": 0.6312,
      "step": 52
    },
    {
      "epoch": 0.019392608854738383,
      "grad_norm": 18.677793502807617,
      "learning_rate": 4.997230428360414e-05,
      "loss": 0.7239,
      "step": 53
    },
    {
      "epoch": 0.019758507135016465,
      "grad_norm": 3.5922183990478516,
      "learning_rate": 4.9963072378138845e-05,
      "loss": 0.4148,
      "step": 54
    },
    {
      "epoch": 0.020124405415294547,
      "grad_norm": 5.828266143798828,
      "learning_rate": 4.995384047267356e-05,
      "loss": 0.5921,
      "step": 55
    },
    {
      "epoch": 0.020490303695572632,
      "grad_norm": 6.240745544433594,
      "learning_rate": 4.994460856720827e-05,
      "loss": 0.5127,
      "step": 56
    },
    {
      "epoch": 0.020856201975850714,
      "grad_norm": 9.371582984924316,
      "learning_rate": 4.993537666174299e-05,
      "loss": 0.6107,
      "step": 57
    },
    {
      "epoch": 0.021222100256128796,
      "grad_norm": 6.266089916229248,
      "learning_rate": 4.99261447562777e-05,
      "loss": 0.4183,
      "step": 58
    },
    {
      "epoch": 0.021587998536406878,
      "grad_norm": 5.715867519378662,
      "learning_rate": 4.991691285081241e-05,
      "loss": 0.722,
      "step": 59
    },
    {
      "epoch": 0.021953896816684963,
      "grad_norm": 5.71608829498291,
      "learning_rate": 4.990768094534712e-05,
      "loss": 0.4184,
      "step": 60
    },
    {
      "epoch": 0.022319795096963045,
      "grad_norm": 13.49493408203125,
      "learning_rate": 4.989844903988183e-05,
      "loss": 0.6887,
      "step": 61
    },
    {
      "epoch": 0.022685693377241127,
      "grad_norm": 9.925760269165039,
      "learning_rate": 4.988921713441654e-05,
      "loss": 0.5921,
      "step": 62
    },
    {
      "epoch": 0.02305159165751921,
      "grad_norm": 2.951352119445801,
      "learning_rate": 4.987998522895126e-05,
      "loss": 0.5208,
      "step": 63
    },
    {
      "epoch": 0.02341748993779729,
      "grad_norm": 3.126063346862793,
      "learning_rate": 4.987075332348597e-05,
      "loss": 0.5138,
      "step": 64
    },
    {
      "epoch": 0.023783388218075376,
      "grad_norm": 3.200230598449707,
      "learning_rate": 4.986152141802068e-05,
      "loss": 0.3901,
      "step": 65
    },
    {
      "epoch": 0.024149286498353458,
      "grad_norm": 3.98761248588562,
      "learning_rate": 4.985228951255539e-05,
      "loss": 0.3063,
      "step": 66
    },
    {
      "epoch": 0.02451518477863154,
      "grad_norm": 13.082633018493652,
      "learning_rate": 4.98430576070901e-05,
      "loss": 0.5999,
      "step": 67
    },
    {
      "epoch": 0.024881083058909622,
      "grad_norm": 8.735843658447266,
      "learning_rate": 4.983382570162482e-05,
      "loss": 0.5696,
      "step": 68
    },
    {
      "epoch": 0.025246981339187707,
      "grad_norm": 11.408510208129883,
      "learning_rate": 4.982459379615953e-05,
      "loss": 0.3812,
      "step": 69
    },
    {
      "epoch": 0.02561287961946579,
      "grad_norm": 5.581221580505371,
      "learning_rate": 4.981536189069424e-05,
      "loss": 0.2529,
      "step": 70
    },
    {
      "epoch": 0.02597877789974387,
      "grad_norm": 8.139554023742676,
      "learning_rate": 4.980612998522896e-05,
      "loss": 0.2999,
      "step": 71
    },
    {
      "epoch": 0.026344676180021953,
      "grad_norm": 47.992488861083984,
      "learning_rate": 4.979689807976367e-05,
      "loss": 1.1513,
      "step": 72
    },
    {
      "epoch": 0.026710574460300035,
      "grad_norm": 40.334007263183594,
      "learning_rate": 4.978766617429837e-05,
      "loss": 0.8844,
      "step": 73
    },
    {
      "epoch": 0.02707647274057812,
      "grad_norm": 9.410490989685059,
      "learning_rate": 4.977843426883309e-05,
      "loss": 0.5059,
      "step": 74
    },
    {
      "epoch": 0.027442371020856202,
      "grad_norm": 6.871063709259033,
      "learning_rate": 4.97692023633678e-05,
      "loss": 0.5701,
      "step": 75
    },
    {
      "epoch": 0.027808269301134284,
      "grad_norm": 4.6191511154174805,
      "learning_rate": 4.975997045790252e-05,
      "loss": 0.2245,
      "step": 76
    },
    {
      "epoch": 0.028174167581412366,
      "grad_norm": 21.528972625732422,
      "learning_rate": 4.975073855243723e-05,
      "loss": 0.7981,
      "step": 77
    },
    {
      "epoch": 0.02854006586169045,
      "grad_norm": 11.714152336120605,
      "learning_rate": 4.974150664697194e-05,
      "loss": 0.6604,
      "step": 78
    },
    {
      "epoch": 0.028905964141968533,
      "grad_norm": 8.612500190734863,
      "learning_rate": 4.973227474150665e-05,
      "loss": 1.0444,
      "step": 79
    },
    {
      "epoch": 0.029271862422246615,
      "grad_norm": 9.675956726074219,
      "learning_rate": 4.972304283604136e-05,
      "loss": 0.4215,
      "step": 80
    },
    {
      "epoch": 0.029637760702524697,
      "grad_norm": 12.019928932189941,
      "learning_rate": 4.971381093057607e-05,
      "loss": 0.484,
      "step": 81
    },
    {
      "epoch": 0.03000365898280278,
      "grad_norm": 4.286660194396973,
      "learning_rate": 4.970457902511079e-05,
      "loss": 0.4106,
      "step": 82
    },
    {
      "epoch": 0.030369557263080865,
      "grad_norm": 2.1215243339538574,
      "learning_rate": 4.96953471196455e-05,
      "loss": 0.2425,
      "step": 83
    },
    {
      "epoch": 0.030735455543358946,
      "grad_norm": 4.8601250648498535,
      "learning_rate": 4.968611521418021e-05,
      "loss": 0.3311,
      "step": 84
    },
    {
      "epoch": 0.03110135382363703,
      "grad_norm": 4.2161383628845215,
      "learning_rate": 4.967688330871492e-05,
      "loss": 0.3088,
      "step": 85
    },
    {
      "epoch": 0.031467252103915114,
      "grad_norm": 8.469701766967773,
      "learning_rate": 4.966765140324963e-05,
      "loss": 0.3771,
      "step": 86
    },
    {
      "epoch": 0.031833150384193196,
      "grad_norm": 7.6794538497924805,
      "learning_rate": 4.965841949778435e-05,
      "loss": 0.4127,
      "step": 87
    },
    {
      "epoch": 0.03219904866447128,
      "grad_norm": 2.2611911296844482,
      "learning_rate": 4.964918759231906e-05,
      "loss": 0.2755,
      "step": 88
    },
    {
      "epoch": 0.03256494694474936,
      "grad_norm": 3.5376501083374023,
      "learning_rate": 4.963995568685377e-05,
      "loss": 0.268,
      "step": 89
    },
    {
      "epoch": 0.03293084522502744,
      "grad_norm": 2.27750825881958,
      "learning_rate": 4.9630723781388485e-05,
      "loss": 0.3045,
      "step": 90
    },
    {
      "epoch": 0.03329674350530552,
      "grad_norm": 3.297470808029175,
      "learning_rate": 4.9621491875923196e-05,
      "loss": 0.3138,
      "step": 91
    },
    {
      "epoch": 0.033662641785583605,
      "grad_norm": 7.017153263092041,
      "learning_rate": 4.96122599704579e-05,
      "loss": 0.4767,
      "step": 92
    },
    {
      "epoch": 0.03402854006586169,
      "grad_norm": 2.3510890007019043,
      "learning_rate": 4.960302806499262e-05,
      "loss": 0.1017,
      "step": 93
    },
    {
      "epoch": 0.034394438346139776,
      "grad_norm": 5.36118221282959,
      "learning_rate": 4.959379615952733e-05,
      "loss": 0.4078,
      "step": 94
    },
    {
      "epoch": 0.03476033662641786,
      "grad_norm": 7.7214813232421875,
      "learning_rate": 4.958456425406204e-05,
      "loss": 0.4831,
      "step": 95
    },
    {
      "epoch": 0.03512623490669594,
      "grad_norm": 3.8735640048980713,
      "learning_rate": 4.9575332348596755e-05,
      "loss": 0.1109,
      "step": 96
    },
    {
      "epoch": 0.03549213318697402,
      "grad_norm": 7.282219886779785,
      "learning_rate": 4.9566100443131466e-05,
      "loss": 0.2307,
      "step": 97
    },
    {
      "epoch": 0.035858031467252104,
      "grad_norm": 9.082653999328613,
      "learning_rate": 4.9556868537666177e-05,
      "loss": 0.1391,
      "step": 98
    },
    {
      "epoch": 0.036223929747530186,
      "grad_norm": 10.205718040466309,
      "learning_rate": 4.954763663220089e-05,
      "loss": 0.6083,
      "step": 99
    },
    {
      "epoch": 0.03658982802780827,
      "grad_norm": 14.97369384765625,
      "learning_rate": 4.95384047267356e-05,
      "loss": 0.5098,
      "step": 100
    },
    {
      "epoch": 0.03695572630808635,
      "grad_norm": 12.733516693115234,
      "learning_rate": 4.9529172821270315e-05,
      "loss": 0.7876,
      "step": 101
    },
    {
      "epoch": 0.03732162458836443,
      "grad_norm": 13.93982982635498,
      "learning_rate": 4.9519940915805025e-05,
      "loss": 0.6761,
      "step": 102
    },
    {
      "epoch": 0.03768752286864252,
      "grad_norm": 13.174092292785645,
      "learning_rate": 4.9510709010339736e-05,
      "loss": 0.3824,
      "step": 103
    },
    {
      "epoch": 0.0380534211489206,
      "grad_norm": 7.5801100730896,
      "learning_rate": 4.9501477104874447e-05,
      "loss": 0.5373,
      "step": 104
    },
    {
      "epoch": 0.038419319429198684,
      "grad_norm": 12.05787181854248,
      "learning_rate": 4.949224519940916e-05,
      "loss": 0.2573,
      "step": 105
    },
    {
      "epoch": 0.038785217709476766,
      "grad_norm": 19.974037170410156,
      "learning_rate": 4.9483013293943874e-05,
      "loss": 0.317,
      "step": 106
    },
    {
      "epoch": 0.03915111598975485,
      "grad_norm": 14.603202819824219,
      "learning_rate": 4.9473781388478585e-05,
      "loss": 0.3298,
      "step": 107
    },
    {
      "epoch": 0.03951701427003293,
      "grad_norm": 7.982493877410889,
      "learning_rate": 4.9464549483013296e-05,
      "loss": 0.4705,
      "step": 108
    },
    {
      "epoch": 0.03988291255031101,
      "grad_norm": 2.5201683044433594,
      "learning_rate": 4.945531757754801e-05,
      "loss": 0.1715,
      "step": 109
    },
    {
      "epoch": 0.040248810830589093,
      "grad_norm": 2.5080223083496094,
      "learning_rate": 4.9446085672082723e-05,
      "loss": 0.1989,
      "step": 110
    },
    {
      "epoch": 0.04061470911086718,
      "grad_norm": 2.381314277648926,
      "learning_rate": 4.943685376661743e-05,
      "loss": 0.3996,
      "step": 111
    },
    {
      "epoch": 0.040980607391145264,
      "grad_norm": 10.304417610168457,
      "learning_rate": 4.9427621861152145e-05,
      "loss": 0.2873,
      "step": 112
    },
    {
      "epoch": 0.041346505671423346,
      "grad_norm": 5.847716808319092,
      "learning_rate": 4.9418389955686855e-05,
      "loss": 0.1286,
      "step": 113
    },
    {
      "epoch": 0.04171240395170143,
      "grad_norm": 3.0420331954956055,
      "learning_rate": 4.9409158050221566e-05,
      "loss": 0.2061,
      "step": 114
    },
    {
      "epoch": 0.04207830223197951,
      "grad_norm": 8.731443405151367,
      "learning_rate": 4.939992614475628e-05,
      "loss": 0.3556,
      "step": 115
    },
    {
      "epoch": 0.04244420051225759,
      "grad_norm": 17.511112213134766,
      "learning_rate": 4.9390694239290994e-05,
      "loss": 0.4478,
      "step": 116
    },
    {
      "epoch": 0.042810098792535674,
      "grad_norm": 7.176655292510986,
      "learning_rate": 4.9381462333825704e-05,
      "loss": 0.3003,
      "step": 117
    },
    {
      "epoch": 0.043175997072813756,
      "grad_norm": 14.813531875610352,
      "learning_rate": 4.9372230428360415e-05,
      "loss": 0.4367,
      "step": 118
    },
    {
      "epoch": 0.04354189535309184,
      "grad_norm": 18.863285064697266,
      "learning_rate": 4.9362998522895125e-05,
      "loss": 0.7858,
      "step": 119
    },
    {
      "epoch": 0.043907793633369926,
      "grad_norm": 6.050473690032959,
      "learning_rate": 4.935376661742984e-05,
      "loss": 0.2778,
      "step": 120
    },
    {
      "epoch": 0.04427369191364801,
      "grad_norm": 4.161172389984131,
      "learning_rate": 4.934453471196455e-05,
      "loss": 0.226,
      "step": 121
    },
    {
      "epoch": 0.04463959019392609,
      "grad_norm": 7.902561664581299,
      "learning_rate": 4.9335302806499264e-05,
      "loss": 0.443,
      "step": 122
    },
    {
      "epoch": 0.04500548847420417,
      "grad_norm": 10.297159194946289,
      "learning_rate": 4.9326070901033974e-05,
      "loss": 0.4977,
      "step": 123
    },
    {
      "epoch": 0.045371386754482254,
      "grad_norm": 2.424184560775757,
      "learning_rate": 4.9316838995568685e-05,
      "loss": 0.3216,
      "step": 124
    },
    {
      "epoch": 0.045737285034760336,
      "grad_norm": 2.369314670562744,
      "learning_rate": 4.9307607090103395e-05,
      "loss": 0.2146,
      "step": 125
    },
    {
      "epoch": 0.04610318331503842,
      "grad_norm": 6.874690532684326,
      "learning_rate": 4.929837518463811e-05,
      "loss": 0.5608,
      "step": 126
    },
    {
      "epoch": 0.0464690815953165,
      "grad_norm": 2.797171115875244,
      "learning_rate": 4.928914327917282e-05,
      "loss": 0.2964,
      "step": 127
    },
    {
      "epoch": 0.04683497987559458,
      "grad_norm": 2.39868426322937,
      "learning_rate": 4.927991137370754e-05,
      "loss": 0.146,
      "step": 128
    },
    {
      "epoch": 0.04720087815587267,
      "grad_norm": 12.649468421936035,
      "learning_rate": 4.927067946824225e-05,
      "loss": 0.5171,
      "step": 129
    },
    {
      "epoch": 0.04756677643615075,
      "grad_norm": 4.011669635772705,
      "learning_rate": 4.9261447562776955e-05,
      "loss": 0.4377,
      "step": 130
    },
    {
      "epoch": 0.047932674716428834,
      "grad_norm": 3.7458395957946777,
      "learning_rate": 4.925221565731167e-05,
      "loss": 0.2211,
      "step": 131
    },
    {
      "epoch": 0.048298572996706916,
      "grad_norm": 3.755655527114868,
      "learning_rate": 4.924298375184638e-05,
      "loss": 0.4016,
      "step": 132
    },
    {
      "epoch": 0.048664471276985,
      "grad_norm": 2.380204677581787,
      "learning_rate": 4.923375184638109e-05,
      "loss": 0.1568,
      "step": 133
    },
    {
      "epoch": 0.04903036955726308,
      "grad_norm": 4.0001220703125,
      "learning_rate": 4.922451994091581e-05,
      "loss": 0.1323,
      "step": 134
    },
    {
      "epoch": 0.04939626783754116,
      "grad_norm": 3.1570353507995605,
      "learning_rate": 4.921528803545052e-05,
      "loss": 0.1312,
      "step": 135
    },
    {
      "epoch": 0.049762166117819244,
      "grad_norm": 3.2308638095855713,
      "learning_rate": 4.920605612998523e-05,
      "loss": 0.5081,
      "step": 136
    },
    {
      "epoch": 0.050128064398097326,
      "grad_norm": 8.479947090148926,
      "learning_rate": 4.919682422451994e-05,
      "loss": 0.2246,
      "step": 137
    },
    {
      "epoch": 0.050493962678375415,
      "grad_norm": 5.625203609466553,
      "learning_rate": 4.918759231905465e-05,
      "loss": 0.4451,
      "step": 138
    },
    {
      "epoch": 0.0508598609586535,
      "grad_norm": 4.000598430633545,
      "learning_rate": 4.917836041358937e-05,
      "loss": 0.2441,
      "step": 139
    },
    {
      "epoch": 0.05122575923893158,
      "grad_norm": 2.281498670578003,
      "learning_rate": 4.916912850812408e-05,
      "loss": 0.1995,
      "step": 140
    },
    {
      "epoch": 0.05159165751920966,
      "grad_norm": 7.903746128082275,
      "learning_rate": 4.915989660265879e-05,
      "loss": 0.4073,
      "step": 141
    },
    {
      "epoch": 0.05195755579948774,
      "grad_norm": 9.161561012268066,
      "learning_rate": 4.91506646971935e-05,
      "loss": 0.2002,
      "step": 142
    },
    {
      "epoch": 0.052323454079765824,
      "grad_norm": 3.4768853187561035,
      "learning_rate": 4.914143279172821e-05,
      "loss": 0.1264,
      "step": 143
    },
    {
      "epoch": 0.052689352360043906,
      "grad_norm": 23.13722038269043,
      "learning_rate": 4.913220088626292e-05,
      "loss": 0.7038,
      "step": 144
    },
    {
      "epoch": 0.05305525064032199,
      "grad_norm": 13.072086334228516,
      "learning_rate": 4.912296898079764e-05,
      "loss": 0.315,
      "step": 145
    },
    {
      "epoch": 0.05342114892060007,
      "grad_norm": 10.033279418945312,
      "learning_rate": 4.911373707533235e-05,
      "loss": 0.2131,
      "step": 146
    },
    {
      "epoch": 0.05378704720087816,
      "grad_norm": 2.152894973754883,
      "learning_rate": 4.910450516986707e-05,
      "loss": 0.1738,
      "step": 147
    },
    {
      "epoch": 0.05415294548115624,
      "grad_norm": 2.812220811843872,
      "learning_rate": 4.909527326440178e-05,
      "loss": 0.1598,
      "step": 148
    },
    {
      "epoch": 0.05451884376143432,
      "grad_norm": 1.048092007637024,
      "learning_rate": 4.908604135893648e-05,
      "loss": 0.2238,
      "step": 149
    },
    {
      "epoch": 0.054884742041712405,
      "grad_norm": 12.10428524017334,
      "learning_rate": 4.90768094534712e-05,
      "loss": 0.3962,
      "step": 150
    },
    {
      "epoch": 0.055250640321990487,
      "grad_norm": 1.1638695001602173,
      "learning_rate": 4.906757754800591e-05,
      "loss": 0.2375,
      "step": 151
    },
    {
      "epoch": 0.05561653860226857,
      "grad_norm": 1.715337872505188,
      "learning_rate": 4.905834564254062e-05,
      "loss": 0.1253,
      "step": 152
    },
    {
      "epoch": 0.05598243688254665,
      "grad_norm": 4.199483394622803,
      "learning_rate": 4.904911373707534e-05,
      "loss": 0.3925,
      "step": 153
    },
    {
      "epoch": 0.05634833516282473,
      "grad_norm": 4.7285919189453125,
      "learning_rate": 4.903988183161005e-05,
      "loss": 0.2692,
      "step": 154
    },
    {
      "epoch": 0.056714233443102814,
      "grad_norm": 2.865082025527954,
      "learning_rate": 4.903064992614476e-05,
      "loss": 0.1675,
      "step": 155
    },
    {
      "epoch": 0.0570801317233809,
      "grad_norm": 2.908289670944214,
      "learning_rate": 4.902141802067947e-05,
      "loss": 0.1282,
      "step": 156
    },
    {
      "epoch": 0.057446030003658985,
      "grad_norm": 4.522146701812744,
      "learning_rate": 4.901218611521418e-05,
      "loss": 0.3633,
      "step": 157
    },
    {
      "epoch": 0.05781192828393707,
      "grad_norm": 1.7177716493606567,
      "learning_rate": 4.90029542097489e-05,
      "loss": 0.0866,
      "step": 158
    },
    {
      "epoch": 0.05817782656421515,
      "grad_norm": 7.0301947593688965,
      "learning_rate": 4.899372230428361e-05,
      "loss": 0.1519,
      "step": 159
    },
    {
      "epoch": 0.05854372484449323,
      "grad_norm": 14.868990898132324,
      "learning_rate": 4.898449039881832e-05,
      "loss": 0.5691,
      "step": 160
    },
    {
      "epoch": 0.05890962312477131,
      "grad_norm": 2.9581832885742188,
      "learning_rate": 4.897525849335303e-05,
      "loss": 0.1187,
      "step": 161
    },
    {
      "epoch": 0.059275521405049394,
      "grad_norm": 5.316128730773926,
      "learning_rate": 4.896602658788774e-05,
      "loss": 0.2189,
      "step": 162
    },
    {
      "epoch": 0.059641419685327476,
      "grad_norm": 2.498837947845459,
      "learning_rate": 4.895679468242245e-05,
      "loss": 0.3424,
      "step": 163
    },
    {
      "epoch": 0.06000731796560556,
      "grad_norm": 4.079292297363281,
      "learning_rate": 4.894756277695717e-05,
      "loss": 0.2501,
      "step": 164
    },
    {
      "epoch": 0.06037321624588365,
      "grad_norm": 2.171116828918457,
      "learning_rate": 4.893833087149188e-05,
      "loss": 0.394,
      "step": 165
    },
    {
      "epoch": 0.06073911452616173,
      "grad_norm": 5.9175825119018555,
      "learning_rate": 4.892909896602659e-05,
      "loss": 0.2374,
      "step": 166
    },
    {
      "epoch": 0.06110501280643981,
      "grad_norm": 4.451068878173828,
      "learning_rate": 4.8919867060561306e-05,
      "loss": 0.1082,
      "step": 167
    },
    {
      "epoch": 0.06147091108671789,
      "grad_norm": 3.5637054443359375,
      "learning_rate": 4.891063515509601e-05,
      "loss": 0.3594,
      "step": 168
    },
    {
      "epoch": 0.061836809366995975,
      "grad_norm": 11.005951881408691,
      "learning_rate": 4.890140324963073e-05,
      "loss": 0.1745,
      "step": 169
    },
    {
      "epoch": 0.06220270764727406,
      "grad_norm": 8.758485794067383,
      "learning_rate": 4.889217134416544e-05,
      "loss": 0.4761,
      "step": 170
    },
    {
      "epoch": 0.06256860592755215,
      "grad_norm": 3.6544129848480225,
      "learning_rate": 4.888293943870015e-05,
      "loss": 0.2754,
      "step": 171
    },
    {
      "epoch": 0.06293450420783023,
      "grad_norm": 5.934715270996094,
      "learning_rate": 4.8873707533234866e-05,
      "loss": 0.2323,
      "step": 172
    },
    {
      "epoch": 0.06330040248810831,
      "grad_norm": 3.1807425022125244,
      "learning_rate": 4.8864475627769576e-05,
      "loss": 0.1222,
      "step": 173
    },
    {
      "epoch": 0.06366630076838639,
      "grad_norm": 5.6125311851501465,
      "learning_rate": 4.885524372230429e-05,
      "loss": 0.2493,
      "step": 174
    },
    {
      "epoch": 0.06403219904866447,
      "grad_norm": 4.502623558044434,
      "learning_rate": 4.8846011816839e-05,
      "loss": 0.0997,
      "step": 175
    },
    {
      "epoch": 0.06439809732894256,
      "grad_norm": 8.692625999450684,
      "learning_rate": 4.883677991137371e-05,
      "loss": 0.1353,
      "step": 176
    },
    {
      "epoch": 0.06476399560922064,
      "grad_norm": 13.937300682067871,
      "learning_rate": 4.8827548005908425e-05,
      "loss": 0.4562,
      "step": 177
    },
    {
      "epoch": 0.06512989388949872,
      "grad_norm": 7.220410346984863,
      "learning_rate": 4.8818316100443136e-05,
      "loss": 0.4244,
      "step": 178
    },
    {
      "epoch": 0.0654957921697768,
      "grad_norm": 5.284034252166748,
      "learning_rate": 4.8809084194977846e-05,
      "loss": 0.2434,
      "step": 179
    },
    {
      "epoch": 0.06586169045005488,
      "grad_norm": 5.376635551452637,
      "learning_rate": 4.879985228951256e-05,
      "loss": 0.6556,
      "step": 180
    },
    {
      "epoch": 0.06622758873033296,
      "grad_norm": 5.865353107452393,
      "learning_rate": 4.879062038404727e-05,
      "loss": 0.1166,
      "step": 181
    },
    {
      "epoch": 0.06659348701061105,
      "grad_norm": 3.814380407333374,
      "learning_rate": 4.878138847858198e-05,
      "loss": 0.1634,
      "step": 182
    },
    {
      "epoch": 0.06695938529088913,
      "grad_norm": 1.9329514503479004,
      "learning_rate": 4.8772156573116695e-05,
      "loss": 0.0675,
      "step": 183
    },
    {
      "epoch": 0.06732528357116721,
      "grad_norm": 7.058513641357422,
      "learning_rate": 4.8762924667651406e-05,
      "loss": 0.3799,
      "step": 184
    },
    {
      "epoch": 0.06769118185144529,
      "grad_norm": 2.051100730895996,
      "learning_rate": 4.8753692762186116e-05,
      "loss": 0.1901,
      "step": 185
    },
    {
      "epoch": 0.06805708013172337,
      "grad_norm": 3.381044387817383,
      "learning_rate": 4.8744460856720834e-05,
      "loss": 0.2781,
      "step": 186
    },
    {
      "epoch": 0.06842297841200147,
      "grad_norm": 2.306873321533203,
      "learning_rate": 4.873522895125554e-05,
      "loss": 0.2881,
      "step": 187
    },
    {
      "epoch": 0.06878887669227955,
      "grad_norm": 5.3051347732543945,
      "learning_rate": 4.8725997045790255e-05,
      "loss": 0.1935,
      "step": 188
    },
    {
      "epoch": 0.06915477497255763,
      "grad_norm": 4.315341949462891,
      "learning_rate": 4.8716765140324965e-05,
      "loss": 0.0894,
      "step": 189
    },
    {
      "epoch": 0.06952067325283572,
      "grad_norm": 10.548255920410156,
      "learning_rate": 4.8707533234859676e-05,
      "loss": 0.3217,
      "step": 190
    },
    {
      "epoch": 0.0698865715331138,
      "grad_norm": 4.472862720489502,
      "learning_rate": 4.869830132939439e-05,
      "loss": 0.1771,
      "step": 191
    },
    {
      "epoch": 0.07025246981339188,
      "grad_norm": 10.484743118286133,
      "learning_rate": 4.8689069423929104e-05,
      "loss": 0.3983,
      "step": 192
    },
    {
      "epoch": 0.07061836809366996,
      "grad_norm": 0.7973571419715881,
      "learning_rate": 4.8679837518463814e-05,
      "loss": 0.0416,
      "step": 193
    },
    {
      "epoch": 0.07098426637394804,
      "grad_norm": 3.8820102214813232,
      "learning_rate": 4.8670605612998525e-05,
      "loss": 0.1116,
      "step": 194
    },
    {
      "epoch": 0.07135016465422613,
      "grad_norm": 5.064569473266602,
      "learning_rate": 4.8661373707533235e-05,
      "loss": 0.2279,
      "step": 195
    },
    {
      "epoch": 0.07171606293450421,
      "grad_norm": 2.254304885864258,
      "learning_rate": 4.8652141802067946e-05,
      "loss": 0.125,
      "step": 196
    },
    {
      "epoch": 0.07208196121478229,
      "grad_norm": 1.2213809490203857,
      "learning_rate": 4.864290989660266e-05,
      "loss": 0.0555,
      "step": 197
    },
    {
      "epoch": 0.07244785949506037,
      "grad_norm": 6.749486923217773,
      "learning_rate": 4.8633677991137374e-05,
      "loss": 0.1215,
      "step": 198
    },
    {
      "epoch": 0.07281375777533845,
      "grad_norm": 7.274303436279297,
      "learning_rate": 4.8624446085672084e-05,
      "loss": 0.2638,
      "step": 199
    },
    {
      "epoch": 0.07317965605561653,
      "grad_norm": 6.351257801055908,
      "learning_rate": 4.8615214180206795e-05,
      "loss": 0.1983,
      "step": 200
    },
    {
      "epoch": 0.07354555433589462,
      "grad_norm": 11.679495811462402,
      "learning_rate": 4.8605982274741505e-05,
      "loss": 0.5467,
      "step": 201
    },
    {
      "epoch": 0.0739114526161727,
      "grad_norm": 4.360590934753418,
      "learning_rate": 4.859675036927622e-05,
      "loss": 0.0557,
      "step": 202
    },
    {
      "epoch": 0.07427735089645078,
      "grad_norm": 6.806382656097412,
      "learning_rate": 4.858751846381093e-05,
      "loss": 0.1673,
      "step": 203
    },
    {
      "epoch": 0.07464324917672886,
      "grad_norm": 2.907670259475708,
      "learning_rate": 4.8578286558345644e-05,
      "loss": 0.0356,
      "step": 204
    },
    {
      "epoch": 0.07500914745700696,
      "grad_norm": 3.492962598800659,
      "learning_rate": 4.856905465288036e-05,
      "loss": 0.048,
      "step": 205
    },
    {
      "epoch": 0.07537504573728504,
      "grad_norm": 8.476929664611816,
      "learning_rate": 4.8559822747415065e-05,
      "loss": 0.1042,
      "step": 206
    },
    {
      "epoch": 0.07574094401756312,
      "grad_norm": 9.719111442565918,
      "learning_rate": 4.855059084194978e-05,
      "loss": 0.4681,
      "step": 207
    },
    {
      "epoch": 0.0761068422978412,
      "grad_norm": 13.863869667053223,
      "learning_rate": 4.854135893648449e-05,
      "loss": 0.6226,
      "step": 208
    },
    {
      "epoch": 0.07647274057811929,
      "grad_norm": 1.3854247331619263,
      "learning_rate": 4.85321270310192e-05,
      "loss": 0.2805,
      "step": 209
    },
    {
      "epoch": 0.07683863885839737,
      "grad_norm": 2.0265049934387207,
      "learning_rate": 4.852289512555392e-05,
      "loss": 0.2828,
      "step": 210
    },
    {
      "epoch": 0.07720453713867545,
      "grad_norm": 2.2022452354431152,
      "learning_rate": 4.851366322008863e-05,
      "loss": 0.0199,
      "step": 211
    },
    {
      "epoch": 0.07757043541895353,
      "grad_norm": 8.413715362548828,
      "learning_rate": 4.850443131462334e-05,
      "loss": 0.0898,
      "step": 212
    },
    {
      "epoch": 0.07793633369923161,
      "grad_norm": 5.957566738128662,
      "learning_rate": 4.849519940915805e-05,
      "loss": 0.0886,
      "step": 213
    },
    {
      "epoch": 0.0783022319795097,
      "grad_norm": 4.596611976623535,
      "learning_rate": 4.848596750369276e-05,
      "loss": 0.1319,
      "step": 214
    },
    {
      "epoch": 0.07866813025978778,
      "grad_norm": 19.640302658081055,
      "learning_rate": 4.847673559822747e-05,
      "loss": 0.5319,
      "step": 215
    },
    {
      "epoch": 0.07903402854006586,
      "grad_norm": 13.456621170043945,
      "learning_rate": 4.846750369276219e-05,
      "loss": 0.3938,
      "step": 216
    },
    {
      "epoch": 0.07939992682034394,
      "grad_norm": 17.173564910888672,
      "learning_rate": 4.84582717872969e-05,
      "loss": 0.6455,
      "step": 217
    },
    {
      "epoch": 0.07976582510062202,
      "grad_norm": 15.207951545715332,
      "learning_rate": 4.844903988183161e-05,
      "loss": 0.8016,
      "step": 218
    },
    {
      "epoch": 0.0801317233809001,
      "grad_norm": 27.557849884033203,
      "learning_rate": 4.843980797636632e-05,
      "loss": 0.4458,
      "step": 219
    },
    {
      "epoch": 0.08049762166117819,
      "grad_norm": 6.136518478393555,
      "learning_rate": 4.843057607090103e-05,
      "loss": 0.1041,
      "step": 220
    },
    {
      "epoch": 0.08086351994145627,
      "grad_norm": 3.698756217956543,
      "learning_rate": 4.842134416543575e-05,
      "loss": 0.4313,
      "step": 221
    },
    {
      "epoch": 0.08122941822173436,
      "grad_norm": 0.46363237500190735,
      "learning_rate": 4.841211225997046e-05,
      "loss": 0.0277,
      "step": 222
    },
    {
      "epoch": 0.08159531650201245,
      "grad_norm": 1.548191785812378,
      "learning_rate": 4.840288035450517e-05,
      "loss": 0.1893,
      "step": 223
    },
    {
      "epoch": 0.08196121478229053,
      "grad_norm": 0.9021000266075134,
      "learning_rate": 4.839364844903989e-05,
      "loss": 0.2276,
      "step": 224
    },
    {
      "epoch": 0.08232711306256861,
      "grad_norm": 9.884462356567383,
      "learning_rate": 4.838441654357459e-05,
      "loss": 0.4641,
      "step": 225
    },
    {
      "epoch": 0.08269301134284669,
      "grad_norm": 2.5808515548706055,
      "learning_rate": 4.83751846381093e-05,
      "loss": 0.0808,
      "step": 226
    },
    {
      "epoch": 0.08305890962312477,
      "grad_norm": 2.3356692790985107,
      "learning_rate": 4.836595273264402e-05,
      "loss": 0.12,
      "step": 227
    },
    {
      "epoch": 0.08342480790340286,
      "grad_norm": 3.923203945159912,
      "learning_rate": 4.835672082717873e-05,
      "loss": 0.3625,
      "step": 228
    },
    {
      "epoch": 0.08379070618368094,
      "grad_norm": 2.7064626216888428,
      "learning_rate": 4.834748892171345e-05,
      "loss": 0.4537,
      "step": 229
    },
    {
      "epoch": 0.08415660446395902,
      "grad_norm": 3.7270255088806152,
      "learning_rate": 4.833825701624816e-05,
      "loss": 0.1931,
      "step": 230
    },
    {
      "epoch": 0.0845225027442371,
      "grad_norm": 4.8914690017700195,
      "learning_rate": 4.832902511078287e-05,
      "loss": 0.3604,
      "step": 231
    },
    {
      "epoch": 0.08488840102451518,
      "grad_norm": 2.3890626430511475,
      "learning_rate": 4.831979320531758e-05,
      "loss": 0.1713,
      "step": 232
    },
    {
      "epoch": 0.08525429930479327,
      "grad_norm": 3.751908540725708,
      "learning_rate": 4.831056129985229e-05,
      "loss": 0.0787,
      "step": 233
    },
    {
      "epoch": 0.08562019758507135,
      "grad_norm": 1.0433484315872192,
      "learning_rate": 4.8301329394387e-05,
      "loss": 0.0757,
      "step": 234
    },
    {
      "epoch": 0.08598609586534943,
      "grad_norm": 6.167947769165039,
      "learning_rate": 4.829209748892172e-05,
      "loss": 0.277,
      "step": 235
    },
    {
      "epoch": 0.08635199414562751,
      "grad_norm": 9.379138946533203,
      "learning_rate": 4.828286558345643e-05,
      "loss": 0.46,
      "step": 236
    },
    {
      "epoch": 0.0867178924259056,
      "grad_norm": 4.78037166595459,
      "learning_rate": 4.827363367799114e-05,
      "loss": 0.1539,
      "step": 237
    },
    {
      "epoch": 0.08708379070618368,
      "grad_norm": 1.2345315217971802,
      "learning_rate": 4.826440177252585e-05,
      "loss": 0.0449,
      "step": 238
    },
    {
      "epoch": 0.08744968898646176,
      "grad_norm": 3.8932247161865234,
      "learning_rate": 4.825516986706056e-05,
      "loss": 0.1113,
      "step": 239
    },
    {
      "epoch": 0.08781558726673985,
      "grad_norm": 5.681131839752197,
      "learning_rate": 4.824593796159528e-05,
      "loss": 0.3163,
      "step": 240
    },
    {
      "epoch": 0.08818148554701793,
      "grad_norm": 4.728951454162598,
      "learning_rate": 4.823670605612999e-05,
      "loss": 0.0704,
      "step": 241
    },
    {
      "epoch": 0.08854738382729602,
      "grad_norm": 4.68056583404541,
      "learning_rate": 4.82274741506647e-05,
      "loss": 0.0824,
      "step": 242
    },
    {
      "epoch": 0.0889132821075741,
      "grad_norm": 19.638029098510742,
      "learning_rate": 4.8218242245199416e-05,
      "loss": 0.405,
      "step": 243
    },
    {
      "epoch": 0.08927918038785218,
      "grad_norm": 14.47612476348877,
      "learning_rate": 4.820901033973412e-05,
      "loss": 0.5463,
      "step": 244
    },
    {
      "epoch": 0.08964507866813026,
      "grad_norm": 17.393314361572266,
      "learning_rate": 4.819977843426883e-05,
      "loss": 0.229,
      "step": 245
    },
    {
      "epoch": 0.09001097694840834,
      "grad_norm": 5.94188928604126,
      "learning_rate": 4.819054652880355e-05,
      "loss": 0.3245,
      "step": 246
    },
    {
      "epoch": 0.09037687522868643,
      "grad_norm": 5.182157516479492,
      "learning_rate": 4.818131462333826e-05,
      "loss": 0.4163,
      "step": 247
    },
    {
      "epoch": 0.09074277350896451,
      "grad_norm": 11.377558708190918,
      "learning_rate": 4.8172082717872976e-05,
      "loss": 0.1869,
      "step": 248
    },
    {
      "epoch": 0.09110867178924259,
      "grad_norm": 2.715696334838867,
      "learning_rate": 4.8162850812407686e-05,
      "loss": 0.0338,
      "step": 249
    },
    {
      "epoch": 0.09147457006952067,
      "grad_norm": 10.855191230773926,
      "learning_rate": 4.81536189069424e-05,
      "loss": 0.1279,
      "step": 250
    },
    {
      "epoch": 0.09184046834979875,
      "grad_norm": 5.091346263885498,
      "learning_rate": 4.814438700147711e-05,
      "loss": 0.1422,
      "step": 251
    },
    {
      "epoch": 0.09220636663007684,
      "grad_norm": 1.8330622911453247,
      "learning_rate": 4.813515509601182e-05,
      "loss": 0.244,
      "step": 252
    },
    {
      "epoch": 0.09257226491035492,
      "grad_norm": 3.5655057430267334,
      "learning_rate": 4.812592319054653e-05,
      "loss": 0.2143,
      "step": 253
    },
    {
      "epoch": 0.092938163190633,
      "grad_norm": 9.727542877197266,
      "learning_rate": 4.8116691285081246e-05,
      "loss": 0.2254,
      "step": 254
    },
    {
      "epoch": 0.09330406147091108,
      "grad_norm": 6.373230457305908,
      "learning_rate": 4.8107459379615956e-05,
      "loss": 0.1954,
      "step": 255
    },
    {
      "epoch": 0.09366995975118916,
      "grad_norm": 6.235262393951416,
      "learning_rate": 4.809822747415067e-05,
      "loss": 0.1864,
      "step": 256
    },
    {
      "epoch": 0.09403585803146725,
      "grad_norm": 2.821021556854248,
      "learning_rate": 4.808899556868538e-05,
      "loss": 0.2585,
      "step": 257
    },
    {
      "epoch": 0.09440175631174534,
      "grad_norm": 9.740594863891602,
      "learning_rate": 4.807976366322009e-05,
      "loss": 0.494,
      "step": 258
    },
    {
      "epoch": 0.09476765459202342,
      "grad_norm": 6.914149761199951,
      "learning_rate": 4.8070531757754805e-05,
      "loss": 0.2085,
      "step": 259
    },
    {
      "epoch": 0.0951335528723015,
      "grad_norm": 3.1155569553375244,
      "learning_rate": 4.8061299852289516e-05,
      "loss": 0.2149,
      "step": 260
    },
    {
      "epoch": 0.09549945115257959,
      "grad_norm": 1.8715332746505737,
      "learning_rate": 4.8052067946824226e-05,
      "loss": 0.0701,
      "step": 261
    },
    {
      "epoch": 0.09586534943285767,
      "grad_norm": 4.00767183303833,
      "learning_rate": 4.8042836041358944e-05,
      "loss": 0.2808,
      "step": 262
    },
    {
      "epoch": 0.09623124771313575,
      "grad_norm": 3.4357683658599854,
      "learning_rate": 4.803360413589365e-05,
      "loss": 0.2629,
      "step": 263
    },
    {
      "epoch": 0.09659714599341383,
      "grad_norm": 2.3887717723846436,
      "learning_rate": 4.802437223042836e-05,
      "loss": 0.1285,
      "step": 264
    },
    {
      "epoch": 0.09696304427369191,
      "grad_norm": 2.129286766052246,
      "learning_rate": 4.8015140324963075e-05,
      "loss": 0.2348,
      "step": 265
    },
    {
      "epoch": 0.09732894255397,
      "grad_norm": 6.198492050170898,
      "learning_rate": 4.8005908419497786e-05,
      "loss": 0.2893,
      "step": 266
    },
    {
      "epoch": 0.09769484083424808,
      "grad_norm": 3.919463872909546,
      "learning_rate": 4.7996676514032496e-05,
      "loss": 0.0944,
      "step": 267
    },
    {
      "epoch": 0.09806073911452616,
      "grad_norm": 4.548121929168701,
      "learning_rate": 4.7987444608567214e-05,
      "loss": 0.1414,
      "step": 268
    },
    {
      "epoch": 0.09842663739480424,
      "grad_norm": 9.471104621887207,
      "learning_rate": 4.7978212703101924e-05,
      "loss": 0.8026,
      "step": 269
    },
    {
      "epoch": 0.09879253567508232,
      "grad_norm": 9.981995582580566,
      "learning_rate": 4.7968980797636635e-05,
      "loss": 0.2315,
      "step": 270
    },
    {
      "epoch": 0.0991584339553604,
      "grad_norm": 8.072916984558105,
      "learning_rate": 4.7959748892171345e-05,
      "loss": 0.3255,
      "step": 271
    },
    {
      "epoch": 0.09952433223563849,
      "grad_norm": 11.252346992492676,
      "learning_rate": 4.7950516986706056e-05,
      "loss": 0.4668,
      "step": 272
    },
    {
      "epoch": 0.09989023051591657,
      "grad_norm": 4.051431179046631,
      "learning_rate": 4.794128508124077e-05,
      "loss": 0.2212,
      "step": 273
    },
    {
      "epoch": 0.10025612879619465,
      "grad_norm": 2.128425359725952,
      "learning_rate": 4.7932053175775484e-05,
      "loss": 0.104,
      "step": 274
    },
    {
      "epoch": 0.10062202707647273,
      "grad_norm": 0.9223390221595764,
      "learning_rate": 4.7922821270310194e-05,
      "loss": 0.0384,
      "step": 275
    },
    {
      "epoch": 0.10098792535675083,
      "grad_norm": 2.2612802982330322,
      "learning_rate": 4.7913589364844905e-05,
      "loss": 0.0538,
      "step": 276
    },
    {
      "epoch": 0.10135382363702891,
      "grad_norm": 3.7372868061065674,
      "learning_rate": 4.7904357459379616e-05,
      "loss": 0.3636,
      "step": 277
    },
    {
      "epoch": 0.101719721917307,
      "grad_norm": 1.3864514827728271,
      "learning_rate": 4.789512555391433e-05,
      "loss": 0.0368,
      "step": 278
    },
    {
      "epoch": 0.10208562019758508,
      "grad_norm": 10.33056354522705,
      "learning_rate": 4.7885893648449043e-05,
      "loss": 0.2238,
      "step": 279
    },
    {
      "epoch": 0.10245151847786316,
      "grad_norm": 6.376560688018799,
      "learning_rate": 4.7876661742983754e-05,
      "loss": 0.1857,
      "step": 280
    },
    {
      "epoch": 0.10281741675814124,
      "grad_norm": 2.153944492340088,
      "learning_rate": 4.786742983751847e-05,
      "loss": 0.2935,
      "step": 281
    },
    {
      "epoch": 0.10318331503841932,
      "grad_norm": 6.125494003295898,
      "learning_rate": 4.7858197932053175e-05,
      "loss": 0.0868,
      "step": 282
    },
    {
      "epoch": 0.1035492133186974,
      "grad_norm": 8.637334823608398,
      "learning_rate": 4.7848966026587886e-05,
      "loss": 0.1462,
      "step": 283
    },
    {
      "epoch": 0.10391511159897548,
      "grad_norm": 5.4953293800354,
      "learning_rate": 4.78397341211226e-05,
      "loss": 0.0524,
      "step": 284
    },
    {
      "epoch": 0.10428100987925357,
      "grad_norm": 8.61683177947998,
      "learning_rate": 4.7830502215657313e-05,
      "loss": 0.0891,
      "step": 285
    },
    {
      "epoch": 0.10464690815953165,
      "grad_norm": 16.717947006225586,
      "learning_rate": 4.7821270310192024e-05,
      "loss": 0.2044,
      "step": 286
    },
    {
      "epoch": 0.10501280643980973,
      "grad_norm": 10.564629554748535,
      "learning_rate": 4.781203840472674e-05,
      "loss": 0.0862,
      "step": 287
    },
    {
      "epoch": 0.10537870472008781,
      "grad_norm": 19.654848098754883,
      "learning_rate": 4.780280649926145e-05,
      "loss": 0.3398,
      "step": 288
    },
    {
      "epoch": 0.1057446030003659,
      "grad_norm": 12.496408462524414,
      "learning_rate": 4.779357459379616e-05,
      "loss": 0.4541,
      "step": 289
    },
    {
      "epoch": 0.10611050128064398,
      "grad_norm": 2.8528127670288086,
      "learning_rate": 4.778434268833087e-05,
      "loss": 0.3511,
      "step": 290
    },
    {
      "epoch": 0.10647639956092206,
      "grad_norm": 2.0429670810699463,
      "learning_rate": 4.7775110782865584e-05,
      "loss": 0.0336,
      "step": 291
    },
    {
      "epoch": 0.10684229784120014,
      "grad_norm": 0.987113893032074,
      "learning_rate": 4.77658788774003e-05,
      "loss": 0.0214,
      "step": 292
    },
    {
      "epoch": 0.10720819612147824,
      "grad_norm": 0.4982430636882782,
      "learning_rate": 4.775664697193501e-05,
      "loss": 0.0149,
      "step": 293
    },
    {
      "epoch": 0.10757409440175632,
      "grad_norm": 9.113299369812012,
      "learning_rate": 4.774741506646972e-05,
      "loss": 0.1912,
      "step": 294
    },
    {
      "epoch": 0.1079399926820344,
      "grad_norm": 3.0378551483154297,
      "learning_rate": 4.773818316100443e-05,
      "loss": 0.0446,
      "step": 295
    },
    {
      "epoch": 0.10830589096231248,
      "grad_norm": 3.919302463531494,
      "learning_rate": 4.772895125553914e-05,
      "loss": 0.1358,
      "step": 296
    },
    {
      "epoch": 0.10867178924259056,
      "grad_norm": 7.631805419921875,
      "learning_rate": 4.7719719350073854e-05,
      "loss": 0.6396,
      "step": 297
    },
    {
      "epoch": 0.10903768752286865,
      "grad_norm": 4.099268436431885,
      "learning_rate": 4.771048744460857e-05,
      "loss": 0.1489,
      "step": 298
    },
    {
      "epoch": 0.10940358580314673,
      "grad_norm": 5.4649858474731445,
      "learning_rate": 4.770125553914328e-05,
      "loss": 0.0887,
      "step": 299
    },
    {
      "epoch": 0.10976948408342481,
      "grad_norm": 8.88396167755127,
      "learning_rate": 4.7692023633678e-05,
      "loss": 0.1761,
      "step": 300
    },
    {
      "epoch": 0.11013538236370289,
      "grad_norm": 4.5620880126953125,
      "learning_rate": 4.76827917282127e-05,
      "loss": 0.0844,
      "step": 301
    },
    {
      "epoch": 0.11050128064398097,
      "grad_norm": 7.4453816413879395,
      "learning_rate": 4.767355982274741e-05,
      "loss": 0.1008,
      "step": 302
    },
    {
      "epoch": 0.11086717892425905,
      "grad_norm": 2.640084981918335,
      "learning_rate": 4.766432791728213e-05,
      "loss": 0.3223,
      "step": 303
    },
    {
      "epoch": 0.11123307720453714,
      "grad_norm": 4.011873722076416,
      "learning_rate": 4.765509601181684e-05,
      "loss": 0.0435,
      "step": 304
    },
    {
      "epoch": 0.11159897548481522,
      "grad_norm": 1.4141103029251099,
      "learning_rate": 4.764586410635155e-05,
      "loss": 0.0364,
      "step": 305
    },
    {
      "epoch": 0.1119648737650933,
      "grad_norm": 10.875825881958008,
      "learning_rate": 4.763663220088627e-05,
      "loss": 0.1171,
      "step": 306
    },
    {
      "epoch": 0.11233077204537138,
      "grad_norm": 4.698378086090088,
      "learning_rate": 4.762740029542098e-05,
      "loss": 0.1993,
      "step": 307
    },
    {
      "epoch": 0.11269667032564946,
      "grad_norm": 5.051383018493652,
      "learning_rate": 4.761816838995569e-05,
      "loss": 0.0366,
      "step": 308
    },
    {
      "epoch": 0.11306256860592755,
      "grad_norm": 18.738624572753906,
      "learning_rate": 4.76089364844904e-05,
      "loss": 0.2846,
      "step": 309
    },
    {
      "epoch": 0.11342846688620563,
      "grad_norm": 5.096854209899902,
      "learning_rate": 4.759970457902511e-05,
      "loss": 0.2994,
      "step": 310
    },
    {
      "epoch": 0.11379436516648372,
      "grad_norm": 1.1445043087005615,
      "learning_rate": 4.759047267355983e-05,
      "loss": 0.0253,
      "step": 311
    },
    {
      "epoch": 0.1141602634467618,
      "grad_norm": 0.3323216438293457,
      "learning_rate": 4.758124076809454e-05,
      "loss": 0.0123,
      "step": 312
    },
    {
      "epoch": 0.11452616172703989,
      "grad_norm": 15.837270736694336,
      "learning_rate": 4.757200886262925e-05,
      "loss": 0.4486,
      "step": 313
    },
    {
      "epoch": 0.11489206000731797,
      "grad_norm": 2.532329797744751,
      "learning_rate": 4.756277695716396e-05,
      "loss": 0.0419,
      "step": 314
    },
    {
      "epoch": 0.11525795828759605,
      "grad_norm": 3.1410083770751953,
      "learning_rate": 4.755354505169867e-05,
      "loss": 0.0295,
      "step": 315
    },
    {
      "epoch": 0.11562385656787413,
      "grad_norm": 6.2131123542785645,
      "learning_rate": 4.754431314623338e-05,
      "loss": 0.0631,
      "step": 316
    },
    {
      "epoch": 0.11598975484815222,
      "grad_norm": 4.796561241149902,
      "learning_rate": 4.75350812407681e-05,
      "loss": 0.2386,
      "step": 317
    },
    {
      "epoch": 0.1163556531284303,
      "grad_norm": 12.927547454833984,
      "learning_rate": 4.752584933530281e-05,
      "loss": 0.1919,
      "step": 318
    },
    {
      "epoch": 0.11672155140870838,
      "grad_norm": 14.62267780303955,
      "learning_rate": 4.7516617429837526e-05,
      "loss": 0.3207,
      "step": 319
    },
    {
      "epoch": 0.11708744968898646,
      "grad_norm": 1.0828193426132202,
      "learning_rate": 4.750738552437223e-05,
      "loss": 0.0119,
      "step": 320
    },
    {
      "epoch": 0.11745334796926454,
      "grad_norm": 0.15686409175395966,
      "learning_rate": 4.749815361890694e-05,
      "loss": 0.0084,
      "step": 321
    },
    {
      "epoch": 0.11781924624954263,
      "grad_norm": 1.1034518480300903,
      "learning_rate": 4.748892171344166e-05,
      "loss": 0.0142,
      "step": 322
    },
    {
      "epoch": 0.11818514452982071,
      "grad_norm": 4.903170585632324,
      "learning_rate": 4.747968980797637e-05,
      "loss": 0.0323,
      "step": 323
    },
    {
      "epoch": 0.11855104281009879,
      "grad_norm": 2.1468453407287598,
      "learning_rate": 4.747045790251108e-05,
      "loss": 0.016,
      "step": 324
    },
    {
      "epoch": 0.11891694109037687,
      "grad_norm": 0.0771903395652771,
      "learning_rate": 4.7461225997045796e-05,
      "loss": 0.0043,
      "step": 325
    },
    {
      "epoch": 0.11928283937065495,
      "grad_norm": 10.148649215698242,
      "learning_rate": 4.745199409158051e-05,
      "loss": 0.244,
      "step": 326
    },
    {
      "epoch": 0.11964873765093303,
      "grad_norm": 14.200273513793945,
      "learning_rate": 4.744276218611521e-05,
      "loss": 0.0742,
      "step": 327
    },
    {
      "epoch": 0.12001463593121112,
      "grad_norm": 0.3587450385093689,
      "learning_rate": 4.743353028064993e-05,
      "loss": 0.008,
      "step": 328
    },
    {
      "epoch": 0.12038053421148921,
      "grad_norm": 0.4920884370803833,
      "learning_rate": 4.742429837518464e-05,
      "loss": 0.0076,
      "step": 329
    },
    {
      "epoch": 0.1207464324917673,
      "grad_norm": 13.664010047912598,
      "learning_rate": 4.7415066469719356e-05,
      "loss": 0.1142,
      "step": 330
    },
    {
      "epoch": 0.12111233077204538,
      "grad_norm": 18.54723358154297,
      "learning_rate": 4.7405834564254066e-05,
      "loss": 0.0925,
      "step": 331
    },
    {
      "epoch": 0.12147822905232346,
      "grad_norm": 0.07781514525413513,
      "learning_rate": 4.739660265878878e-05,
      "loss": 0.0039,
      "step": 332
    },
    {
      "epoch": 0.12184412733260154,
      "grad_norm": 16.35199546813965,
      "learning_rate": 4.738737075332349e-05,
      "loss": 0.1042,
      "step": 333
    },
    {
      "epoch": 0.12221002561287962,
      "grad_norm": 23.42154884338379,
      "learning_rate": 4.73781388478582e-05,
      "loss": 0.561,
      "step": 334
    },
    {
      "epoch": 0.1225759238931577,
      "grad_norm": 13.687488555908203,
      "learning_rate": 4.736890694239291e-05,
      "loss": 0.6268,
      "step": 335
    },
    {
      "epoch": 0.12294182217343579,
      "grad_norm": 0.07808573544025421,
      "learning_rate": 4.7359675036927626e-05,
      "loss": 0.004,
      "step": 336
    },
    {
      "epoch": 0.12330772045371387,
      "grad_norm": 30.600919723510742,
      "learning_rate": 4.7350443131462337e-05,
      "loss": 0.3575,
      "step": 337
    },
    {
      "epoch": 0.12367361873399195,
      "grad_norm": 0.2529444396495819,
      "learning_rate": 4.7341211225997054e-05,
      "loss": 0.0034,
      "step": 338
    },
    {
      "epoch": 0.12403951701427003,
      "grad_norm": 3.3660635948181152,
      "learning_rate": 4.733197932053176e-05,
      "loss": 0.3857,
      "step": 339
    },
    {
      "epoch": 0.12440541529454811,
      "grad_norm": 0.06258987635374069,
      "learning_rate": 4.732274741506647e-05,
      "loss": 0.0031,
      "step": 340
    },
    {
      "epoch": 0.1247713135748262,
      "grad_norm": 38.63154983520508,
      "learning_rate": 4.7313515509601186e-05,
      "loss": 0.4314,
      "step": 341
    },
    {
      "epoch": 0.1251372118551043,
      "grad_norm": 19.615665435791016,
      "learning_rate": 4.7304283604135896e-05,
      "loss": 0.3552,
      "step": 342
    },
    {
      "epoch": 0.12550311013538237,
      "grad_norm": 2.362658977508545,
      "learning_rate": 4.729505169867061e-05,
      "loss": 0.3399,
      "step": 343
    },
    {
      "epoch": 0.12586900841566045,
      "grad_norm": 11.018267631530762,
      "learning_rate": 4.7285819793205324e-05,
      "loss": 0.3477,
      "step": 344
    },
    {
      "epoch": 0.12623490669593854,
      "grad_norm": 20.012357711791992,
      "learning_rate": 4.7276587887740034e-05,
      "loss": 0.1579,
      "step": 345
    },
    {
      "epoch": 0.12660080497621662,
      "grad_norm": 5.392115592956543,
      "learning_rate": 4.726735598227474e-05,
      "loss": 0.5438,
      "step": 346
    },
    {
      "epoch": 0.1269667032564947,
      "grad_norm": 29.174524307250977,
      "learning_rate": 4.7258124076809456e-05,
      "loss": 0.308,
      "step": 347
    },
    {
      "epoch": 0.12733260153677278,
      "grad_norm": 8.563344955444336,
      "learning_rate": 4.7248892171344166e-05,
      "loss": 0.0701,
      "step": 348
    },
    {
      "epoch": 0.12769849981705086,
      "grad_norm": 11.756417274475098,
      "learning_rate": 4.7239660265878883e-05,
      "loss": 0.0958,
      "step": 349
    },
    {
      "epoch": 0.12806439809732895,
      "grad_norm": 0.8567541241645813,
      "learning_rate": 4.7230428360413594e-05,
      "loss": 0.0223,
      "step": 350
    },
    {
      "epoch": 0.12843029637760703,
      "grad_norm": 5.636238098144531,
      "learning_rate": 4.7221196454948305e-05,
      "loss": 0.0427,
      "step": 351
    },
    {
      "epoch": 0.1287961946578851,
      "grad_norm": 9.30325698852539,
      "learning_rate": 4.7211964549483015e-05,
      "loss": 0.2241,
      "step": 352
    },
    {
      "epoch": 0.1291620929381632,
      "grad_norm": 28.706192016601562,
      "learning_rate": 4.7202732644017726e-05,
      "loss": 0.5304,
      "step": 353
    },
    {
      "epoch": 0.12952799121844127,
      "grad_norm": 14.375612258911133,
      "learning_rate": 4.7193500738552436e-05,
      "loss": 0.2398,
      "step": 354
    },
    {
      "epoch": 0.12989388949871936,
      "grad_norm": 1.1089600324630737,
      "learning_rate": 4.7184268833087154e-05,
      "loss": 0.0217,
      "step": 355
    },
    {
      "epoch": 0.13025978777899744,
      "grad_norm": 3.2736690044403076,
      "learning_rate": 4.7175036927621864e-05,
      "loss": 0.0325,
      "step": 356
    },
    {
      "epoch": 0.13062568605927552,
      "grad_norm": 8.307621002197266,
      "learning_rate": 4.7165805022156575e-05,
      "loss": 0.0886,
      "step": 357
    },
    {
      "epoch": 0.1309915843395536,
      "grad_norm": 0.6972277760505676,
      "learning_rate": 4.7156573116691285e-05,
      "loss": 0.0196,
      "step": 358
    },
    {
      "epoch": 0.13135748261983168,
      "grad_norm": 0.16398581862449646,
      "learning_rate": 4.7147341211225996e-05,
      "loss": 0.009,
      "step": 359
    },
    {
      "epoch": 0.13172338090010977,
      "grad_norm": 0.6507591009140015,
      "learning_rate": 4.713810930576071e-05,
      "loss": 0.0138,
      "step": 360
    },
    {
      "epoch": 0.13208927918038785,
      "grad_norm": 0.19375014305114746,
      "learning_rate": 4.7128877400295424e-05,
      "loss": 0.01,
      "step": 361
    },
    {
      "epoch": 0.13245517746066593,
      "grad_norm": 10.574409484863281,
      "learning_rate": 4.7119645494830134e-05,
      "loss": 0.0982,
      "step": 362
    },
    {
      "epoch": 0.132821075740944,
      "grad_norm": 5.764728546142578,
      "learning_rate": 4.711041358936485e-05,
      "loss": 0.0552,
      "step": 363
    },
    {
      "epoch": 0.1331869740212221,
      "grad_norm": 2.9735467433929443,
      "learning_rate": 4.710118168389956e-05,
      "loss": 0.3162,
      "step": 364
    },
    {
      "epoch": 0.13355287230150018,
      "grad_norm": 0.5062566995620728,
      "learning_rate": 4.7091949778434266e-05,
      "loss": 0.0077,
      "step": 365
    },
    {
      "epoch": 0.13391877058177826,
      "grad_norm": 11.19738483428955,
      "learning_rate": 4.708271787296898e-05,
      "loss": 0.6332,
      "step": 366
    },
    {
      "epoch": 0.13428466886205634,
      "grad_norm": 0.08789832890033722,
      "learning_rate": 4.7073485967503694e-05,
      "loss": 0.0047,
      "step": 367
    },
    {
      "epoch": 0.13465056714233442,
      "grad_norm": 0.691762387752533,
      "learning_rate": 4.7064254062038404e-05,
      "loss": 0.0089,
      "step": 368
    },
    {
      "epoch": 0.1350164654226125,
      "grad_norm": 4.507549285888672,
      "learning_rate": 4.705502215657312e-05,
      "loss": 0.0222,
      "step": 369
    },
    {
      "epoch": 0.13538236370289058,
      "grad_norm": 6.829446315765381,
      "learning_rate": 4.704579025110783e-05,
      "loss": 0.2749,
      "step": 370
    },
    {
      "epoch": 0.13574826198316867,
      "grad_norm": 1.9549098014831543,
      "learning_rate": 4.703655834564254e-05,
      "loss": 0.318,
      "step": 371
    },
    {
      "epoch": 0.13611416026344675,
      "grad_norm": 2.233151912689209,
      "learning_rate": 4.702732644017725e-05,
      "loss": 0.3763,
      "step": 372
    },
    {
      "epoch": 0.13648005854372486,
      "grad_norm": 0.14229841530323029,
      "learning_rate": 4.7018094534711964e-05,
      "loss": 0.0048,
      "step": 373
    },
    {
      "epoch": 0.13684595682400294,
      "grad_norm": 10.492718696594238,
      "learning_rate": 4.700886262924668e-05,
      "loss": 0.2269,
      "step": 374
    },
    {
      "epoch": 0.13721185510428102,
      "grad_norm": 15.197863578796387,
      "learning_rate": 4.699963072378139e-05,
      "loss": 0.178,
      "step": 375
    },
    {
      "epoch": 0.1375777533845591,
      "grad_norm": 7.673487186431885,
      "learning_rate": 4.69903988183161e-05,
      "loss": 0.0588,
      "step": 376
    },
    {
      "epoch": 0.13794365166483719,
      "grad_norm": 0.19446665048599243,
      "learning_rate": 4.698116691285081e-05,
      "loss": 0.0084,
      "step": 377
    },
    {
      "epoch": 0.13830954994511527,
      "grad_norm": 1.5933423042297363,
      "learning_rate": 4.697193500738552e-05,
      "loss": 0.2825,
      "step": 378
    },
    {
      "epoch": 0.13867544822539335,
      "grad_norm": 3.8021178245544434,
      "learning_rate": 4.696270310192024e-05,
      "loss": 0.3146,
      "step": 379
    },
    {
      "epoch": 0.13904134650567143,
      "grad_norm": 3.984395980834961,
      "learning_rate": 4.695347119645495e-05,
      "loss": 0.1791,
      "step": 380
    },
    {
      "epoch": 0.1394072447859495,
      "grad_norm": 9.201112747192383,
      "learning_rate": 4.694423929098966e-05,
      "loss": 0.4981,
      "step": 381
    },
    {
      "epoch": 0.1397731430662276,
      "grad_norm": 1.092607855796814,
      "learning_rate": 4.693500738552438e-05,
      "loss": 0.0215,
      "step": 382
    },
    {
      "epoch": 0.14013904134650568,
      "grad_norm": 6.693210601806641,
      "learning_rate": 4.692577548005909e-05,
      "loss": 0.1421,
      "step": 383
    },
    {
      "epoch": 0.14050493962678376,
      "grad_norm": 7.836849689483643,
      "learning_rate": 4.691654357459379e-05,
      "loss": 0.1465,
      "step": 384
    },
    {
      "epoch": 0.14087083790706184,
      "grad_norm": 2.9226388931274414,
      "learning_rate": 4.690731166912851e-05,
      "loss": 0.0695,
      "step": 385
    },
    {
      "epoch": 0.14123673618733992,
      "grad_norm": 7.029890060424805,
      "learning_rate": 4.689807976366322e-05,
      "loss": 0.1181,
      "step": 386
    },
    {
      "epoch": 0.141602634467618,
      "grad_norm": 5.178413391113281,
      "learning_rate": 4.688884785819793e-05,
      "loss": 0.08,
      "step": 387
    },
    {
      "epoch": 0.1419685327478961,
      "grad_norm": 10.055627822875977,
      "learning_rate": 4.687961595273265e-05,
      "loss": 0.3617,
      "step": 388
    },
    {
      "epoch": 0.14233443102817417,
      "grad_norm": 5.91233491897583,
      "learning_rate": 4.687038404726736e-05,
      "loss": 0.3681,
      "step": 389
    },
    {
      "epoch": 0.14270032930845225,
      "grad_norm": 6.537895202636719,
      "learning_rate": 4.686115214180207e-05,
      "loss": 0.097,
      "step": 390
    },
    {
      "epoch": 0.14306622758873033,
      "grad_norm": 0.3971320688724518,
      "learning_rate": 4.685192023633678e-05,
      "loss": 0.0227,
      "step": 391
    },
    {
      "epoch": 0.14343212586900841,
      "grad_norm": 3.753081798553467,
      "learning_rate": 4.684268833087149e-05,
      "loss": 0.074,
      "step": 392
    },
    {
      "epoch": 0.1437980241492865,
      "grad_norm": 3.366687774658203,
      "learning_rate": 4.683345642540621e-05,
      "loss": 0.3088,
      "step": 393
    },
    {
      "epoch": 0.14416392242956458,
      "grad_norm": 4.0805344581604,
      "learning_rate": 4.682422451994092e-05,
      "loss": 0.0711,
      "step": 394
    },
    {
      "epoch": 0.14452982070984266,
      "grad_norm": 0.7477601170539856,
      "learning_rate": 4.681499261447563e-05,
      "loss": 0.0359,
      "step": 395
    },
    {
      "epoch": 0.14489571899012074,
      "grad_norm": 0.4010894298553467,
      "learning_rate": 4.680576070901034e-05,
      "loss": 0.0244,
      "step": 396
    },
    {
      "epoch": 0.14526161727039882,
      "grad_norm": 3.0773062705993652,
      "learning_rate": 4.679652880354505e-05,
      "loss": 0.2049,
      "step": 397
    },
    {
      "epoch": 0.1456275155506769,
      "grad_norm": 5.221221923828125,
      "learning_rate": 4.678729689807976e-05,
      "loss": 0.2586,
      "step": 398
    },
    {
      "epoch": 0.145993413830955,
      "grad_norm": 7.145312309265137,
      "learning_rate": 4.677806499261448e-05,
      "loss": 0.2957,
      "step": 399
    },
    {
      "epoch": 0.14635931211123307,
      "grad_norm": 0.3888159394264221,
      "learning_rate": 4.676883308714919e-05,
      "loss": 0.0174,
      "step": 400
    },
    {
      "epoch": 0.14672521039151115,
      "grad_norm": 6.894836902618408,
      "learning_rate": 4.6759601181683907e-05,
      "loss": 0.129,
      "step": 401
    },
    {
      "epoch": 0.14709110867178923,
      "grad_norm": 0.7734101414680481,
      "learning_rate": 4.675036927621862e-05,
      "loss": 0.0241,
      "step": 402
    },
    {
      "epoch": 0.14745700695206732,
      "grad_norm": 6.139435768127441,
      "learning_rate": 4.674113737075332e-05,
      "loss": 0.1641,
      "step": 403
    },
    {
      "epoch": 0.1478229052323454,
      "grad_norm": 4.359033107757568,
      "learning_rate": 4.673190546528804e-05,
      "loss": 0.2692,
      "step": 404
    },
    {
      "epoch": 0.14818880351262348,
      "grad_norm": 4.155320644378662,
      "learning_rate": 4.672267355982275e-05,
      "loss": 0.2017,
      "step": 405
    },
    {
      "epoch": 0.14855470179290156,
      "grad_norm": 0.3719627559185028,
      "learning_rate": 4.671344165435746e-05,
      "loss": 0.0167,
      "step": 406
    },
    {
      "epoch": 0.14892060007317964,
      "grad_norm": 7.402838706970215,
      "learning_rate": 4.670420974889218e-05,
      "loss": 0.472,
      "step": 407
    },
    {
      "epoch": 0.14928649835345773,
      "grad_norm": 4.168835639953613,
      "learning_rate": 4.669497784342689e-05,
      "loss": 0.0559,
      "step": 408
    },
    {
      "epoch": 0.14965239663373583,
      "grad_norm": 11.947415351867676,
      "learning_rate": 4.66857459379616e-05,
      "loss": 0.1645,
      "step": 409
    },
    {
      "epoch": 0.15001829491401392,
      "grad_norm": 14.421212196350098,
      "learning_rate": 4.667651403249631e-05,
      "loss": 0.4925,
      "step": 410
    },
    {
      "epoch": 0.150384193194292,
      "grad_norm": 0.6077454090118408,
      "learning_rate": 4.666728212703102e-05,
      "loss": 0.0158,
      "step": 411
    },
    {
      "epoch": 0.15075009147457008,
      "grad_norm": 3.870619058609009,
      "learning_rate": 4.6658050221565736e-05,
      "loss": 0.0938,
      "step": 412
    },
    {
      "epoch": 0.15111598975484816,
      "grad_norm": 6.593410968780518,
      "learning_rate": 4.664881831610045e-05,
      "loss": 0.105,
      "step": 413
    },
    {
      "epoch": 0.15148188803512624,
      "grad_norm": 4.441373348236084,
      "learning_rate": 4.663958641063516e-05,
      "loss": 0.3026,
      "step": 414
    },
    {
      "epoch": 0.15184778631540433,
      "grad_norm": 3.330977439880371,
      "learning_rate": 4.663035450516987e-05,
      "loss": 0.6215,
      "step": 415
    },
    {
      "epoch": 0.1522136845956824,
      "grad_norm": 5.540907382965088,
      "learning_rate": 4.662112259970458e-05,
      "loss": 0.0864,
      "step": 416
    },
    {
      "epoch": 0.1525795828759605,
      "grad_norm": 6.051027297973633,
      "learning_rate": 4.661189069423929e-05,
      "loss": 0.0989,
      "step": 417
    },
    {
      "epoch": 0.15294548115623857,
      "grad_norm": 3.366992950439453,
      "learning_rate": 4.6602658788774006e-05,
      "loss": 0.1467,
      "step": 418
    },
    {
      "epoch": 0.15331137943651665,
      "grad_norm": 6.294758319854736,
      "learning_rate": 4.659342688330872e-05,
      "loss": 0.0868,
      "step": 419
    },
    {
      "epoch": 0.15367727771679474,
      "grad_norm": 0.2799408435821533,
      "learning_rate": 4.6584194977843434e-05,
      "loss": 0.0139,
      "step": 420
    },
    {
      "epoch": 0.15404317599707282,
      "grad_norm": 6.442307949066162,
      "learning_rate": 4.6574963072378145e-05,
      "loss": 0.2645,
      "step": 421
    },
    {
      "epoch": 0.1544090742773509,
      "grad_norm": 3.851797580718994,
      "learning_rate": 4.656573116691285e-05,
      "loss": 0.2725,
      "step": 422
    },
    {
      "epoch": 0.15477497255762898,
      "grad_norm": 1.648223876953125,
      "learning_rate": 4.6556499261447566e-05,
      "loss": 0.0356,
      "step": 423
    },
    {
      "epoch": 0.15514087083790706,
      "grad_norm": 4.276697635650635,
      "learning_rate": 4.6547267355982276e-05,
      "loss": 0.0921,
      "step": 424
    },
    {
      "epoch": 0.15550676911818515,
      "grad_norm": 0.8921048045158386,
      "learning_rate": 4.653803545051699e-05,
      "loss": 0.0215,
      "step": 425
    },
    {
      "epoch": 0.15587266739846323,
      "grad_norm": 5.977614402770996,
      "learning_rate": 4.6528803545051704e-05,
      "loss": 0.3688,
      "step": 426
    },
    {
      "epoch": 0.1562385656787413,
      "grad_norm": 0.4278334975242615,
      "learning_rate": 4.6519571639586415e-05,
      "loss": 0.0172,
      "step": 427
    },
    {
      "epoch": 0.1566044639590194,
      "grad_norm": 0.703199028968811,
      "learning_rate": 4.6510339734121125e-05,
      "loss": 0.024,
      "step": 428
    },
    {
      "epoch": 0.15697036223929747,
      "grad_norm": 5.58139181137085,
      "learning_rate": 4.6501107828655836e-05,
      "loss": 0.0451,
      "step": 429
    },
    {
      "epoch": 0.15733626051957555,
      "grad_norm": 2.4946577548980713,
      "learning_rate": 4.6491875923190546e-05,
      "loss": 0.0323,
      "step": 430
    },
    {
      "epoch": 0.15770215879985364,
      "grad_norm": 3.348285436630249,
      "learning_rate": 4.6482644017725264e-05,
      "loss": 0.0366,
      "step": 431
    },
    {
      "epoch": 0.15806805708013172,
      "grad_norm": 5.555301189422607,
      "learning_rate": 4.6473412112259974e-05,
      "loss": 0.1819,
      "step": 432
    },
    {
      "epoch": 0.1584339553604098,
      "grad_norm": 0.16571837663650513,
      "learning_rate": 4.6464180206794685e-05,
      "loss": 0.008,
      "step": 433
    },
    {
      "epoch": 0.15879985364068788,
      "grad_norm": 10.0337553024292,
      "learning_rate": 4.6454948301329395e-05,
      "loss": 0.657,
      "step": 434
    },
    {
      "epoch": 0.15916575192096596,
      "grad_norm": 1.4510560035705566,
      "learning_rate": 4.6445716395864106e-05,
      "loss": 0.2787,
      "step": 435
    },
    {
      "epoch": 0.15953165020124405,
      "grad_norm": 1.9514161348342896,
      "learning_rate": 4.6436484490398816e-05,
      "loss": 0.0306,
      "step": 436
    },
    {
      "epoch": 0.15989754848152213,
      "grad_norm": 8.637591361999512,
      "learning_rate": 4.6427252584933534e-05,
      "loss": 0.3661,
      "step": 437
    },
    {
      "epoch": 0.1602634467618002,
      "grad_norm": 1.4057106971740723,
      "learning_rate": 4.6418020679468244e-05,
      "loss": 0.0184,
      "step": 438
    },
    {
      "epoch": 0.1606293450420783,
      "grad_norm": 0.3630636930465698,
      "learning_rate": 4.6408788774002955e-05,
      "loss": 0.0105,
      "step": 439
    },
    {
      "epoch": 0.16099524332235637,
      "grad_norm": 0.23338459432125092,
      "learning_rate": 4.639955686853767e-05,
      "loss": 0.0092,
      "step": 440
    },
    {
      "epoch": 0.16136114160263446,
      "grad_norm": 8.446063995361328,
      "learning_rate": 4.6390324963072376e-05,
      "loss": 0.0608,
      "step": 441
    },
    {
      "epoch": 0.16172703988291254,
      "grad_norm": 1.273791790008545,
      "learning_rate": 4.638109305760709e-05,
      "loss": 0.0128,
      "step": 442
    },
    {
      "epoch": 0.16209293816319062,
      "grad_norm": 13.506211280822754,
      "learning_rate": 4.6371861152141804e-05,
      "loss": 0.2093,
      "step": 443
    },
    {
      "epoch": 0.16245883644346873,
      "grad_norm": 6.349762439727783,
      "learning_rate": 4.6362629246676514e-05,
      "loss": 0.0438,
      "step": 444
    },
    {
      "epoch": 0.1628247347237468,
      "grad_norm": 7.71196985244751,
      "learning_rate": 4.635339734121123e-05,
      "loss": 0.2313,
      "step": 445
    },
    {
      "epoch": 0.1631906330040249,
      "grad_norm": 12.637846946716309,
      "learning_rate": 4.634416543574594e-05,
      "loss": 0.1357,
      "step": 446
    },
    {
      "epoch": 0.16355653128430298,
      "grad_norm": 4.154121398925781,
      "learning_rate": 4.633493353028065e-05,
      "loss": 0.2808,
      "step": 447
    },
    {
      "epoch": 0.16392242956458106,
      "grad_norm": 0.12696464359760284,
      "learning_rate": 4.632570162481536e-05,
      "loss": 0.0068,
      "step": 448
    },
    {
      "epoch": 0.16428832784485914,
      "grad_norm": 1.5046718120574951,
      "learning_rate": 4.6316469719350074e-05,
      "loss": 0.0195,
      "step": 449
    },
    {
      "epoch": 0.16465422612513722,
      "grad_norm": 1.863145112991333,
      "learning_rate": 4.630723781388479e-05,
      "loss": 0.0244,
      "step": 450
    },
    {
      "epoch": 0.1650201244054153,
      "grad_norm": 0.13451513648033142,
      "learning_rate": 4.62980059084195e-05,
      "loss": 0.0069,
      "step": 451
    },
    {
      "epoch": 0.16538602268569338,
      "grad_norm": 6.539358139038086,
      "learning_rate": 4.628877400295421e-05,
      "loss": 0.0644,
      "step": 452
    },
    {
      "epoch": 0.16575192096597147,
      "grad_norm": 4.220997333526611,
      "learning_rate": 4.627954209748892e-05,
      "loss": 0.0462,
      "step": 453
    },
    {
      "epoch": 0.16611781924624955,
      "grad_norm": 10.935047149658203,
      "learning_rate": 4.6270310192023633e-05,
      "loss": 0.3722,
      "step": 454
    },
    {
      "epoch": 0.16648371752652763,
      "grad_norm": 0.25666293501853943,
      "learning_rate": 4.6261078286558344e-05,
      "loss": 0.0082,
      "step": 455
    },
    {
      "epoch": 0.1668496158068057,
      "grad_norm": 5.781463146209717,
      "learning_rate": 4.625184638109306e-05,
      "loss": 0.0863,
      "step": 456
    },
    {
      "epoch": 0.1672155140870838,
      "grad_norm": 4.454761028289795,
      "learning_rate": 4.624261447562777e-05,
      "loss": 0.2226,
      "step": 457
    },
    {
      "epoch": 0.16758141236736188,
      "grad_norm": 0.20291152596473694,
      "learning_rate": 4.623338257016248e-05,
      "loss": 0.0067,
      "step": 458
    },
    {
      "epoch": 0.16794731064763996,
      "grad_norm": 12.490324020385742,
      "learning_rate": 4.62241506646972e-05,
      "loss": 0.4174,
      "step": 459
    },
    {
      "epoch": 0.16831320892791804,
      "grad_norm": 8.733274459838867,
      "learning_rate": 4.6214918759231903e-05,
      "loss": 0.2076,
      "step": 460
    },
    {
      "epoch": 0.16867910720819612,
      "grad_norm": 0.1145176887512207,
      "learning_rate": 4.620568685376662e-05,
      "loss": 0.0053,
      "step": 461
    },
    {
      "epoch": 0.1690450054884742,
      "grad_norm": 5.052664756774902,
      "learning_rate": 4.619645494830133e-05,
      "loss": 0.255,
      "step": 462
    },
    {
      "epoch": 0.16941090376875229,
      "grad_norm": 5.9484543800354,
      "learning_rate": 4.618722304283604e-05,
      "loss": 0.0732,
      "step": 463
    },
    {
      "epoch": 0.16977680204903037,
      "grad_norm": 0.405955046415329,
      "learning_rate": 4.617799113737076e-05,
      "loss": 0.0083,
      "step": 464
    },
    {
      "epoch": 0.17014270032930845,
      "grad_norm": 6.962174415588379,
      "learning_rate": 4.616875923190547e-05,
      "loss": 0.7544,
      "step": 465
    },
    {
      "epoch": 0.17050859860958653,
      "grad_norm": 0.23931777477264404,
      "learning_rate": 4.615952732644018e-05,
      "loss": 0.0086,
      "step": 466
    },
    {
      "epoch": 0.1708744968898646,
      "grad_norm": 2.8751015663146973,
      "learning_rate": 4.615029542097489e-05,
      "loss": 0.4058,
      "step": 467
    },
    {
      "epoch": 0.1712403951701427,
      "grad_norm": 7.774903774261475,
      "learning_rate": 4.61410635155096e-05,
      "loss": 0.1419,
      "step": 468
    },
    {
      "epoch": 0.17160629345042078,
      "grad_norm": 0.46243008971214294,
      "learning_rate": 4.613183161004431e-05,
      "loss": 0.0122,
      "step": 469
    },
    {
      "epoch": 0.17197219173069886,
      "grad_norm": 0.5731741786003113,
      "learning_rate": 4.612259970457903e-05,
      "loss": 0.0129,
      "step": 470
    },
    {
      "epoch": 0.17233809001097694,
      "grad_norm": 3.591078042984009,
      "learning_rate": 4.611336779911374e-05,
      "loss": 0.2493,
      "step": 471
    },
    {
      "epoch": 0.17270398829125502,
      "grad_norm": 2.7823872566223145,
      "learning_rate": 4.610413589364845e-05,
      "loss": 0.0653,
      "step": 472
    },
    {
      "epoch": 0.1730698865715331,
      "grad_norm": 5.125853538513184,
      "learning_rate": 4.609490398818316e-05,
      "loss": 0.3136,
      "step": 473
    },
    {
      "epoch": 0.1734357848518112,
      "grad_norm": 2.3016879558563232,
      "learning_rate": 4.608567208271787e-05,
      "loss": 0.1388,
      "step": 474
    },
    {
      "epoch": 0.17380168313208927,
      "grad_norm": 1.124167561531067,
      "learning_rate": 4.607644017725259e-05,
      "loss": 0.0323,
      "step": 475
    },
    {
      "epoch": 0.17416758141236735,
      "grad_norm": 2.1380996704101562,
      "learning_rate": 4.60672082717873e-05,
      "loss": 0.0296,
      "step": 476
    },
    {
      "epoch": 0.17453347969264543,
      "grad_norm": 2.1867241859436035,
      "learning_rate": 4.605797636632201e-05,
      "loss": 0.0592,
      "step": 477
    },
    {
      "epoch": 0.17489937797292351,
      "grad_norm": 0.903039813041687,
      "learning_rate": 4.604874446085673e-05,
      "loss": 0.0397,
      "step": 478
    },
    {
      "epoch": 0.1752652762532016,
      "grad_norm": 8.74166202545166,
      "learning_rate": 4.603951255539143e-05,
      "loss": 0.2576,
      "step": 479
    },
    {
      "epoch": 0.1756311745334797,
      "grad_norm": 5.153055191040039,
      "learning_rate": 4.603028064992615e-05,
      "loss": 0.1784,
      "step": 480
    },
    {
      "epoch": 0.1759970728137578,
      "grad_norm": 3.156764268875122,
      "learning_rate": 4.602104874446086e-05,
      "loss": 0.1608,
      "step": 481
    },
    {
      "epoch": 0.17636297109403587,
      "grad_norm": 4.0984272956848145,
      "learning_rate": 4.601181683899557e-05,
      "loss": 0.31,
      "step": 482
    },
    {
      "epoch": 0.17672886937431395,
      "grad_norm": 1.493376612663269,
      "learning_rate": 4.600258493353029e-05,
      "loss": 0.0303,
      "step": 483
    },
    {
      "epoch": 0.17709476765459203,
      "grad_norm": 0.13054679334163666,
      "learning_rate": 4.5993353028065e-05,
      "loss": 0.0052,
      "step": 484
    },
    {
      "epoch": 0.17746066593487012,
      "grad_norm": 10.061956405639648,
      "learning_rate": 4.598412112259971e-05,
      "loss": 0.503,
      "step": 485
    },
    {
      "epoch": 0.1778265642151482,
      "grad_norm": 0.15059182047843933,
      "learning_rate": 4.597488921713442e-05,
      "loss": 0.0057,
      "step": 486
    },
    {
      "epoch": 0.17819246249542628,
      "grad_norm": 0.29584190249443054,
      "learning_rate": 4.596565731166913e-05,
      "loss": 0.0075,
      "step": 487
    },
    {
      "epoch": 0.17855836077570436,
      "grad_norm": 0.13720914721488953,
      "learning_rate": 4.595642540620384e-05,
      "loss": 0.0042,
      "step": 488
    },
    {
      "epoch": 0.17892425905598244,
      "grad_norm": 2.388336658477783,
      "learning_rate": 4.594719350073856e-05,
      "loss": 0.036,
      "step": 489
    },
    {
      "epoch": 0.17929015733626052,
      "grad_norm": 6.114902973175049,
      "learning_rate": 4.593796159527327e-05,
      "loss": 0.0545,
      "step": 490
    },
    {
      "epoch": 0.1796560556165386,
      "grad_norm": 7.434106826782227,
      "learning_rate": 4.592872968980798e-05,
      "loss": 0.2039,
      "step": 491
    },
    {
      "epoch": 0.1800219538968167,
      "grad_norm": 13.407135009765625,
      "learning_rate": 4.591949778434269e-05,
      "loss": 0.1674,
      "step": 492
    },
    {
      "epoch": 0.18038785217709477,
      "grad_norm": 0.1715432107448578,
      "learning_rate": 4.59102658788774e-05,
      "loss": 0.0052,
      "step": 493
    },
    {
      "epoch": 0.18075375045737285,
      "grad_norm": 3.9685895442962646,
      "learning_rate": 4.5901033973412116e-05,
      "loss": 0.0365,
      "step": 494
    },
    {
      "epoch": 0.18111964873765093,
      "grad_norm": 0.6665824055671692,
      "learning_rate": 4.589180206794683e-05,
      "loss": 0.008,
      "step": 495
    },
    {
      "epoch": 0.18148554701792902,
      "grad_norm": 6.588708877563477,
      "learning_rate": 4.588257016248154e-05,
      "loss": 0.3311,
      "step": 496
    },
    {
      "epoch": 0.1818514452982071,
      "grad_norm": 10.899481773376465,
      "learning_rate": 4.5873338257016255e-05,
      "loss": 0.1805,
      "step": 497
    },
    {
      "epoch": 0.18221734357848518,
      "grad_norm": 25.255037307739258,
      "learning_rate": 4.586410635155096e-05,
      "loss": 0.3907,
      "step": 498
    },
    {
      "epoch": 0.18258324185876326,
      "grad_norm": 0.8403284549713135,
      "learning_rate": 4.585487444608567e-05,
      "loss": 0.0075,
      "step": 499
    },
    {
      "epoch": 0.18294914013904134,
      "grad_norm": 10.528026580810547,
      "learning_rate": 4.5845642540620386e-05,
      "loss": 0.1746,
      "step": 500
    },
    {
      "epoch": 0.18331503841931943,
      "grad_norm": 5.861608982086182,
      "learning_rate": 4.58364106351551e-05,
      "loss": 0.0389,
      "step": 501
    },
    {
      "epoch": 0.1836809366995975,
      "grad_norm": 0.12022686004638672,
      "learning_rate": 4.5827178729689814e-05,
      "loss": 0.0032,
      "step": 502
    },
    {
      "epoch": 0.1840468349798756,
      "grad_norm": 2.0724809169769287,
      "learning_rate": 4.5817946824224525e-05,
      "loss": 0.3364,
      "step": 503
    },
    {
      "epoch": 0.18441273326015367,
      "grad_norm": 7.485904216766357,
      "learning_rate": 4.5808714918759235e-05,
      "loss": 0.2989,
      "step": 504
    },
    {
      "epoch": 0.18477863154043175,
      "grad_norm": 6.987902641296387,
      "learning_rate": 4.5799483013293946e-05,
      "loss": 0.0518,
      "step": 505
    },
    {
      "epoch": 0.18514452982070984,
      "grad_norm": 3.796520709991455,
      "learning_rate": 4.5790251107828657e-05,
      "loss": 0.0239,
      "step": 506
    },
    {
      "epoch": 0.18551042810098792,
      "grad_norm": 20.287263870239258,
      "learning_rate": 4.578101920236337e-05,
      "loss": 0.2041,
      "step": 507
    },
    {
      "epoch": 0.185876326381266,
      "grad_norm": 14.597356796264648,
      "learning_rate": 4.5771787296898084e-05,
      "loss": 0.102,
      "step": 508
    },
    {
      "epoch": 0.18624222466154408,
      "grad_norm": 10.59138298034668,
      "learning_rate": 4.5762555391432795e-05,
      "loss": 0.0552,
      "step": 509
    },
    {
      "epoch": 0.18660812294182216,
      "grad_norm": 15.18314266204834,
      "learning_rate": 4.5753323485967505e-05,
      "loss": 0.3271,
      "step": 510
    },
    {
      "epoch": 0.18697402122210025,
      "grad_norm": 4.658552169799805,
      "learning_rate": 4.5744091580502216e-05,
      "loss": 0.4675,
      "step": 511
    },
    {
      "epoch": 0.18733991950237833,
      "grad_norm": 8.360968589782715,
      "learning_rate": 4.5734859675036927e-05,
      "loss": 0.1386,
      "step": 512
    },
    {
      "epoch": 0.1877058177826564,
      "grad_norm": 2.059699535369873,
      "learning_rate": 4.5725627769571644e-05,
      "loss": 0.0194,
      "step": 513
    },
    {
      "epoch": 0.1880717160629345,
      "grad_norm": 5.146965503692627,
      "learning_rate": 4.5716395864106354e-05,
      "loss": 0.1983,
      "step": 514
    },
    {
      "epoch": 0.1884376143432126,
      "grad_norm": 2.3216943740844727,
      "learning_rate": 4.5707163958641065e-05,
      "loss": 0.0134,
      "step": 515
    },
    {
      "epoch": 0.18880351262349068,
      "grad_norm": 0.2633092701435089,
      "learning_rate": 4.569793205317578e-05,
      "loss": 0.0066,
      "step": 516
    },
    {
      "epoch": 0.18916941090376876,
      "grad_norm": 11.077272415161133,
      "learning_rate": 4.5688700147710486e-05,
      "loss": 0.4363,
      "step": 517
    },
    {
      "epoch": 0.18953530918404685,
      "grad_norm": 0.6071349382400513,
      "learning_rate": 4.56794682422452e-05,
      "loss": 0.0122,
      "step": 518
    },
    {
      "epoch": 0.18990120746432493,
      "grad_norm": 0.1338202804327011,
      "learning_rate": 4.5670236336779914e-05,
      "loss": 0.0064,
      "step": 519
    },
    {
      "epoch": 0.190267105744603,
      "grad_norm": 0.7613650560379028,
      "learning_rate": 4.5661004431314625e-05,
      "loss": 0.0085,
      "step": 520
    },
    {
      "epoch": 0.1906330040248811,
      "grad_norm": 0.15354494750499725,
      "learning_rate": 4.565177252584934e-05,
      "loss": 0.0071,
      "step": 521
    },
    {
      "epoch": 0.19099890230515917,
      "grad_norm": 0.11032671481370926,
      "learning_rate": 4.564254062038405e-05,
      "loss": 0.0053,
      "step": 522
    },
    {
      "epoch": 0.19136480058543726,
      "grad_norm": 0.11537329852581024,
      "learning_rate": 4.563330871491876e-05,
      "loss": 0.005,
      "step": 523
    },
    {
      "epoch": 0.19173069886571534,
      "grad_norm": 1.5206363201141357,
      "learning_rate": 4.5624076809453473e-05,
      "loss": 0.3262,
      "step": 524
    },
    {
      "epoch": 0.19209659714599342,
      "grad_norm": 0.11555661261081696,
      "learning_rate": 4.5614844903988184e-05,
      "loss": 0.0051,
      "step": 525
    },
    {
      "epoch": 0.1924624954262715,
      "grad_norm": 5.388230323791504,
      "learning_rate": 4.5605612998522895e-05,
      "loss": 0.0325,
      "step": 526
    },
    {
      "epoch": 0.19282839370654958,
      "grad_norm": 0.5199119448661804,
      "learning_rate": 4.559638109305761e-05,
      "loss": 0.0107,
      "step": 527
    },
    {
      "epoch": 0.19319429198682767,
      "grad_norm": 1.3073394298553467,
      "learning_rate": 4.558714918759232e-05,
      "loss": 0.016,
      "step": 528
    },
    {
      "epoch": 0.19356019026710575,
      "grad_norm": 0.26064297556877136,
      "learning_rate": 4.557791728212703e-05,
      "loss": 0.0071,
      "step": 529
    },
    {
      "epoch": 0.19392608854738383,
      "grad_norm": 0.9716894626617432,
      "learning_rate": 4.5568685376661744e-05,
      "loss": 0.0134,
      "step": 530
    },
    {
      "epoch": 0.1942919868276619,
      "grad_norm": 0.07819928973913193,
      "learning_rate": 4.5559453471196454e-05,
      "loss": 0.0034,
      "step": 531
    },
    {
      "epoch": 0.19465788510794,
      "grad_norm": 0.0836121067404747,
      "learning_rate": 4.555022156573117e-05,
      "loss": 0.0042,
      "step": 532
    },
    {
      "epoch": 0.19502378338821807,
      "grad_norm": 1.4463802576065063,
      "learning_rate": 4.554098966026588e-05,
      "loss": 0.0104,
      "step": 533
    },
    {
      "epoch": 0.19538968166849616,
      "grad_norm": 0.1617857962846756,
      "learning_rate": 4.553175775480059e-05,
      "loss": 0.0068,
      "step": 534
    },
    {
      "epoch": 0.19575557994877424,
      "grad_norm": 0.09998009353876114,
      "learning_rate": 4.552252584933531e-05,
      "loss": 0.0047,
      "step": 535
    },
    {
      "epoch": 0.19612147822905232,
      "grad_norm": 3.2490317821502686,
      "learning_rate": 4.5513293943870014e-05,
      "loss": 0.6065,
      "step": 536
    },
    {
      "epoch": 0.1964873765093304,
      "grad_norm": 0.20827965438365936,
      "learning_rate": 4.5504062038404724e-05,
      "loss": 0.0045,
      "step": 537
    },
    {
      "epoch": 0.19685327478960848,
      "grad_norm": 4.608050346374512,
      "learning_rate": 4.549483013293944e-05,
      "loss": 0.5678,
      "step": 538
    },
    {
      "epoch": 0.19721917306988657,
      "grad_norm": 0.153542622923851,
      "learning_rate": 4.548559822747415e-05,
      "loss": 0.0058,
      "step": 539
    },
    {
      "epoch": 0.19758507135016465,
      "grad_norm": 4.029941082000732,
      "learning_rate": 4.547636632200886e-05,
      "loss": 0.2446,
      "step": 540
    },
    {
      "epoch": 0.19795096963044273,
      "grad_norm": 0.14893275499343872,
      "learning_rate": 4.546713441654358e-05,
      "loss": 0.0066,
      "step": 541
    },
    {
      "epoch": 0.1983168679107208,
      "grad_norm": 0.17854970693588257,
      "learning_rate": 4.545790251107829e-05,
      "loss": 0.0068,
      "step": 542
    },
    {
      "epoch": 0.1986827661909989,
      "grad_norm": 15.0172700881958,
      "learning_rate": 4.5448670605613e-05,
      "loss": 0.4391,
      "step": 543
    },
    {
      "epoch": 0.19904866447127698,
      "grad_norm": 5.249223709106445,
      "learning_rate": 4.543943870014771e-05,
      "loss": 0.6956,
      "step": 544
    },
    {
      "epoch": 0.19941456275155506,
      "grad_norm": 0.16011573374271393,
      "learning_rate": 4.543020679468242e-05,
      "loss": 0.0075,
      "step": 545
    },
    {
      "epoch": 0.19978046103183314,
      "grad_norm": 18.595529556274414,
      "learning_rate": 4.542097488921714e-05,
      "loss": 0.3659,
      "step": 546
    },
    {
      "epoch": 0.20014635931211122,
      "grad_norm": 19.873056411743164,
      "learning_rate": 4.541174298375185e-05,
      "loss": 0.0856,
      "step": 547
    },
    {
      "epoch": 0.2005122575923893,
      "grad_norm": 16.15496253967285,
      "learning_rate": 4.540251107828656e-05,
      "loss": 0.3196,
      "step": 548
    },
    {
      "epoch": 0.20087815587266739,
      "grad_norm": 9.238576889038086,
      "learning_rate": 4.539327917282127e-05,
      "loss": 0.4044,
      "step": 549
    },
    {
      "epoch": 0.20124405415294547,
      "grad_norm": 1.3066864013671875,
      "learning_rate": 4.538404726735598e-05,
      "loss": 0.0147,
      "step": 550
    },
    {
      "epoch": 0.20160995243322358,
      "grad_norm": 11.52284049987793,
      "learning_rate": 4.53748153618907e-05,
      "loss": 0.1558,
      "step": 551
    },
    {
      "epoch": 0.20197585071350166,
      "grad_norm": 11.777478218078613,
      "learning_rate": 4.536558345642541e-05,
      "loss": 0.0416,
      "step": 552
    },
    {
      "epoch": 0.20234174899377974,
      "grad_norm": 26.471080780029297,
      "learning_rate": 4.535635155096012e-05,
      "loss": 0.7018,
      "step": 553
    },
    {
      "epoch": 0.20270764727405782,
      "grad_norm": 0.34573259949684143,
      "learning_rate": 4.534711964549484e-05,
      "loss": 0.0115,
      "step": 554
    },
    {
      "epoch": 0.2030735455543359,
      "grad_norm": 3.018653392791748,
      "learning_rate": 4.533788774002954e-05,
      "loss": 0.534,
      "step": 555
    },
    {
      "epoch": 0.203439443834614,
      "grad_norm": 13.340129852294922,
      "learning_rate": 4.532865583456425e-05,
      "loss": 0.5904,
      "step": 556
    },
    {
      "epoch": 0.20380534211489207,
      "grad_norm": 0.22244854271411896,
      "learning_rate": 4.531942392909897e-05,
      "loss": 0.0123,
      "step": 557
    },
    {
      "epoch": 0.20417124039517015,
      "grad_norm": 1.83343505859375,
      "learning_rate": 4.531019202363368e-05,
      "loss": 0.252,
      "step": 558
    },
    {
      "epoch": 0.20453713867544823,
      "grad_norm": 3.391674280166626,
      "learning_rate": 4.530096011816839e-05,
      "loss": 0.034,
      "step": 559
    },
    {
      "epoch": 0.20490303695572631,
      "grad_norm": 5.464221954345703,
      "learning_rate": 4.529172821270311e-05,
      "loss": 0.046,
      "step": 560
    },
    {
      "epoch": 0.2052689352360044,
      "grad_norm": 0.49637115001678467,
      "learning_rate": 4.528249630723782e-05,
      "loss": 0.0138,
      "step": 561
    },
    {
      "epoch": 0.20563483351628248,
      "grad_norm": 1.1236003637313843,
      "learning_rate": 4.527326440177253e-05,
      "loss": 0.0289,
      "step": 562
    },
    {
      "epoch": 0.20600073179656056,
      "grad_norm": 0.731807291507721,
      "learning_rate": 4.526403249630724e-05,
      "loss": 0.0283,
      "step": 563
    },
    {
      "epoch": 0.20636663007683864,
      "grad_norm": 1.9525943994522095,
      "learning_rate": 4.525480059084195e-05,
      "loss": 0.0308,
      "step": 564
    },
    {
      "epoch": 0.20673252835711672,
      "grad_norm": 10.01131534576416,
      "learning_rate": 4.524556868537667e-05,
      "loss": 0.352,
      "step": 565
    },
    {
      "epoch": 0.2070984266373948,
      "grad_norm": 0.30211442708969116,
      "learning_rate": 4.523633677991138e-05,
      "loss": 0.014,
      "step": 566
    },
    {
      "epoch": 0.2074643249176729,
      "grad_norm": 0.2518497109413147,
      "learning_rate": 4.522710487444609e-05,
      "loss": 0.0126,
      "step": 567
    },
    {
      "epoch": 0.20783022319795097,
      "grad_norm": 0.6619564890861511,
      "learning_rate": 4.52178729689808e-05,
      "loss": 0.0175,
      "step": 568
    },
    {
      "epoch": 0.20819612147822905,
      "grad_norm": 0.38418325781822205,
      "learning_rate": 4.520864106351551e-05,
      "loss": 0.0138,
      "step": 569
    },
    {
      "epoch": 0.20856201975850713,
      "grad_norm": 10.007390975952148,
      "learning_rate": 4.519940915805022e-05,
      "loss": 0.0474,
      "step": 570
    },
    {
      "epoch": 0.20892791803878522,
      "grad_norm": 4.983364105224609,
      "learning_rate": 4.519017725258494e-05,
      "loss": 0.0287,
      "step": 571
    },
    {
      "epoch": 0.2092938163190633,
      "grad_norm": 0.1652548760175705,
      "learning_rate": 4.518094534711965e-05,
      "loss": 0.009,
      "step": 572
    },
    {
      "epoch": 0.20965971459934138,
      "grad_norm": 1.3850018978118896,
      "learning_rate": 4.5171713441654365e-05,
      "loss": 0.2527,
      "step": 573
    },
    {
      "epoch": 0.21002561287961946,
      "grad_norm": 11.287447929382324,
      "learning_rate": 4.516248153618907e-05,
      "loss": 0.5642,
      "step": 574
    },
    {
      "epoch": 0.21039151115989754,
      "grad_norm": 0.8530198931694031,
      "learning_rate": 4.515324963072378e-05,
      "loss": 0.0124,
      "step": 575
    },
    {
      "epoch": 0.21075740944017562,
      "grad_norm": 1.4588704109191895,
      "learning_rate": 4.5144017725258497e-05,
      "loss": 0.2768,
      "step": 576
    },
    {
      "epoch": 0.2111233077204537,
      "grad_norm": 1.5726388692855835,
      "learning_rate": 4.513478581979321e-05,
      "loss": 0.2769,
      "step": 577
    },
    {
      "epoch": 0.2114892060007318,
      "grad_norm": 2.906630516052246,
      "learning_rate": 4.512555391432792e-05,
      "loss": 0.5722,
      "step": 578
    },
    {
      "epoch": 0.21185510428100987,
      "grad_norm": 3.723716974258423,
      "learning_rate": 4.5116322008862635e-05,
      "loss": 0.2409,
      "step": 579
    },
    {
      "epoch": 0.21222100256128795,
      "grad_norm": 1.3847191333770752,
      "learning_rate": 4.5107090103397346e-05,
      "loss": 0.2591,
      "step": 580
    },
    {
      "epoch": 0.21258690084156603,
      "grad_norm": 0.3347782492637634,
      "learning_rate": 4.5097858197932056e-05,
      "loss": 0.0176,
      "step": 581
    },
    {
      "epoch": 0.21295279912184412,
      "grad_norm": 5.475038051605225,
      "learning_rate": 4.508862629246677e-05,
      "loss": 0.4799,
      "step": 582
    },
    {
      "epoch": 0.2133186974021222,
      "grad_norm": 7.805726051330566,
      "learning_rate": 4.507939438700148e-05,
      "loss": 0.1786,
      "step": 583
    },
    {
      "epoch": 0.21368459568240028,
      "grad_norm": 5.708698272705078,
      "learning_rate": 4.5070162481536195e-05,
      "loss": 0.058,
      "step": 584
    },
    {
      "epoch": 0.21405049396267836,
      "grad_norm": 12.116691589355469,
      "learning_rate": 4.5060930576070905e-05,
      "loss": 0.1254,
      "step": 585
    },
    {
      "epoch": 0.21441639224295647,
      "grad_norm": 4.275583267211914,
      "learning_rate": 4.5051698670605616e-05,
      "loss": 0.3281,
      "step": 586
    },
    {
      "epoch": 0.21478229052323455,
      "grad_norm": 5.908675670623779,
      "learning_rate": 4.5042466765140326e-05,
      "loss": 0.0541,
      "step": 587
    },
    {
      "epoch": 0.21514818880351264,
      "grad_norm": 5.301957607269287,
      "learning_rate": 4.503323485967504e-05,
      "loss": 0.2003,
      "step": 588
    },
    {
      "epoch": 0.21551408708379072,
      "grad_norm": 7.374704360961914,
      "learning_rate": 4.502400295420975e-05,
      "loss": 0.0998,
      "step": 589
    },
    {
      "epoch": 0.2158799853640688,
      "grad_norm": 0.5230259299278259,
      "learning_rate": 4.5014771048744465e-05,
      "loss": 0.0314,
      "step": 590
    },
    {
      "epoch": 0.21624588364434688,
      "grad_norm": 0.8740494251251221,
      "learning_rate": 4.5005539143279175e-05,
      "loss": 0.0518,
      "step": 591
    },
    {
      "epoch": 0.21661178192462496,
      "grad_norm": 5.007817268371582,
      "learning_rate": 4.499630723781389e-05,
      "loss": 0.1156,
      "step": 592
    },
    {
      "epoch": 0.21697768020490305,
      "grad_norm": 0.7800716161727905,
      "learning_rate": 4.4987075332348596e-05,
      "loss": 0.0455,
      "step": 593
    },
    {
      "epoch": 0.21734357848518113,
      "grad_norm": 9.219416618347168,
      "learning_rate": 4.497784342688331e-05,
      "loss": 0.1527,
      "step": 594
    },
    {
      "epoch": 0.2177094767654592,
      "grad_norm": 1.4322075843811035,
      "learning_rate": 4.4968611521418024e-05,
      "loss": 0.2004,
      "step": 595
    },
    {
      "epoch": 0.2180753750457373,
      "grad_norm": 5.347282886505127,
      "learning_rate": 4.4959379615952735e-05,
      "loss": 0.1073,
      "step": 596
    },
    {
      "epoch": 0.21844127332601537,
      "grad_norm": 0.47180721163749695,
      "learning_rate": 4.4950147710487445e-05,
      "loss": 0.0285,
      "step": 597
    },
    {
      "epoch": 0.21880717160629345,
      "grad_norm": 6.947980880737305,
      "learning_rate": 4.494091580502216e-05,
      "loss": 0.0766,
      "step": 598
    },
    {
      "epoch": 0.21917306988657154,
      "grad_norm": 12.473930358886719,
      "learning_rate": 4.493168389955687e-05,
      "loss": 0.1718,
      "step": 599
    },
    {
      "epoch": 0.21953896816684962,
      "grad_norm": 1.2170685529708862,
      "learning_rate": 4.492245199409158e-05,
      "loss": 0.2271,
      "step": 600
    },
    {
      "epoch": 0.2199048664471277,
      "grad_norm": 1.8411980867385864,
      "learning_rate": 4.4913220088626294e-05,
      "loss": 0.2024,
      "step": 601
    },
    {
      "epoch": 0.22027076472740578,
      "grad_norm": 0.28615662455558777,
      "learning_rate": 4.4903988183161005e-05,
      "loss": 0.0152,
      "step": 602
    },
    {
      "epoch": 0.22063666300768386,
      "grad_norm": 1.3339136838912964,
      "learning_rate": 4.489475627769572e-05,
      "loss": 0.0275,
      "step": 603
    },
    {
      "epoch": 0.22100256128796195,
      "grad_norm": 3.871006488800049,
      "learning_rate": 4.488552437223043e-05,
      "loss": 0.0549,
      "step": 604
    },
    {
      "epoch": 0.22136845956824003,
      "grad_norm": 1.8293951749801636,
      "learning_rate": 4.487629246676514e-05,
      "loss": 0.0407,
      "step": 605
    },
    {
      "epoch": 0.2217343578485181,
      "grad_norm": 0.9477755427360535,
      "learning_rate": 4.4867060561299854e-05,
      "loss": 0.0306,
      "step": 606
    },
    {
      "epoch": 0.2221002561287962,
      "grad_norm": 2.1268911361694336,
      "learning_rate": 4.4857828655834564e-05,
      "loss": 0.0336,
      "step": 607
    },
    {
      "epoch": 0.22246615440907427,
      "grad_norm": 6.130687236785889,
      "learning_rate": 4.4848596750369275e-05,
      "loss": 0.0526,
      "step": 608
    },
    {
      "epoch": 0.22283205268935236,
      "grad_norm": 4.372295379638672,
      "learning_rate": 4.483936484490399e-05,
      "loss": 0.2234,
      "step": 609
    },
    {
      "epoch": 0.22319795096963044,
      "grad_norm": 0.25932151079177856,
      "learning_rate": 4.48301329394387e-05,
      "loss": 0.0128,
      "step": 610
    },
    {
      "epoch": 0.22356384924990852,
      "grad_norm": 0.20802706480026245,
      "learning_rate": 4.482090103397342e-05,
      "loss": 0.0101,
      "step": 611
    },
    {
      "epoch": 0.2239297475301866,
      "grad_norm": 0.19513404369354248,
      "learning_rate": 4.4811669128508124e-05,
      "loss": 0.0102,
      "step": 612
    },
    {
      "epoch": 0.22429564581046468,
      "grad_norm": 2.44295334815979,
      "learning_rate": 4.4802437223042834e-05,
      "loss": 0.2334,
      "step": 613
    },
    {
      "epoch": 0.22466154409074277,
      "grad_norm": 1.203397274017334,
      "learning_rate": 4.479320531757755e-05,
      "loss": 0.2466,
      "step": 614
    },
    {
      "epoch": 0.22502744237102085,
      "grad_norm": 10.697369575500488,
      "learning_rate": 4.478397341211226e-05,
      "loss": 0.4089,
      "step": 615
    },
    {
      "epoch": 0.22539334065129893,
      "grad_norm": 0.4532126486301422,
      "learning_rate": 4.477474150664697e-05,
      "loss": 0.0149,
      "step": 616
    },
    {
      "epoch": 0.225759238931577,
      "grad_norm": 5.113574504852295,
      "learning_rate": 4.476550960118169e-05,
      "loss": 0.1364,
      "step": 617
    },
    {
      "epoch": 0.2261251372118551,
      "grad_norm": 0.21063631772994995,
      "learning_rate": 4.47562776957164e-05,
      "loss": 0.0091,
      "step": 618
    },
    {
      "epoch": 0.22649103549213317,
      "grad_norm": 15.739707946777344,
      "learning_rate": 4.4747045790251104e-05,
      "loss": 0.198,
      "step": 619
    },
    {
      "epoch": 0.22685693377241126,
      "grad_norm": 7.596031665802002,
      "learning_rate": 4.473781388478582e-05,
      "loss": 0.0395,
      "step": 620
    },
    {
      "epoch": 0.22722283205268934,
      "grad_norm": 0.384515643119812,
      "learning_rate": 4.472858197932053e-05,
      "loss": 0.0176,
      "step": 621
    },
    {
      "epoch": 0.22758873033296745,
      "grad_norm": 9.800819396972656,
      "learning_rate": 4.471935007385525e-05,
      "loss": 0.1137,
      "step": 622
    },
    {
      "epoch": 0.22795462861324553,
      "grad_norm": 1.6263811588287354,
      "learning_rate": 4.471011816838996e-05,
      "loss": 0.2228,
      "step": 623
    },
    {
      "epoch": 0.2283205268935236,
      "grad_norm": 4.159555435180664,
      "learning_rate": 4.470088626292467e-05,
      "loss": 0.1662,
      "step": 624
    },
    {
      "epoch": 0.2286864251738017,
      "grad_norm": 9.096314430236816,
      "learning_rate": 4.469165435745938e-05,
      "loss": 0.3217,
      "step": 625
    },
    {
      "epoch": 0.22905232345407978,
      "grad_norm": 0.7873957753181458,
      "learning_rate": 4.468242245199409e-05,
      "loss": 0.0184,
      "step": 626
    },
    {
      "epoch": 0.22941822173435786,
      "grad_norm": 3.2737371921539307,
      "learning_rate": 4.46731905465288e-05,
      "loss": 0.0391,
      "step": 627
    },
    {
      "epoch": 0.22978412001463594,
      "grad_norm": 1.7988051176071167,
      "learning_rate": 4.466395864106352e-05,
      "loss": 0.0281,
      "step": 628
    },
    {
      "epoch": 0.23015001829491402,
      "grad_norm": 0.6102665662765503,
      "learning_rate": 4.465472673559823e-05,
      "loss": 0.0196,
      "step": 629
    },
    {
      "epoch": 0.2305159165751921,
      "grad_norm": 1.5422698259353638,
      "learning_rate": 4.464549483013294e-05,
      "loss": 0.2903,
      "step": 630
    },
    {
      "epoch": 0.23088181485547019,
      "grad_norm": 1.6843873262405396,
      "learning_rate": 4.463626292466765e-05,
      "loss": 0.0217,
      "step": 631
    },
    {
      "epoch": 0.23124771313574827,
      "grad_norm": 0.1489533632993698,
      "learning_rate": 4.462703101920236e-05,
      "loss": 0.0074,
      "step": 632
    },
    {
      "epoch": 0.23161361141602635,
      "grad_norm": 0.14786334335803986,
      "learning_rate": 4.461779911373708e-05,
      "loss": 0.0068,
      "step": 633
    },
    {
      "epoch": 0.23197950969630443,
      "grad_norm": 0.5003981590270996,
      "learning_rate": 4.460856720827179e-05,
      "loss": 0.015,
      "step": 634
    },
    {
      "epoch": 0.2323454079765825,
      "grad_norm": 6.827515125274658,
      "learning_rate": 4.45993353028065e-05,
      "loss": 0.1704,
      "step": 635
    },
    {
      "epoch": 0.2327113062568606,
      "grad_norm": 3.8238956928253174,
      "learning_rate": 4.459010339734122e-05,
      "loss": 0.4233,
      "step": 636
    },
    {
      "epoch": 0.23307720453713868,
      "grad_norm": 1.8423371315002441,
      "learning_rate": 4.458087149187593e-05,
      "loss": 0.0161,
      "step": 637
    },
    {
      "epoch": 0.23344310281741676,
      "grad_norm": 11.312756538391113,
      "learning_rate": 4.457163958641063e-05,
      "loss": 0.0855,
      "step": 638
    },
    {
      "epoch": 0.23380900109769484,
      "grad_norm": 0.15040363371372223,
      "learning_rate": 4.456240768094535e-05,
      "loss": 0.0084,
      "step": 639
    },
    {
      "epoch": 0.23417489937797292,
      "grad_norm": 0.42416223883628845,
      "learning_rate": 4.455317577548006e-05,
      "loss": 0.0126,
      "step": 640
    },
    {
      "epoch": 0.234540797658251,
      "grad_norm": 9.063898086547852,
      "learning_rate": 4.454394387001477e-05,
      "loss": 0.099,
      "step": 641
    },
    {
      "epoch": 0.2349066959385291,
      "grad_norm": 5.819077491760254,
      "learning_rate": 4.453471196454949e-05,
      "loss": 0.2249,
      "step": 642
    },
    {
      "epoch": 0.23527259421880717,
      "grad_norm": 8.631905555725098,
      "learning_rate": 4.45254800590842e-05,
      "loss": 0.1061,
      "step": 643
    },
    {
      "epoch": 0.23563849249908525,
      "grad_norm": 2.666074752807617,
      "learning_rate": 4.451624815361891e-05,
      "loss": 0.2288,
      "step": 644
    },
    {
      "epoch": 0.23600439077936333,
      "grad_norm": 4.505793571472168,
      "learning_rate": 4.450701624815362e-05,
      "loss": 0.4242,
      "step": 645
    },
    {
      "epoch": 0.23637028905964141,
      "grad_norm": 0.2516579031944275,
      "learning_rate": 4.449778434268833e-05,
      "loss": 0.0106,
      "step": 646
    },
    {
      "epoch": 0.2367361873399195,
      "grad_norm": 0.19839970767498016,
      "learning_rate": 4.448855243722305e-05,
      "loss": 0.0109,
      "step": 647
    },
    {
      "epoch": 0.23710208562019758,
      "grad_norm": 8.525782585144043,
      "learning_rate": 4.447932053175776e-05,
      "loss": 0.1117,
      "step": 648
    },
    {
      "epoch": 0.23746798390047566,
      "grad_norm": 0.28021231293678284,
      "learning_rate": 4.447008862629247e-05,
      "loss": 0.0121,
      "step": 649
    },
    {
      "epoch": 0.23783388218075374,
      "grad_norm": 1.2163045406341553,
      "learning_rate": 4.446085672082718e-05,
      "loss": 0.0211,
      "step": 650
    },
    {
      "epoch": 0.23819978046103182,
      "grad_norm": 1.6750534772872925,
      "learning_rate": 4.445162481536189e-05,
      "loss": 0.0245,
      "step": 651
    },
    {
      "epoch": 0.2385656787413099,
      "grad_norm": 2.5718905925750732,
      "learning_rate": 4.444239290989661e-05,
      "loss": 0.038,
      "step": 652
    },
    {
      "epoch": 0.238931577021588,
      "grad_norm": 0.30700480937957764,
      "learning_rate": 4.443316100443132e-05,
      "loss": 0.0128,
      "step": 653
    },
    {
      "epoch": 0.23929747530186607,
      "grad_norm": 0.4240199625492096,
      "learning_rate": 4.442392909896603e-05,
      "loss": 0.0103,
      "step": 654
    },
    {
      "epoch": 0.23966337358214415,
      "grad_norm": 14.331607818603516,
      "learning_rate": 4.4414697193500745e-05,
      "loss": 0.226,
      "step": 655
    },
    {
      "epoch": 0.24002927186242223,
      "grad_norm": 0.10051868110895157,
      "learning_rate": 4.4405465288035456e-05,
      "loss": 0.0053,
      "step": 656
    },
    {
      "epoch": 0.24039517014270034,
      "grad_norm": 0.17112113535404205,
      "learning_rate": 4.439623338257016e-05,
      "loss": 0.006,
      "step": 657
    },
    {
      "epoch": 0.24076106842297842,
      "grad_norm": 0.20505978167057037,
      "learning_rate": 4.438700147710488e-05,
      "loss": 0.0072,
      "step": 658
    },
    {
      "epoch": 0.2411269667032565,
      "grad_norm": 6.901309490203857,
      "learning_rate": 4.437776957163959e-05,
      "loss": 0.2006,
      "step": 659
    },
    {
      "epoch": 0.2414928649835346,
      "grad_norm": 0.2857922911643982,
      "learning_rate": 4.43685376661743e-05,
      "loss": 0.0057,
      "step": 660
    },
    {
      "epoch": 0.24185876326381267,
      "grad_norm": 1.638012170791626,
      "learning_rate": 4.4359305760709015e-05,
      "loss": 0.3599,
      "step": 661
    },
    {
      "epoch": 0.24222466154409075,
      "grad_norm": 0.056637365370988846,
      "learning_rate": 4.4350073855243726e-05,
      "loss": 0.0028,
      "step": 662
    },
    {
      "epoch": 0.24259055982436883,
      "grad_norm": 0.26021745800971985,
      "learning_rate": 4.4340841949778436e-05,
      "loss": 0.004,
      "step": 663
    },
    {
      "epoch": 0.24295645810464692,
      "grad_norm": 0.1207181066274643,
      "learning_rate": 4.433161004431315e-05,
      "loss": 0.0046,
      "step": 664
    },
    {
      "epoch": 0.243322356384925,
      "grad_norm": 6.714619159698486,
      "learning_rate": 4.432237813884786e-05,
      "loss": 0.0245,
      "step": 665
    },
    {
      "epoch": 0.24368825466520308,
      "grad_norm": 0.1516495794057846,
      "learning_rate": 4.4313146233382575e-05,
      "loss": 0.0053,
      "step": 666
    },
    {
      "epoch": 0.24405415294548116,
      "grad_norm": 10.478436470031738,
      "learning_rate": 4.4303914327917285e-05,
      "loss": 0.1196,
      "step": 667
    },
    {
      "epoch": 0.24442005122575924,
      "grad_norm": 0.06204340234398842,
      "learning_rate": 4.4294682422451996e-05,
      "loss": 0.003,
      "step": 668
    },
    {
      "epoch": 0.24478594950603733,
      "grad_norm": 0.05848762020468712,
      "learning_rate": 4.4285450516986706e-05,
      "loss": 0.0028,
      "step": 669
    },
    {
      "epoch": 0.2451518477863154,
      "grad_norm": 9.174349784851074,
      "learning_rate": 4.427621861152142e-05,
      "loss": 0.163,
      "step": 670
    },
    {
      "epoch": 0.2455177460665935,
      "grad_norm": 3.145625352859497,
      "learning_rate": 4.426698670605613e-05,
      "loss": 0.4599,
      "step": 671
    },
    {
      "epoch": 0.24588364434687157,
      "grad_norm": 7.060890197753906,
      "learning_rate": 4.4257754800590845e-05,
      "loss": 0.478,
      "step": 672
    },
    {
      "epoch": 0.24624954262714965,
      "grad_norm": 2.554048538208008,
      "learning_rate": 4.4248522895125555e-05,
      "loss": 0.2609,
      "step": 673
    },
    {
      "epoch": 0.24661544090742774,
      "grad_norm": 0.258926123380661,
      "learning_rate": 4.423929098966027e-05,
      "loss": 0.007,
      "step": 674
    },
    {
      "epoch": 0.24698133918770582,
      "grad_norm": 5.998249530792236,
      "learning_rate": 4.423005908419498e-05,
      "loss": 0.0363,
      "step": 675
    },
    {
      "epoch": 0.2473472374679839,
      "grad_norm": 21.713909149169922,
      "learning_rate": 4.422082717872969e-05,
      "loss": 0.441,
      "step": 676
    },
    {
      "epoch": 0.24771313574826198,
      "grad_norm": 1.389819622039795,
      "learning_rate": 4.4211595273264404e-05,
      "loss": 0.0189,
      "step": 677
    },
    {
      "epoch": 0.24807903402854006,
      "grad_norm": 16.423236846923828,
      "learning_rate": 4.4202363367799115e-05,
      "loss": 0.5846,
      "step": 678
    },
    {
      "epoch": 0.24844493230881814,
      "grad_norm": 5.936838150024414,
      "learning_rate": 4.4193131462333825e-05,
      "loss": 0.0365,
      "step": 679
    },
    {
      "epoch": 0.24881083058909623,
      "grad_norm": 0.11947967112064362,
      "learning_rate": 4.418389955686854e-05,
      "loss": 0.0058,
      "step": 680
    },
    {
      "epoch": 0.2491767288693743,
      "grad_norm": 17.216337203979492,
      "learning_rate": 4.417466765140325e-05,
      "loss": 0.1672,
      "step": 681
    },
    {
      "epoch": 0.2495426271496524,
      "grad_norm": 0.1200060173869133,
      "learning_rate": 4.4165435745937964e-05,
      "loss": 0.0054,
      "step": 682
    },
    {
      "epoch": 0.24990852542993047,
      "grad_norm": 7.53611946105957,
      "learning_rate": 4.4156203840472674e-05,
      "loss": 0.2182,
      "step": 683
    },
    {
      "epoch": 0.2502744237102086,
      "grad_norm": 8.952842712402344,
      "learning_rate": 4.4146971935007385e-05,
      "loss": 0.2035,
      "step": 684
    },
    {
      "epoch": 0.25064032199048664,
      "grad_norm": 0.09135531634092331,
      "learning_rate": 4.41377400295421e-05,
      "loss": 0.0049,
      "step": 685
    },
    {
      "epoch": 0.25100622027076475,
      "grad_norm": 0.270955890417099,
      "learning_rate": 4.412850812407681e-05,
      "loss": 0.0067,
      "step": 686
    },
    {
      "epoch": 0.2513721185510428,
      "grad_norm": 0.32131659984588623,
      "learning_rate": 4.411927621861152e-05,
      "loss": 0.0083,
      "step": 687
    },
    {
      "epoch": 0.2517380168313209,
      "grad_norm": 1.9748631715774536,
      "learning_rate": 4.4110044313146234e-05,
      "loss": 0.2945,
      "step": 688
    },
    {
      "epoch": 0.25210391511159896,
      "grad_norm": 0.3931090831756592,
      "learning_rate": 4.4100812407680944e-05,
      "loss": 0.0089,
      "step": 689
    },
    {
      "epoch": 0.2524698133918771,
      "grad_norm": 2.148806095123291,
      "learning_rate": 4.4091580502215655e-05,
      "loss": 0.3133,
      "step": 690
    },
    {
      "epoch": 0.25283571167215513,
      "grad_norm": 9.439386367797852,
      "learning_rate": 4.408234859675037e-05,
      "loss": 0.3013,
      "step": 691
    },
    {
      "epoch": 0.25320160995243324,
      "grad_norm": 1.7202483415603638,
      "learning_rate": 4.407311669128508e-05,
      "loss": 0.0354,
      "step": 692
    },
    {
      "epoch": 0.2535675082327113,
      "grad_norm": 0.44638994336128235,
      "learning_rate": 4.40638847858198e-05,
      "loss": 0.0107,
      "step": 693
    },
    {
      "epoch": 0.2539334065129894,
      "grad_norm": 6.441300392150879,
      "learning_rate": 4.405465288035451e-05,
      "loss": 0.1108,
      "step": 694
    },
    {
      "epoch": 0.25429930479326746,
      "grad_norm": 2.5426435470581055,
      "learning_rate": 4.4045420974889215e-05,
      "loss": 0.0244,
      "step": 695
    },
    {
      "epoch": 0.25466520307354557,
      "grad_norm": 0.8721680045127869,
      "learning_rate": 4.403618906942393e-05,
      "loss": 0.0191,
      "step": 696
    },
    {
      "epoch": 0.2550311013538236,
      "grad_norm": 5.9050469398498535,
      "learning_rate": 4.402695716395864e-05,
      "loss": 0.0544,
      "step": 697
    },
    {
      "epoch": 0.25539699963410173,
      "grad_norm": 4.307048320770264,
      "learning_rate": 4.401772525849335e-05,
      "loss": 0.0372,
      "step": 698
    },
    {
      "epoch": 0.2557628979143798,
      "grad_norm": 0.49246060848236084,
      "learning_rate": 4.400849335302807e-05,
      "loss": 0.0128,
      "step": 699
    },
    {
      "epoch": 0.2561287961946579,
      "grad_norm": 0.7605961561203003,
      "learning_rate": 4.399926144756278e-05,
      "loss": 0.0158,
      "step": 700
    },
    {
      "epoch": 0.25649469447493595,
      "grad_norm": 0.14861799776554108,
      "learning_rate": 4.399002954209749e-05,
      "loss": 0.007,
      "step": 701
    },
    {
      "epoch": 0.25686059275521406,
      "grad_norm": 12.686286926269531,
      "learning_rate": 4.39807976366322e-05,
      "loss": 0.0483,
      "step": 702
    },
    {
      "epoch": 0.2572264910354921,
      "grad_norm": 8.501293182373047,
      "learning_rate": 4.397156573116691e-05,
      "loss": 0.0912,
      "step": 703
    },
    {
      "epoch": 0.2575923893157702,
      "grad_norm": 6.342442989349365,
      "learning_rate": 4.396233382570163e-05,
      "loss": 0.5096,
      "step": 704
    },
    {
      "epoch": 0.2579582875960483,
      "grad_norm": 14.682680130004883,
      "learning_rate": 4.395310192023634e-05,
      "loss": 0.279,
      "step": 705
    },
    {
      "epoch": 0.2583241858763264,
      "grad_norm": 0.15779368579387665,
      "learning_rate": 4.394387001477105e-05,
      "loss": 0.004,
      "step": 706
    },
    {
      "epoch": 0.25869008415660444,
      "grad_norm": 8.119088172912598,
      "learning_rate": 4.393463810930576e-05,
      "loss": 0.0733,
      "step": 707
    },
    {
      "epoch": 0.25905598243688255,
      "grad_norm": 1.776018738746643,
      "learning_rate": 4.392540620384047e-05,
      "loss": 0.2843,
      "step": 708
    },
    {
      "epoch": 0.2594218807171606,
      "grad_norm": 0.11289187520742416,
      "learning_rate": 4.391617429837518e-05,
      "loss": 0.0057,
      "step": 709
    },
    {
      "epoch": 0.2597877789974387,
      "grad_norm": 13.540393829345703,
      "learning_rate": 4.39069423929099e-05,
      "loss": 0.0598,
      "step": 710
    },
    {
      "epoch": 0.2601536772777168,
      "grad_norm": 7.577687740325928,
      "learning_rate": 4.389771048744461e-05,
      "loss": 0.3737,
      "step": 711
    },
    {
      "epoch": 0.2605195755579949,
      "grad_norm": 2.124162435531616,
      "learning_rate": 4.388847858197933e-05,
      "loss": 0.4693,
      "step": 712
    },
    {
      "epoch": 0.260885473838273,
      "grad_norm": 0.24201132357120514,
      "learning_rate": 4.387924667651404e-05,
      "loss": 0.0103,
      "step": 713
    },
    {
      "epoch": 0.26125137211855104,
      "grad_norm": 0.23779793083667755,
      "learning_rate": 4.387001477104874e-05,
      "loss": 0.01,
      "step": 714
    },
    {
      "epoch": 0.26161727039882915,
      "grad_norm": 0.18722230195999146,
      "learning_rate": 4.386078286558346e-05,
      "loss": 0.0093,
      "step": 715
    },
    {
      "epoch": 0.2619831686791072,
      "grad_norm": 0.733869194984436,
      "learning_rate": 4.385155096011817e-05,
      "loss": 0.0185,
      "step": 716
    },
    {
      "epoch": 0.2623490669593853,
      "grad_norm": 0.2586134672164917,
      "learning_rate": 4.384231905465288e-05,
      "loss": 0.0113,
      "step": 717
    },
    {
      "epoch": 0.26271496523966337,
      "grad_norm": 7.728055953979492,
      "learning_rate": 4.38330871491876e-05,
      "loss": 0.3239,
      "step": 718
    },
    {
      "epoch": 0.2630808635199415,
      "grad_norm": 0.21208502352237701,
      "learning_rate": 4.382385524372231e-05,
      "loss": 0.0084,
      "step": 719
    },
    {
      "epoch": 0.26344676180021953,
      "grad_norm": 2.3218748569488525,
      "learning_rate": 4.381462333825702e-05,
      "loss": 0.3625,
      "step": 720
    },
    {
      "epoch": 0.26381266008049764,
      "grad_norm": 5.535512447357178,
      "learning_rate": 4.380539143279173e-05,
      "loss": 0.2155,
      "step": 721
    },
    {
      "epoch": 0.2641785583607757,
      "grad_norm": 0.16081471741199493,
      "learning_rate": 4.379615952732644e-05,
      "loss": 0.0088,
      "step": 722
    },
    {
      "epoch": 0.2645444566410538,
      "grad_norm": 0.5056748986244202,
      "learning_rate": 4.378692762186116e-05,
      "loss": 0.0109,
      "step": 723
    },
    {
      "epoch": 0.26491035492133186,
      "grad_norm": 7.1390910148620605,
      "learning_rate": 4.377769571639587e-05,
      "loss": 0.1818,
      "step": 724
    },
    {
      "epoch": 0.26527625320160997,
      "grad_norm": 0.6540467143058777,
      "learning_rate": 4.376846381093058e-05,
      "loss": 0.0116,
      "step": 725
    },
    {
      "epoch": 0.265642151481888,
      "grad_norm": 4.5336456298828125,
      "learning_rate": 4.375923190546529e-05,
      "loss": 0.2049,
      "step": 726
    },
    {
      "epoch": 0.26600804976216613,
      "grad_norm": 0.27774617075920105,
      "learning_rate": 4.375e-05,
      "loss": 0.013,
      "step": 727
    },
    {
      "epoch": 0.2663739480424442,
      "grad_norm": 0.9181379675865173,
      "learning_rate": 4.374076809453471e-05,
      "loss": 0.0129,
      "step": 728
    },
    {
      "epoch": 0.2667398463227223,
      "grad_norm": 6.656878471374512,
      "learning_rate": 4.373153618906943e-05,
      "loss": 0.079,
      "step": 729
    },
    {
      "epoch": 0.26710574460300035,
      "grad_norm": 0.3995917737483978,
      "learning_rate": 4.372230428360414e-05,
      "loss": 0.0119,
      "step": 730
    },
    {
      "epoch": 0.26747164288327846,
      "grad_norm": 11.521944046020508,
      "learning_rate": 4.371307237813885e-05,
      "loss": 0.1189,
      "step": 731
    },
    {
      "epoch": 0.2678375411635565,
      "grad_norm": 0.39537692070007324,
      "learning_rate": 4.3703840472673566e-05,
      "loss": 0.0102,
      "step": 732
    },
    {
      "epoch": 0.2682034394438346,
      "grad_norm": 10.266562461853027,
      "learning_rate": 4.369460856720827e-05,
      "loss": 0.0894,
      "step": 733
    },
    {
      "epoch": 0.2685693377241127,
      "grad_norm": 0.1480223685503006,
      "learning_rate": 4.368537666174299e-05,
      "loss": 0.0072,
      "step": 734
    },
    {
      "epoch": 0.2689352360043908,
      "grad_norm": 11.450078964233398,
      "learning_rate": 4.36761447562777e-05,
      "loss": 0.075,
      "step": 735
    },
    {
      "epoch": 0.26930113428466884,
      "grad_norm": 0.09111589938402176,
      "learning_rate": 4.366691285081241e-05,
      "loss": 0.005,
      "step": 736
    },
    {
      "epoch": 0.26966703256494695,
      "grad_norm": 0.09983330219984055,
      "learning_rate": 4.3657680945347125e-05,
      "loss": 0.0057,
      "step": 737
    },
    {
      "epoch": 0.270032930845225,
      "grad_norm": 1.7703642845153809,
      "learning_rate": 4.3648449039881836e-05,
      "loss": 0.2989,
      "step": 738
    },
    {
      "epoch": 0.2703988291255031,
      "grad_norm": 11.172018051147461,
      "learning_rate": 4.3639217134416546e-05,
      "loss": 0.7553,
      "step": 739
    },
    {
      "epoch": 0.27076472740578117,
      "grad_norm": 7.068397045135498,
      "learning_rate": 4.362998522895126e-05,
      "loss": 0.1819,
      "step": 740
    },
    {
      "epoch": 0.2711306256860593,
      "grad_norm": 0.10632814466953278,
      "learning_rate": 4.362075332348597e-05,
      "loss": 0.0052,
      "step": 741
    },
    {
      "epoch": 0.27149652396633733,
      "grad_norm": 6.836369037628174,
      "learning_rate": 4.361152141802068e-05,
      "loss": 0.0304,
      "step": 742
    },
    {
      "epoch": 0.27186242224661544,
      "grad_norm": 12.645634651184082,
      "learning_rate": 4.3602289512555395e-05,
      "loss": 0.1136,
      "step": 743
    },
    {
      "epoch": 0.2722283205268935,
      "grad_norm": 0.18206921219825745,
      "learning_rate": 4.3593057607090106e-05,
      "loss": 0.0076,
      "step": 744
    },
    {
      "epoch": 0.2725942188071716,
      "grad_norm": 0.5035207867622375,
      "learning_rate": 4.3583825701624817e-05,
      "loss": 0.0108,
      "step": 745
    },
    {
      "epoch": 0.2729601170874497,
      "grad_norm": 0.11500576138496399,
      "learning_rate": 4.357459379615953e-05,
      "loss": 0.0048,
      "step": 746
    },
    {
      "epoch": 0.27332601536772777,
      "grad_norm": 7.850056171417236,
      "learning_rate": 4.356536189069424e-05,
      "loss": 0.0524,
      "step": 747
    },
    {
      "epoch": 0.2736919136480059,
      "grad_norm": 0.33676964044570923,
      "learning_rate": 4.3556129985228955e-05,
      "loss": 0.0073,
      "step": 748
    },
    {
      "epoch": 0.27405781192828393,
      "grad_norm": 1.4884247779846191,
      "learning_rate": 4.3546898079763666e-05,
      "loss": 0.0142,
      "step": 749
    },
    {
      "epoch": 0.27442371020856204,
      "grad_norm": 0.23959441483020782,
      "learning_rate": 4.3537666174298376e-05,
      "loss": 0.0046,
      "step": 750
    },
    {
      "epoch": 0.2747896084888401,
      "grad_norm": 13.565336227416992,
      "learning_rate": 4.3528434268833093e-05,
      "loss": 0.1484,
      "step": 751
    },
    {
      "epoch": 0.2751555067691182,
      "grad_norm": 0.15688569843769073,
      "learning_rate": 4.35192023633678e-05,
      "loss": 0.0055,
      "step": 752
    },
    {
      "epoch": 0.27552140504939626,
      "grad_norm": 0.04977249726653099,
      "learning_rate": 4.3509970457902514e-05,
      "loss": 0.0028,
      "step": 753
    },
    {
      "epoch": 0.27588730332967437,
      "grad_norm": 0.4615950286388397,
      "learning_rate": 4.3500738552437225e-05,
      "loss": 0.0059,
      "step": 754
    },
    {
      "epoch": 0.2762532016099524,
      "grad_norm": 3.1199259757995605,
      "learning_rate": 4.3491506646971936e-05,
      "loss": 0.3184,
      "step": 755
    },
    {
      "epoch": 0.27661909989023054,
      "grad_norm": 0.07847648113965988,
      "learning_rate": 4.348227474150665e-05,
      "loss": 0.004,
      "step": 756
    },
    {
      "epoch": 0.2769849981705086,
      "grad_norm": 0.05812175199389458,
      "learning_rate": 4.3473042836041363e-05,
      "loss": 0.0032,
      "step": 757
    },
    {
      "epoch": 0.2773508964507867,
      "grad_norm": 0.1831536442041397,
      "learning_rate": 4.3463810930576074e-05,
      "loss": 0.003,
      "step": 758
    },
    {
      "epoch": 0.27771679473106475,
      "grad_norm": 9.484633445739746,
      "learning_rate": 4.3454579025110785e-05,
      "loss": 0.5081,
      "step": 759
    },
    {
      "epoch": 0.27808269301134286,
      "grad_norm": 4.974143981933594,
      "learning_rate": 4.3445347119645495e-05,
      "loss": 0.2697,
      "step": 760
    },
    {
      "epoch": 0.2784485912916209,
      "grad_norm": 0.2132636159658432,
      "learning_rate": 4.3436115214180206e-05,
      "loss": 0.0053,
      "step": 761
    },
    {
      "epoch": 0.278814489571899,
      "grad_norm": 2.9623777866363525,
      "learning_rate": 4.342688330871492e-05,
      "loss": 0.2959,
      "step": 762
    },
    {
      "epoch": 0.2791803878521771,
      "grad_norm": 0.6543756127357483,
      "learning_rate": 4.3417651403249634e-05,
      "loss": 0.008,
      "step": 763
    },
    {
      "epoch": 0.2795462861324552,
      "grad_norm": 5.600554466247559,
      "learning_rate": 4.3408419497784344e-05,
      "loss": 0.1588,
      "step": 764
    },
    {
      "epoch": 0.27991218441273324,
      "grad_norm": 3.478074550628662,
      "learning_rate": 4.3399187592319055e-05,
      "loss": 0.2448,
      "step": 765
    },
    {
      "epoch": 0.28027808269301135,
      "grad_norm": 0.37956172227859497,
      "learning_rate": 4.3389955686853765e-05,
      "loss": 0.0094,
      "step": 766
    },
    {
      "epoch": 0.2806439809732894,
      "grad_norm": 0.40919414162635803,
      "learning_rate": 4.338072378138848e-05,
      "loss": 0.009,
      "step": 767
    },
    {
      "epoch": 0.2810098792535675,
      "grad_norm": 13.810245513916016,
      "learning_rate": 4.337149187592319e-05,
      "loss": 0.1978,
      "step": 768
    },
    {
      "epoch": 0.2813757775338456,
      "grad_norm": 0.14242930710315704,
      "learning_rate": 4.3362259970457904e-05,
      "loss": 0.0053,
      "step": 769
    },
    {
      "epoch": 0.2817416758141237,
      "grad_norm": 9.18985366821289,
      "learning_rate": 4.335302806499262e-05,
      "loss": 0.6733,
      "step": 770
    },
    {
      "epoch": 0.28210757409440174,
      "grad_norm": 0.12775175273418427,
      "learning_rate": 4.3343796159527325e-05,
      "loss": 0.0039,
      "step": 771
    },
    {
      "epoch": 0.28247347237467985,
      "grad_norm": 6.997625350952148,
      "learning_rate": 4.3334564254062035e-05,
      "loss": 0.354,
      "step": 772
    },
    {
      "epoch": 0.2828393706549579,
      "grad_norm": 24.58887481689453,
      "learning_rate": 4.332533234859675e-05,
      "loss": 0.4063,
      "step": 773
    },
    {
      "epoch": 0.283205268935236,
      "grad_norm": 0.49375119805336,
      "learning_rate": 4.331610044313146e-05,
      "loss": 0.0129,
      "step": 774
    },
    {
      "epoch": 0.28357116721551406,
      "grad_norm": 0.25038033723831177,
      "learning_rate": 4.330686853766618e-05,
      "loss": 0.0076,
      "step": 775
    },
    {
      "epoch": 0.2839370654957922,
      "grad_norm": 0.4139404594898224,
      "learning_rate": 4.329763663220089e-05,
      "loss": 0.0094,
      "step": 776
    },
    {
      "epoch": 0.2843029637760702,
      "grad_norm": 5.054896354675293,
      "learning_rate": 4.32884047267356e-05,
      "loss": 0.0625,
      "step": 777
    },
    {
      "epoch": 0.28466886205634834,
      "grad_norm": 1.8373545408248901,
      "learning_rate": 4.327917282127031e-05,
      "loss": 0.0236,
      "step": 778
    },
    {
      "epoch": 0.2850347603366264,
      "grad_norm": 3.4502599239349365,
      "learning_rate": 4.326994091580502e-05,
      "loss": 0.3138,
      "step": 779
    },
    {
      "epoch": 0.2854006586169045,
      "grad_norm": 0.563351035118103,
      "learning_rate": 4.326070901033973e-05,
      "loss": 0.0098,
      "step": 780
    },
    {
      "epoch": 0.2857665568971826,
      "grad_norm": 4.968235492706299,
      "learning_rate": 4.325147710487445e-05,
      "loss": 0.2592,
      "step": 781
    },
    {
      "epoch": 0.28613245517746067,
      "grad_norm": 10.451277732849121,
      "learning_rate": 4.324224519940916e-05,
      "loss": 0.4751,
      "step": 782
    },
    {
      "epoch": 0.2864983534577388,
      "grad_norm": 3.2935452461242676,
      "learning_rate": 4.323301329394387e-05,
      "loss": 0.2796,
      "step": 783
    },
    {
      "epoch": 0.28686425173801683,
      "grad_norm": 2.873380422592163,
      "learning_rate": 4.322378138847858e-05,
      "loss": 0.2053,
      "step": 784
    },
    {
      "epoch": 0.28723015001829494,
      "grad_norm": 14.185912132263184,
      "learning_rate": 4.321454948301329e-05,
      "loss": 0.3256,
      "step": 785
    },
    {
      "epoch": 0.287596048298573,
      "grad_norm": 4.222096920013428,
      "learning_rate": 4.320531757754801e-05,
      "loss": 0.2914,
      "step": 786
    },
    {
      "epoch": 0.2879619465788511,
      "grad_norm": 0.8522122502326965,
      "learning_rate": 4.319608567208272e-05,
      "loss": 0.0196,
      "step": 787
    },
    {
      "epoch": 0.28832784485912916,
      "grad_norm": 3.7310521602630615,
      "learning_rate": 4.318685376661743e-05,
      "loss": 0.1238,
      "step": 788
    },
    {
      "epoch": 0.28869374313940727,
      "grad_norm": 7.651554584503174,
      "learning_rate": 4.317762186115215e-05,
      "loss": 0.0818,
      "step": 789
    },
    {
      "epoch": 0.2890596414196853,
      "grad_norm": 2.6233723163604736,
      "learning_rate": 4.316838995568685e-05,
      "loss": 0.0485,
      "step": 790
    },
    {
      "epoch": 0.28942553969996343,
      "grad_norm": 1.5726457834243774,
      "learning_rate": 4.315915805022156e-05,
      "loss": 0.0417,
      "step": 791
    },
    {
      "epoch": 0.2897914379802415,
      "grad_norm": 3.5479040145874023,
      "learning_rate": 4.314992614475628e-05,
      "loss": 0.3901,
      "step": 792
    },
    {
      "epoch": 0.2901573362605196,
      "grad_norm": 2.039557456970215,
      "learning_rate": 4.314069423929099e-05,
      "loss": 0.0505,
      "step": 793
    },
    {
      "epoch": 0.29052323454079765,
      "grad_norm": 10.660993576049805,
      "learning_rate": 4.313146233382571e-05,
      "loss": 0.1047,
      "step": 794
    },
    {
      "epoch": 0.29088913282107576,
      "grad_norm": 13.064915657043457,
      "learning_rate": 4.312223042836042e-05,
      "loss": 0.234,
      "step": 795
    },
    {
      "epoch": 0.2912550311013538,
      "grad_norm": 0.7895233035087585,
      "learning_rate": 4.311299852289513e-05,
      "loss": 0.0222,
      "step": 796
    },
    {
      "epoch": 0.2916209293816319,
      "grad_norm": 12.933052062988281,
      "learning_rate": 4.310376661742984e-05,
      "loss": 0.1559,
      "step": 797
    },
    {
      "epoch": 0.29198682766191,
      "grad_norm": 0.4522295296192169,
      "learning_rate": 4.309453471196455e-05,
      "loss": 0.015,
      "step": 798
    },
    {
      "epoch": 0.2923527259421881,
      "grad_norm": 4.3508782386779785,
      "learning_rate": 4.308530280649926e-05,
      "loss": 0.1459,
      "step": 799
    },
    {
      "epoch": 0.29271862422246614,
      "grad_norm": 3.260099172592163,
      "learning_rate": 4.307607090103398e-05,
      "loss": 0.0366,
      "step": 800
    },
    {
      "epoch": 0.29308452250274425,
      "grad_norm": 0.7427045702934265,
      "learning_rate": 4.306683899556869e-05,
      "loss": 0.011,
      "step": 801
    },
    {
      "epoch": 0.2934504207830223,
      "grad_norm": 0.2809814512729645,
      "learning_rate": 4.30576070901034e-05,
      "loss": 0.0135,
      "step": 802
    },
    {
      "epoch": 0.2938163190633004,
      "grad_norm": 9.654035568237305,
      "learning_rate": 4.304837518463811e-05,
      "loss": 0.3757,
      "step": 803
    },
    {
      "epoch": 0.29418221734357847,
      "grad_norm": 0.14009711146354675,
      "learning_rate": 4.303914327917282e-05,
      "loss": 0.0076,
      "step": 804
    },
    {
      "epoch": 0.2945481156238566,
      "grad_norm": 0.11661780625581741,
      "learning_rate": 4.302991137370754e-05,
      "loss": 0.0062,
      "step": 805
    },
    {
      "epoch": 0.29491401390413463,
      "grad_norm": 0.11140665411949158,
      "learning_rate": 4.302067946824225e-05,
      "loss": 0.006,
      "step": 806
    },
    {
      "epoch": 0.29527991218441274,
      "grad_norm": 9.745707511901855,
      "learning_rate": 4.301144756277696e-05,
      "loss": 0.1739,
      "step": 807
    },
    {
      "epoch": 0.2956458104646908,
      "grad_norm": 0.2756563127040863,
      "learning_rate": 4.3002215657311676e-05,
      "loss": 0.0081,
      "step": 808
    },
    {
      "epoch": 0.2960117087449689,
      "grad_norm": 2.013319730758667,
      "learning_rate": 4.299298375184638e-05,
      "loss": 0.2956,
      "step": 809
    },
    {
      "epoch": 0.29637760702524696,
      "grad_norm": 0.12393741309642792,
      "learning_rate": 4.298375184638109e-05,
      "loss": 0.0055,
      "step": 810
    },
    {
      "epoch": 0.29674350530552507,
      "grad_norm": 0.12953771650791168,
      "learning_rate": 4.297451994091581e-05,
      "loss": 0.0055,
      "step": 811
    },
    {
      "epoch": 0.2971094035858031,
      "grad_norm": 5.879860877990723,
      "learning_rate": 4.296528803545052e-05,
      "loss": 0.0865,
      "step": 812
    },
    {
      "epoch": 0.29747530186608123,
      "grad_norm": 0.08665817230939865,
      "learning_rate": 4.2956056129985236e-05,
      "loss": 0.0039,
      "step": 813
    },
    {
      "epoch": 0.2978412001463593,
      "grad_norm": 0.8561637997627258,
      "learning_rate": 4.2946824224519946e-05,
      "loss": 0.0082,
      "step": 814
    },
    {
      "epoch": 0.2982070984266374,
      "grad_norm": 2.8339734077453613,
      "learning_rate": 4.293759231905466e-05,
      "loss": 0.2972,
      "step": 815
    },
    {
      "epoch": 0.29857299670691545,
      "grad_norm": 3.5401244163513184,
      "learning_rate": 4.292836041358937e-05,
      "loss": 0.3959,
      "step": 816
    },
    {
      "epoch": 0.29893889498719356,
      "grad_norm": 1.963788628578186,
      "learning_rate": 4.291912850812408e-05,
      "loss": 0.2473,
      "step": 817
    },
    {
      "epoch": 0.29930479326747167,
      "grad_norm": 23.93262481689453,
      "learning_rate": 4.290989660265879e-05,
      "loss": 0.2489,
      "step": 818
    },
    {
      "epoch": 0.2996706915477497,
      "grad_norm": 9.00779914855957,
      "learning_rate": 4.2900664697193506e-05,
      "loss": 0.1049,
      "step": 819
    },
    {
      "epoch": 0.30003658982802783,
      "grad_norm": 4.645883560180664,
      "learning_rate": 4.2891432791728216e-05,
      "loss": 0.2802,
      "step": 820
    },
    {
      "epoch": 0.3004024881083059,
      "grad_norm": 0.9032303094863892,
      "learning_rate": 4.288220088626293e-05,
      "loss": 0.0251,
      "step": 821
    },
    {
      "epoch": 0.300768386388584,
      "grad_norm": 3.586804151535034,
      "learning_rate": 4.287296898079764e-05,
      "loss": 0.3304,
      "step": 822
    },
    {
      "epoch": 0.30113428466886205,
      "grad_norm": 2.9988784790039062,
      "learning_rate": 4.286373707533235e-05,
      "loss": 0.019,
      "step": 823
    },
    {
      "epoch": 0.30150018294914016,
      "grad_norm": 11.604828834533691,
      "learning_rate": 4.2854505169867065e-05,
      "loss": 0.1705,
      "step": 824
    },
    {
      "epoch": 0.3018660812294182,
      "grad_norm": 3.0763938426971436,
      "learning_rate": 4.2845273264401776e-05,
      "loss": 0.1759,
      "step": 825
    },
    {
      "epoch": 0.3022319795096963,
      "grad_norm": 0.20633219182491302,
      "learning_rate": 4.2836041358936486e-05,
      "loss": 0.0111,
      "step": 826
    },
    {
      "epoch": 0.3025978777899744,
      "grad_norm": 0.14696624875068665,
      "learning_rate": 4.2826809453471204e-05,
      "loss": 0.0057,
      "step": 827
    },
    {
      "epoch": 0.3029637760702525,
      "grad_norm": 2.0830259323120117,
      "learning_rate": 4.281757754800591e-05,
      "loss": 0.2629,
      "step": 828
    },
    {
      "epoch": 0.30332967435053054,
      "grad_norm": 9.28265380859375,
      "learning_rate": 4.280834564254062e-05,
      "loss": 0.1216,
      "step": 829
    },
    {
      "epoch": 0.30369557263080865,
      "grad_norm": 0.4286320209503174,
      "learning_rate": 4.2799113737075335e-05,
      "loss": 0.0115,
      "step": 830
    },
    {
      "epoch": 0.3040614709110867,
      "grad_norm": 2.1925480365753174,
      "learning_rate": 4.2789881831610046e-05,
      "loss": 0.0321,
      "step": 831
    },
    {
      "epoch": 0.3044273691913648,
      "grad_norm": 1.1456354856491089,
      "learning_rate": 4.2780649926144756e-05,
      "loss": 0.0265,
      "step": 832
    },
    {
      "epoch": 0.30479326747164287,
      "grad_norm": 12.703558921813965,
      "learning_rate": 4.2771418020679474e-05,
      "loss": 0.6136,
      "step": 833
    },
    {
      "epoch": 0.305159165751921,
      "grad_norm": 0.22882463037967682,
      "learning_rate": 4.2762186115214184e-05,
      "loss": 0.0107,
      "step": 834
    },
    {
      "epoch": 0.30552506403219903,
      "grad_norm": 5.065133571624756,
      "learning_rate": 4.2752954209748895e-05,
      "loss": 0.1502,
      "step": 835
    },
    {
      "epoch": 0.30589096231247714,
      "grad_norm": 5.123943328857422,
      "learning_rate": 4.2743722304283605e-05,
      "loss": 0.3983,
      "step": 836
    },
    {
      "epoch": 0.3062568605927552,
      "grad_norm": 6.499660968780518,
      "learning_rate": 4.2734490398818316e-05,
      "loss": 0.0637,
      "step": 837
    },
    {
      "epoch": 0.3066227588730333,
      "grad_norm": 8.890316009521484,
      "learning_rate": 4.272525849335303e-05,
      "loss": 0.527,
      "step": 838
    },
    {
      "epoch": 0.30698865715331136,
      "grad_norm": 10.303182601928711,
      "learning_rate": 4.2716026587887744e-05,
      "loss": 0.0723,
      "step": 839
    },
    {
      "epoch": 0.30735455543358947,
      "grad_norm": 6.685085773468018,
      "learning_rate": 4.2706794682422454e-05,
      "loss": 0.1729,
      "step": 840
    },
    {
      "epoch": 0.3077204537138675,
      "grad_norm": 11.104071617126465,
      "learning_rate": 4.2697562776957165e-05,
      "loss": 0.2057,
      "step": 841
    },
    {
      "epoch": 0.30808635199414564,
      "grad_norm": 8.930619239807129,
      "learning_rate": 4.2688330871491875e-05,
      "loss": 0.3362,
      "step": 842
    },
    {
      "epoch": 0.3084522502744237,
      "grad_norm": 10.296478271484375,
      "learning_rate": 4.2679098966026586e-05,
      "loss": 0.4335,
      "step": 843
    },
    {
      "epoch": 0.3088181485547018,
      "grad_norm": 1.0993674993515015,
      "learning_rate": 4.26698670605613e-05,
      "loss": 0.0306,
      "step": 844
    },
    {
      "epoch": 0.30918404683497985,
      "grad_norm": 0.7165756821632385,
      "learning_rate": 4.2660635155096014e-05,
      "loss": 0.0209,
      "step": 845
    },
    {
      "epoch": 0.30954994511525796,
      "grad_norm": 2.836857795715332,
      "learning_rate": 4.265140324963073e-05,
      "loss": 0.2307,
      "step": 846
    },
    {
      "epoch": 0.309915843395536,
      "grad_norm": 1.2475972175598145,
      "learning_rate": 4.2642171344165435e-05,
      "loss": 0.0179,
      "step": 847
    },
    {
      "epoch": 0.3102817416758141,
      "grad_norm": 0.3214273452758789,
      "learning_rate": 4.2632939438700145e-05,
      "loss": 0.0095,
      "step": 848
    },
    {
      "epoch": 0.3106476399560922,
      "grad_norm": 3.310028314590454,
      "learning_rate": 4.262370753323486e-05,
      "loss": 0.1964,
      "step": 849
    },
    {
      "epoch": 0.3110135382363703,
      "grad_norm": 0.8299633860588074,
      "learning_rate": 4.261447562776957e-05,
      "loss": 0.0162,
      "step": 850
    },
    {
      "epoch": 0.31137943651664834,
      "grad_norm": 0.33223116397857666,
      "learning_rate": 4.2605243722304284e-05,
      "loss": 0.0119,
      "step": 851
    },
    {
      "epoch": 0.31174533479692645,
      "grad_norm": 0.1418132483959198,
      "learning_rate": 4.2596011816839e-05,
      "loss": 0.0036,
      "step": 852
    },
    {
      "epoch": 0.31211123307720456,
      "grad_norm": 5.953847885131836,
      "learning_rate": 4.258677991137371e-05,
      "loss": 0.0429,
      "step": 853
    },
    {
      "epoch": 0.3124771313574826,
      "grad_norm": 1.1526981592178345,
      "learning_rate": 4.257754800590842e-05,
      "loss": 0.0113,
      "step": 854
    },
    {
      "epoch": 0.31284302963776073,
      "grad_norm": 3.0206921100616455,
      "learning_rate": 4.256831610044313e-05,
      "loss": 0.2024,
      "step": 855
    },
    {
      "epoch": 0.3132089279180388,
      "grad_norm": 5.872127532958984,
      "learning_rate": 4.255908419497784e-05,
      "loss": 0.0483,
      "step": 856
    },
    {
      "epoch": 0.3135748261983169,
      "grad_norm": 8.823299407958984,
      "learning_rate": 4.254985228951256e-05,
      "loss": 0.2652,
      "step": 857
    },
    {
      "epoch": 0.31394072447859495,
      "grad_norm": 6.618396759033203,
      "learning_rate": 4.254062038404727e-05,
      "loss": 0.0505,
      "step": 858
    },
    {
      "epoch": 0.31430662275887306,
      "grad_norm": 0.12702369689941406,
      "learning_rate": 4.253138847858198e-05,
      "loss": 0.0041,
      "step": 859
    },
    {
      "epoch": 0.3146725210391511,
      "grad_norm": 3.6588196754455566,
      "learning_rate": 4.252215657311669e-05,
      "loss": 0.151,
      "step": 860
    },
    {
      "epoch": 0.3150384193194292,
      "grad_norm": 6.8194684982299805,
      "learning_rate": 4.25129246676514e-05,
      "loss": 0.1011,
      "step": 861
    },
    {
      "epoch": 0.3154043175997073,
      "grad_norm": 0.48103317618370056,
      "learning_rate": 4.2503692762186113e-05,
      "loss": 0.0115,
      "step": 862
    },
    {
      "epoch": 0.3157702158799854,
      "grad_norm": 0.6979906558990479,
      "learning_rate": 4.249446085672083e-05,
      "loss": 0.0158,
      "step": 863
    },
    {
      "epoch": 0.31613611416026344,
      "grad_norm": 3.0553035736083984,
      "learning_rate": 4.248522895125554e-05,
      "loss": 0.0302,
      "step": 864
    },
    {
      "epoch": 0.31650201244054155,
      "grad_norm": 15.799670219421387,
      "learning_rate": 4.247599704579026e-05,
      "loss": 0.0798,
      "step": 865
    },
    {
      "epoch": 0.3168679107208196,
      "grad_norm": 17.519180297851562,
      "learning_rate": 4.246676514032496e-05,
      "loss": 0.335,
      "step": 866
    },
    {
      "epoch": 0.3172338090010977,
      "grad_norm": 0.12433397024869919,
      "learning_rate": 4.245753323485967e-05,
      "loss": 0.0031,
      "step": 867
    },
    {
      "epoch": 0.31759970728137576,
      "grad_norm": 0.6639726161956787,
      "learning_rate": 4.244830132939439e-05,
      "loss": 0.0176,
      "step": 868
    },
    {
      "epoch": 0.3179656055616539,
      "grad_norm": 1.6115977764129639,
      "learning_rate": 4.24390694239291e-05,
      "loss": 0.0189,
      "step": 869
    },
    {
      "epoch": 0.31833150384193193,
      "grad_norm": 0.2786010503768921,
      "learning_rate": 4.242983751846381e-05,
      "loss": 0.0038,
      "step": 870
    },
    {
      "epoch": 0.31869740212221004,
      "grad_norm": 0.9847123622894287,
      "learning_rate": 4.242060561299853e-05,
      "loss": 0.0119,
      "step": 871
    },
    {
      "epoch": 0.3190633004024881,
      "grad_norm": 0.10187321901321411,
      "learning_rate": 4.241137370753324e-05,
      "loss": 0.0029,
      "step": 872
    },
    {
      "epoch": 0.3194291986827662,
      "grad_norm": 0.13231530785560608,
      "learning_rate": 4.240214180206794e-05,
      "loss": 0.0024,
      "step": 873
    },
    {
      "epoch": 0.31979509696304426,
      "grad_norm": 9.184962272644043,
      "learning_rate": 4.239290989660266e-05,
      "loss": 0.0757,
      "step": 874
    },
    {
      "epoch": 0.32016099524332237,
      "grad_norm": 0.06314680725336075,
      "learning_rate": 4.238367799113737e-05,
      "loss": 0.0016,
      "step": 875
    },
    {
      "epoch": 0.3205268935236004,
      "grad_norm": 0.07915227860212326,
      "learning_rate": 4.237444608567209e-05,
      "loss": 0.0015,
      "step": 876
    },
    {
      "epoch": 0.32089279180387853,
      "grad_norm": 0.3652683198451996,
      "learning_rate": 4.23652141802068e-05,
      "loss": 0.0035,
      "step": 877
    },
    {
      "epoch": 0.3212586900841566,
      "grad_norm": 5.143102169036865,
      "learning_rate": 4.235598227474151e-05,
      "loss": 0.0231,
      "step": 878
    },
    {
      "epoch": 0.3216245883644347,
      "grad_norm": 0.01918194070458412,
      "learning_rate": 4.234675036927622e-05,
      "loss": 0.001,
      "step": 879
    },
    {
      "epoch": 0.32199048664471275,
      "grad_norm": 13.662834167480469,
      "learning_rate": 4.233751846381093e-05,
      "loss": 0.7833,
      "step": 880
    },
    {
      "epoch": 0.32235638492499086,
      "grad_norm": 17.606929779052734,
      "learning_rate": 4.232828655834564e-05,
      "loss": 0.3796,
      "step": 881
    },
    {
      "epoch": 0.3227222832052689,
      "grad_norm": 6.864554405212402,
      "learning_rate": 4.231905465288036e-05,
      "loss": 0.1865,
      "step": 882
    },
    {
      "epoch": 0.323088181485547,
      "grad_norm": 0.027468470856547356,
      "learning_rate": 4.230982274741507e-05,
      "loss": 0.0014,
      "step": 883
    },
    {
      "epoch": 0.3234540797658251,
      "grad_norm": 0.020711755380034447,
      "learning_rate": 4.2300590841949786e-05,
      "loss": 0.001,
      "step": 884
    },
    {
      "epoch": 0.3238199780461032,
      "grad_norm": 0.016014622524380684,
      "learning_rate": 4.229135893648449e-05,
      "loss": 0.0008,
      "step": 885
    },
    {
      "epoch": 0.32418587632638124,
      "grad_norm": 0.025596410036087036,
      "learning_rate": 4.22821270310192e-05,
      "loss": 0.0012,
      "step": 886
    },
    {
      "epoch": 0.32455177460665935,
      "grad_norm": 2.5447287559509277,
      "learning_rate": 4.227289512555392e-05,
      "loss": 0.3879,
      "step": 887
    },
    {
      "epoch": 0.32491767288693746,
      "grad_norm": 4.213871955871582,
      "learning_rate": 4.226366322008863e-05,
      "loss": 0.2653,
      "step": 888
    },
    {
      "epoch": 0.3252835711672155,
      "grad_norm": 0.029904110357165337,
      "learning_rate": 4.225443131462334e-05,
      "loss": 0.0014,
      "step": 889
    },
    {
      "epoch": 0.3256494694474936,
      "grad_norm": 4.056912899017334,
      "learning_rate": 4.2245199409158056e-05,
      "loss": 0.254,
      "step": 890
    },
    {
      "epoch": 0.3260153677277717,
      "grad_norm": 0.47452089190483093,
      "learning_rate": 4.223596750369277e-05,
      "loss": 0.0093,
      "step": 891
    },
    {
      "epoch": 0.3263812660080498,
      "grad_norm": 0.4697895050048828,
      "learning_rate": 4.222673559822747e-05,
      "loss": 0.0122,
      "step": 892
    },
    {
      "epoch": 0.32674716428832784,
      "grad_norm": 3.5024478435516357,
      "learning_rate": 4.221750369276219e-05,
      "loss": 0.0221,
      "step": 893
    },
    {
      "epoch": 0.32711306256860595,
      "grad_norm": 27.428380966186523,
      "learning_rate": 4.22082717872969e-05,
      "loss": 0.0552,
      "step": 894
    },
    {
      "epoch": 0.327478960848884,
      "grad_norm": 4.951428413391113,
      "learning_rate": 4.2199039881831616e-05,
      "loss": 0.0294,
      "step": 895
    },
    {
      "epoch": 0.3278448591291621,
      "grad_norm": 0.262608140707016,
      "learning_rate": 4.2189807976366326e-05,
      "loss": 0.0066,
      "step": 896
    },
    {
      "epoch": 0.32821075740944017,
      "grad_norm": 7.539300441741943,
      "learning_rate": 4.218057607090104e-05,
      "loss": 0.2702,
      "step": 897
    },
    {
      "epoch": 0.3285766556897183,
      "grad_norm": 0.17076116800308228,
      "learning_rate": 4.217134416543575e-05,
      "loss": 0.006,
      "step": 898
    },
    {
      "epoch": 0.32894255396999633,
      "grad_norm": 0.5843523144721985,
      "learning_rate": 4.216211225997046e-05,
      "loss": 0.0131,
      "step": 899
    },
    {
      "epoch": 0.32930845225027444,
      "grad_norm": 1.6738287210464478,
      "learning_rate": 4.215288035450517e-05,
      "loss": 0.2955,
      "step": 900
    },
    {
      "epoch": 0.3296743505305525,
      "grad_norm": 10.328688621520996,
      "learning_rate": 4.2143648449039886e-05,
      "loss": 0.3262,
      "step": 901
    },
    {
      "epoch": 0.3300402488108306,
      "grad_norm": 4.340680122375488,
      "learning_rate": 4.2134416543574596e-05,
      "loss": 0.4174,
      "step": 902
    },
    {
      "epoch": 0.33040614709110866,
      "grad_norm": 5.44715690612793,
      "learning_rate": 4.212518463810931e-05,
      "loss": 0.4264,
      "step": 903
    },
    {
      "epoch": 0.33077204537138677,
      "grad_norm": 0.17621739208698273,
      "learning_rate": 4.211595273264402e-05,
      "loss": 0.0086,
      "step": 904
    },
    {
      "epoch": 0.3311379436516648,
      "grad_norm": 0.21430566906929016,
      "learning_rate": 4.210672082717873e-05,
      "loss": 0.0105,
      "step": 905
    },
    {
      "epoch": 0.33150384193194293,
      "grad_norm": 0.17872819304466248,
      "learning_rate": 4.2097488921713445e-05,
      "loss": 0.0091,
      "step": 906
    },
    {
      "epoch": 0.331869740212221,
      "grad_norm": 0.19284336268901825,
      "learning_rate": 4.2088257016248156e-05,
      "loss": 0.0094,
      "step": 907
    },
    {
      "epoch": 0.3322356384924991,
      "grad_norm": 2.066490650177002,
      "learning_rate": 4.2079025110782866e-05,
      "loss": 0.2351,
      "step": 908
    },
    {
      "epoch": 0.33260153677277715,
      "grad_norm": 0.3203682005405426,
      "learning_rate": 4.2069793205317584e-05,
      "loss": 0.0114,
      "step": 909
    },
    {
      "epoch": 0.33296743505305526,
      "grad_norm": 4.0050201416015625,
      "learning_rate": 4.2060561299852294e-05,
      "loss": 0.1177,
      "step": 910
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 9.730587005615234,
      "learning_rate": 4.2051329394387e-05,
      "loss": 0.0711,
      "step": 911
    },
    {
      "epoch": 0.3336992316136114,
      "grad_norm": 5.25230598449707,
      "learning_rate": 4.2042097488921715e-05,
      "loss": 0.3769,
      "step": 912
    },
    {
      "epoch": 0.3340651298938895,
      "grad_norm": 0.3802769184112549,
      "learning_rate": 4.2032865583456426e-05,
      "loss": 0.0195,
      "step": 913
    },
    {
      "epoch": 0.3344310281741676,
      "grad_norm": 9.43958854675293,
      "learning_rate": 4.202363367799114e-05,
      "loss": 0.0695,
      "step": 914
    },
    {
      "epoch": 0.33479692645444564,
      "grad_norm": 1.2840224504470825,
      "learning_rate": 4.2014401772525854e-05,
      "loss": 0.184,
      "step": 915
    },
    {
      "epoch": 0.33516282473472375,
      "grad_norm": 1.7646077871322632,
      "learning_rate": 4.2005169867060564e-05,
      "loss": 0.1759,
      "step": 916
    },
    {
      "epoch": 0.3355287230150018,
      "grad_norm": 0.3265163004398346,
      "learning_rate": 4.1995937961595275e-05,
      "loss": 0.0164,
      "step": 917
    },
    {
      "epoch": 0.3358946212952799,
      "grad_norm": 0.7729976177215576,
      "learning_rate": 4.1986706056129985e-05,
      "loss": 0.0312,
      "step": 918
    },
    {
      "epoch": 0.33626051957555797,
      "grad_norm": 5.820679664611816,
      "learning_rate": 4.1977474150664696e-05,
      "loss": 0.0676,
      "step": 919
    },
    {
      "epoch": 0.3366264178558361,
      "grad_norm": 2.185911178588867,
      "learning_rate": 4.196824224519941e-05,
      "loss": 0.2117,
      "step": 920
    },
    {
      "epoch": 0.33699231613611413,
      "grad_norm": 8.281041145324707,
      "learning_rate": 4.1959010339734124e-05,
      "loss": 0.0976,
      "step": 921
    },
    {
      "epoch": 0.33735821441639224,
      "grad_norm": 1.176608681678772,
      "learning_rate": 4.1949778434268834e-05,
      "loss": 0.0334,
      "step": 922
    },
    {
      "epoch": 0.33772411269667035,
      "grad_norm": 10.489508628845215,
      "learning_rate": 4.1940546528803545e-05,
      "loss": 0.2709,
      "step": 923
    },
    {
      "epoch": 0.3380900109769484,
      "grad_norm": 1.8659950494766235,
      "learning_rate": 4.1931314623338256e-05,
      "loss": 0.1178,
      "step": 924
    },
    {
      "epoch": 0.3384559092572265,
      "grad_norm": 3.0841116905212402,
      "learning_rate": 4.192208271787297e-05,
      "loss": 0.1206,
      "step": 925
    },
    {
      "epoch": 0.33882180753750457,
      "grad_norm": 0.5099377036094666,
      "learning_rate": 4.1912850812407683e-05,
      "loss": 0.0187,
      "step": 926
    },
    {
      "epoch": 0.3391877058177827,
      "grad_norm": 0.08344763517379761,
      "learning_rate": 4.1903618906942394e-05,
      "loss": 0.0039,
      "step": 927
    },
    {
      "epoch": 0.33955360409806074,
      "grad_norm": 0.37805426120758057,
      "learning_rate": 4.189438700147711e-05,
      "loss": 0.0119,
      "step": 928
    },
    {
      "epoch": 0.33991950237833884,
      "grad_norm": 0.12279301136732101,
      "learning_rate": 4.188515509601182e-05,
      "loss": 0.0053,
      "step": 929
    },
    {
      "epoch": 0.3402854006586169,
      "grad_norm": 0.20451591908931732,
      "learning_rate": 4.1875923190546526e-05,
      "loss": 0.0054,
      "step": 930
    },
    {
      "epoch": 0.340651298938895,
      "grad_norm": 0.24601837992668152,
      "learning_rate": 4.186669128508124e-05,
      "loss": 0.0126,
      "step": 931
    },
    {
      "epoch": 0.34101719721917306,
      "grad_norm": 0.29463937878608704,
      "learning_rate": 4.1857459379615953e-05,
      "loss": 0.0072,
      "step": 932
    },
    {
      "epoch": 0.3413830954994512,
      "grad_norm": 0.16110862791538239,
      "learning_rate": 4.1848227474150664e-05,
      "loss": 0.0027,
      "step": 933
    },
    {
      "epoch": 0.3417489937797292,
      "grad_norm": 2.6849591732025146,
      "learning_rate": 4.183899556868538e-05,
      "loss": 0.342,
      "step": 934
    },
    {
      "epoch": 0.34211489206000734,
      "grad_norm": 15.033845901489258,
      "learning_rate": 4.182976366322009e-05,
      "loss": 0.447,
      "step": 935
    },
    {
      "epoch": 0.3424807903402854,
      "grad_norm": 3.231109857559204,
      "learning_rate": 4.18205317577548e-05,
      "loss": 0.1342,
      "step": 936
    },
    {
      "epoch": 0.3428466886205635,
      "grad_norm": 0.16607841849327087,
      "learning_rate": 4.181129985228951e-05,
      "loss": 0.0026,
      "step": 937
    },
    {
      "epoch": 0.34321258690084155,
      "grad_norm": 8.016613006591797,
      "learning_rate": 4.1802067946824224e-05,
      "loss": 0.2031,
      "step": 938
    },
    {
      "epoch": 0.34357848518111966,
      "grad_norm": 0.06342919170856476,
      "learning_rate": 4.179283604135894e-05,
      "loss": 0.003,
      "step": 939
    },
    {
      "epoch": 0.3439443834613977,
      "grad_norm": 0.07668245583772659,
      "learning_rate": 4.178360413589365e-05,
      "loss": 0.0032,
      "step": 940
    },
    {
      "epoch": 0.3443102817416758,
      "grad_norm": 0.13029389083385468,
      "learning_rate": 4.177437223042836e-05,
      "loss": 0.0049,
      "step": 941
    },
    {
      "epoch": 0.3446761800219539,
      "grad_norm": 13.94611930847168,
      "learning_rate": 4.176514032496307e-05,
      "loss": 0.4685,
      "step": 942
    },
    {
      "epoch": 0.345042078302232,
      "grad_norm": 2.634760856628418,
      "learning_rate": 4.175590841949778e-05,
      "loss": 0.1351,
      "step": 943
    },
    {
      "epoch": 0.34540797658251005,
      "grad_norm": 6.603567600250244,
      "learning_rate": 4.1746676514032494e-05,
      "loss": 0.2834,
      "step": 944
    },
    {
      "epoch": 0.34577387486278816,
      "grad_norm": 2.846691608428955,
      "learning_rate": 4.173744460856721e-05,
      "loss": 0.124,
      "step": 945
    },
    {
      "epoch": 0.3461397731430662,
      "grad_norm": 1.5698655843734741,
      "learning_rate": 4.172821270310192e-05,
      "loss": 0.029,
      "step": 946
    },
    {
      "epoch": 0.3465056714233443,
      "grad_norm": 7.124452590942383,
      "learning_rate": 4.171898079763664e-05,
      "loss": 0.155,
      "step": 947
    },
    {
      "epoch": 0.3468715697036224,
      "grad_norm": 12.108977317810059,
      "learning_rate": 4.170974889217135e-05,
      "loss": 0.518,
      "step": 948
    },
    {
      "epoch": 0.3472374679839005,
      "grad_norm": 0.14088739454746246,
      "learning_rate": 4.170051698670605e-05,
      "loss": 0.0032,
      "step": 949
    },
    {
      "epoch": 0.34760336626417854,
      "grad_norm": 0.11110912263393402,
      "learning_rate": 4.169128508124077e-05,
      "loss": 0.003,
      "step": 950
    },
    {
      "epoch": 0.34796926454445665,
      "grad_norm": 2.912101984024048,
      "learning_rate": 4.168205317577548e-05,
      "loss": 0.026,
      "step": 951
    },
    {
      "epoch": 0.3483351628247347,
      "grad_norm": 0.9555459022521973,
      "learning_rate": 4.167282127031019e-05,
      "loss": 0.0207,
      "step": 952
    },
    {
      "epoch": 0.3487010611050128,
      "grad_norm": 0.5668696165084839,
      "learning_rate": 4.166358936484491e-05,
      "loss": 0.0111,
      "step": 953
    },
    {
      "epoch": 0.34906695938529086,
      "grad_norm": 0.5411545634269714,
      "learning_rate": 4.165435745937962e-05,
      "loss": 0.0045,
      "step": 954
    },
    {
      "epoch": 0.349432857665569,
      "grad_norm": 3.488837242126465,
      "learning_rate": 4.164512555391433e-05,
      "loss": 0.3389,
      "step": 955
    },
    {
      "epoch": 0.34979875594584703,
      "grad_norm": 0.046477578580379486,
      "learning_rate": 4.163589364844904e-05,
      "loss": 0.002,
      "step": 956
    },
    {
      "epoch": 0.35016465422612514,
      "grad_norm": 0.5394270420074463,
      "learning_rate": 4.162666174298375e-05,
      "loss": 0.0054,
      "step": 957
    },
    {
      "epoch": 0.3505305525064032,
      "grad_norm": 6.8887104988098145,
      "learning_rate": 4.161742983751847e-05,
      "loss": 0.0573,
      "step": 958
    },
    {
      "epoch": 0.3508964507866813,
      "grad_norm": 12.002660751342773,
      "learning_rate": 4.160819793205318e-05,
      "loss": 0.2114,
      "step": 959
    },
    {
      "epoch": 0.3512623490669594,
      "grad_norm": 5.266103267669678,
      "learning_rate": 4.159896602658789e-05,
      "loss": 0.4259,
      "step": 960
    },
    {
      "epoch": 0.35162824734723747,
      "grad_norm": 0.04505537450313568,
      "learning_rate": 4.15897341211226e-05,
      "loss": 0.0019,
      "step": 961
    },
    {
      "epoch": 0.3519941456275156,
      "grad_norm": 0.05941678211092949,
      "learning_rate": 4.158050221565731e-05,
      "loss": 0.0027,
      "step": 962
    },
    {
      "epoch": 0.35236004390779363,
      "grad_norm": 11.481810569763184,
      "learning_rate": 4.157127031019202e-05,
      "loss": 0.2996,
      "step": 963
    },
    {
      "epoch": 0.35272594218807174,
      "grad_norm": 4.511000633239746,
      "learning_rate": 4.156203840472674e-05,
      "loss": 0.0392,
      "step": 964
    },
    {
      "epoch": 0.3530918404683498,
      "grad_norm": 3.9767673015594482,
      "learning_rate": 4.155280649926145e-05,
      "loss": 0.3525,
      "step": 965
    },
    {
      "epoch": 0.3534577387486279,
      "grad_norm": 2.3915116786956787,
      "learning_rate": 4.1543574593796166e-05,
      "loss": 0.0157,
      "step": 966
    },
    {
      "epoch": 0.35382363702890596,
      "grad_norm": 1.3343653678894043,
      "learning_rate": 4.153434268833088e-05,
      "loss": 0.019,
      "step": 967
    },
    {
      "epoch": 0.35418953530918407,
      "grad_norm": 1.6655802726745605,
      "learning_rate": 4.152511078286558e-05,
      "loss": 0.0321,
      "step": 968
    },
    {
      "epoch": 0.3545554335894621,
      "grad_norm": 6.148624897003174,
      "learning_rate": 4.15158788774003e-05,
      "loss": 0.0528,
      "step": 969
    },
    {
      "epoch": 0.35492133186974023,
      "grad_norm": 1.785171389579773,
      "learning_rate": 4.150664697193501e-05,
      "loss": 0.0294,
      "step": 970
    },
    {
      "epoch": 0.3552872301500183,
      "grad_norm": 3.0469236373901367,
      "learning_rate": 4.149741506646972e-05,
      "loss": 0.1316,
      "step": 971
    },
    {
      "epoch": 0.3556531284302964,
      "grad_norm": 1.247483491897583,
      "learning_rate": 4.1488183161004436e-05,
      "loss": 0.0234,
      "step": 972
    },
    {
      "epoch": 0.35601902671057445,
      "grad_norm": 0.24128255248069763,
      "learning_rate": 4.147895125553915e-05,
      "loss": 0.0075,
      "step": 973
    },
    {
      "epoch": 0.35638492499085256,
      "grad_norm": 3.5430667400360107,
      "learning_rate": 4.146971935007386e-05,
      "loss": 0.0874,
      "step": 974
    },
    {
      "epoch": 0.3567508232711306,
      "grad_norm": 1.1845123767852783,
      "learning_rate": 4.146048744460857e-05,
      "loss": 0.0144,
      "step": 975
    },
    {
      "epoch": 0.3571167215514087,
      "grad_norm": 8.72227668762207,
      "learning_rate": 4.145125553914328e-05,
      "loss": 0.0927,
      "step": 976
    },
    {
      "epoch": 0.3574826198316868,
      "grad_norm": 0.6935221552848816,
      "learning_rate": 4.1442023633677996e-05,
      "loss": 0.0165,
      "step": 977
    },
    {
      "epoch": 0.3578485181119649,
      "grad_norm": 0.09220503270626068,
      "learning_rate": 4.1432791728212707e-05,
      "loss": 0.0025,
      "step": 978
    },
    {
      "epoch": 0.35821441639224294,
      "grad_norm": 0.1706591546535492,
      "learning_rate": 4.142355982274742e-05,
      "loss": 0.0036,
      "step": 979
    },
    {
      "epoch": 0.35858031467252105,
      "grad_norm": 0.8770192265510559,
      "learning_rate": 4.141432791728213e-05,
      "loss": 0.0076,
      "step": 980
    },
    {
      "epoch": 0.3589462129527991,
      "grad_norm": 16.02445411682129,
      "learning_rate": 4.140509601181684e-05,
      "loss": 0.1587,
      "step": 981
    },
    {
      "epoch": 0.3593121112330772,
      "grad_norm": 5.186642646789551,
      "learning_rate": 4.139586410635155e-05,
      "loss": 0.0797,
      "step": 982
    },
    {
      "epoch": 0.35967800951335527,
      "grad_norm": 10.273032188415527,
      "learning_rate": 4.1386632200886266e-05,
      "loss": 0.0671,
      "step": 983
    },
    {
      "epoch": 0.3600439077936334,
      "grad_norm": 0.0318329855799675,
      "learning_rate": 4.1377400295420977e-05,
      "loss": 0.0015,
      "step": 984
    },
    {
      "epoch": 0.36040980607391143,
      "grad_norm": 4.671662330627441,
      "learning_rate": 4.1368168389955694e-05,
      "loss": 0.0217,
      "step": 985
    },
    {
      "epoch": 0.36077570435418954,
      "grad_norm": 6.3985466957092285,
      "learning_rate": 4.1358936484490404e-05,
      "loss": 0.5273,
      "step": 986
    },
    {
      "epoch": 0.3611416026344676,
      "grad_norm": 9.09765338897705,
      "learning_rate": 4.134970457902511e-05,
      "loss": 0.4272,
      "step": 987
    },
    {
      "epoch": 0.3615075009147457,
      "grad_norm": 4.957180500030518,
      "learning_rate": 4.1340472673559826e-05,
      "loss": 0.4049,
      "step": 988
    },
    {
      "epoch": 0.36187339919502376,
      "grad_norm": 15.18438720703125,
      "learning_rate": 4.1331240768094536e-05,
      "loss": 0.2831,
      "step": 989
    },
    {
      "epoch": 0.36223929747530187,
      "grad_norm": 1.040940761566162,
      "learning_rate": 4.132200886262925e-05,
      "loss": 0.0063,
      "step": 990
    },
    {
      "epoch": 0.3626051957555799,
      "grad_norm": 0.07895149290561676,
      "learning_rate": 4.1312776957163964e-05,
      "loss": 0.0033,
      "step": 991
    },
    {
      "epoch": 0.36297109403585803,
      "grad_norm": 0.03405885025858879,
      "learning_rate": 4.1303545051698675e-05,
      "loss": 0.0015,
      "step": 992
    },
    {
      "epoch": 0.3633369923161361,
      "grad_norm": 5.3680620193481445,
      "learning_rate": 4.1294313146233385e-05,
      "loss": 0.048,
      "step": 993
    },
    {
      "epoch": 0.3637028905964142,
      "grad_norm": 5.473161220550537,
      "learning_rate": 4.1285081240768096e-05,
      "loss": 0.2479,
      "step": 994
    },
    {
      "epoch": 0.3640687888766923,
      "grad_norm": 27.187541961669922,
      "learning_rate": 4.1275849335302806e-05,
      "loss": 0.0807,
      "step": 995
    },
    {
      "epoch": 0.36443468715697036,
      "grad_norm": 11.781102180480957,
      "learning_rate": 4.1266617429837523e-05,
      "loss": 0.8383,
      "step": 996
    },
    {
      "epoch": 0.36480058543724847,
      "grad_norm": 0.9358559250831604,
      "learning_rate": 4.1257385524372234e-05,
      "loss": 0.0121,
      "step": 997
    },
    {
      "epoch": 0.3651664837175265,
      "grad_norm": 2.6479907035827637,
      "learning_rate": 4.1248153618906945e-05,
      "loss": 0.0275,
      "step": 998
    },
    {
      "epoch": 0.36553238199780463,
      "grad_norm": 13.451425552368164,
      "learning_rate": 4.1238921713441655e-05,
      "loss": 0.1608,
      "step": 999
    },
    {
      "epoch": 0.3658982802780827,
      "grad_norm": 0.40925171971321106,
      "learning_rate": 4.1229689807976366e-05,
      "loss": 0.0051,
      "step": 1000
    },
    {
      "epoch": 0.3662641785583608,
      "grad_norm": 8.32685661315918,
      "learning_rate": 4.1220457902511076e-05,
      "loss": 0.1982,
      "step": 1001
    },
    {
      "epoch": 0.36663007683863885,
      "grad_norm": 6.726290225982666,
      "learning_rate": 4.1211225997045794e-05,
      "loss": 0.066,
      "step": 1002
    },
    {
      "epoch": 0.36699597511891696,
      "grad_norm": 0.1533229500055313,
      "learning_rate": 4.1201994091580504e-05,
      "loss": 0.0046,
      "step": 1003
    },
    {
      "epoch": 0.367361873399195,
      "grad_norm": 0.0854213759303093,
      "learning_rate": 4.1192762186115215e-05,
      "loss": 0.0036,
      "step": 1004
    },
    {
      "epoch": 0.3677277716794731,
      "grad_norm": 0.051320724189281464,
      "learning_rate": 4.118353028064993e-05,
      "loss": 0.0025,
      "step": 1005
    },
    {
      "epoch": 0.3680936699597512,
      "grad_norm": 0.1051122397184372,
      "learning_rate": 4.1174298375184636e-05,
      "loss": 0.0035,
      "step": 1006
    },
    {
      "epoch": 0.3684595682400293,
      "grad_norm": 4.5644636154174805,
      "learning_rate": 4.116506646971935e-05,
      "loss": 0.2962,
      "step": 1007
    },
    {
      "epoch": 0.36882546652030734,
      "grad_norm": 0.13988977670669556,
      "learning_rate": 4.1155834564254064e-05,
      "loss": 0.004,
      "step": 1008
    },
    {
      "epoch": 0.36919136480058545,
      "grad_norm": 7.689339637756348,
      "learning_rate": 4.1146602658788774e-05,
      "loss": 0.0758,
      "step": 1009
    },
    {
      "epoch": 0.3695572630808635,
      "grad_norm": 0.8980519771575928,
      "learning_rate": 4.113737075332349e-05,
      "loss": 0.0052,
      "step": 1010
    },
    {
      "epoch": 0.3699231613611416,
      "grad_norm": 0.2337864488363266,
      "learning_rate": 4.11281388478582e-05,
      "loss": 0.0068,
      "step": 1011
    },
    {
      "epoch": 0.37028905964141967,
      "grad_norm": 5.069775104522705,
      "learning_rate": 4.111890694239291e-05,
      "loss": 0.0376,
      "step": 1012
    },
    {
      "epoch": 0.3706549579216978,
      "grad_norm": 3.673839807510376,
      "learning_rate": 4.110967503692762e-05,
      "loss": 0.0192,
      "step": 1013
    },
    {
      "epoch": 0.37102085620197583,
      "grad_norm": 0.047454264014959335,
      "learning_rate": 4.1100443131462334e-05,
      "loss": 0.002,
      "step": 1014
    },
    {
      "epoch": 0.37138675448225394,
      "grad_norm": 0.06536351144313812,
      "learning_rate": 4.109121122599705e-05,
      "loss": 0.0027,
      "step": 1015
    },
    {
      "epoch": 0.371752652762532,
      "grad_norm": 0.3378661572933197,
      "learning_rate": 4.108197932053176e-05,
      "loss": 0.0044,
      "step": 1016
    },
    {
      "epoch": 0.3721185510428101,
      "grad_norm": 3.7964723110198975,
      "learning_rate": 4.107274741506647e-05,
      "loss": 0.2711,
      "step": 1017
    },
    {
      "epoch": 0.37248444932308816,
      "grad_norm": 14.646235466003418,
      "learning_rate": 4.106351550960118e-05,
      "loss": 0.2035,
      "step": 1018
    },
    {
      "epoch": 0.37285034760336627,
      "grad_norm": 1.6742963790893555,
      "learning_rate": 4.105428360413589e-05,
      "loss": 0.0098,
      "step": 1019
    },
    {
      "epoch": 0.3732162458836443,
      "grad_norm": 1.3170477151870728,
      "learning_rate": 4.1045051698670604e-05,
      "loss": 0.0134,
      "step": 1020
    },
    {
      "epoch": 0.37358214416392244,
      "grad_norm": 0.05941144376993179,
      "learning_rate": 4.103581979320532e-05,
      "loss": 0.0019,
      "step": 1021
    },
    {
      "epoch": 0.3739480424442005,
      "grad_norm": 0.19910386204719543,
      "learning_rate": 4.102658788774003e-05,
      "loss": 0.0034,
      "step": 1022
    },
    {
      "epoch": 0.3743139407244786,
      "grad_norm": 0.1489187628030777,
      "learning_rate": 4.101735598227474e-05,
      "loss": 0.0039,
      "step": 1023
    },
    {
      "epoch": 0.37467983900475665,
      "grad_norm": 0.046633630990982056,
      "learning_rate": 4.100812407680946e-05,
      "loss": 0.0017,
      "step": 1024
    },
    {
      "epoch": 0.37504573728503476,
      "grad_norm": 3.710054874420166,
      "learning_rate": 4.099889217134416e-05,
      "loss": 0.0272,
      "step": 1025
    },
    {
      "epoch": 0.3754116355653128,
      "grad_norm": 0.1605396568775177,
      "learning_rate": 4.098966026587888e-05,
      "loss": 0.0041,
      "step": 1026
    },
    {
      "epoch": 0.3757775338455909,
      "grad_norm": 0.24894875288009644,
      "learning_rate": 4.098042836041359e-05,
      "loss": 0.0054,
      "step": 1027
    },
    {
      "epoch": 0.376143432125869,
      "grad_norm": 0.06940975785255432,
      "learning_rate": 4.09711964549483e-05,
      "loss": 0.0021,
      "step": 1028
    },
    {
      "epoch": 0.3765093304061471,
      "grad_norm": 0.12690334022045135,
      "learning_rate": 4.096196454948302e-05,
      "loss": 0.0034,
      "step": 1029
    },
    {
      "epoch": 0.3768752286864252,
      "grad_norm": 0.08864274621009827,
      "learning_rate": 4.095273264401773e-05,
      "loss": 0.0019,
      "step": 1030
    },
    {
      "epoch": 0.37724112696670326,
      "grad_norm": 10.395519256591797,
      "learning_rate": 4.094350073855244e-05,
      "loss": 0.2947,
      "step": 1031
    },
    {
      "epoch": 0.37760702524698136,
      "grad_norm": 0.0278640016913414,
      "learning_rate": 4.093426883308715e-05,
      "loss": 0.0012,
      "step": 1032
    },
    {
      "epoch": 0.3779729235272594,
      "grad_norm": 2.120166778564453,
      "learning_rate": 4.092503692762186e-05,
      "loss": 0.4155,
      "step": 1033
    },
    {
      "epoch": 0.37833882180753753,
      "grad_norm": 12.121076583862305,
      "learning_rate": 4.091580502215657e-05,
      "loss": 0.5795,
      "step": 1034
    },
    {
      "epoch": 0.3787047200878156,
      "grad_norm": 0.02488723397254944,
      "learning_rate": 4.090657311669129e-05,
      "loss": 0.0011,
      "step": 1035
    },
    {
      "epoch": 0.3790706183680937,
      "grad_norm": 0.14913852512836456,
      "learning_rate": 4.0897341211226e-05,
      "loss": 0.003,
      "step": 1036
    },
    {
      "epoch": 0.37943651664837175,
      "grad_norm": 10.712076187133789,
      "learning_rate": 4.088810930576071e-05,
      "loss": 0.0471,
      "step": 1037
    },
    {
      "epoch": 0.37980241492864986,
      "grad_norm": 0.04998331889510155,
      "learning_rate": 4.087887740029542e-05,
      "loss": 0.0022,
      "step": 1038
    },
    {
      "epoch": 0.3801683132089279,
      "grad_norm": 0.46509504318237305,
      "learning_rate": 4.086964549483013e-05,
      "loss": 0.0058,
      "step": 1039
    },
    {
      "epoch": 0.380534211489206,
      "grad_norm": 0.07815847545862198,
      "learning_rate": 4.086041358936485e-05,
      "loss": 0.0032,
      "step": 1040
    },
    {
      "epoch": 0.3809001097694841,
      "grad_norm": 0.06630205363035202,
      "learning_rate": 4.085118168389956e-05,
      "loss": 0.0029,
      "step": 1041
    },
    {
      "epoch": 0.3812660080497622,
      "grad_norm": 0.07005126774311066,
      "learning_rate": 4.084194977843427e-05,
      "loss": 0.0026,
      "step": 1042
    },
    {
      "epoch": 0.38163190633004024,
      "grad_norm": 4.415816307067871,
      "learning_rate": 4.083271787296899e-05,
      "loss": 0.0372,
      "step": 1043
    },
    {
      "epoch": 0.38199780461031835,
      "grad_norm": 0.1729428470134735,
      "learning_rate": 4.082348596750369e-05,
      "loss": 0.0038,
      "step": 1044
    },
    {
      "epoch": 0.3823637028905964,
      "grad_norm": 0.16942986845970154,
      "learning_rate": 4.08142540620384e-05,
      "loss": 0.0039,
      "step": 1045
    },
    {
      "epoch": 0.3827296011708745,
      "grad_norm": 11.091999053955078,
      "learning_rate": 4.080502215657312e-05,
      "loss": 0.0932,
      "step": 1046
    },
    {
      "epoch": 0.38309549945115257,
      "grad_norm": 6.709338665008545,
      "learning_rate": 4.079579025110783e-05,
      "loss": 0.1922,
      "step": 1047
    },
    {
      "epoch": 0.3834613977314307,
      "grad_norm": 0.08116057515144348,
      "learning_rate": 4.0786558345642547e-05,
      "loss": 0.0035,
      "step": 1048
    },
    {
      "epoch": 0.38382729601170873,
      "grad_norm": 0.07751990109682083,
      "learning_rate": 4.077732644017726e-05,
      "loss": 0.0032,
      "step": 1049
    },
    {
      "epoch": 0.38419319429198684,
      "grad_norm": 0.2783143222332001,
      "learning_rate": 4.076809453471197e-05,
      "loss": 0.0053,
      "step": 1050
    },
    {
      "epoch": 0.3845590925722649,
      "grad_norm": 0.2303263545036316,
      "learning_rate": 4.075886262924668e-05,
      "loss": 0.0047,
      "step": 1051
    },
    {
      "epoch": 0.384924990852543,
      "grad_norm": 2.656034469604492,
      "learning_rate": 4.074963072378139e-05,
      "loss": 0.2104,
      "step": 1052
    },
    {
      "epoch": 0.38529088913282106,
      "grad_norm": 0.042622990906238556,
      "learning_rate": 4.07403988183161e-05,
      "loss": 0.0018,
      "step": 1053
    },
    {
      "epoch": 0.38565678741309917,
      "grad_norm": 0.04788623005151749,
      "learning_rate": 4.073116691285082e-05,
      "loss": 0.0021,
      "step": 1054
    },
    {
      "epoch": 0.3860226856933772,
      "grad_norm": 0.14751717448234558,
      "learning_rate": 4.072193500738553e-05,
      "loss": 0.0037,
      "step": 1055
    },
    {
      "epoch": 0.38638858397365533,
      "grad_norm": 0.20411700010299683,
      "learning_rate": 4.071270310192024e-05,
      "loss": 0.0027,
      "step": 1056
    },
    {
      "epoch": 0.3867544822539334,
      "grad_norm": 0.05106980726122856,
      "learning_rate": 4.070347119645495e-05,
      "loss": 0.0019,
      "step": 1057
    },
    {
      "epoch": 0.3871203805342115,
      "grad_norm": 0.03249678388237953,
      "learning_rate": 4.069423929098966e-05,
      "loss": 0.0014,
      "step": 1058
    },
    {
      "epoch": 0.38748627881448955,
      "grad_norm": 0.038931798189878464,
      "learning_rate": 4.0685007385524376e-05,
      "loss": 0.0017,
      "step": 1059
    },
    {
      "epoch": 0.38785217709476766,
      "grad_norm": 10.631426811218262,
      "learning_rate": 4.067577548005909e-05,
      "loss": 0.3062,
      "step": 1060
    },
    {
      "epoch": 0.3882180753750457,
      "grad_norm": 1.075602650642395,
      "learning_rate": 4.06665435745938e-05,
      "loss": 0.0069,
      "step": 1061
    },
    {
      "epoch": 0.3885839736553238,
      "grad_norm": 0.23518750071525574,
      "learning_rate": 4.0657311669128515e-05,
      "loss": 0.0034,
      "step": 1062
    },
    {
      "epoch": 0.3889498719356019,
      "grad_norm": 0.7352984547615051,
      "learning_rate": 4.064807976366322e-05,
      "loss": 0.0077,
      "step": 1063
    },
    {
      "epoch": 0.38931577021588,
      "grad_norm": 0.03127619996666908,
      "learning_rate": 4.063884785819793e-05,
      "loss": 0.0012,
      "step": 1064
    },
    {
      "epoch": 0.3896816684961581,
      "grad_norm": 4.288915634155273,
      "learning_rate": 4.0629615952732646e-05,
      "loss": 0.3281,
      "step": 1065
    },
    {
      "epoch": 0.39004756677643615,
      "grad_norm": 0.09351296722888947,
      "learning_rate": 4.062038404726736e-05,
      "loss": 0.0018,
      "step": 1066
    },
    {
      "epoch": 0.39041346505671426,
      "grad_norm": 0.08429578691720963,
      "learning_rate": 4.0611152141802074e-05,
      "loss": 0.0019,
      "step": 1067
    },
    {
      "epoch": 0.3907793633369923,
      "grad_norm": 3.2427845001220703,
      "learning_rate": 4.0601920236336785e-05,
      "loss": 0.0262,
      "step": 1068
    },
    {
      "epoch": 0.3911452616172704,
      "grad_norm": 19.370996475219727,
      "learning_rate": 4.0592688330871495e-05,
      "loss": 0.0544,
      "step": 1069
    },
    {
      "epoch": 0.3915111598975485,
      "grad_norm": 3.9821882247924805,
      "learning_rate": 4.0583456425406206e-05,
      "loss": 0.0086,
      "step": 1070
    },
    {
      "epoch": 0.3918770581778266,
      "grad_norm": 5.952430725097656,
      "learning_rate": 4.0574224519940916e-05,
      "loss": 0.3333,
      "step": 1071
    },
    {
      "epoch": 0.39224295645810464,
      "grad_norm": 0.050988439470529556,
      "learning_rate": 4.056499261447563e-05,
      "loss": 0.0014,
      "step": 1072
    },
    {
      "epoch": 0.39260885473838275,
      "grad_norm": 1.651959776878357,
      "learning_rate": 4.0555760709010344e-05,
      "loss": 0.0113,
      "step": 1073
    },
    {
      "epoch": 0.3929747530186608,
      "grad_norm": 0.3487679064273834,
      "learning_rate": 4.0546528803545055e-05,
      "loss": 0.0055,
      "step": 1074
    },
    {
      "epoch": 0.3933406512989389,
      "grad_norm": 7.346160888671875,
      "learning_rate": 4.0537296898079765e-05,
      "loss": 0.2925,
      "step": 1075
    },
    {
      "epoch": 0.39370654957921697,
      "grad_norm": 0.08664988726377487,
      "learning_rate": 4.0528064992614476e-05,
      "loss": 0.0026,
      "step": 1076
    },
    {
      "epoch": 0.3940724478594951,
      "grad_norm": 7.933176517486572,
      "learning_rate": 4.0518833087149186e-05,
      "loss": 0.1565,
      "step": 1077
    },
    {
      "epoch": 0.39443834613977313,
      "grad_norm": 24.324234008789062,
      "learning_rate": 4.0509601181683904e-05,
      "loss": 0.5873,
      "step": 1078
    },
    {
      "epoch": 0.39480424442005124,
      "grad_norm": 1.7350136041641235,
      "learning_rate": 4.0500369276218614e-05,
      "loss": 0.0205,
      "step": 1079
    },
    {
      "epoch": 0.3951701427003293,
      "grad_norm": 6.619609832763672,
      "learning_rate": 4.0491137370753325e-05,
      "loss": 0.2158,
      "step": 1080
    },
    {
      "epoch": 0.3955360409806074,
      "grad_norm": 3.0774197578430176,
      "learning_rate": 4.048190546528804e-05,
      "loss": 0.1205,
      "step": 1081
    },
    {
      "epoch": 0.39590193926088546,
      "grad_norm": 9.972490310668945,
      "learning_rate": 4.0472673559822746e-05,
      "loss": 0.0335,
      "step": 1082
    },
    {
      "epoch": 0.39626783754116357,
      "grad_norm": 0.07835379242897034,
      "learning_rate": 4.0463441654357456e-05,
      "loss": 0.0028,
      "step": 1083
    },
    {
      "epoch": 0.3966337358214416,
      "grad_norm": 0.4001957178115845,
      "learning_rate": 4.0454209748892174e-05,
      "loss": 0.0133,
      "step": 1084
    },
    {
      "epoch": 0.39699963410171973,
      "grad_norm": 0.20275981724262238,
      "learning_rate": 4.0444977843426884e-05,
      "loss": 0.0064,
      "step": 1085
    },
    {
      "epoch": 0.3973655323819978,
      "grad_norm": 0.8840126991271973,
      "learning_rate": 4.04357459379616e-05,
      "loss": 0.0126,
      "step": 1086
    },
    {
      "epoch": 0.3977314306622759,
      "grad_norm": 0.09633747488260269,
      "learning_rate": 4.042651403249631e-05,
      "loss": 0.0035,
      "step": 1087
    },
    {
      "epoch": 0.39809732894255395,
      "grad_norm": 0.11163196712732315,
      "learning_rate": 4.041728212703102e-05,
      "loss": 0.0026,
      "step": 1088
    },
    {
      "epoch": 0.39846322722283206,
      "grad_norm": 0.1849939376115799,
      "learning_rate": 4.040805022156573e-05,
      "loss": 0.005,
      "step": 1089
    },
    {
      "epoch": 0.3988291255031101,
      "grad_norm": 6.894286155700684,
      "learning_rate": 4.0398818316100444e-05,
      "loss": 0.0397,
      "step": 1090
    },
    {
      "epoch": 0.3991950237833882,
      "grad_norm": 5.927282810211182,
      "learning_rate": 4.0389586410635154e-05,
      "loss": 0.201,
      "step": 1091
    },
    {
      "epoch": 0.3995609220636663,
      "grad_norm": 4.895211696624756,
      "learning_rate": 4.038035450516987e-05,
      "loss": 0.0184,
      "step": 1092
    },
    {
      "epoch": 0.3999268203439444,
      "grad_norm": 12.577080726623535,
      "learning_rate": 4.037112259970458e-05,
      "loss": 0.1165,
      "step": 1093
    },
    {
      "epoch": 0.40029271862422244,
      "grad_norm": 4.352025508880615,
      "learning_rate": 4.036189069423929e-05,
      "loss": 0.3279,
      "step": 1094
    },
    {
      "epoch": 0.40065861690450055,
      "grad_norm": 0.07762026786804199,
      "learning_rate": 4.0352658788774e-05,
      "loss": 0.002,
      "step": 1095
    },
    {
      "epoch": 0.4010245151847786,
      "grad_norm": 0.07013195008039474,
      "learning_rate": 4.0343426883308714e-05,
      "loss": 0.0016,
      "step": 1096
    },
    {
      "epoch": 0.4013904134650567,
      "grad_norm": 0.4645220935344696,
      "learning_rate": 4.033419497784343e-05,
      "loss": 0.0076,
      "step": 1097
    },
    {
      "epoch": 0.40175631174533477,
      "grad_norm": 0.036648768931627274,
      "learning_rate": 4.032496307237814e-05,
      "loss": 0.0013,
      "step": 1098
    },
    {
      "epoch": 0.4021222100256129,
      "grad_norm": 21.25423240661621,
      "learning_rate": 4.031573116691285e-05,
      "loss": 0.0757,
      "step": 1099
    },
    {
      "epoch": 0.40248810830589093,
      "grad_norm": 3.405017375946045,
      "learning_rate": 4.030649926144757e-05,
      "loss": 0.1457,
      "step": 1100
    },
    {
      "epoch": 0.40285400658616904,
      "grad_norm": 0.18299447000026703,
      "learning_rate": 4.0297267355982273e-05,
      "loss": 0.0029,
      "step": 1101
    },
    {
      "epoch": 0.40321990486644715,
      "grad_norm": 5.899768829345703,
      "learning_rate": 4.0288035450516984e-05,
      "loss": 0.2361,
      "step": 1102
    },
    {
      "epoch": 0.4035858031467252,
      "grad_norm": 7.46808385848999,
      "learning_rate": 4.02788035450517e-05,
      "loss": 0.076,
      "step": 1103
    },
    {
      "epoch": 0.4039517014270033,
      "grad_norm": 4.352721691131592,
      "learning_rate": 4.026957163958641e-05,
      "loss": 0.1545,
      "step": 1104
    },
    {
      "epoch": 0.40431759970728137,
      "grad_norm": 0.49551576375961304,
      "learning_rate": 4.026033973412112e-05,
      "loss": 0.0069,
      "step": 1105
    },
    {
      "epoch": 0.4046834979875595,
      "grad_norm": 1.4333606958389282,
      "learning_rate": 4.025110782865584e-05,
      "loss": 0.0228,
      "step": 1106
    },
    {
      "epoch": 0.40504939626783754,
      "grad_norm": 13.542362213134766,
      "learning_rate": 4.024187592319055e-05,
      "loss": 0.8997,
      "step": 1107
    },
    {
      "epoch": 0.40541529454811565,
      "grad_norm": 5.511876583099365,
      "learning_rate": 4.023264401772526e-05,
      "loss": 0.0986,
      "step": 1108
    },
    {
      "epoch": 0.4057811928283937,
      "grad_norm": 5.350011825561523,
      "learning_rate": 4.022341211225997e-05,
      "loss": 0.4581,
      "step": 1109
    },
    {
      "epoch": 0.4061470911086718,
      "grad_norm": 7.817060470581055,
      "learning_rate": 4.021418020679468e-05,
      "loss": 0.1456,
      "step": 1110
    },
    {
      "epoch": 0.40651298938894986,
      "grad_norm": 6.1051201820373535,
      "learning_rate": 4.02049483013294e-05,
      "loss": 0.2446,
      "step": 1111
    },
    {
      "epoch": 0.406878887669228,
      "grad_norm": 7.982057094573975,
      "learning_rate": 4.019571639586411e-05,
      "loss": 0.0526,
      "step": 1112
    },
    {
      "epoch": 0.407244785949506,
      "grad_norm": 0.20629152655601501,
      "learning_rate": 4.018648449039882e-05,
      "loss": 0.0057,
      "step": 1113
    },
    {
      "epoch": 0.40761068422978414,
      "grad_norm": 0.3805106580257416,
      "learning_rate": 4.017725258493353e-05,
      "loss": 0.0093,
      "step": 1114
    },
    {
      "epoch": 0.4079765825100622,
      "grad_norm": 2.890380620956421,
      "learning_rate": 4.016802067946824e-05,
      "loss": 0.0726,
      "step": 1115
    },
    {
      "epoch": 0.4083424807903403,
      "grad_norm": 1.4824457168579102,
      "learning_rate": 4.015878877400295e-05,
      "loss": 0.039,
      "step": 1116
    },
    {
      "epoch": 0.40870837907061836,
      "grad_norm": 0.6200398802757263,
      "learning_rate": 4.014955686853767e-05,
      "loss": 0.0119,
      "step": 1117
    },
    {
      "epoch": 0.40907427735089646,
      "grad_norm": 6.757983207702637,
      "learning_rate": 4.014032496307238e-05,
      "loss": 0.1138,
      "step": 1118
    },
    {
      "epoch": 0.4094401756311745,
      "grad_norm": 1.7632249593734741,
      "learning_rate": 4.01310930576071e-05,
      "loss": 0.0277,
      "step": 1119
    },
    {
      "epoch": 0.40980607391145263,
      "grad_norm": 0.7212395071983337,
      "learning_rate": 4.01218611521418e-05,
      "loss": 0.0145,
      "step": 1120
    },
    {
      "epoch": 0.4101719721917307,
      "grad_norm": 0.08072499930858612,
      "learning_rate": 4.011262924667651e-05,
      "loss": 0.0025,
      "step": 1121
    },
    {
      "epoch": 0.4105378704720088,
      "grad_norm": 0.04622556269168854,
      "learning_rate": 4.010339734121123e-05,
      "loss": 0.0018,
      "step": 1122
    },
    {
      "epoch": 0.41090376875228685,
      "grad_norm": 0.17953188717365265,
      "learning_rate": 4.009416543574594e-05,
      "loss": 0.0031,
      "step": 1123
    },
    {
      "epoch": 0.41126966703256496,
      "grad_norm": 0.9770980477333069,
      "learning_rate": 4.008493353028065e-05,
      "loss": 0.01,
      "step": 1124
    },
    {
      "epoch": 0.411635565312843,
      "grad_norm": 3.5089118480682373,
      "learning_rate": 4.007570162481537e-05,
      "loss": 0.0241,
      "step": 1125
    },
    {
      "epoch": 0.4120014635931211,
      "grad_norm": 0.05057087913155556,
      "learning_rate": 4.006646971935008e-05,
      "loss": 0.0019,
      "step": 1126
    },
    {
      "epoch": 0.4123673618733992,
      "grad_norm": 3.600890874862671,
      "learning_rate": 4.005723781388479e-05,
      "loss": 0.0157,
      "step": 1127
    },
    {
      "epoch": 0.4127332601536773,
      "grad_norm": 12.232937812805176,
      "learning_rate": 4.00480059084195e-05,
      "loss": 0.3518,
      "step": 1128
    },
    {
      "epoch": 0.41309915843395534,
      "grad_norm": 18.242259979248047,
      "learning_rate": 4.003877400295421e-05,
      "loss": 0.064,
      "step": 1129
    },
    {
      "epoch": 0.41346505671423345,
      "grad_norm": 7.734925746917725,
      "learning_rate": 4.002954209748893e-05,
      "loss": 0.2772,
      "step": 1130
    },
    {
      "epoch": 0.4138309549945115,
      "grad_norm": 0.024013636633753777,
      "learning_rate": 4.002031019202364e-05,
      "loss": 0.0011,
      "step": 1131
    },
    {
      "epoch": 0.4141968532747896,
      "grad_norm": 0.310443252325058,
      "learning_rate": 4.001107828655835e-05,
      "loss": 0.003,
      "step": 1132
    },
    {
      "epoch": 0.41456275155506767,
      "grad_norm": 0.13218051195144653,
      "learning_rate": 4.000184638109306e-05,
      "loss": 0.003,
      "step": 1133
    },
    {
      "epoch": 0.4149286498353458,
      "grad_norm": 0.04382622614502907,
      "learning_rate": 3.999261447562777e-05,
      "loss": 0.0011,
      "step": 1134
    },
    {
      "epoch": 0.41529454811562383,
      "grad_norm": 16.166122436523438,
      "learning_rate": 3.998338257016248e-05,
      "loss": 0.5551,
      "step": 1135
    },
    {
      "epoch": 0.41566044639590194,
      "grad_norm": 25.633615493774414,
      "learning_rate": 3.99741506646972e-05,
      "loss": 0.1323,
      "step": 1136
    },
    {
      "epoch": 0.41602634467618005,
      "grad_norm": 5.366091728210449,
      "learning_rate": 3.996491875923191e-05,
      "loss": 0.0277,
      "step": 1137
    },
    {
      "epoch": 0.4163922429564581,
      "grad_norm": 8.715492248535156,
      "learning_rate": 3.9955686853766625e-05,
      "loss": 0.1723,
      "step": 1138
    },
    {
      "epoch": 0.4167581412367362,
      "grad_norm": 0.03458299860358238,
      "learning_rate": 3.994645494830133e-05,
      "loss": 0.0012,
      "step": 1139
    },
    {
      "epoch": 0.41712403951701427,
      "grad_norm": 0.12975144386291504,
      "learning_rate": 3.993722304283604e-05,
      "loss": 0.0027,
      "step": 1140
    },
    {
      "epoch": 0.4174899377972924,
      "grad_norm": 0.06648088246583939,
      "learning_rate": 3.9927991137370756e-05,
      "loss": 0.0015,
      "step": 1141
    },
    {
      "epoch": 0.41785583607757043,
      "grad_norm": 0.023735854774713516,
      "learning_rate": 3.991875923190547e-05,
      "loss": 0.001,
      "step": 1142
    },
    {
      "epoch": 0.41822173435784854,
      "grad_norm": 0.028355365619063377,
      "learning_rate": 3.990952732644018e-05,
      "loss": 0.0011,
      "step": 1143
    },
    {
      "epoch": 0.4185876326381266,
      "grad_norm": 0.16657502949237823,
      "learning_rate": 3.9900295420974895e-05,
      "loss": 0.003,
      "step": 1144
    },
    {
      "epoch": 0.4189535309184047,
      "grad_norm": 0.06898552179336548,
      "learning_rate": 3.9891063515509605e-05,
      "loss": 0.0026,
      "step": 1145
    },
    {
      "epoch": 0.41931942919868276,
      "grad_norm": 7.733545303344727,
      "learning_rate": 3.988183161004431e-05,
      "loss": 0.3557,
      "step": 1146
    },
    {
      "epoch": 0.41968532747896087,
      "grad_norm": 0.013278177939355373,
      "learning_rate": 3.9872599704579026e-05,
      "loss": 0.0006,
      "step": 1147
    },
    {
      "epoch": 0.4200512257592389,
      "grad_norm": 0.0931662917137146,
      "learning_rate": 3.986336779911374e-05,
      "loss": 0.002,
      "step": 1148
    },
    {
      "epoch": 0.42041712403951703,
      "grad_norm": 15.87153434753418,
      "learning_rate": 3.9854135893648454e-05,
      "loss": 0.4723,
      "step": 1149
    },
    {
      "epoch": 0.4207830223197951,
      "grad_norm": 1.2654225826263428,
      "learning_rate": 3.9844903988183165e-05,
      "loss": 0.0125,
      "step": 1150
    },
    {
      "epoch": 0.4211489206000732,
      "grad_norm": 15.004364967346191,
      "learning_rate": 3.9835672082717875e-05,
      "loss": 0.3154,
      "step": 1151
    },
    {
      "epoch": 0.42151481888035125,
      "grad_norm": 0.9311321377754211,
      "learning_rate": 3.9826440177252586e-05,
      "loss": 0.0047,
      "step": 1152
    },
    {
      "epoch": 0.42188071716062936,
      "grad_norm": 7.6834211349487305,
      "learning_rate": 3.9817208271787297e-05,
      "loss": 0.4245,
      "step": 1153
    },
    {
      "epoch": 0.4222466154409074,
      "grad_norm": 4.322671413421631,
      "learning_rate": 3.980797636632201e-05,
      "loss": 0.2818,
      "step": 1154
    },
    {
      "epoch": 0.4226125137211855,
      "grad_norm": 0.07536523044109344,
      "learning_rate": 3.9798744460856724e-05,
      "loss": 0.0018,
      "step": 1155
    },
    {
      "epoch": 0.4229784120014636,
      "grad_norm": 14.127447128295898,
      "learning_rate": 3.9789512555391435e-05,
      "loss": 0.2403,
      "step": 1156
    },
    {
      "epoch": 0.4233443102817417,
      "grad_norm": 0.1759198009967804,
      "learning_rate": 3.978028064992615e-05,
      "loss": 0.005,
      "step": 1157
    },
    {
      "epoch": 0.42371020856201974,
      "grad_norm": 0.16080474853515625,
      "learning_rate": 3.9771048744460856e-05,
      "loss": 0.0044,
      "step": 1158
    },
    {
      "epoch": 0.42407610684229785,
      "grad_norm": 20.74578094482422,
      "learning_rate": 3.9761816838995567e-05,
      "loss": 0.3181,
      "step": 1159
    },
    {
      "epoch": 0.4244420051225759,
      "grad_norm": 0.16249944269657135,
      "learning_rate": 3.9752584933530284e-05,
      "loss": 0.0043,
      "step": 1160
    },
    {
      "epoch": 0.424807903402854,
      "grad_norm": 0.22054506838321686,
      "learning_rate": 3.9743353028064994e-05,
      "loss": 0.0076,
      "step": 1161
    },
    {
      "epoch": 0.42517380168313207,
      "grad_norm": 3.364871025085449,
      "learning_rate": 3.9734121122599705e-05,
      "loss": 0.2916,
      "step": 1162
    },
    {
      "epoch": 0.4255396999634102,
      "grad_norm": 0.25769856572151184,
      "learning_rate": 3.972488921713442e-05,
      "loss": 0.0096,
      "step": 1163
    },
    {
      "epoch": 0.42590559824368823,
      "grad_norm": 0.15762844681739807,
      "learning_rate": 3.971565731166913e-05,
      "loss": 0.0058,
      "step": 1164
    },
    {
      "epoch": 0.42627149652396634,
      "grad_norm": 0.2093752771615982,
      "learning_rate": 3.970642540620384e-05,
      "loss": 0.0079,
      "step": 1165
    },
    {
      "epoch": 0.4266373948042444,
      "grad_norm": 4.953813076019287,
      "learning_rate": 3.9697193500738554e-05,
      "loss": 0.1915,
      "step": 1166
    },
    {
      "epoch": 0.4270032930845225,
      "grad_norm": 0.16138537228107452,
      "learning_rate": 3.9687961595273265e-05,
      "loss": 0.0046,
      "step": 1167
    },
    {
      "epoch": 0.42736919136480056,
      "grad_norm": 9.875404357910156,
      "learning_rate": 3.967872968980798e-05,
      "loss": 0.2579,
      "step": 1168
    },
    {
      "epoch": 0.42773508964507867,
      "grad_norm": 13.786026954650879,
      "learning_rate": 3.966949778434269e-05,
      "loss": 0.1798,
      "step": 1169
    },
    {
      "epoch": 0.4281009879253567,
      "grad_norm": 0.07719779759645462,
      "learning_rate": 3.96602658788774e-05,
      "loss": 0.0021,
      "step": 1170
    },
    {
      "epoch": 0.42846688620563483,
      "grad_norm": 0.11802233755588531,
      "learning_rate": 3.9651033973412114e-05,
      "loss": 0.0043,
      "step": 1171
    },
    {
      "epoch": 0.42883278448591294,
      "grad_norm": 0.36728787422180176,
      "learning_rate": 3.9641802067946824e-05,
      "loss": 0.0059,
      "step": 1172
    },
    {
      "epoch": 0.429198682766191,
      "grad_norm": 4.359315872192383,
      "learning_rate": 3.9632570162481535e-05,
      "loss": 0.7061,
      "step": 1173
    },
    {
      "epoch": 0.4295645810464691,
      "grad_norm": 0.04867352172732353,
      "learning_rate": 3.962333825701625e-05,
      "loss": 0.0014,
      "step": 1174
    },
    {
      "epoch": 0.42993047932674716,
      "grad_norm": 0.21736280620098114,
      "learning_rate": 3.961410635155096e-05,
      "loss": 0.0037,
      "step": 1175
    },
    {
      "epoch": 0.43029637760702527,
      "grad_norm": 8.852428436279297,
      "learning_rate": 3.960487444608567e-05,
      "loss": 0.5042,
      "step": 1176
    },
    {
      "epoch": 0.4306622758873033,
      "grad_norm": 0.2607687711715698,
      "learning_rate": 3.9595642540620384e-05,
      "loss": 0.0062,
      "step": 1177
    },
    {
      "epoch": 0.43102817416758143,
      "grad_norm": 0.032460931688547134,
      "learning_rate": 3.9586410635155094e-05,
      "loss": 0.0013,
      "step": 1178
    },
    {
      "epoch": 0.4313940724478595,
      "grad_norm": 0.16705337166786194,
      "learning_rate": 3.957717872968981e-05,
      "loss": 0.0043,
      "step": 1179
    },
    {
      "epoch": 0.4317599707281376,
      "grad_norm": 0.2114899456501007,
      "learning_rate": 3.956794682422452e-05,
      "loss": 0.0052,
      "step": 1180
    },
    {
      "epoch": 0.43212586900841565,
      "grad_norm": 0.13643358647823334,
      "learning_rate": 3.955871491875923e-05,
      "loss": 0.0045,
      "step": 1181
    },
    {
      "epoch": 0.43249176728869376,
      "grad_norm": 0.11843207478523254,
      "learning_rate": 3.954948301329395e-05,
      "loss": 0.0036,
      "step": 1182
    },
    {
      "epoch": 0.4328576655689718,
      "grad_norm": 0.10413416475057602,
      "learning_rate": 3.954025110782866e-05,
      "loss": 0.0031,
      "step": 1183
    },
    {
      "epoch": 0.4332235638492499,
      "grad_norm": 0.13455623388290405,
      "learning_rate": 3.9531019202363364e-05,
      "loss": 0.0043,
      "step": 1184
    },
    {
      "epoch": 0.433589462129528,
      "grad_norm": 6.681205749511719,
      "learning_rate": 3.952178729689808e-05,
      "loss": 0.5934,
      "step": 1185
    },
    {
      "epoch": 0.4339553604098061,
      "grad_norm": 4.692196369171143,
      "learning_rate": 3.951255539143279e-05,
      "loss": 0.0169,
      "step": 1186
    },
    {
      "epoch": 0.43432125869008414,
      "grad_norm": 0.11078128963708878,
      "learning_rate": 3.950332348596751e-05,
      "loss": 0.0026,
      "step": 1187
    },
    {
      "epoch": 0.43468715697036225,
      "grad_norm": 0.07518862187862396,
      "learning_rate": 3.949409158050222e-05,
      "loss": 0.0018,
      "step": 1188
    },
    {
      "epoch": 0.4350530552506403,
      "grad_norm": 5.672361373901367,
      "learning_rate": 3.948485967503693e-05,
      "loss": 0.0379,
      "step": 1189
    },
    {
      "epoch": 0.4354189535309184,
      "grad_norm": 20.077848434448242,
      "learning_rate": 3.947562776957164e-05,
      "loss": 0.2536,
      "step": 1190
    },
    {
      "epoch": 0.43578485181119647,
      "grad_norm": 0.43670853972435,
      "learning_rate": 3.946639586410635e-05,
      "loss": 0.012,
      "step": 1191
    },
    {
      "epoch": 0.4361507500914746,
      "grad_norm": 3.007561683654785,
      "learning_rate": 3.945716395864106e-05,
      "loss": 0.1361,
      "step": 1192
    },
    {
      "epoch": 0.43651664837175264,
      "grad_norm": 3.784693479537964,
      "learning_rate": 3.944793205317578e-05,
      "loss": 0.0365,
      "step": 1193
    },
    {
      "epoch": 0.43688254665203075,
      "grad_norm": 10.449925422668457,
      "learning_rate": 3.943870014771049e-05,
      "loss": 0.2137,
      "step": 1194
    },
    {
      "epoch": 0.4372484449323088,
      "grad_norm": 12.99402904510498,
      "learning_rate": 3.94294682422452e-05,
      "loss": 0.2137,
      "step": 1195
    },
    {
      "epoch": 0.4376143432125869,
      "grad_norm": 0.10513614118099213,
      "learning_rate": 3.942023633677991e-05,
      "loss": 0.0031,
      "step": 1196
    },
    {
      "epoch": 0.43798024149286496,
      "grad_norm": 3.429821729660034,
      "learning_rate": 3.941100443131462e-05,
      "loss": 0.2589,
      "step": 1197
    },
    {
      "epoch": 0.4383461397731431,
      "grad_norm": 0.3231624960899353,
      "learning_rate": 3.940177252584934e-05,
      "loss": 0.0061,
      "step": 1198
    },
    {
      "epoch": 0.4387120380534211,
      "grad_norm": 1.9943333864212036,
      "learning_rate": 3.939254062038405e-05,
      "loss": 0.1964,
      "step": 1199
    },
    {
      "epoch": 0.43907793633369924,
      "grad_norm": 1.9658881425857544,
      "learning_rate": 3.938330871491876e-05,
      "loss": 0.2241,
      "step": 1200
    },
    {
      "epoch": 0.4394438346139773,
      "grad_norm": 2.3353099822998047,
      "learning_rate": 3.937407680945348e-05,
      "loss": 0.1151,
      "step": 1201
    },
    {
      "epoch": 0.4398097328942554,
      "grad_norm": 2.3817670345306396,
      "learning_rate": 3.936484490398819e-05,
      "loss": 0.0444,
      "step": 1202
    },
    {
      "epoch": 0.44017563117453345,
      "grad_norm": 0.916593074798584,
      "learning_rate": 3.935561299852289e-05,
      "loss": 0.0295,
      "step": 1203
    },
    {
      "epoch": 0.44054152945481156,
      "grad_norm": 3.840327739715576,
      "learning_rate": 3.934638109305761e-05,
      "loss": 0.184,
      "step": 1204
    },
    {
      "epoch": 0.4409074277350896,
      "grad_norm": 2.428466796875,
      "learning_rate": 3.933714918759232e-05,
      "loss": 0.0706,
      "step": 1205
    },
    {
      "epoch": 0.44127332601536773,
      "grad_norm": 1.2786184549331665,
      "learning_rate": 3.932791728212703e-05,
      "loss": 0.0589,
      "step": 1206
    },
    {
      "epoch": 0.44163922429564584,
      "grad_norm": 1.162455677986145,
      "learning_rate": 3.931868537666175e-05,
      "loss": 0.0276,
      "step": 1207
    },
    {
      "epoch": 0.4420051225759239,
      "grad_norm": 3.185669183731079,
      "learning_rate": 3.930945347119646e-05,
      "loss": 0.0886,
      "step": 1208
    },
    {
      "epoch": 0.442371020856202,
      "grad_norm": 1.625535011291504,
      "learning_rate": 3.930022156573117e-05,
      "loss": 0.0415,
      "step": 1209
    },
    {
      "epoch": 0.44273691913648006,
      "grad_norm": 1.8044788837432861,
      "learning_rate": 3.929098966026588e-05,
      "loss": 0.0373,
      "step": 1210
    },
    {
      "epoch": 0.44310281741675817,
      "grad_norm": 15.793051719665527,
      "learning_rate": 3.928175775480059e-05,
      "loss": 0.3017,
      "step": 1211
    },
    {
      "epoch": 0.4434687156970362,
      "grad_norm": 4.2168474197387695,
      "learning_rate": 3.927252584933531e-05,
      "loss": 0.0399,
      "step": 1212
    },
    {
      "epoch": 0.44383461397731433,
      "grad_norm": 0.48735034465789795,
      "learning_rate": 3.926329394387002e-05,
      "loss": 0.0095,
      "step": 1213
    },
    {
      "epoch": 0.4442005122575924,
      "grad_norm": 0.07142732292413712,
      "learning_rate": 3.925406203840473e-05,
      "loss": 0.0015,
      "step": 1214
    },
    {
      "epoch": 0.4445664105378705,
      "grad_norm": 0.2552352249622345,
      "learning_rate": 3.924483013293944e-05,
      "loss": 0.0027,
      "step": 1215
    },
    {
      "epoch": 0.44493230881814855,
      "grad_norm": 3.586160659790039,
      "learning_rate": 3.923559822747415e-05,
      "loss": 0.1188,
      "step": 1216
    },
    {
      "epoch": 0.44529820709842666,
      "grad_norm": 0.031382057815790176,
      "learning_rate": 3.922636632200886e-05,
      "loss": 0.0011,
      "step": 1217
    },
    {
      "epoch": 0.4456641053787047,
      "grad_norm": 17.77608871459961,
      "learning_rate": 3.921713441654358e-05,
      "loss": 0.1177,
      "step": 1218
    },
    {
      "epoch": 0.4460300036589828,
      "grad_norm": 0.019583869725465775,
      "learning_rate": 3.920790251107829e-05,
      "loss": 0.0008,
      "step": 1219
    },
    {
      "epoch": 0.4463959019392609,
      "grad_norm": 0.022599974647164345,
      "learning_rate": 3.9198670605613005e-05,
      "loss": 0.0009,
      "step": 1220
    },
    {
      "epoch": 0.446761800219539,
      "grad_norm": 0.019005849957466125,
      "learning_rate": 3.9189438700147716e-05,
      "loss": 0.0008,
      "step": 1221
    },
    {
      "epoch": 0.44712769849981704,
      "grad_norm": 11.915019989013672,
      "learning_rate": 3.918020679468242e-05,
      "loss": 0.3449,
      "step": 1222
    },
    {
      "epoch": 0.44749359678009515,
      "grad_norm": 2.0722880363464355,
      "learning_rate": 3.917097488921714e-05,
      "loss": 0.0086,
      "step": 1223
    },
    {
      "epoch": 0.4478594950603732,
      "grad_norm": 0.014529752545058727,
      "learning_rate": 3.916174298375185e-05,
      "loss": 0.0006,
      "step": 1224
    },
    {
      "epoch": 0.4482253933406513,
      "grad_norm": 0.015977563336491585,
      "learning_rate": 3.915251107828656e-05,
      "loss": 0.0008,
      "step": 1225
    },
    {
      "epoch": 0.44859129162092937,
      "grad_norm": 0.031095772981643677,
      "learning_rate": 3.9143279172821275e-05,
      "loss": 0.0009,
      "step": 1226
    },
    {
      "epoch": 0.4489571899012075,
      "grad_norm": 0.25909677147865295,
      "learning_rate": 3.9134047267355986e-05,
      "loss": 0.0035,
      "step": 1227
    },
    {
      "epoch": 0.44932308818148553,
      "grad_norm": 0.014730390161275864,
      "learning_rate": 3.9124815361890696e-05,
      "loss": 0.0007,
      "step": 1228
    },
    {
      "epoch": 0.44968898646176364,
      "grad_norm": 0.013329708017408848,
      "learning_rate": 3.911558345642541e-05,
      "loss": 0.0006,
      "step": 1229
    },
    {
      "epoch": 0.4500548847420417,
      "grad_norm": 0.015133745968341827,
      "learning_rate": 3.910635155096012e-05,
      "loss": 0.0007,
      "step": 1230
    },
    {
      "epoch": 0.4504207830223198,
      "grad_norm": 0.01651792973279953,
      "learning_rate": 3.9097119645494835e-05,
      "loss": 0.0006,
      "step": 1231
    },
    {
      "epoch": 0.45078668130259786,
      "grad_norm": 0.033523645251989365,
      "learning_rate": 3.9087887740029545e-05,
      "loss": 0.0008,
      "step": 1232
    },
    {
      "epoch": 0.45115257958287597,
      "grad_norm": 3.5376527309417725,
      "learning_rate": 3.9078655834564256e-05,
      "loss": 0.0212,
      "step": 1233
    },
    {
      "epoch": 0.451518477863154,
      "grad_norm": 0.010612846352159977,
      "learning_rate": 3.9069423929098966e-05,
      "loss": 0.0005,
      "step": 1234
    },
    {
      "epoch": 0.45188437614343213,
      "grad_norm": 6.878598690032959,
      "learning_rate": 3.906019202363368e-05,
      "loss": 0.7213,
      "step": 1235
    },
    {
      "epoch": 0.4522502744237102,
      "grad_norm": 6.975252151489258,
      "learning_rate": 3.905096011816839e-05,
      "loss": 0.2954,
      "step": 1236
    },
    {
      "epoch": 0.4526161727039883,
      "grad_norm": 1.4294393062591553,
      "learning_rate": 3.9041728212703105e-05,
      "loss": 0.0053,
      "step": 1237
    },
    {
      "epoch": 0.45298207098426635,
      "grad_norm": 0.5245462656021118,
      "learning_rate": 3.9032496307237815e-05,
      "loss": 0.0037,
      "step": 1238
    },
    {
      "epoch": 0.45334796926454446,
      "grad_norm": 0.010950948111712933,
      "learning_rate": 3.902326440177253e-05,
      "loss": 0.0005,
      "step": 1239
    },
    {
      "epoch": 0.4537138675448225,
      "grad_norm": 0.01507834903895855,
      "learning_rate": 3.901403249630724e-05,
      "loss": 0.0006,
      "step": 1240
    },
    {
      "epoch": 0.4540797658251006,
      "grad_norm": 0.06422032415866852,
      "learning_rate": 3.900480059084195e-05,
      "loss": 0.0018,
      "step": 1241
    },
    {
      "epoch": 0.4544456641053787,
      "grad_norm": 0.013024665415287018,
      "learning_rate": 3.8995568685376664e-05,
      "loss": 0.0006,
      "step": 1242
    },
    {
      "epoch": 0.4548115623856568,
      "grad_norm": 0.15486672520637512,
      "learning_rate": 3.8986336779911375e-05,
      "loss": 0.0032,
      "step": 1243
    },
    {
      "epoch": 0.4551774606659349,
      "grad_norm": 3.3943631649017334,
      "learning_rate": 3.8977104874446085e-05,
      "loss": 0.2018,
      "step": 1244
    },
    {
      "epoch": 0.45554335894621295,
      "grad_norm": 13.925997734069824,
      "learning_rate": 3.89678729689808e-05,
      "loss": 0.0431,
      "step": 1245
    },
    {
      "epoch": 0.45590925722649106,
      "grad_norm": 0.03509340062737465,
      "learning_rate": 3.895864106351551e-05,
      "loss": 0.0012,
      "step": 1246
    },
    {
      "epoch": 0.4562751555067691,
      "grad_norm": 4.9563398361206055,
      "learning_rate": 3.8949409158050224e-05,
      "loss": 0.3721,
      "step": 1247
    },
    {
      "epoch": 0.4566410537870472,
      "grad_norm": 0.021506518125534058,
      "learning_rate": 3.8940177252584934e-05,
      "loss": 0.0008,
      "step": 1248
    },
    {
      "epoch": 0.4570069520673253,
      "grad_norm": 20.067646026611328,
      "learning_rate": 3.8930945347119645e-05,
      "loss": 0.29,
      "step": 1249
    },
    {
      "epoch": 0.4573728503476034,
      "grad_norm": 8.435562133789062,
      "learning_rate": 3.892171344165436e-05,
      "loss": 0.0539,
      "step": 1250
    },
    {
      "epoch": 0.45773874862788144,
      "grad_norm": 0.5992914438247681,
      "learning_rate": 3.891248153618907e-05,
      "loss": 0.0061,
      "step": 1251
    },
    {
      "epoch": 0.45810464690815955,
      "grad_norm": 0.29836106300354004,
      "learning_rate": 3.890324963072378e-05,
      "loss": 0.0035,
      "step": 1252
    },
    {
      "epoch": 0.4584705451884376,
      "grad_norm": 0.019277971237897873,
      "learning_rate": 3.8894017725258494e-05,
      "loss": 0.0008,
      "step": 1253
    },
    {
      "epoch": 0.4588364434687157,
      "grad_norm": 0.0703313946723938,
      "learning_rate": 3.8884785819793204e-05,
      "loss": 0.0023,
      "step": 1254
    },
    {
      "epoch": 0.45920234174899377,
      "grad_norm": 6.0354509353637695,
      "learning_rate": 3.8875553914327915e-05,
      "loss": 0.2217,
      "step": 1255
    },
    {
      "epoch": 0.4595682400292719,
      "grad_norm": 0.21379120647907257,
      "learning_rate": 3.886632200886263e-05,
      "loss": 0.0026,
      "step": 1256
    },
    {
      "epoch": 0.45993413830954993,
      "grad_norm": 0.997417151927948,
      "learning_rate": 3.885709010339734e-05,
      "loss": 0.0137,
      "step": 1257
    },
    {
      "epoch": 0.46030003658982804,
      "grad_norm": 0.6503170728683472,
      "learning_rate": 3.884785819793206e-05,
      "loss": 0.013,
      "step": 1258
    },
    {
      "epoch": 0.4606659348701061,
      "grad_norm": 0.06624758243560791,
      "learning_rate": 3.883862629246677e-05,
      "loss": 0.0015,
      "step": 1259
    },
    {
      "epoch": 0.4610318331503842,
      "grad_norm": 10.145898818969727,
      "learning_rate": 3.8829394387001474e-05,
      "loss": 0.1879,
      "step": 1260
    },
    {
      "epoch": 0.46139773143066226,
      "grad_norm": 0.406902939081192,
      "learning_rate": 3.882016248153619e-05,
      "loss": 0.0034,
      "step": 1261
    },
    {
      "epoch": 0.46176362971094037,
      "grad_norm": 0.016662435606122017,
      "learning_rate": 3.88109305760709e-05,
      "loss": 0.0008,
      "step": 1262
    },
    {
      "epoch": 0.4621295279912184,
      "grad_norm": 0.043328769505023956,
      "learning_rate": 3.880169867060561e-05,
      "loss": 0.0013,
      "step": 1263
    },
    {
      "epoch": 0.46249542627149653,
      "grad_norm": 7.523042678833008,
      "learning_rate": 3.879246676514033e-05,
      "loss": 0.3927,
      "step": 1264
    },
    {
      "epoch": 0.4628613245517746,
      "grad_norm": 0.035744212567806244,
      "learning_rate": 3.878323485967504e-05,
      "loss": 0.0012,
      "step": 1265
    },
    {
      "epoch": 0.4632272228320527,
      "grad_norm": 0.04558466747403145,
      "learning_rate": 3.877400295420975e-05,
      "loss": 0.0015,
      "step": 1266
    },
    {
      "epoch": 0.46359312111233075,
      "grad_norm": 0.034076981246471405,
      "learning_rate": 3.876477104874446e-05,
      "loss": 0.001,
      "step": 1267
    },
    {
      "epoch": 0.46395901939260886,
      "grad_norm": 0.012562642805278301,
      "learning_rate": 3.875553914327917e-05,
      "loss": 0.0005,
      "step": 1268
    },
    {
      "epoch": 0.4643249176728869,
      "grad_norm": 0.06436911225318909,
      "learning_rate": 3.874630723781389e-05,
      "loss": 0.0015,
      "step": 1269
    },
    {
      "epoch": 0.464690815953165,
      "grad_norm": 0.6137872934341431,
      "learning_rate": 3.87370753323486e-05,
      "loss": 0.0068,
      "step": 1270
    },
    {
      "epoch": 0.4650567142334431,
      "grad_norm": 5.8158369064331055,
      "learning_rate": 3.872784342688331e-05,
      "loss": 0.1883,
      "step": 1271
    },
    {
      "epoch": 0.4654226125137212,
      "grad_norm": 10.996541023254395,
      "learning_rate": 3.871861152141802e-05,
      "loss": 0.4284,
      "step": 1272
    },
    {
      "epoch": 0.46578851079399924,
      "grad_norm": 0.14878645539283752,
      "learning_rate": 3.870937961595273e-05,
      "loss": 0.0022,
      "step": 1273
    },
    {
      "epoch": 0.46615440907427735,
      "grad_norm": 5.017612457275391,
      "learning_rate": 3.870014771048744e-05,
      "loss": 0.1802,
      "step": 1274
    },
    {
      "epoch": 0.4665203073545554,
      "grad_norm": 4.712726593017578,
      "learning_rate": 3.869091580502216e-05,
      "loss": 0.0216,
      "step": 1275
    },
    {
      "epoch": 0.4668862056348335,
      "grad_norm": 0.12758804857730865,
      "learning_rate": 3.868168389955687e-05,
      "loss": 0.0028,
      "step": 1276
    },
    {
      "epoch": 0.46725210391511157,
      "grad_norm": 14.776644706726074,
      "learning_rate": 3.867245199409158e-05,
      "loss": 0.1293,
      "step": 1277
    },
    {
      "epoch": 0.4676180021953897,
      "grad_norm": 6.7290496826171875,
      "learning_rate": 3.86632200886263e-05,
      "loss": 0.0246,
      "step": 1278
    },
    {
      "epoch": 0.4679839004756678,
      "grad_norm": 0.03661499544978142,
      "learning_rate": 3.8653988183161e-05,
      "loss": 0.0013,
      "step": 1279
    },
    {
      "epoch": 0.46834979875594585,
      "grad_norm": 0.22214923799037933,
      "learning_rate": 3.864475627769572e-05,
      "loss": 0.0046,
      "step": 1280
    },
    {
      "epoch": 0.46871569703622395,
      "grad_norm": 0.6291002631187439,
      "learning_rate": 3.863552437223043e-05,
      "loss": 0.0081,
      "step": 1281
    },
    {
      "epoch": 0.469081595316502,
      "grad_norm": 0.04873746261000633,
      "learning_rate": 3.862629246676514e-05,
      "loss": 0.0012,
      "step": 1282
    },
    {
      "epoch": 0.4694474935967801,
      "grad_norm": 0.04482816532254219,
      "learning_rate": 3.861706056129986e-05,
      "loss": 0.0011,
      "step": 1283
    },
    {
      "epoch": 0.4698133918770582,
      "grad_norm": 0.024066675454378128,
      "learning_rate": 3.860782865583457e-05,
      "loss": 0.0009,
      "step": 1284
    },
    {
      "epoch": 0.4701792901573363,
      "grad_norm": 3.443962335586548,
      "learning_rate": 3.859859675036928e-05,
      "loss": 0.2056,
      "step": 1285
    },
    {
      "epoch": 0.47054518843761434,
      "grad_norm": 0.07571911066770554,
      "learning_rate": 3.858936484490399e-05,
      "loss": 0.0018,
      "step": 1286
    },
    {
      "epoch": 0.47091108671789245,
      "grad_norm": 0.30610793828964233,
      "learning_rate": 3.85801329394387e-05,
      "loss": 0.0061,
      "step": 1287
    },
    {
      "epoch": 0.4712769849981705,
      "grad_norm": 0.013288182206451893,
      "learning_rate": 3.857090103397342e-05,
      "loss": 0.0006,
      "step": 1288
    },
    {
      "epoch": 0.4716428832784486,
      "grad_norm": 0.2297922819852829,
      "learning_rate": 3.856166912850813e-05,
      "loss": 0.0032,
      "step": 1289
    },
    {
      "epoch": 0.47200878155872666,
      "grad_norm": 0.061011020094156265,
      "learning_rate": 3.855243722304284e-05,
      "loss": 0.0011,
      "step": 1290
    },
    {
      "epoch": 0.4723746798390048,
      "grad_norm": 0.6329822540283203,
      "learning_rate": 3.854320531757755e-05,
      "loss": 0.0066,
      "step": 1291
    },
    {
      "epoch": 0.47274057811928283,
      "grad_norm": 0.21964676678180695,
      "learning_rate": 3.853397341211226e-05,
      "loss": 0.0033,
      "step": 1292
    },
    {
      "epoch": 0.47310647639956094,
      "grad_norm": 0.15711137652397156,
      "learning_rate": 3.852474150664697e-05,
      "loss": 0.002,
      "step": 1293
    },
    {
      "epoch": 0.473472374679839,
      "grad_norm": 0.026697413995862007,
      "learning_rate": 3.851550960118169e-05,
      "loss": 0.0008,
      "step": 1294
    },
    {
      "epoch": 0.4738382729601171,
      "grad_norm": 0.06383597105741501,
      "learning_rate": 3.85062776957164e-05,
      "loss": 0.0013,
      "step": 1295
    },
    {
      "epoch": 0.47420417124039516,
      "grad_norm": 0.3952502906322479,
      "learning_rate": 3.849704579025111e-05,
      "loss": 0.0029,
      "step": 1296
    },
    {
      "epoch": 0.47457006952067327,
      "grad_norm": 6.0150651931762695,
      "learning_rate": 3.8487813884785826e-05,
      "loss": 0.019,
      "step": 1297
    },
    {
      "epoch": 0.4749359678009513,
      "grad_norm": 7.233732223510742,
      "learning_rate": 3.847858197932053e-05,
      "loss": 0.2072,
      "step": 1298
    },
    {
      "epoch": 0.47530186608122943,
      "grad_norm": 15.199966430664062,
      "learning_rate": 3.846935007385525e-05,
      "loss": 0.1388,
      "step": 1299
    },
    {
      "epoch": 0.4756677643615075,
      "grad_norm": 2.8507449626922607,
      "learning_rate": 3.846011816838996e-05,
      "loss": 0.0164,
      "step": 1300
    },
    {
      "epoch": 0.4760336626417856,
      "grad_norm": 8.760549545288086,
      "learning_rate": 3.845088626292467e-05,
      "loss": 0.2191,
      "step": 1301
    },
    {
      "epoch": 0.47639956092206365,
      "grad_norm": 4.0211992263793945,
      "learning_rate": 3.8441654357459385e-05,
      "loss": 0.5259,
      "step": 1302
    },
    {
      "epoch": 0.47676545920234176,
      "grad_norm": 0.06700951606035233,
      "learning_rate": 3.8432422451994096e-05,
      "loss": 0.0012,
      "step": 1303
    },
    {
      "epoch": 0.4771313574826198,
      "grad_norm": 11.548433303833008,
      "learning_rate": 3.8423190546528806e-05,
      "loss": 0.1661,
      "step": 1304
    },
    {
      "epoch": 0.4774972557628979,
      "grad_norm": 8.242792129516602,
      "learning_rate": 3.841395864106352e-05,
      "loss": 0.4201,
      "step": 1305
    },
    {
      "epoch": 0.477863154043176,
      "grad_norm": 0.056117840111255646,
      "learning_rate": 3.840472673559823e-05,
      "loss": 0.0014,
      "step": 1306
    },
    {
      "epoch": 0.4782290523234541,
      "grad_norm": 6.458941459655762,
      "learning_rate": 3.839549483013294e-05,
      "loss": 0.4038,
      "step": 1307
    },
    {
      "epoch": 0.47859495060373214,
      "grad_norm": 7.255924224853516,
      "learning_rate": 3.8386262924667655e-05,
      "loss": 0.3036,
      "step": 1308
    },
    {
      "epoch": 0.47896084888401025,
      "grad_norm": 4.03804349899292,
      "learning_rate": 3.8377031019202366e-05,
      "loss": 0.0919,
      "step": 1309
    },
    {
      "epoch": 0.4793267471642883,
      "grad_norm": 4.292646884918213,
      "learning_rate": 3.8367799113737076e-05,
      "loss": 0.3578,
      "step": 1310
    },
    {
      "epoch": 0.4796926454445664,
      "grad_norm": 1.5990495681762695,
      "learning_rate": 3.835856720827179e-05,
      "loss": 0.1164,
      "step": 1311
    },
    {
      "epoch": 0.48005854372484447,
      "grad_norm": 0.43767380714416504,
      "learning_rate": 3.83493353028065e-05,
      "loss": 0.0156,
      "step": 1312
    },
    {
      "epoch": 0.4804244420051226,
      "grad_norm": 2.316401720046997,
      "learning_rate": 3.8340103397341215e-05,
      "loss": 0.0432,
      "step": 1313
    },
    {
      "epoch": 0.4807903402854007,
      "grad_norm": 0.1680735945701599,
      "learning_rate": 3.8330871491875925e-05,
      "loss": 0.0058,
      "step": 1314
    },
    {
      "epoch": 0.48115623856567874,
      "grad_norm": 0.7046979665756226,
      "learning_rate": 3.8321639586410636e-05,
      "loss": 0.038,
      "step": 1315
    },
    {
      "epoch": 0.48152213684595685,
      "grad_norm": 1.065530776977539,
      "learning_rate": 3.831240768094535e-05,
      "loss": 0.0426,
      "step": 1316
    },
    {
      "epoch": 0.4818880351262349,
      "grad_norm": 0.3406153917312622,
      "learning_rate": 3.830317577548006e-05,
      "loss": 0.0115,
      "step": 1317
    },
    {
      "epoch": 0.482253933406513,
      "grad_norm": 0.48240768909454346,
      "learning_rate": 3.829394387001477e-05,
      "loss": 0.0154,
      "step": 1318
    },
    {
      "epoch": 0.48261983168679107,
      "grad_norm": 0.4768892228603363,
      "learning_rate": 3.8284711964549485e-05,
      "loss": 0.0114,
      "step": 1319
    },
    {
      "epoch": 0.4829857299670692,
      "grad_norm": 0.31146150827407837,
      "learning_rate": 3.8275480059084195e-05,
      "loss": 0.0097,
      "step": 1320
    },
    {
      "epoch": 0.48335162824734723,
      "grad_norm": 0.10258610546588898,
      "learning_rate": 3.826624815361891e-05,
      "loss": 0.0025,
      "step": 1321
    },
    {
      "epoch": 0.48371752652762534,
      "grad_norm": 9.567091941833496,
      "learning_rate": 3.825701624815362e-05,
      "loss": 0.0895,
      "step": 1322
    },
    {
      "epoch": 0.4840834248079034,
      "grad_norm": 11.533585548400879,
      "learning_rate": 3.8247784342688334e-05,
      "loss": 0.1086,
      "step": 1323
    },
    {
      "epoch": 0.4844493230881815,
      "grad_norm": 0.1087312325835228,
      "learning_rate": 3.8238552437223044e-05,
      "loss": 0.0024,
      "step": 1324
    },
    {
      "epoch": 0.48481522136845956,
      "grad_norm": 0.2303243726491928,
      "learning_rate": 3.8229320531757755e-05,
      "loss": 0.004,
      "step": 1325
    },
    {
      "epoch": 0.48518111964873767,
      "grad_norm": 0.03224661573767662,
      "learning_rate": 3.8220088626292465e-05,
      "loss": 0.0012,
      "step": 1326
    },
    {
      "epoch": 0.4855470179290157,
      "grad_norm": 0.02021127939224243,
      "learning_rate": 3.821085672082718e-05,
      "loss": 0.0007,
      "step": 1327
    },
    {
      "epoch": 0.48591291620929383,
      "grad_norm": 0.012686673551797867,
      "learning_rate": 3.820162481536189e-05,
      "loss": 0.0006,
      "step": 1328
    },
    {
      "epoch": 0.4862788144895719,
      "grad_norm": 0.037212811410427094,
      "learning_rate": 3.8192392909896604e-05,
      "loss": 0.0009,
      "step": 1329
    },
    {
      "epoch": 0.48664471276985,
      "grad_norm": 0.011553024873137474,
      "learning_rate": 3.8183161004431314e-05,
      "loss": 0.0005,
      "step": 1330
    },
    {
      "epoch": 0.48701061105012805,
      "grad_norm": 0.027502525597810745,
      "learning_rate": 3.8173929098966025e-05,
      "loss": 0.0007,
      "step": 1331
    },
    {
      "epoch": 0.48737650933040616,
      "grad_norm": 5.734894752502441,
      "learning_rate": 3.816469719350074e-05,
      "loss": 0.3541,
      "step": 1332
    },
    {
      "epoch": 0.4877424076106842,
      "grad_norm": 0.8268874883651733,
      "learning_rate": 3.815546528803545e-05,
      "loss": 0.0092,
      "step": 1333
    },
    {
      "epoch": 0.4881083058909623,
      "grad_norm": 0.4903489947319031,
      "learning_rate": 3.8146233382570163e-05,
      "loss": 0.0041,
      "step": 1334
    },
    {
      "epoch": 0.4884742041712404,
      "grad_norm": 0.06850067526102066,
      "learning_rate": 3.813700147710488e-05,
      "loss": 0.001,
      "step": 1335
    },
    {
      "epoch": 0.4888401024515185,
      "grad_norm": 7.346166133880615,
      "learning_rate": 3.8127769571639585e-05,
      "loss": 1.0631,
      "step": 1336
    },
    {
      "epoch": 0.48920600073179654,
      "grad_norm": 0.01184085849672556,
      "learning_rate": 3.8118537666174295e-05,
      "loss": 0.0005,
      "step": 1337
    },
    {
      "epoch": 0.48957189901207465,
      "grad_norm": 7.031966209411621,
      "learning_rate": 3.810930576070901e-05,
      "loss": 0.5367,
      "step": 1338
    },
    {
      "epoch": 0.4899377972923527,
      "grad_norm": 3.0458192825317383,
      "learning_rate": 3.810007385524372e-05,
      "loss": 0.3129,
      "step": 1339
    },
    {
      "epoch": 0.4903036955726308,
      "grad_norm": 4.330775260925293,
      "learning_rate": 3.809084194977844e-05,
      "loss": 0.199,
      "step": 1340
    },
    {
      "epoch": 0.49066959385290887,
      "grad_norm": 0.14177671074867249,
      "learning_rate": 3.808161004431315e-05,
      "loss": 0.003,
      "step": 1341
    },
    {
      "epoch": 0.491035492133187,
      "grad_norm": 0.03721010684967041,
      "learning_rate": 3.807237813884786e-05,
      "loss": 0.0014,
      "step": 1342
    },
    {
      "epoch": 0.49140139041346503,
      "grad_norm": 5.938660621643066,
      "learning_rate": 3.806314623338257e-05,
      "loss": 0.1044,
      "step": 1343
    },
    {
      "epoch": 0.49176728869374314,
      "grad_norm": 0.17518708109855652,
      "learning_rate": 3.805391432791728e-05,
      "loss": 0.004,
      "step": 1344
    },
    {
      "epoch": 0.4921331869740212,
      "grad_norm": 2.817235231399536,
      "learning_rate": 3.804468242245199e-05,
      "loss": 0.2113,
      "step": 1345
    },
    {
      "epoch": 0.4924990852542993,
      "grad_norm": 0.43708550930023193,
      "learning_rate": 3.803545051698671e-05,
      "loss": 0.0134,
      "step": 1346
    },
    {
      "epoch": 0.49286498353457736,
      "grad_norm": 0.24049822986125946,
      "learning_rate": 3.802621861152142e-05,
      "loss": 0.0059,
      "step": 1347
    },
    {
      "epoch": 0.49323088181485547,
      "grad_norm": 3.1215367317199707,
      "learning_rate": 3.801698670605613e-05,
      "loss": 0.2695,
      "step": 1348
    },
    {
      "epoch": 0.4935967800951336,
      "grad_norm": 6.729610919952393,
      "learning_rate": 3.800775480059084e-05,
      "loss": 0.1228,
      "step": 1349
    },
    {
      "epoch": 0.49396267837541163,
      "grad_norm": 0.9442467093467712,
      "learning_rate": 3.799852289512555e-05,
      "loss": 0.0305,
      "step": 1350
    },
    {
      "epoch": 0.49432857665568974,
      "grad_norm": 0.26898929476737976,
      "learning_rate": 3.798929098966027e-05,
      "loss": 0.0078,
      "step": 1351
    },
    {
      "epoch": 0.4946944749359678,
      "grad_norm": 0.2631794810295105,
      "learning_rate": 3.798005908419498e-05,
      "loss": 0.0062,
      "step": 1352
    },
    {
      "epoch": 0.4950603732162459,
      "grad_norm": 0.44249677658081055,
      "learning_rate": 3.797082717872969e-05,
      "loss": 0.0129,
      "step": 1353
    },
    {
      "epoch": 0.49542627149652396,
      "grad_norm": 0.6496700644493103,
      "learning_rate": 3.796159527326441e-05,
      "loss": 0.0208,
      "step": 1354
    },
    {
      "epoch": 0.49579216977680207,
      "grad_norm": 0.2770628035068512,
      "learning_rate": 3.795236336779911e-05,
      "loss": 0.0061,
      "step": 1355
    },
    {
      "epoch": 0.4961580680570801,
      "grad_norm": 0.3411470949649811,
      "learning_rate": 3.794313146233382e-05,
      "loss": 0.008,
      "step": 1356
    },
    {
      "epoch": 0.49652396633735824,
      "grad_norm": 0.27558228373527527,
      "learning_rate": 3.793389955686854e-05,
      "loss": 0.0069,
      "step": 1357
    },
    {
      "epoch": 0.4968898646176363,
      "grad_norm": 2.013265371322632,
      "learning_rate": 3.792466765140325e-05,
      "loss": 0.4168,
      "step": 1358
    },
    {
      "epoch": 0.4972557628979144,
      "grad_norm": 0.06993257254362106,
      "learning_rate": 3.791543574593797e-05,
      "loss": 0.0021,
      "step": 1359
    },
    {
      "epoch": 0.49762166117819245,
      "grad_norm": 5.14052677154541,
      "learning_rate": 3.790620384047268e-05,
      "loss": 0.0477,
      "step": 1360
    },
    {
      "epoch": 0.49798755945847056,
      "grad_norm": 3.9058687686920166,
      "learning_rate": 3.789697193500739e-05,
      "loss": 0.3372,
      "step": 1361
    },
    {
      "epoch": 0.4983534577387486,
      "grad_norm": 0.46714112162590027,
      "learning_rate": 3.78877400295421e-05,
      "loss": 0.0161,
      "step": 1362
    },
    {
      "epoch": 0.4987193560190267,
      "grad_norm": 0.3185409605503082,
      "learning_rate": 3.787850812407681e-05,
      "loss": 0.0133,
      "step": 1363
    },
    {
      "epoch": 0.4990852542993048,
      "grad_norm": 0.8244525194168091,
      "learning_rate": 3.786927621861152e-05,
      "loss": 0.013,
      "step": 1364
    },
    {
      "epoch": 0.4994511525795829,
      "grad_norm": 0.13586550951004028,
      "learning_rate": 3.786004431314624e-05,
      "loss": 0.0048,
      "step": 1365
    },
    {
      "epoch": 0.49981705085986095,
      "grad_norm": 0.9259798526763916,
      "learning_rate": 3.785081240768095e-05,
      "loss": 0.0076,
      "step": 1366
    },
    {
      "epoch": 0.500182949140139,
      "grad_norm": 10.25019645690918,
      "learning_rate": 3.784158050221566e-05,
      "loss": 0.072,
      "step": 1367
    },
    {
      "epoch": 0.5005488474204172,
      "grad_norm": 1.1607707738876343,
      "learning_rate": 3.783234859675037e-05,
      "loss": 0.0089,
      "step": 1368
    },
    {
      "epoch": 0.5009147457006952,
      "grad_norm": 11.86716079711914,
      "learning_rate": 3.782311669128508e-05,
      "loss": 0.1385,
      "step": 1369
    },
    {
      "epoch": 0.5012806439809733,
      "grad_norm": 1.5070074796676636,
      "learning_rate": 3.78138847858198e-05,
      "loss": 0.0149,
      "step": 1370
    },
    {
      "epoch": 0.5016465422612514,
      "grad_norm": 0.03033948875963688,
      "learning_rate": 3.780465288035451e-05,
      "loss": 0.0014,
      "step": 1371
    },
    {
      "epoch": 0.5020124405415295,
      "grad_norm": 0.5274598598480225,
      "learning_rate": 3.779542097488922e-05,
      "loss": 0.0072,
      "step": 1372
    },
    {
      "epoch": 0.5023783388218075,
      "grad_norm": 3.510751485824585,
      "learning_rate": 3.7786189069423936e-05,
      "loss": 0.1019,
      "step": 1373
    },
    {
      "epoch": 0.5027442371020856,
      "grad_norm": 0.06734927743673325,
      "learning_rate": 3.777695716395864e-05,
      "loss": 0.0027,
      "step": 1374
    },
    {
      "epoch": 0.5031101353823637,
      "grad_norm": 8.664851188659668,
      "learning_rate": 3.776772525849335e-05,
      "loss": 0.204,
      "step": 1375
    },
    {
      "epoch": 0.5034760336626418,
      "grad_norm": 0.09268740564584732,
      "learning_rate": 3.775849335302807e-05,
      "loss": 0.0036,
      "step": 1376
    },
    {
      "epoch": 0.5038419319429198,
      "grad_norm": 3.6655993461608887,
      "learning_rate": 3.774926144756278e-05,
      "loss": 0.1946,
      "step": 1377
    },
    {
      "epoch": 0.5042078302231979,
      "grad_norm": 0.15884904563426971,
      "learning_rate": 3.774002954209749e-05,
      "loss": 0.0041,
      "step": 1378
    },
    {
      "epoch": 0.504573728503476,
      "grad_norm": 7.067004680633545,
      "learning_rate": 3.7730797636632206e-05,
      "loss": 0.035,
      "step": 1379
    },
    {
      "epoch": 0.5049396267837541,
      "grad_norm": 0.07183433324098587,
      "learning_rate": 3.7721565731166916e-05,
      "loss": 0.0029,
      "step": 1380
    },
    {
      "epoch": 0.5053055250640321,
      "grad_norm": 0.13491563498973846,
      "learning_rate": 3.771233382570163e-05,
      "loss": 0.0038,
      "step": 1381
    },
    {
      "epoch": 0.5056714233443103,
      "grad_norm": 0.05412260442972183,
      "learning_rate": 3.770310192023634e-05,
      "loss": 0.0021,
      "step": 1382
    },
    {
      "epoch": 0.5060373216245884,
      "grad_norm": 4.304403305053711,
      "learning_rate": 3.769387001477105e-05,
      "loss": 0.0306,
      "step": 1383
    },
    {
      "epoch": 0.5064032199048665,
      "grad_norm": 5.938301086425781,
      "learning_rate": 3.7684638109305765e-05,
      "loss": 0.0197,
      "step": 1384
    },
    {
      "epoch": 0.5067691181851446,
      "grad_norm": 0.3922494649887085,
      "learning_rate": 3.7675406203840476e-05,
      "loss": 0.0062,
      "step": 1385
    },
    {
      "epoch": 0.5071350164654226,
      "grad_norm": 4.727331638336182,
      "learning_rate": 3.7666174298375187e-05,
      "loss": 0.0377,
      "step": 1386
    },
    {
      "epoch": 0.5075009147457007,
      "grad_norm": 0.020003730431199074,
      "learning_rate": 3.76569423929099e-05,
      "loss": 0.0009,
      "step": 1387
    },
    {
      "epoch": 0.5078668130259788,
      "grad_norm": 0.07309814542531967,
      "learning_rate": 3.764771048744461e-05,
      "loss": 0.0026,
      "step": 1388
    },
    {
      "epoch": 0.5082327113062569,
      "grad_norm": 0.0853487104177475,
      "learning_rate": 3.7638478581979325e-05,
      "loss": 0.0019,
      "step": 1389
    },
    {
      "epoch": 0.5085986095865349,
      "grad_norm": 0.07859499007463455,
      "learning_rate": 3.7629246676514035e-05,
      "loss": 0.002,
      "step": 1390
    },
    {
      "epoch": 0.508964507866813,
      "grad_norm": 0.03898546099662781,
      "learning_rate": 3.7620014771048746e-05,
      "loss": 0.0012,
      "step": 1391
    },
    {
      "epoch": 0.5093304061470911,
      "grad_norm": 0.12995600700378418,
      "learning_rate": 3.761078286558346e-05,
      "loss": 0.0028,
      "step": 1392
    },
    {
      "epoch": 0.5096963044273692,
      "grad_norm": 9.133201599121094,
      "learning_rate": 3.760155096011817e-05,
      "loss": 0.0465,
      "step": 1393
    },
    {
      "epoch": 0.5100622027076472,
      "grad_norm": 0.5658473968505859,
      "learning_rate": 3.759231905465288e-05,
      "loss": 0.0047,
      "step": 1394
    },
    {
      "epoch": 0.5104281009879253,
      "grad_norm": 0.0679624155163765,
      "learning_rate": 3.7583087149187595e-05,
      "loss": 0.0016,
      "step": 1395
    },
    {
      "epoch": 0.5107939992682035,
      "grad_norm": 0.01758832111954689,
      "learning_rate": 3.7573855243722306e-05,
      "loss": 0.0007,
      "step": 1396
    },
    {
      "epoch": 0.5111598975484816,
      "grad_norm": 0.17185908555984497,
      "learning_rate": 3.7564623338257016e-05,
      "loss": 0.0029,
      "step": 1397
    },
    {
      "epoch": 0.5115257958287596,
      "grad_norm": 6.745234966278076,
      "learning_rate": 3.7555391432791733e-05,
      "loss": 0.0156,
      "step": 1398
    },
    {
      "epoch": 0.5118916941090377,
      "grad_norm": 0.009595319628715515,
      "learning_rate": 3.7546159527326444e-05,
      "loss": 0.0004,
      "step": 1399
    },
    {
      "epoch": 0.5122575923893158,
      "grad_norm": 4.358095645904541,
      "learning_rate": 3.7536927621861155e-05,
      "loss": 0.0151,
      "step": 1400
    },
    {
      "epoch": 0.5126234906695939,
      "grad_norm": 7.120147705078125,
      "learning_rate": 3.7527695716395865e-05,
      "loss": 0.3462,
      "step": 1401
    },
    {
      "epoch": 0.5129893889498719,
      "grad_norm": 19.997760772705078,
      "learning_rate": 3.7518463810930576e-05,
      "loss": 0.365,
      "step": 1402
    },
    {
      "epoch": 0.51335528723015,
      "grad_norm": 0.0068123782984912395,
      "learning_rate": 3.750923190546529e-05,
      "loss": 0.0003,
      "step": 1403
    },
    {
      "epoch": 0.5137211855104281,
      "grad_norm": 0.023148231208324432,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.0006,
      "step": 1404
    },
    {
      "epoch": 0.5140870837907062,
      "grad_norm": 13.539356231689453,
      "learning_rate": 3.7490768094534714e-05,
      "loss": 0.1658,
      "step": 1405
    },
    {
      "epoch": 0.5144529820709842,
      "grad_norm": 0.07005525380373001,
      "learning_rate": 3.7481536189069425e-05,
      "loss": 0.0013,
      "step": 1406
    },
    {
      "epoch": 0.5148188803512623,
      "grad_norm": 0.10167434811592102,
      "learning_rate": 3.7472304283604135e-05,
      "loss": 0.0019,
      "step": 1407
    },
    {
      "epoch": 0.5151847786315404,
      "grad_norm": 0.18654881417751312,
      "learning_rate": 3.7463072378138846e-05,
      "loss": 0.0023,
      "step": 1408
    },
    {
      "epoch": 0.5155506769118186,
      "grad_norm": 0.13818861544132233,
      "learning_rate": 3.745384047267356e-05,
      "loss": 0.0026,
      "step": 1409
    },
    {
      "epoch": 0.5159165751920965,
      "grad_norm": 0.09647787362337112,
      "learning_rate": 3.7444608567208274e-05,
      "loss": 0.0021,
      "step": 1410
    },
    {
      "epoch": 0.5162824734723747,
      "grad_norm": 29.2469425201416,
      "learning_rate": 3.743537666174299e-05,
      "loss": 0.0443,
      "step": 1411
    },
    {
      "epoch": 0.5166483717526528,
      "grad_norm": 9.464944839477539,
      "learning_rate": 3.7426144756277695e-05,
      "loss": 0.0242,
      "step": 1412
    },
    {
      "epoch": 0.5170142700329309,
      "grad_norm": 0.04764889180660248,
      "learning_rate": 3.7416912850812405e-05,
      "loss": 0.0013,
      "step": 1413
    },
    {
      "epoch": 0.5173801683132089,
      "grad_norm": 4.456536293029785,
      "learning_rate": 3.740768094534712e-05,
      "loss": 0.4624,
      "step": 1414
    },
    {
      "epoch": 0.517746066593487,
      "grad_norm": 5.02349853515625,
      "learning_rate": 3.739844903988183e-05,
      "loss": 0.7676,
      "step": 1415
    },
    {
      "epoch": 0.5181119648737651,
      "grad_norm": 0.07159440219402313,
      "learning_rate": 3.7389217134416544e-05,
      "loss": 0.0019,
      "step": 1416
    },
    {
      "epoch": 0.5184778631540432,
      "grad_norm": 0.19156764447689056,
      "learning_rate": 3.737998522895126e-05,
      "loss": 0.0024,
      "step": 1417
    },
    {
      "epoch": 0.5188437614343212,
      "grad_norm": 10.195390701293945,
      "learning_rate": 3.737075332348597e-05,
      "loss": 0.5118,
      "step": 1418
    },
    {
      "epoch": 0.5192096597145993,
      "grad_norm": 0.9091480374336243,
      "learning_rate": 3.7361521418020675e-05,
      "loss": 0.007,
      "step": 1419
    },
    {
      "epoch": 0.5195755579948774,
      "grad_norm": 21.69405746459961,
      "learning_rate": 3.735228951255539e-05,
      "loss": 0.3997,
      "step": 1420
    },
    {
      "epoch": 0.5199414562751555,
      "grad_norm": 0.05149007961153984,
      "learning_rate": 3.73430576070901e-05,
      "loss": 0.002,
      "step": 1421
    },
    {
      "epoch": 0.5203073545554336,
      "grad_norm": 4.977615833282471,
      "learning_rate": 3.733382570162482e-05,
      "loss": 0.2174,
      "step": 1422
    },
    {
      "epoch": 0.5206732528357116,
      "grad_norm": 0.2834159731864929,
      "learning_rate": 3.732459379615953e-05,
      "loss": 0.006,
      "step": 1423
    },
    {
      "epoch": 0.5210391511159898,
      "grad_norm": 2.6549558639526367,
      "learning_rate": 3.731536189069424e-05,
      "loss": 0.0145,
      "step": 1424
    },
    {
      "epoch": 0.5214050493962679,
      "grad_norm": 1.0006746053695679,
      "learning_rate": 3.730612998522895e-05,
      "loss": 0.0173,
      "step": 1425
    },
    {
      "epoch": 0.521770947676546,
      "grad_norm": 0.09285387396812439,
      "learning_rate": 3.729689807976366e-05,
      "loss": 0.004,
      "step": 1426
    },
    {
      "epoch": 0.522136845956824,
      "grad_norm": 0.05323071777820587,
      "learning_rate": 3.728766617429837e-05,
      "loss": 0.0024,
      "step": 1427
    },
    {
      "epoch": 0.5225027442371021,
      "grad_norm": 0.06877884268760681,
      "learning_rate": 3.727843426883309e-05,
      "loss": 0.0027,
      "step": 1428
    },
    {
      "epoch": 0.5228686425173802,
      "grad_norm": 2.7634692192077637,
      "learning_rate": 3.72692023633678e-05,
      "loss": 0.0147,
      "step": 1429
    },
    {
      "epoch": 0.5232345407976583,
      "grad_norm": 0.05011645331978798,
      "learning_rate": 3.725997045790252e-05,
      "loss": 0.0018,
      "step": 1430
    },
    {
      "epoch": 0.5236004390779363,
      "grad_norm": 11.655360221862793,
      "learning_rate": 3.725073855243722e-05,
      "loss": 0.1697,
      "step": 1431
    },
    {
      "epoch": 0.5239663373582144,
      "grad_norm": 0.05959645286202431,
      "learning_rate": 3.724150664697193e-05,
      "loss": 0.0021,
      "step": 1432
    },
    {
      "epoch": 0.5243322356384925,
      "grad_norm": 1.9126650094985962,
      "learning_rate": 3.723227474150665e-05,
      "loss": 0.2967,
      "step": 1433
    },
    {
      "epoch": 0.5246981339187706,
      "grad_norm": 0.5846908688545227,
      "learning_rate": 3.722304283604136e-05,
      "loss": 0.0082,
      "step": 1434
    },
    {
      "epoch": 0.5250640321990486,
      "grad_norm": 0.04063507169485092,
      "learning_rate": 3.721381093057607e-05,
      "loss": 0.0017,
      "step": 1435
    },
    {
      "epoch": 0.5254299304793267,
      "grad_norm": 0.06333126127719879,
      "learning_rate": 3.720457902511079e-05,
      "loss": 0.0025,
      "step": 1436
    },
    {
      "epoch": 0.5257958287596048,
      "grad_norm": 0.1643441915512085,
      "learning_rate": 3.71953471196455e-05,
      "loss": 0.0053,
      "step": 1437
    },
    {
      "epoch": 0.526161727039883,
      "grad_norm": 0.06259298324584961,
      "learning_rate": 3.71861152141802e-05,
      "loss": 0.0026,
      "step": 1438
    },
    {
      "epoch": 0.526527625320161,
      "grad_norm": 0.6461468935012817,
      "learning_rate": 3.717688330871492e-05,
      "loss": 0.009,
      "step": 1439
    },
    {
      "epoch": 0.5268935236004391,
      "grad_norm": 13.072718620300293,
      "learning_rate": 3.716765140324963e-05,
      "loss": 0.1918,
      "step": 1440
    },
    {
      "epoch": 0.5272594218807172,
      "grad_norm": 8.290454864501953,
      "learning_rate": 3.715841949778435e-05,
      "loss": 0.202,
      "step": 1441
    },
    {
      "epoch": 0.5276253201609953,
      "grad_norm": 0.07948648929595947,
      "learning_rate": 3.714918759231906e-05,
      "loss": 0.0025,
      "step": 1442
    },
    {
      "epoch": 0.5279912184412733,
      "grad_norm": 0.11116126924753189,
      "learning_rate": 3.713995568685377e-05,
      "loss": 0.0036,
      "step": 1443
    },
    {
      "epoch": 0.5283571167215514,
      "grad_norm": 0.04947759583592415,
      "learning_rate": 3.713072378138848e-05,
      "loss": 0.0014,
      "step": 1444
    },
    {
      "epoch": 0.5287230150018295,
      "grad_norm": 0.7223473787307739,
      "learning_rate": 3.712149187592319e-05,
      "loss": 0.0059,
      "step": 1445
    },
    {
      "epoch": 0.5290889132821076,
      "grad_norm": 16.546201705932617,
      "learning_rate": 3.71122599704579e-05,
      "loss": 0.4297,
      "step": 1446
    },
    {
      "epoch": 0.5294548115623856,
      "grad_norm": 0.021644359454512596,
      "learning_rate": 3.710302806499262e-05,
      "loss": 0.0008,
      "step": 1447
    },
    {
      "epoch": 0.5298207098426637,
      "grad_norm": 0.6369462609291077,
      "learning_rate": 3.709379615952733e-05,
      "loss": 0.0074,
      "step": 1448
    },
    {
      "epoch": 0.5301866081229418,
      "grad_norm": 0.09089614450931549,
      "learning_rate": 3.708456425406204e-05,
      "loss": 0.0028,
      "step": 1449
    },
    {
      "epoch": 0.5305525064032199,
      "grad_norm": 0.03358077630400658,
      "learning_rate": 3.707533234859675e-05,
      "loss": 0.0012,
      "step": 1450
    },
    {
      "epoch": 0.5309184046834979,
      "grad_norm": 0.028674017637968063,
      "learning_rate": 3.706610044313146e-05,
      "loss": 0.0011,
      "step": 1451
    },
    {
      "epoch": 0.531284302963776,
      "grad_norm": 33.01530838012695,
      "learning_rate": 3.705686853766618e-05,
      "loss": 0.2843,
      "step": 1452
    },
    {
      "epoch": 0.5316502012440542,
      "grad_norm": 30.223981857299805,
      "learning_rate": 3.704763663220089e-05,
      "loss": 0.2304,
      "step": 1453
    },
    {
      "epoch": 0.5320160995243323,
      "grad_norm": 0.07279881089925766,
      "learning_rate": 3.70384047267356e-05,
      "loss": 0.002,
      "step": 1454
    },
    {
      "epoch": 0.5323819978046103,
      "grad_norm": 0.07510687410831451,
      "learning_rate": 3.7029172821270316e-05,
      "loss": 0.0019,
      "step": 1455
    },
    {
      "epoch": 0.5327478960848884,
      "grad_norm": 15.411957740783691,
      "learning_rate": 3.7019940915805027e-05,
      "loss": 0.2197,
      "step": 1456
    },
    {
      "epoch": 0.5331137943651665,
      "grad_norm": 0.05898379534482956,
      "learning_rate": 3.701070901033973e-05,
      "loss": 0.0014,
      "step": 1457
    },
    {
      "epoch": 0.5334796926454446,
      "grad_norm": 0.04036230221390724,
      "learning_rate": 3.700147710487445e-05,
      "loss": 0.0009,
      "step": 1458
    },
    {
      "epoch": 0.5338455909257227,
      "grad_norm": 0.0337895005941391,
      "learning_rate": 3.699224519940916e-05,
      "loss": 0.0011,
      "step": 1459
    },
    {
      "epoch": 0.5342114892060007,
      "grad_norm": 5.443002223968506,
      "learning_rate": 3.6983013293943876e-05,
      "loss": 0.2233,
      "step": 1460
    },
    {
      "epoch": 0.5345773874862788,
      "grad_norm": 23.393171310424805,
      "learning_rate": 3.6973781388478586e-05,
      "loss": 0.1214,
      "step": 1461
    },
    {
      "epoch": 0.5349432857665569,
      "grad_norm": 0.0864652767777443,
      "learning_rate": 3.69645494830133e-05,
      "loss": 0.0021,
      "step": 1462
    },
    {
      "epoch": 0.535309184046835,
      "grad_norm": 0.2038879096508026,
      "learning_rate": 3.695531757754801e-05,
      "loss": 0.0041,
      "step": 1463
    },
    {
      "epoch": 0.535675082327113,
      "grad_norm": 10.953527450561523,
      "learning_rate": 3.694608567208272e-05,
      "loss": 0.1214,
      "step": 1464
    },
    {
      "epoch": 0.5360409806073911,
      "grad_norm": 11.903515815734863,
      "learning_rate": 3.693685376661743e-05,
      "loss": 0.1978,
      "step": 1465
    },
    {
      "epoch": 0.5364068788876692,
      "grad_norm": 3.946101188659668,
      "learning_rate": 3.6927621861152146e-05,
      "loss": 0.2976,
      "step": 1466
    },
    {
      "epoch": 0.5367727771679474,
      "grad_norm": 6.879022598266602,
      "learning_rate": 3.6918389955686856e-05,
      "loss": 0.1652,
      "step": 1467
    },
    {
      "epoch": 0.5371386754482254,
      "grad_norm": 6.046897888183594,
      "learning_rate": 3.690915805022157e-05,
      "loss": 0.0286,
      "step": 1468
    },
    {
      "epoch": 0.5375045737285035,
      "grad_norm": 9.75317096710205,
      "learning_rate": 3.689992614475628e-05,
      "loss": 0.0243,
      "step": 1469
    },
    {
      "epoch": 0.5378704720087816,
      "grad_norm": 0.07495024055242538,
      "learning_rate": 3.689069423929099e-05,
      "loss": 0.0012,
      "step": 1470
    },
    {
      "epoch": 0.5382363702890597,
      "grad_norm": 3.5706560611724854,
      "learning_rate": 3.6881462333825705e-05,
      "loss": 0.0183,
      "step": 1471
    },
    {
      "epoch": 0.5386022685693377,
      "grad_norm": 5.253696441650391,
      "learning_rate": 3.6872230428360416e-05,
      "loss": 0.2313,
      "step": 1472
    },
    {
      "epoch": 0.5389681668496158,
      "grad_norm": 7.120874881744385,
      "learning_rate": 3.6862998522895126e-05,
      "loss": 0.1523,
      "step": 1473
    },
    {
      "epoch": 0.5393340651298939,
      "grad_norm": 0.03661790490150452,
      "learning_rate": 3.6853766617429844e-05,
      "loss": 0.0013,
      "step": 1474
    },
    {
      "epoch": 0.539699963410172,
      "grad_norm": 1.719605565071106,
      "learning_rate": 3.6844534711964554e-05,
      "loss": 0.0127,
      "step": 1475
    },
    {
      "epoch": 0.54006586169045,
      "grad_norm": 0.14737622439861298,
      "learning_rate": 3.683530280649926e-05,
      "loss": 0.0022,
      "step": 1476
    },
    {
      "epoch": 0.5404317599707281,
      "grad_norm": 0.28409069776535034,
      "learning_rate": 3.6826070901033975e-05,
      "loss": 0.0043,
      "step": 1477
    },
    {
      "epoch": 0.5407976582510062,
      "grad_norm": 5.308727741241455,
      "learning_rate": 3.6816838995568686e-05,
      "loss": 0.265,
      "step": 1478
    },
    {
      "epoch": 0.5411635565312843,
      "grad_norm": 0.16749106347560883,
      "learning_rate": 3.6807607090103396e-05,
      "loss": 0.0031,
      "step": 1479
    },
    {
      "epoch": 0.5415294548115623,
      "grad_norm": 0.06920225918292999,
      "learning_rate": 3.6798375184638114e-05,
      "loss": 0.0016,
      "step": 1480
    },
    {
      "epoch": 0.5418953530918404,
      "grad_norm": 8.161705017089844,
      "learning_rate": 3.6789143279172824e-05,
      "loss": 0.2158,
      "step": 1481
    },
    {
      "epoch": 0.5422612513721186,
      "grad_norm": 0.021400993689894676,
      "learning_rate": 3.6779911373707535e-05,
      "loss": 0.0006,
      "step": 1482
    },
    {
      "epoch": 0.5426271496523967,
      "grad_norm": 0.1332070380449295,
      "learning_rate": 3.6770679468242245e-05,
      "loss": 0.0027,
      "step": 1483
    },
    {
      "epoch": 0.5429930479326747,
      "grad_norm": 0.8259792327880859,
      "learning_rate": 3.6761447562776956e-05,
      "loss": 0.0172,
      "step": 1484
    },
    {
      "epoch": 0.5433589462129528,
      "grad_norm": 0.23839232325553894,
      "learning_rate": 3.675221565731167e-05,
      "loss": 0.0071,
      "step": 1485
    },
    {
      "epoch": 0.5437248444932309,
      "grad_norm": 0.0994175598025322,
      "learning_rate": 3.6742983751846384e-05,
      "loss": 0.0027,
      "step": 1486
    },
    {
      "epoch": 0.544090742773509,
      "grad_norm": 0.5886296033859253,
      "learning_rate": 3.6733751846381094e-05,
      "loss": 0.0065,
      "step": 1487
    },
    {
      "epoch": 0.544456641053787,
      "grad_norm": 0.014311675913631916,
      "learning_rate": 3.6724519940915805e-05,
      "loss": 0.0005,
      "step": 1488
    },
    {
      "epoch": 0.5448225393340651,
      "grad_norm": 0.15214568376541138,
      "learning_rate": 3.6715288035450515e-05,
      "loss": 0.0042,
      "step": 1489
    },
    {
      "epoch": 0.5451884376143432,
      "grad_norm": 0.2541103661060333,
      "learning_rate": 3.670605612998523e-05,
      "loss": 0.0054,
      "step": 1490
    },
    {
      "epoch": 0.5455543358946213,
      "grad_norm": 0.009985091164708138,
      "learning_rate": 3.669682422451994e-05,
      "loss": 0.0004,
      "step": 1491
    },
    {
      "epoch": 0.5459202341748994,
      "grad_norm": 0.3967941403388977,
      "learning_rate": 3.6687592319054654e-05,
      "loss": 0.0099,
      "step": 1492
    },
    {
      "epoch": 0.5462861324551774,
      "grad_norm": 0.008313601836562157,
      "learning_rate": 3.667836041358937e-05,
      "loss": 0.0004,
      "step": 1493
    },
    {
      "epoch": 0.5466520307354555,
      "grad_norm": 0.014807811938226223,
      "learning_rate": 3.666912850812408e-05,
      "loss": 0.0005,
      "step": 1494
    },
    {
      "epoch": 0.5470179290157337,
      "grad_norm": 0.11144400388002396,
      "learning_rate": 3.6659896602658785e-05,
      "loss": 0.0036,
      "step": 1495
    },
    {
      "epoch": 0.5473838272960118,
      "grad_norm": 8.968151092529297,
      "learning_rate": 3.66506646971935e-05,
      "loss": 0.9088,
      "step": 1496
    },
    {
      "epoch": 0.5477497255762898,
      "grad_norm": 0.3024086356163025,
      "learning_rate": 3.664143279172821e-05,
      "loss": 0.0035,
      "step": 1497
    },
    {
      "epoch": 0.5481156238565679,
      "grad_norm": 2.276575803756714,
      "learning_rate": 3.6632200886262924e-05,
      "loss": 0.0183,
      "step": 1498
    },
    {
      "epoch": 0.548481522136846,
      "grad_norm": 0.20214639604091644,
      "learning_rate": 3.662296898079764e-05,
      "loss": 0.0031,
      "step": 1499
    },
    {
      "epoch": 0.5488474204171241,
      "grad_norm": 0.05630571395158768,
      "learning_rate": 3.661373707533235e-05,
      "loss": 0.0013,
      "step": 1500
    },
    {
      "epoch": 0.5492133186974021,
      "grad_norm": 0.009857316501438618,
      "learning_rate": 3.660450516986706e-05,
      "loss": 0.0004,
      "step": 1501
    },
    {
      "epoch": 0.5495792169776802,
      "grad_norm": 7.784295082092285,
      "learning_rate": 3.659527326440177e-05,
      "loss": 0.2574,
      "step": 1502
    },
    {
      "epoch": 0.5499451152579583,
      "grad_norm": 0.35504740476608276,
      "learning_rate": 3.658604135893648e-05,
      "loss": 0.0037,
      "step": 1503
    },
    {
      "epoch": 0.5503110135382364,
      "grad_norm": 11.637709617614746,
      "learning_rate": 3.65768094534712e-05,
      "loss": 0.2775,
      "step": 1504
    },
    {
      "epoch": 0.5506769118185144,
      "grad_norm": 5.025627613067627,
      "learning_rate": 3.656757754800591e-05,
      "loss": 0.2737,
      "step": 1505
    },
    {
      "epoch": 0.5510428100987925,
      "grad_norm": 6.47990083694458,
      "learning_rate": 3.655834564254062e-05,
      "loss": 0.2347,
      "step": 1506
    },
    {
      "epoch": 0.5514087083790706,
      "grad_norm": 5.112481594085693,
      "learning_rate": 3.654911373707533e-05,
      "loss": 0.2226,
      "step": 1507
    },
    {
      "epoch": 0.5517746066593487,
      "grad_norm": 3.856335163116455,
      "learning_rate": 3.653988183161004e-05,
      "loss": 0.287,
      "step": 1508
    },
    {
      "epoch": 0.5521405049396267,
      "grad_norm": 12.30361270904541,
      "learning_rate": 3.6530649926144753e-05,
      "loss": 0.4081,
      "step": 1509
    },
    {
      "epoch": 0.5525064032199049,
      "grad_norm": 0.04710201919078827,
      "learning_rate": 3.652141802067947e-05,
      "loss": 0.0009,
      "step": 1510
    },
    {
      "epoch": 0.552872301500183,
      "grad_norm": 5.909361839294434,
      "learning_rate": 3.651218611521418e-05,
      "loss": 0.0323,
      "step": 1511
    },
    {
      "epoch": 0.5532381997804611,
      "grad_norm": 0.12832415103912354,
      "learning_rate": 3.65029542097489e-05,
      "loss": 0.0026,
      "step": 1512
    },
    {
      "epoch": 0.5536040980607391,
      "grad_norm": 0.10952223837375641,
      "learning_rate": 3.649372230428361e-05,
      "loss": 0.0032,
      "step": 1513
    },
    {
      "epoch": 0.5539699963410172,
      "grad_norm": 0.06249422952532768,
      "learning_rate": 3.648449039881831e-05,
      "loss": 0.0015,
      "step": 1514
    },
    {
      "epoch": 0.5543358946212953,
      "grad_norm": 10.455747604370117,
      "learning_rate": 3.647525849335303e-05,
      "loss": 0.1349,
      "step": 1515
    },
    {
      "epoch": 0.5547017929015734,
      "grad_norm": 0.1010805144906044,
      "learning_rate": 3.646602658788774e-05,
      "loss": 0.0024,
      "step": 1516
    },
    {
      "epoch": 0.5550676911818514,
      "grad_norm": 2.265300750732422,
      "learning_rate": 3.645679468242245e-05,
      "loss": 0.4671,
      "step": 1517
    },
    {
      "epoch": 0.5554335894621295,
      "grad_norm": 3.768357753753662,
      "learning_rate": 3.644756277695717e-05,
      "loss": 0.3258,
      "step": 1518
    },
    {
      "epoch": 0.5557994877424076,
      "grad_norm": 2.168823480606079,
      "learning_rate": 3.643833087149188e-05,
      "loss": 0.0215,
      "step": 1519
    },
    {
      "epoch": 0.5561653860226857,
      "grad_norm": 0.3727436661720276,
      "learning_rate": 3.642909896602659e-05,
      "loss": 0.0107,
      "step": 1520
    },
    {
      "epoch": 0.5565312843029637,
      "grad_norm": 15.068596839904785,
      "learning_rate": 3.64198670605613e-05,
      "loss": 0.2103,
      "step": 1521
    },
    {
      "epoch": 0.5568971825832418,
      "grad_norm": 4.677650451660156,
      "learning_rate": 3.641063515509601e-05,
      "loss": 0.036,
      "step": 1522
    },
    {
      "epoch": 0.5572630808635199,
      "grad_norm": 0.22105756402015686,
      "learning_rate": 3.640140324963073e-05,
      "loss": 0.0077,
      "step": 1523
    },
    {
      "epoch": 0.557628979143798,
      "grad_norm": 0.24196060001850128,
      "learning_rate": 3.639217134416544e-05,
      "loss": 0.0095,
      "step": 1524
    },
    {
      "epoch": 0.557994877424076,
      "grad_norm": 0.2568657398223877,
      "learning_rate": 3.638293943870015e-05,
      "loss": 0.0113,
      "step": 1525
    },
    {
      "epoch": 0.5583607757043542,
      "grad_norm": 0.6022541522979736,
      "learning_rate": 3.637370753323486e-05,
      "loss": 0.0242,
      "step": 1526
    },
    {
      "epoch": 0.5587266739846323,
      "grad_norm": 0.46020397543907166,
      "learning_rate": 3.636447562776957e-05,
      "loss": 0.0128,
      "step": 1527
    },
    {
      "epoch": 0.5590925722649104,
      "grad_norm": 0.3349471092224121,
      "learning_rate": 3.635524372230428e-05,
      "loss": 0.0091,
      "step": 1528
    },
    {
      "epoch": 0.5594584705451885,
      "grad_norm": 0.42725670337677,
      "learning_rate": 3.6346011816839e-05,
      "loss": 0.0141,
      "step": 1529
    },
    {
      "epoch": 0.5598243688254665,
      "grad_norm": 3.1901113986968994,
      "learning_rate": 3.633677991137371e-05,
      "loss": 0.2505,
      "step": 1530
    },
    {
      "epoch": 0.5601902671057446,
      "grad_norm": 7.261070251464844,
      "learning_rate": 3.6327548005908426e-05,
      "loss": 0.0661,
      "step": 1531
    },
    {
      "epoch": 0.5605561653860227,
      "grad_norm": 0.2161572426557541,
      "learning_rate": 3.631831610044314e-05,
      "loss": 0.0058,
      "step": 1532
    },
    {
      "epoch": 0.5609220636663008,
      "grad_norm": 0.36283034086227417,
      "learning_rate": 3.630908419497784e-05,
      "loss": 0.0106,
      "step": 1533
    },
    {
      "epoch": 0.5612879619465788,
      "grad_norm": 0.10130497068166733,
      "learning_rate": 3.629985228951256e-05,
      "loss": 0.0032,
      "step": 1534
    },
    {
      "epoch": 0.5616538602268569,
      "grad_norm": 0.4873511791229248,
      "learning_rate": 3.629062038404727e-05,
      "loss": 0.0133,
      "step": 1535
    },
    {
      "epoch": 0.562019758507135,
      "grad_norm": 0.14861899614334106,
      "learning_rate": 3.628138847858198e-05,
      "loss": 0.0057,
      "step": 1536
    },
    {
      "epoch": 0.5623856567874131,
      "grad_norm": 0.11051257699728012,
      "learning_rate": 3.6272156573116696e-05,
      "loss": 0.0033,
      "step": 1537
    },
    {
      "epoch": 0.5627515550676911,
      "grad_norm": 0.15536223351955414,
      "learning_rate": 3.626292466765141e-05,
      "loss": 0.0039,
      "step": 1538
    },
    {
      "epoch": 0.5631174533479693,
      "grad_norm": 0.03673846647143364,
      "learning_rate": 3.625369276218612e-05,
      "loss": 0.0009,
      "step": 1539
    },
    {
      "epoch": 0.5634833516282474,
      "grad_norm": 0.10334387421607971,
      "learning_rate": 3.624446085672083e-05,
      "loss": 0.003,
      "step": 1540
    },
    {
      "epoch": 0.5638492499085255,
      "grad_norm": 2.2085139751434326,
      "learning_rate": 3.623522895125554e-05,
      "loss": 0.3057,
      "step": 1541
    },
    {
      "epoch": 0.5642151481888035,
      "grad_norm": 0.6572049260139465,
      "learning_rate": 3.6225997045790256e-05,
      "loss": 0.0061,
      "step": 1542
    },
    {
      "epoch": 0.5645810464690816,
      "grad_norm": 2.2034878730773926,
      "learning_rate": 3.6216765140324966e-05,
      "loss": 0.3305,
      "step": 1543
    },
    {
      "epoch": 0.5649469447493597,
      "grad_norm": 0.25187617540359497,
      "learning_rate": 3.620753323485968e-05,
      "loss": 0.0064,
      "step": 1544
    },
    {
      "epoch": 0.5653128430296378,
      "grad_norm": 0.1565401256084442,
      "learning_rate": 3.619830132939439e-05,
      "loss": 0.0032,
      "step": 1545
    },
    {
      "epoch": 0.5656787413099158,
      "grad_norm": 7.665265083312988,
      "learning_rate": 3.61890694239291e-05,
      "loss": 0.4541,
      "step": 1546
    },
    {
      "epoch": 0.5660446395901939,
      "grad_norm": 0.042922090739011765,
      "learning_rate": 3.617983751846381e-05,
      "loss": 0.0017,
      "step": 1547
    },
    {
      "epoch": 0.566410537870472,
      "grad_norm": 0.05249130725860596,
      "learning_rate": 3.6170605612998526e-05,
      "loss": 0.0021,
      "step": 1548
    },
    {
      "epoch": 0.5667764361507501,
      "grad_norm": 5.880438327789307,
      "learning_rate": 3.6161373707533236e-05,
      "loss": 0.3801,
      "step": 1549
    },
    {
      "epoch": 0.5671423344310281,
      "grad_norm": 4.673778533935547,
      "learning_rate": 3.615214180206795e-05,
      "loss": 0.2536,
      "step": 1550
    },
    {
      "epoch": 0.5675082327113062,
      "grad_norm": 0.14447808265686035,
      "learning_rate": 3.6142909896602664e-05,
      "loss": 0.0045,
      "step": 1551
    },
    {
      "epoch": 0.5678741309915843,
      "grad_norm": 5.4394001960754395,
      "learning_rate": 3.613367799113737e-05,
      "loss": 0.4998,
      "step": 1552
    },
    {
      "epoch": 0.5682400292718625,
      "grad_norm": 1.5891755819320679,
      "learning_rate": 3.6124446085672085e-05,
      "loss": 0.0117,
      "step": 1553
    },
    {
      "epoch": 0.5686059275521405,
      "grad_norm": 0.18298755586147308,
      "learning_rate": 3.6115214180206796e-05,
      "loss": 0.0055,
      "step": 1554
    },
    {
      "epoch": 0.5689718258324186,
      "grad_norm": 11.477229118347168,
      "learning_rate": 3.6105982274741506e-05,
      "loss": 0.1517,
      "step": 1555
    },
    {
      "epoch": 0.5693377241126967,
      "grad_norm": 2.648244619369507,
      "learning_rate": 3.6096750369276224e-05,
      "loss": 0.0191,
      "step": 1556
    },
    {
      "epoch": 0.5697036223929748,
      "grad_norm": 0.2772919237613678,
      "learning_rate": 3.6087518463810934e-05,
      "loss": 0.0072,
      "step": 1557
    },
    {
      "epoch": 0.5700695206732528,
      "grad_norm": 5.00523042678833,
      "learning_rate": 3.6078286558345645e-05,
      "loss": 0.0513,
      "step": 1558
    },
    {
      "epoch": 0.5704354189535309,
      "grad_norm": 0.09256617724895477,
      "learning_rate": 3.6069054652880355e-05,
      "loss": 0.0043,
      "step": 1559
    },
    {
      "epoch": 0.570801317233809,
      "grad_norm": 0.12920765578746796,
      "learning_rate": 3.6059822747415066e-05,
      "loss": 0.0065,
      "step": 1560
    },
    {
      "epoch": 0.5711672155140871,
      "grad_norm": 0.34720900654792786,
      "learning_rate": 3.605059084194978e-05,
      "loss": 0.0114,
      "step": 1561
    },
    {
      "epoch": 0.5715331137943652,
      "grad_norm": 4.683207035064697,
      "learning_rate": 3.6041358936484494e-05,
      "loss": 0.1704,
      "step": 1562
    },
    {
      "epoch": 0.5718990120746432,
      "grad_norm": 0.20044700801372528,
      "learning_rate": 3.6032127031019204e-05,
      "loss": 0.0079,
      "step": 1563
    },
    {
      "epoch": 0.5722649103549213,
      "grad_norm": 0.08738037198781967,
      "learning_rate": 3.6022895125553915e-05,
      "loss": 0.004,
      "step": 1564
    },
    {
      "epoch": 0.5726308086351994,
      "grad_norm": 0.13110361993312836,
      "learning_rate": 3.6013663220088626e-05,
      "loss": 0.0057,
      "step": 1565
    },
    {
      "epoch": 0.5729967069154775,
      "grad_norm": 0.29846999049186707,
      "learning_rate": 3.6004431314623336e-05,
      "loss": 0.0123,
      "step": 1566
    },
    {
      "epoch": 0.5733626051957555,
      "grad_norm": 0.07365771383047104,
      "learning_rate": 3.599519940915805e-05,
      "loss": 0.0037,
      "step": 1567
    },
    {
      "epoch": 0.5737285034760337,
      "grad_norm": 0.06067100539803505,
      "learning_rate": 3.5985967503692764e-05,
      "loss": 0.0024,
      "step": 1568
    },
    {
      "epoch": 0.5740944017563118,
      "grad_norm": 0.47825363278388977,
      "learning_rate": 3.5976735598227474e-05,
      "loss": 0.007,
      "step": 1569
    },
    {
      "epoch": 0.5744603000365899,
      "grad_norm": 0.1319751888513565,
      "learning_rate": 3.596750369276219e-05,
      "loss": 0.0048,
      "step": 1570
    },
    {
      "epoch": 0.5748261983168679,
      "grad_norm": 0.22319431602954865,
      "learning_rate": 3.5958271787296896e-05,
      "loss": 0.0048,
      "step": 1571
    },
    {
      "epoch": 0.575192096597146,
      "grad_norm": 0.11146293580532074,
      "learning_rate": 3.594903988183161e-05,
      "loss": 0.0028,
      "step": 1572
    },
    {
      "epoch": 0.5755579948774241,
      "grad_norm": 0.05511124059557915,
      "learning_rate": 3.5939807976366323e-05,
      "loss": 0.0014,
      "step": 1573
    },
    {
      "epoch": 0.5759238931577022,
      "grad_norm": 0.311850368976593,
      "learning_rate": 3.5930576070901034e-05,
      "loss": 0.0051,
      "step": 1574
    },
    {
      "epoch": 0.5762897914379802,
      "grad_norm": 0.015732966363430023,
      "learning_rate": 3.592134416543575e-05,
      "loss": 0.0007,
      "step": 1575
    },
    {
      "epoch": 0.5766556897182583,
      "grad_norm": 4.517889022827148,
      "learning_rate": 3.591211225997046e-05,
      "loss": 0.2872,
      "step": 1576
    },
    {
      "epoch": 0.5770215879985364,
      "grad_norm": 2.122664451599121,
      "learning_rate": 3.590288035450517e-05,
      "loss": 0.2093,
      "step": 1577
    },
    {
      "epoch": 0.5773874862788145,
      "grad_norm": 0.028291162103414536,
      "learning_rate": 3.589364844903988e-05,
      "loss": 0.0008,
      "step": 1578
    },
    {
      "epoch": 0.5777533845590925,
      "grad_norm": 0.15396198630332947,
      "learning_rate": 3.5884416543574594e-05,
      "loss": 0.0068,
      "step": 1579
    },
    {
      "epoch": 0.5781192828393706,
      "grad_norm": 0.015935799106955528,
      "learning_rate": 3.5875184638109304e-05,
      "loss": 0.0008,
      "step": 1580
    },
    {
      "epoch": 0.5784851811196488,
      "grad_norm": 16.218400955200195,
      "learning_rate": 3.586595273264402e-05,
      "loss": 0.0875,
      "step": 1581
    },
    {
      "epoch": 0.5788510793999269,
      "grad_norm": 0.016175083816051483,
      "learning_rate": 3.585672082717873e-05,
      "loss": 0.0007,
      "step": 1582
    },
    {
      "epoch": 0.5792169776802049,
      "grad_norm": 0.08691269159317017,
      "learning_rate": 3.584748892171344e-05,
      "loss": 0.0027,
      "step": 1583
    },
    {
      "epoch": 0.579582875960483,
      "grad_norm": 0.02231532149016857,
      "learning_rate": 3.583825701624815e-05,
      "loss": 0.001,
      "step": 1584
    },
    {
      "epoch": 0.5799487742407611,
      "grad_norm": 0.10394210368394852,
      "learning_rate": 3.5829025110782864e-05,
      "loss": 0.0025,
      "step": 1585
    },
    {
      "epoch": 0.5803146725210392,
      "grad_norm": 1.4002411365509033,
      "learning_rate": 3.581979320531758e-05,
      "loss": 0.0047,
      "step": 1586
    },
    {
      "epoch": 0.5806805708013172,
      "grad_norm": 1.0296266078948975,
      "learning_rate": 3.581056129985229e-05,
      "loss": 0.0065,
      "step": 1587
    },
    {
      "epoch": 0.5810464690815953,
      "grad_norm": 27.868446350097656,
      "learning_rate": 3.5801329394387e-05,
      "loss": 0.093,
      "step": 1588
    },
    {
      "epoch": 0.5814123673618734,
      "grad_norm": 0.5047295689582825,
      "learning_rate": 3.579209748892172e-05,
      "loss": 0.0068,
      "step": 1589
    },
    {
      "epoch": 0.5817782656421515,
      "grad_norm": 0.18923740088939667,
      "learning_rate": 3.578286558345642e-05,
      "loss": 0.0049,
      "step": 1590
    },
    {
      "epoch": 0.5821441639224295,
      "grad_norm": 21.018875122070312,
      "learning_rate": 3.577363367799114e-05,
      "loss": 0.3034,
      "step": 1591
    },
    {
      "epoch": 0.5825100622027076,
      "grad_norm": 0.19440962374210358,
      "learning_rate": 3.576440177252585e-05,
      "loss": 0.0054,
      "step": 1592
    },
    {
      "epoch": 0.5828759604829857,
      "grad_norm": 0.1839352548122406,
      "learning_rate": 3.575516986706056e-05,
      "loss": 0.0044,
      "step": 1593
    },
    {
      "epoch": 0.5832418587632638,
      "grad_norm": 0.1376362442970276,
      "learning_rate": 3.574593796159528e-05,
      "loss": 0.0026,
      "step": 1594
    },
    {
      "epoch": 0.5836077570435418,
      "grad_norm": 7.213294506072998,
      "learning_rate": 3.573670605612999e-05,
      "loss": 0.2746,
      "step": 1595
    },
    {
      "epoch": 0.58397365532382,
      "grad_norm": 8.992840766906738,
      "learning_rate": 3.57274741506647e-05,
      "loss": 0.2455,
      "step": 1596
    },
    {
      "epoch": 0.5843395536040981,
      "grad_norm": 2.7276628017425537,
      "learning_rate": 3.571824224519941e-05,
      "loss": 0.0092,
      "step": 1597
    },
    {
      "epoch": 0.5847054518843762,
      "grad_norm": 0.7506815791130066,
      "learning_rate": 3.570901033973412e-05,
      "loss": 0.0061,
      "step": 1598
    },
    {
      "epoch": 0.5850713501646543,
      "grad_norm": 0.2168024778366089,
      "learning_rate": 3.569977843426883e-05,
      "loss": 0.0045,
      "step": 1599
    },
    {
      "epoch": 0.5854372484449323,
      "grad_norm": 16.04590606689453,
      "learning_rate": 3.569054652880355e-05,
      "loss": 0.3935,
      "step": 1600
    },
    {
      "epoch": 0.5858031467252104,
      "grad_norm": 0.014774763025343418,
      "learning_rate": 3.568131462333826e-05,
      "loss": 0.0007,
      "step": 1601
    },
    {
      "epoch": 0.5861690450054885,
      "grad_norm": 0.011461077257990837,
      "learning_rate": 3.567208271787297e-05,
      "loss": 0.0005,
      "step": 1602
    },
    {
      "epoch": 0.5865349432857666,
      "grad_norm": 0.02863563597202301,
      "learning_rate": 3.566285081240768e-05,
      "loss": 0.0006,
      "step": 1603
    },
    {
      "epoch": 0.5869008415660446,
      "grad_norm": 19.030044555664062,
      "learning_rate": 3.565361890694239e-05,
      "loss": 0.1175,
      "step": 1604
    },
    {
      "epoch": 0.5872667398463227,
      "grad_norm": 4.4480485916137695,
      "learning_rate": 3.564438700147711e-05,
      "loss": 0.3192,
      "step": 1605
    },
    {
      "epoch": 0.5876326381266008,
      "grad_norm": 8.133711814880371,
      "learning_rate": 3.563515509601182e-05,
      "loss": 0.1986,
      "step": 1606
    },
    {
      "epoch": 0.5879985364068789,
      "grad_norm": 0.0374026894569397,
      "learning_rate": 3.562592319054653e-05,
      "loss": 0.0008,
      "step": 1607
    },
    {
      "epoch": 0.5883644346871569,
      "grad_norm": 0.7735370993614197,
      "learning_rate": 3.561669128508125e-05,
      "loss": 0.0066,
      "step": 1608
    },
    {
      "epoch": 0.588730332967435,
      "grad_norm": 2.4382166862487793,
      "learning_rate": 3.560745937961595e-05,
      "loss": 0.244,
      "step": 1609
    },
    {
      "epoch": 0.5890962312477132,
      "grad_norm": 0.041882310062646866,
      "learning_rate": 3.559822747415066e-05,
      "loss": 0.0009,
      "step": 1610
    },
    {
      "epoch": 0.5894621295279913,
      "grad_norm": 0.102878637611866,
      "learning_rate": 3.558899556868538e-05,
      "loss": 0.0027,
      "step": 1611
    },
    {
      "epoch": 0.5898280278082693,
      "grad_norm": 0.060838133096694946,
      "learning_rate": 3.557976366322009e-05,
      "loss": 0.0017,
      "step": 1612
    },
    {
      "epoch": 0.5901939260885474,
      "grad_norm": 4.598404407501221,
      "learning_rate": 3.5570531757754806e-05,
      "loss": 0.3611,
      "step": 1613
    },
    {
      "epoch": 0.5905598243688255,
      "grad_norm": 0.2286708652973175,
      "learning_rate": 3.556129985228952e-05,
      "loss": 0.0058,
      "step": 1614
    },
    {
      "epoch": 0.5909257226491036,
      "grad_norm": 1.647876501083374,
      "learning_rate": 3.555206794682423e-05,
      "loss": 0.0092,
      "step": 1615
    },
    {
      "epoch": 0.5912916209293816,
      "grad_norm": 8.39512825012207,
      "learning_rate": 3.554283604135894e-05,
      "loss": 0.0483,
      "step": 1616
    },
    {
      "epoch": 0.5916575192096597,
      "grad_norm": 13.840848922729492,
      "learning_rate": 3.553360413589365e-05,
      "loss": 0.4018,
      "step": 1617
    },
    {
      "epoch": 0.5920234174899378,
      "grad_norm": 0.10516417771577835,
      "learning_rate": 3.552437223042836e-05,
      "loss": 0.002,
      "step": 1618
    },
    {
      "epoch": 0.5923893157702159,
      "grad_norm": 0.013969548046588898,
      "learning_rate": 3.5515140324963076e-05,
      "loss": 0.0006,
      "step": 1619
    },
    {
      "epoch": 0.5927552140504939,
      "grad_norm": 0.0573345385491848,
      "learning_rate": 3.550590841949779e-05,
      "loss": 0.0015,
      "step": 1620
    },
    {
      "epoch": 0.593121112330772,
      "grad_norm": 15.786429405212402,
      "learning_rate": 3.54966765140325e-05,
      "loss": 0.2721,
      "step": 1621
    },
    {
      "epoch": 0.5934870106110501,
      "grad_norm": 0.17725870013237,
      "learning_rate": 3.548744460856721e-05,
      "loss": 0.0049,
      "step": 1622
    },
    {
      "epoch": 0.5938529088913282,
      "grad_norm": 0.2492249757051468,
      "learning_rate": 3.547821270310192e-05,
      "loss": 0.0072,
      "step": 1623
    },
    {
      "epoch": 0.5942188071716062,
      "grad_norm": 0.01002165675163269,
      "learning_rate": 3.5468980797636636e-05,
      "loss": 0.0005,
      "step": 1624
    },
    {
      "epoch": 0.5945847054518844,
      "grad_norm": 13.894981384277344,
      "learning_rate": 3.5459748892171347e-05,
      "loss": 0.0752,
      "step": 1625
    },
    {
      "epoch": 0.5949506037321625,
      "grad_norm": 2.692758321762085,
      "learning_rate": 3.545051698670606e-05,
      "loss": 0.0259,
      "step": 1626
    },
    {
      "epoch": 0.5953165020124406,
      "grad_norm": 0.06644020974636078,
      "learning_rate": 3.5441285081240774e-05,
      "loss": 0.0021,
      "step": 1627
    },
    {
      "epoch": 0.5956824002927186,
      "grad_norm": 4.4485063552856445,
      "learning_rate": 3.543205317577548e-05,
      "loss": 0.3255,
      "step": 1628
    },
    {
      "epoch": 0.5960482985729967,
      "grad_norm": 0.2506342828273773,
      "learning_rate": 3.542282127031019e-05,
      "loss": 0.0075,
      "step": 1629
    },
    {
      "epoch": 0.5964141968532748,
      "grad_norm": 9.614272117614746,
      "learning_rate": 3.5413589364844906e-05,
      "loss": 0.0529,
      "step": 1630
    },
    {
      "epoch": 0.5967800951335529,
      "grad_norm": 0.1616407185792923,
      "learning_rate": 3.5404357459379617e-05,
      "loss": 0.0017,
      "step": 1631
    },
    {
      "epoch": 0.5971459934138309,
      "grad_norm": 0.18763920664787292,
      "learning_rate": 3.5395125553914334e-05,
      "loss": 0.0037,
      "step": 1632
    },
    {
      "epoch": 0.597511891694109,
      "grad_norm": 0.11606403440237045,
      "learning_rate": 3.5385893648449044e-05,
      "loss": 0.0038,
      "step": 1633
    },
    {
      "epoch": 0.5978777899743871,
      "grad_norm": 0.16100211441516876,
      "learning_rate": 3.5376661742983755e-05,
      "loss": 0.0036,
      "step": 1634
    },
    {
      "epoch": 0.5982436882546652,
      "grad_norm": 0.020122099667787552,
      "learning_rate": 3.5367429837518466e-05,
      "loss": 0.0007,
      "step": 1635
    },
    {
      "epoch": 0.5986095865349433,
      "grad_norm": 0.13860857486724854,
      "learning_rate": 3.5358197932053176e-05,
      "loss": 0.002,
      "step": 1636
    },
    {
      "epoch": 0.5989754848152213,
      "grad_norm": 0.08262746781110764,
      "learning_rate": 3.534896602658789e-05,
      "loss": 0.0012,
      "step": 1637
    },
    {
      "epoch": 0.5993413830954994,
      "grad_norm": 0.04641322046518326,
      "learning_rate": 3.5339734121122604e-05,
      "loss": 0.0011,
      "step": 1638
    },
    {
      "epoch": 0.5997072813757776,
      "grad_norm": 0.025550957769155502,
      "learning_rate": 3.5330502215657315e-05,
      "loss": 0.0009,
      "step": 1639
    },
    {
      "epoch": 0.6000731796560557,
      "grad_norm": 17.21181297302246,
      "learning_rate": 3.5321270310192025e-05,
      "loss": 0.2375,
      "step": 1640
    },
    {
      "epoch": 0.6004390779363337,
      "grad_norm": 0.09702955186367035,
      "learning_rate": 3.5312038404726736e-05,
      "loss": 0.0016,
      "step": 1641
    },
    {
      "epoch": 0.6008049762166118,
      "grad_norm": 0.01651882566511631,
      "learning_rate": 3.5302806499261446e-05,
      "loss": 0.0005,
      "step": 1642
    },
    {
      "epoch": 0.6011708744968899,
      "grad_norm": 6.848742485046387,
      "learning_rate": 3.5293574593796164e-05,
      "loss": 0.128,
      "step": 1643
    },
    {
      "epoch": 0.601536772777168,
      "grad_norm": 4.541436195373535,
      "learning_rate": 3.5284342688330874e-05,
      "loss": 0.0346,
      "step": 1644
    },
    {
      "epoch": 0.601902671057446,
      "grad_norm": 0.022551411762833595,
      "learning_rate": 3.5275110782865585e-05,
      "loss": 0.0005,
      "step": 1645
    },
    {
      "epoch": 0.6022685693377241,
      "grad_norm": 0.040925562381744385,
      "learning_rate": 3.52658788774003e-05,
      "loss": 0.0006,
      "step": 1646
    },
    {
      "epoch": 0.6026344676180022,
      "grad_norm": 7.117559432983398,
      "learning_rate": 3.5256646971935006e-05,
      "loss": 0.2204,
      "step": 1647
    },
    {
      "epoch": 0.6030003658982803,
      "grad_norm": 0.358246773481369,
      "learning_rate": 3.5247415066469716e-05,
      "loss": 0.0064,
      "step": 1648
    },
    {
      "epoch": 0.6033662641785583,
      "grad_norm": 7.544600963592529,
      "learning_rate": 3.5238183161004434e-05,
      "loss": 0.1134,
      "step": 1649
    },
    {
      "epoch": 0.6037321624588364,
      "grad_norm": 0.29983657598495483,
      "learning_rate": 3.5228951255539144e-05,
      "loss": 0.002,
      "step": 1650
    },
    {
      "epoch": 0.6040980607391145,
      "grad_norm": 5.825004577636719,
      "learning_rate": 3.5219719350073855e-05,
      "loss": 0.0189,
      "step": 1651
    },
    {
      "epoch": 0.6044639590193926,
      "grad_norm": 19.720367431640625,
      "learning_rate": 3.521048744460857e-05,
      "loss": 0.1358,
      "step": 1652
    },
    {
      "epoch": 0.6048298572996706,
      "grad_norm": 0.6108738780021667,
      "learning_rate": 3.520125553914328e-05,
      "loss": 0.006,
      "step": 1653
    },
    {
      "epoch": 0.6051957555799488,
      "grad_norm": 4.12486457824707,
      "learning_rate": 3.519202363367799e-05,
      "loss": 0.0257,
      "step": 1654
    },
    {
      "epoch": 0.6055616538602269,
      "grad_norm": 0.37412169575691223,
      "learning_rate": 3.5182791728212704e-05,
      "loss": 0.0049,
      "step": 1655
    },
    {
      "epoch": 0.605927552140505,
      "grad_norm": 0.561143696308136,
      "learning_rate": 3.5173559822747414e-05,
      "loss": 0.0049,
      "step": 1656
    },
    {
      "epoch": 0.606293450420783,
      "grad_norm": 4.786396503448486,
      "learning_rate": 3.516432791728213e-05,
      "loss": 0.5673,
      "step": 1657
    },
    {
      "epoch": 0.6066593487010611,
      "grad_norm": 3.5947256088256836,
      "learning_rate": 3.515509601181684e-05,
      "loss": 0.4977,
      "step": 1658
    },
    {
      "epoch": 0.6070252469813392,
      "grad_norm": 21.173099517822266,
      "learning_rate": 3.514586410635155e-05,
      "loss": 0.0669,
      "step": 1659
    },
    {
      "epoch": 0.6073911452616173,
      "grad_norm": 0.4468458592891693,
      "learning_rate": 3.513663220088626e-05,
      "loss": 0.0029,
      "step": 1660
    },
    {
      "epoch": 0.6077570435418953,
      "grad_norm": 0.02744411490857601,
      "learning_rate": 3.5127400295420974e-05,
      "loss": 0.001,
      "step": 1661
    },
    {
      "epoch": 0.6081229418221734,
      "grad_norm": 17.882553100585938,
      "learning_rate": 3.511816838995569e-05,
      "loss": 0.6243,
      "step": 1662
    },
    {
      "epoch": 0.6084888401024515,
      "grad_norm": 6.474165439605713,
      "learning_rate": 3.51089364844904e-05,
      "loss": 0.0358,
      "step": 1663
    },
    {
      "epoch": 0.6088547383827296,
      "grad_norm": 0.03518104925751686,
      "learning_rate": 3.509970457902511e-05,
      "loss": 0.0013,
      "step": 1664
    },
    {
      "epoch": 0.6092206366630076,
      "grad_norm": 0.3670008182525635,
      "learning_rate": 3.509047267355983e-05,
      "loss": 0.0056,
      "step": 1665
    },
    {
      "epoch": 0.6095865349432857,
      "grad_norm": 15.849623680114746,
      "learning_rate": 3.508124076809453e-05,
      "loss": 0.0558,
      "step": 1666
    },
    {
      "epoch": 0.6099524332235639,
      "grad_norm": 0.4076971113681793,
      "learning_rate": 3.5072008862629244e-05,
      "loss": 0.0047,
      "step": 1667
    },
    {
      "epoch": 0.610318331503842,
      "grad_norm": 0.06417033076286316,
      "learning_rate": 3.506277695716396e-05,
      "loss": 0.0021,
      "step": 1668
    },
    {
      "epoch": 0.6106842297841201,
      "grad_norm": 2.0178186893463135,
      "learning_rate": 3.505354505169867e-05,
      "loss": 0.5209,
      "step": 1669
    },
    {
      "epoch": 0.6110501280643981,
      "grad_norm": 0.07279079407453537,
      "learning_rate": 3.504431314623338e-05,
      "loss": 0.0025,
      "step": 1670
    },
    {
      "epoch": 0.6114160263446762,
      "grad_norm": 5.333530426025391,
      "learning_rate": 3.50350812407681e-05,
      "loss": 0.2294,
      "step": 1671
    },
    {
      "epoch": 0.6117819246249543,
      "grad_norm": 8.902758598327637,
      "learning_rate": 3.502584933530281e-05,
      "loss": 0.0247,
      "step": 1672
    },
    {
      "epoch": 0.6121478229052324,
      "grad_norm": 3.709810256958008,
      "learning_rate": 3.501661742983752e-05,
      "loss": 0.1085,
      "step": 1673
    },
    {
      "epoch": 0.6125137211855104,
      "grad_norm": 0.30023303627967834,
      "learning_rate": 3.500738552437223e-05,
      "loss": 0.0059,
      "step": 1674
    },
    {
      "epoch": 0.6128796194657885,
      "grad_norm": 0.8619508743286133,
      "learning_rate": 3.499815361890694e-05,
      "loss": 0.0058,
      "step": 1675
    },
    {
      "epoch": 0.6132455177460666,
      "grad_norm": 12.110743522644043,
      "learning_rate": 3.498892171344166e-05,
      "loss": 0.0662,
      "step": 1676
    },
    {
      "epoch": 0.6136114160263447,
      "grad_norm": 3.8003180027008057,
      "learning_rate": 3.497968980797637e-05,
      "loss": 0.0309,
      "step": 1677
    },
    {
      "epoch": 0.6139773143066227,
      "grad_norm": 6.457098484039307,
      "learning_rate": 3.497045790251108e-05,
      "loss": 0.3697,
      "step": 1678
    },
    {
      "epoch": 0.6143432125869008,
      "grad_norm": 12.057703971862793,
      "learning_rate": 3.496122599704579e-05,
      "loss": 0.4418,
      "step": 1679
    },
    {
      "epoch": 0.6147091108671789,
      "grad_norm": 17.62173843383789,
      "learning_rate": 3.49519940915805e-05,
      "loss": 0.2337,
      "step": 1680
    },
    {
      "epoch": 0.615075009147457,
      "grad_norm": 11.02550983428955,
      "learning_rate": 3.494276218611521e-05,
      "loss": 0.3246,
      "step": 1681
    },
    {
      "epoch": 0.615440907427735,
      "grad_norm": 6.039980411529541,
      "learning_rate": 3.493353028064993e-05,
      "loss": 0.2412,
      "step": 1682
    },
    {
      "epoch": 0.6158068057080132,
      "grad_norm": 5.312479019165039,
      "learning_rate": 3.492429837518464e-05,
      "loss": 0.4927,
      "step": 1683
    },
    {
      "epoch": 0.6161727039882913,
      "grad_norm": 8.07598876953125,
      "learning_rate": 3.491506646971936e-05,
      "loss": 0.0758,
      "step": 1684
    },
    {
      "epoch": 0.6165386022685694,
      "grad_norm": 1.9812343120574951,
      "learning_rate": 3.490583456425406e-05,
      "loss": 0.0315,
      "step": 1685
    },
    {
      "epoch": 0.6169045005488474,
      "grad_norm": 0.3632737100124359,
      "learning_rate": 3.489660265878877e-05,
      "loss": 0.0085,
      "step": 1686
    },
    {
      "epoch": 0.6172703988291255,
      "grad_norm": 0.08036711812019348,
      "learning_rate": 3.488737075332349e-05,
      "loss": 0.0033,
      "step": 1687
    },
    {
      "epoch": 0.6176362971094036,
      "grad_norm": 1.036484718322754,
      "learning_rate": 3.48781388478582e-05,
      "loss": 0.0143,
      "step": 1688
    },
    {
      "epoch": 0.6180021953896817,
      "grad_norm": 6.222764015197754,
      "learning_rate": 3.486890694239291e-05,
      "loss": 0.0569,
      "step": 1689
    },
    {
      "epoch": 0.6183680936699597,
      "grad_norm": 0.3536378741264343,
      "learning_rate": 3.485967503692763e-05,
      "loss": 0.0111,
      "step": 1690
    },
    {
      "epoch": 0.6187339919502378,
      "grad_norm": 5.665469646453857,
      "learning_rate": 3.485044313146234e-05,
      "loss": 1.0118,
      "step": 1691
    },
    {
      "epoch": 0.6190998902305159,
      "grad_norm": 6.791797637939453,
      "learning_rate": 3.484121122599705e-05,
      "loss": 0.0964,
      "step": 1692
    },
    {
      "epoch": 0.619465788510794,
      "grad_norm": 0.15811903774738312,
      "learning_rate": 3.483197932053176e-05,
      "loss": 0.0062,
      "step": 1693
    },
    {
      "epoch": 0.619831686791072,
      "grad_norm": 0.3813837170600891,
      "learning_rate": 3.482274741506647e-05,
      "loss": 0.0076,
      "step": 1694
    },
    {
      "epoch": 0.6201975850713501,
      "grad_norm": 0.521360456943512,
      "learning_rate": 3.481351550960119e-05,
      "loss": 0.0129,
      "step": 1695
    },
    {
      "epoch": 0.6205634833516283,
      "grad_norm": 0.0943724736571312,
      "learning_rate": 3.48042836041359e-05,
      "loss": 0.0041,
      "step": 1696
    },
    {
      "epoch": 0.6209293816319064,
      "grad_norm": 0.11658386886119843,
      "learning_rate": 3.479505169867061e-05,
      "loss": 0.0056,
      "step": 1697
    },
    {
      "epoch": 0.6212952799121844,
      "grad_norm": 0.11272852122783661,
      "learning_rate": 3.478581979320532e-05,
      "loss": 0.0056,
      "step": 1698
    },
    {
      "epoch": 0.6216611781924625,
      "grad_norm": 5.274896621704102,
      "learning_rate": 3.477658788774003e-05,
      "loss": 0.1907,
      "step": 1699
    },
    {
      "epoch": 0.6220270764727406,
      "grad_norm": 0.19216595590114594,
      "learning_rate": 3.476735598227474e-05,
      "loss": 0.0053,
      "step": 1700
    },
    {
      "epoch": 0.6223929747530187,
      "grad_norm": 0.05587845668196678,
      "learning_rate": 3.475812407680946e-05,
      "loss": 0.0025,
      "step": 1701
    },
    {
      "epoch": 0.6227588730332967,
      "grad_norm": 0.2651483714580536,
      "learning_rate": 3.474889217134417e-05,
      "loss": 0.0094,
      "step": 1702
    },
    {
      "epoch": 0.6231247713135748,
      "grad_norm": 0.061928849667310715,
      "learning_rate": 3.4739660265878885e-05,
      "loss": 0.0027,
      "step": 1703
    },
    {
      "epoch": 0.6234906695938529,
      "grad_norm": 10.862753868103027,
      "learning_rate": 3.473042836041359e-05,
      "loss": 0.0807,
      "step": 1704
    },
    {
      "epoch": 0.623856567874131,
      "grad_norm": 0.09097117185592651,
      "learning_rate": 3.47211964549483e-05,
      "loss": 0.0034,
      "step": 1705
    },
    {
      "epoch": 0.6242224661544091,
      "grad_norm": 0.04328175261616707,
      "learning_rate": 3.4711964549483016e-05,
      "loss": 0.0019,
      "step": 1706
    },
    {
      "epoch": 0.6245883644346871,
      "grad_norm": 0.030152972787618637,
      "learning_rate": 3.470273264401773e-05,
      "loss": 0.0014,
      "step": 1707
    },
    {
      "epoch": 0.6249542627149652,
      "grad_norm": 0.038234371691942215,
      "learning_rate": 3.469350073855244e-05,
      "loss": 0.0017,
      "step": 1708
    },
    {
      "epoch": 0.6253201609952433,
      "grad_norm": 1.6830037832260132,
      "learning_rate": 3.4684268833087155e-05,
      "loss": 0.3577,
      "step": 1709
    },
    {
      "epoch": 0.6256860592755215,
      "grad_norm": 19.411304473876953,
      "learning_rate": 3.4675036927621865e-05,
      "loss": 0.367,
      "step": 1710
    },
    {
      "epoch": 0.6260519575557995,
      "grad_norm": 17.40985107421875,
      "learning_rate": 3.466580502215657e-05,
      "loss": 0.1445,
      "step": 1711
    },
    {
      "epoch": 0.6264178558360776,
      "grad_norm": 0.06075405701994896,
      "learning_rate": 3.4656573116691286e-05,
      "loss": 0.0025,
      "step": 1712
    },
    {
      "epoch": 0.6267837541163557,
      "grad_norm": 0.04370778426527977,
      "learning_rate": 3.4647341211226e-05,
      "loss": 0.0015,
      "step": 1713
    },
    {
      "epoch": 0.6271496523966338,
      "grad_norm": 0.6328319311141968,
      "learning_rate": 3.4638109305760714e-05,
      "loss": 0.01,
      "step": 1714
    },
    {
      "epoch": 0.6275155506769118,
      "grad_norm": 3.547624111175537,
      "learning_rate": 3.4628877400295425e-05,
      "loss": 0.1629,
      "step": 1715
    },
    {
      "epoch": 0.6278814489571899,
      "grad_norm": 0.09052564203739166,
      "learning_rate": 3.4619645494830135e-05,
      "loss": 0.004,
      "step": 1716
    },
    {
      "epoch": 0.628247347237468,
      "grad_norm": 0.23188455402851105,
      "learning_rate": 3.4610413589364846e-05,
      "loss": 0.0042,
      "step": 1717
    },
    {
      "epoch": 0.6286132455177461,
      "grad_norm": 6.605299949645996,
      "learning_rate": 3.4601181683899556e-05,
      "loss": 0.1737,
      "step": 1718
    },
    {
      "epoch": 0.6289791437980241,
      "grad_norm": 2.8719913959503174,
      "learning_rate": 3.459194977843427e-05,
      "loss": 0.0326,
      "step": 1719
    },
    {
      "epoch": 0.6293450420783022,
      "grad_norm": 0.22679345309734344,
      "learning_rate": 3.4582717872968984e-05,
      "loss": 0.003,
      "step": 1720
    },
    {
      "epoch": 0.6297109403585803,
      "grad_norm": 13.395760536193848,
      "learning_rate": 3.4573485967503695e-05,
      "loss": 0.0651,
      "step": 1721
    },
    {
      "epoch": 0.6300768386388584,
      "grad_norm": 9.333120346069336,
      "learning_rate": 3.4564254062038405e-05,
      "loss": 0.0764,
      "step": 1722
    },
    {
      "epoch": 0.6304427369191364,
      "grad_norm": 0.31455016136169434,
      "learning_rate": 3.4555022156573116e-05,
      "loss": 0.0125,
      "step": 1723
    },
    {
      "epoch": 0.6308086351994145,
      "grad_norm": 6.979921817779541,
      "learning_rate": 3.4545790251107826e-05,
      "loss": 0.2274,
      "step": 1724
    },
    {
      "epoch": 0.6311745334796927,
      "grad_norm": 0.562045693397522,
      "learning_rate": 3.4536558345642544e-05,
      "loss": 0.0085,
      "step": 1725
    },
    {
      "epoch": 0.6315404317599708,
      "grad_norm": 17.943666458129883,
      "learning_rate": 3.4527326440177254e-05,
      "loss": 0.247,
      "step": 1726
    },
    {
      "epoch": 0.6319063300402488,
      "grad_norm": 1.9151774644851685,
      "learning_rate": 3.4518094534711965e-05,
      "loss": 0.3234,
      "step": 1727
    },
    {
      "epoch": 0.6322722283205269,
      "grad_norm": 0.2944060266017914,
      "learning_rate": 3.450886262924668e-05,
      "loss": 0.0044,
      "step": 1728
    },
    {
      "epoch": 0.632638126600805,
      "grad_norm": 0.07447700947523117,
      "learning_rate": 3.449963072378139e-05,
      "loss": 0.0029,
      "step": 1729
    },
    {
      "epoch": 0.6330040248810831,
      "grad_norm": 0.08498266339302063,
      "learning_rate": 3.4490398818316096e-05,
      "loss": 0.0035,
      "step": 1730
    },
    {
      "epoch": 0.6333699231613611,
      "grad_norm": 0.2772301137447357,
      "learning_rate": 3.4481166912850814e-05,
      "loss": 0.0062,
      "step": 1731
    },
    {
      "epoch": 0.6337358214416392,
      "grad_norm": 7.799281597137451,
      "learning_rate": 3.4471935007385524e-05,
      "loss": 0.0517,
      "step": 1732
    },
    {
      "epoch": 0.6341017197219173,
      "grad_norm": 0.038405682891607285,
      "learning_rate": 3.446270310192024e-05,
      "loss": 0.0017,
      "step": 1733
    },
    {
      "epoch": 0.6344676180021954,
      "grad_norm": 0.06110316142439842,
      "learning_rate": 3.445347119645495e-05,
      "loss": 0.0025,
      "step": 1734
    },
    {
      "epoch": 0.6348335162824734,
      "grad_norm": 1.4766452312469482,
      "learning_rate": 3.444423929098966e-05,
      "loss": 0.0094,
      "step": 1735
    },
    {
      "epoch": 0.6351994145627515,
      "grad_norm": 0.32715341448783875,
      "learning_rate": 3.443500738552437e-05,
      "loss": 0.0057,
      "step": 1736
    },
    {
      "epoch": 0.6355653128430296,
      "grad_norm": 6.5042724609375,
      "learning_rate": 3.4425775480059084e-05,
      "loss": 0.1758,
      "step": 1737
    },
    {
      "epoch": 0.6359312111233077,
      "grad_norm": 0.37121185660362244,
      "learning_rate": 3.4416543574593794e-05,
      "loss": 0.0104,
      "step": 1738
    },
    {
      "epoch": 0.6362971094035857,
      "grad_norm": 0.3805115520954132,
      "learning_rate": 3.440731166912851e-05,
      "loss": 0.0049,
      "step": 1739
    },
    {
      "epoch": 0.6366630076838639,
      "grad_norm": 0.09585121273994446,
      "learning_rate": 3.439807976366322e-05,
      "loss": 0.0022,
      "step": 1740
    },
    {
      "epoch": 0.637028905964142,
      "grad_norm": 0.3960326611995697,
      "learning_rate": 3.438884785819793e-05,
      "loss": 0.0102,
      "step": 1741
    },
    {
      "epoch": 0.6373948042444201,
      "grad_norm": 0.03077881783246994,
      "learning_rate": 3.4379615952732643e-05,
      "loss": 0.0014,
      "step": 1742
    },
    {
      "epoch": 0.6377607025246982,
      "grad_norm": 0.3834626078605652,
      "learning_rate": 3.4370384047267354e-05,
      "loss": 0.0051,
      "step": 1743
    },
    {
      "epoch": 0.6381266008049762,
      "grad_norm": 0.048785362392663956,
      "learning_rate": 3.436115214180207e-05,
      "loss": 0.0021,
      "step": 1744
    },
    {
      "epoch": 0.6384924990852543,
      "grad_norm": 0.16431903839111328,
      "learning_rate": 3.435192023633678e-05,
      "loss": 0.0024,
      "step": 1745
    },
    {
      "epoch": 0.6388583973655324,
      "grad_norm": 11.765022277832031,
      "learning_rate": 3.434268833087149e-05,
      "loss": 0.1908,
      "step": 1746
    },
    {
      "epoch": 0.6392242956458105,
      "grad_norm": 0.16348864138126373,
      "learning_rate": 3.433345642540621e-05,
      "loss": 0.0033,
      "step": 1747
    },
    {
      "epoch": 0.6395901939260885,
      "grad_norm": 0.015306392684578896,
      "learning_rate": 3.432422451994092e-05,
      "loss": 0.0007,
      "step": 1748
    },
    {
      "epoch": 0.6399560922063666,
      "grad_norm": 4.852347373962402,
      "learning_rate": 3.4314992614475624e-05,
      "loss": 0.0249,
      "step": 1749
    },
    {
      "epoch": 0.6403219904866447,
      "grad_norm": 0.15562810003757477,
      "learning_rate": 3.430576070901034e-05,
      "loss": 0.0027,
      "step": 1750
    },
    {
      "epoch": 0.6406878887669228,
      "grad_norm": 0.6118643283843994,
      "learning_rate": 3.429652880354505e-05,
      "loss": 0.0056,
      "step": 1751
    },
    {
      "epoch": 0.6410537870472008,
      "grad_norm": 1.3762414455413818,
      "learning_rate": 3.428729689807976e-05,
      "loss": 0.0119,
      "step": 1752
    },
    {
      "epoch": 0.641419685327479,
      "grad_norm": 0.020770283415913582,
      "learning_rate": 3.427806499261448e-05,
      "loss": 0.0009,
      "step": 1753
    },
    {
      "epoch": 0.6417855836077571,
      "grad_norm": 0.20660309493541718,
      "learning_rate": 3.426883308714919e-05,
      "loss": 0.0069,
      "step": 1754
    },
    {
      "epoch": 0.6421514818880352,
      "grad_norm": 0.7010750770568848,
      "learning_rate": 3.42596011816839e-05,
      "loss": 0.0044,
      "step": 1755
    },
    {
      "epoch": 0.6425173801683132,
      "grad_norm": 9.604109764099121,
      "learning_rate": 3.425036927621861e-05,
      "loss": 0.6335,
      "step": 1756
    },
    {
      "epoch": 0.6428832784485913,
      "grad_norm": 0.035269759595394135,
      "learning_rate": 3.424113737075332e-05,
      "loss": 0.0013,
      "step": 1757
    },
    {
      "epoch": 0.6432491767288694,
      "grad_norm": 14.733721733093262,
      "learning_rate": 3.423190546528804e-05,
      "loss": 0.6484,
      "step": 1758
    },
    {
      "epoch": 0.6436150750091475,
      "grad_norm": 0.2856011390686035,
      "learning_rate": 3.422267355982275e-05,
      "loss": 0.0021,
      "step": 1759
    },
    {
      "epoch": 0.6439809732894255,
      "grad_norm": 0.03441529721021652,
      "learning_rate": 3.421344165435746e-05,
      "loss": 0.001,
      "step": 1760
    },
    {
      "epoch": 0.6443468715697036,
      "grad_norm": 1.0984729528427124,
      "learning_rate": 3.420420974889217e-05,
      "loss": 0.0039,
      "step": 1761
    },
    {
      "epoch": 0.6447127698499817,
      "grad_norm": 0.05392618849873543,
      "learning_rate": 3.419497784342688e-05,
      "loss": 0.0012,
      "step": 1762
    },
    {
      "epoch": 0.6450786681302598,
      "grad_norm": 3.7648003101348877,
      "learning_rate": 3.41857459379616e-05,
      "loss": 0.0154,
      "step": 1763
    },
    {
      "epoch": 0.6454445664105378,
      "grad_norm": 0.017250679433345795,
      "learning_rate": 3.417651403249631e-05,
      "loss": 0.0007,
      "step": 1764
    },
    {
      "epoch": 0.6458104646908159,
      "grad_norm": 0.03448290377855301,
      "learning_rate": 3.416728212703102e-05,
      "loss": 0.0009,
      "step": 1765
    },
    {
      "epoch": 0.646176362971094,
      "grad_norm": 19.706708908081055,
      "learning_rate": 3.415805022156574e-05,
      "loss": 0.1085,
      "step": 1766
    },
    {
      "epoch": 0.6465422612513722,
      "grad_norm": 0.023248782381415367,
      "learning_rate": 3.414881831610045e-05,
      "loss": 0.0009,
      "step": 1767
    },
    {
      "epoch": 0.6469081595316502,
      "grad_norm": 0.02743319608271122,
      "learning_rate": 3.413958641063515e-05,
      "loss": 0.0011,
      "step": 1768
    },
    {
      "epoch": 0.6472740578119283,
      "grad_norm": 0.21334435045719147,
      "learning_rate": 3.413035450516987e-05,
      "loss": 0.0023,
      "step": 1769
    },
    {
      "epoch": 0.6476399560922064,
      "grad_norm": 6.50756311416626,
      "learning_rate": 3.412112259970458e-05,
      "loss": 0.1888,
      "step": 1770
    },
    {
      "epoch": 0.6480058543724845,
      "grad_norm": 1.8221888542175293,
      "learning_rate": 3.411189069423929e-05,
      "loss": 0.4328,
      "step": 1771
    },
    {
      "epoch": 0.6483717526527625,
      "grad_norm": 0.024741996079683304,
      "learning_rate": 3.410265878877401e-05,
      "loss": 0.001,
      "step": 1772
    },
    {
      "epoch": 0.6487376509330406,
      "grad_norm": 14.389887809753418,
      "learning_rate": 3.409342688330872e-05,
      "loss": 0.3641,
      "step": 1773
    },
    {
      "epoch": 0.6491035492133187,
      "grad_norm": 1.7164256572723389,
      "learning_rate": 3.408419497784343e-05,
      "loss": 0.3939,
      "step": 1774
    },
    {
      "epoch": 0.6494694474935968,
      "grad_norm": 0.07257683575153351,
      "learning_rate": 3.407496307237814e-05,
      "loss": 0.002,
      "step": 1775
    },
    {
      "epoch": 0.6498353457738749,
      "grad_norm": 13.503174781799316,
      "learning_rate": 3.406573116691285e-05,
      "loss": 0.7964,
      "step": 1776
    },
    {
      "epoch": 0.6502012440541529,
      "grad_norm": 9.497919082641602,
      "learning_rate": 3.405649926144757e-05,
      "loss": 0.0192,
      "step": 1777
    },
    {
      "epoch": 0.650567142334431,
      "grad_norm": 3.6109678745269775,
      "learning_rate": 3.404726735598228e-05,
      "loss": 0.0152,
      "step": 1778
    },
    {
      "epoch": 0.6509330406147091,
      "grad_norm": 0.0795847475528717,
      "learning_rate": 3.403803545051699e-05,
      "loss": 0.0036,
      "step": 1779
    },
    {
      "epoch": 0.6512989388949872,
      "grad_norm": 0.058518704026937485,
      "learning_rate": 3.40288035450517e-05,
      "loss": 0.0027,
      "step": 1780
    },
    {
      "epoch": 0.6516648371752652,
      "grad_norm": 0.07155395299196243,
      "learning_rate": 3.401957163958641e-05,
      "loss": 0.0031,
      "step": 1781
    },
    {
      "epoch": 0.6520307354555434,
      "grad_norm": 0.1307418793439865,
      "learning_rate": 3.401033973412112e-05,
      "loss": 0.006,
      "step": 1782
    },
    {
      "epoch": 0.6523966337358215,
      "grad_norm": 0.32797276973724365,
      "learning_rate": 3.400110782865584e-05,
      "loss": 0.0072,
      "step": 1783
    },
    {
      "epoch": 0.6527625320160996,
      "grad_norm": 10.739622116088867,
      "learning_rate": 3.399187592319055e-05,
      "loss": 0.245,
      "step": 1784
    },
    {
      "epoch": 0.6531284302963776,
      "grad_norm": 1.5566368103027344,
      "learning_rate": 3.3982644017725265e-05,
      "loss": 0.3259,
      "step": 1785
    },
    {
      "epoch": 0.6534943285766557,
      "grad_norm": 0.12430987507104874,
      "learning_rate": 3.3973412112259975e-05,
      "loss": 0.006,
      "step": 1786
    },
    {
      "epoch": 0.6538602268569338,
      "grad_norm": 8.547719955444336,
      "learning_rate": 3.396418020679468e-05,
      "loss": 0.1308,
      "step": 1787
    },
    {
      "epoch": 0.6542261251372119,
      "grad_norm": 0.3498539626598358,
      "learning_rate": 3.3954948301329396e-05,
      "loss": 0.0136,
      "step": 1788
    },
    {
      "epoch": 0.6545920234174899,
      "grad_norm": 8.66677474975586,
      "learning_rate": 3.394571639586411e-05,
      "loss": 0.5293,
      "step": 1789
    },
    {
      "epoch": 0.654957921697768,
      "grad_norm": 0.08361295610666275,
      "learning_rate": 3.393648449039882e-05,
      "loss": 0.0041,
      "step": 1790
    },
    {
      "epoch": 0.6553238199780461,
      "grad_norm": 3.5422871112823486,
      "learning_rate": 3.3927252584933535e-05,
      "loss": 0.0186,
      "step": 1791
    },
    {
      "epoch": 0.6556897182583242,
      "grad_norm": 14.356572151184082,
      "learning_rate": 3.3918020679468245e-05,
      "loss": 0.2594,
      "step": 1792
    },
    {
      "epoch": 0.6560556165386022,
      "grad_norm": 0.9495532512664795,
      "learning_rate": 3.3908788774002956e-05,
      "loss": 0.0112,
      "step": 1793
    },
    {
      "epoch": 0.6564215148188803,
      "grad_norm": 0.13768404722213745,
      "learning_rate": 3.3899556868537666e-05,
      "loss": 0.0069,
      "step": 1794
    },
    {
      "epoch": 0.6567874130991584,
      "grad_norm": 38.79499435424805,
      "learning_rate": 3.389032496307238e-05,
      "loss": 0.1096,
      "step": 1795
    },
    {
      "epoch": 0.6571533113794366,
      "grad_norm": 0.2672179937362671,
      "learning_rate": 3.3881093057607094e-05,
      "loss": 0.0128,
      "step": 1796
    },
    {
      "epoch": 0.6575192096597146,
      "grad_norm": 0.05661989375948906,
      "learning_rate": 3.3871861152141805e-05,
      "loss": 0.0027,
      "step": 1797
    },
    {
      "epoch": 0.6578851079399927,
      "grad_norm": 9.080422401428223,
      "learning_rate": 3.3862629246676515e-05,
      "loss": 0.3408,
      "step": 1798
    },
    {
      "epoch": 0.6582510062202708,
      "grad_norm": 0.14758937060832977,
      "learning_rate": 3.3853397341211226e-05,
      "loss": 0.0072,
      "step": 1799
    },
    {
      "epoch": 0.6586169045005489,
      "grad_norm": 0.1177501305937767,
      "learning_rate": 3.3844165435745937e-05,
      "loss": 0.0056,
      "step": 1800
    },
    {
      "epoch": 0.6589828027808269,
      "grad_norm": 0.5786716938018799,
      "learning_rate": 3.383493353028065e-05,
      "loss": 0.0092,
      "step": 1801
    },
    {
      "epoch": 0.659348701061105,
      "grad_norm": 0.15868617594242096,
      "learning_rate": 3.3825701624815364e-05,
      "loss": 0.006,
      "step": 1802
    },
    {
      "epoch": 0.6597145993413831,
      "grad_norm": 0.11999347060918808,
      "learning_rate": 3.3816469719350075e-05,
      "loss": 0.0058,
      "step": 1803
    },
    {
      "epoch": 0.6600804976216612,
      "grad_norm": 0.08219894766807556,
      "learning_rate": 3.380723781388479e-05,
      "loss": 0.0039,
      "step": 1804
    },
    {
      "epoch": 0.6604463959019392,
      "grad_norm": 0.08849170804023743,
      "learning_rate": 3.37980059084195e-05,
      "loss": 0.0042,
      "step": 1805
    },
    {
      "epoch": 0.6608122941822173,
      "grad_norm": 0.060526296496391296,
      "learning_rate": 3.378877400295421e-05,
      "loss": 0.0028,
      "step": 1806
    },
    {
      "epoch": 0.6611781924624954,
      "grad_norm": 0.06284617632627487,
      "learning_rate": 3.3779542097488924e-05,
      "loss": 0.0028,
      "step": 1807
    },
    {
      "epoch": 0.6615440907427735,
      "grad_norm": 15.257250785827637,
      "learning_rate": 3.3770310192023635e-05,
      "loss": 0.2286,
      "step": 1808
    },
    {
      "epoch": 0.6619099890230515,
      "grad_norm": 0.050841622054576874,
      "learning_rate": 3.3761078286558345e-05,
      "loss": 0.0024,
      "step": 1809
    },
    {
      "epoch": 0.6622758873033296,
      "grad_norm": 0.05846324563026428,
      "learning_rate": 3.375184638109306e-05,
      "loss": 0.0027,
      "step": 1810
    },
    {
      "epoch": 0.6626417855836078,
      "grad_norm": 0.060851432383060455,
      "learning_rate": 3.374261447562777e-05,
      "loss": 0.0025,
      "step": 1811
    },
    {
      "epoch": 0.6630076838638859,
      "grad_norm": 0.05348796024918556,
      "learning_rate": 3.3733382570162483e-05,
      "loss": 0.0024,
      "step": 1812
    },
    {
      "epoch": 0.663373582144164,
      "grad_norm": 0.061012301594018936,
      "learning_rate": 3.3724150664697194e-05,
      "loss": 0.0028,
      "step": 1813
    },
    {
      "epoch": 0.663739480424442,
      "grad_norm": 0.033401742577552795,
      "learning_rate": 3.3714918759231905e-05,
      "loss": 0.0015,
      "step": 1814
    },
    {
      "epoch": 0.6641053787047201,
      "grad_norm": 0.5958781242370605,
      "learning_rate": 3.370568685376662e-05,
      "loss": 0.008,
      "step": 1815
    },
    {
      "epoch": 0.6644712769849982,
      "grad_norm": 0.03343209996819496,
      "learning_rate": 3.369645494830133e-05,
      "loss": 0.0013,
      "step": 1816
    },
    {
      "epoch": 0.6648371752652763,
      "grad_norm": 0.04383024573326111,
      "learning_rate": 3.368722304283604e-05,
      "loss": 0.0019,
      "step": 1817
    },
    {
      "epoch": 0.6652030735455543,
      "grad_norm": 11.24096393585205,
      "learning_rate": 3.3677991137370754e-05,
      "loss": 0.0282,
      "step": 1818
    },
    {
      "epoch": 0.6655689718258324,
      "grad_norm": 0.022848132997751236,
      "learning_rate": 3.3668759231905464e-05,
      "loss": 0.001,
      "step": 1819
    },
    {
      "epoch": 0.6659348701061105,
      "grad_norm": 0.04609430953860283,
      "learning_rate": 3.3659527326440175e-05,
      "loss": 0.0017,
      "step": 1820
    },
    {
      "epoch": 0.6663007683863886,
      "grad_norm": 0.026485424488782883,
      "learning_rate": 3.365029542097489e-05,
      "loss": 0.0011,
      "step": 1821
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.027113977819681168,
      "learning_rate": 3.36410635155096e-05,
      "loss": 0.0012,
      "step": 1822
    },
    {
      "epoch": 0.6670325649469447,
      "grad_norm": 1.81159508228302,
      "learning_rate": 3.363183161004431e-05,
      "loss": 0.3672,
      "step": 1823
    },
    {
      "epoch": 0.6673984632272228,
      "grad_norm": 0.06169215589761734,
      "learning_rate": 3.362259970457903e-05,
      "loss": 0.0022,
      "step": 1824
    },
    {
      "epoch": 0.667764361507501,
      "grad_norm": 0.013893542811274529,
      "learning_rate": 3.3613367799113734e-05,
      "loss": 0.0007,
      "step": 1825
    },
    {
      "epoch": 0.668130259787779,
      "grad_norm": 10.944665908813477,
      "learning_rate": 3.360413589364845e-05,
      "loss": 0.1903,
      "step": 1826
    },
    {
      "epoch": 0.6684961580680571,
      "grad_norm": 0.2778090834617615,
      "learning_rate": 3.359490398818316e-05,
      "loss": 0.0033,
      "step": 1827
    },
    {
      "epoch": 0.6688620563483352,
      "grad_norm": 0.46647149324417114,
      "learning_rate": 3.358567208271787e-05,
      "loss": 0.0041,
      "step": 1828
    },
    {
      "epoch": 0.6692279546286133,
      "grad_norm": 0.5924484729766846,
      "learning_rate": 3.357644017725259e-05,
      "loss": 0.0034,
      "step": 1829
    },
    {
      "epoch": 0.6695938529088913,
      "grad_norm": 0.019465163350105286,
      "learning_rate": 3.35672082717873e-05,
      "loss": 0.0009,
      "step": 1830
    },
    {
      "epoch": 0.6699597511891694,
      "grad_norm": 0.02097506821155548,
      "learning_rate": 3.355797636632201e-05,
      "loss": 0.001,
      "step": 1831
    },
    {
      "epoch": 0.6703256494694475,
      "grad_norm": 1.1733852624893188,
      "learning_rate": 3.354874446085672e-05,
      "loss": 0.0125,
      "step": 1832
    },
    {
      "epoch": 0.6706915477497256,
      "grad_norm": 6.161455154418945,
      "learning_rate": 3.353951255539143e-05,
      "loss": 0.27,
      "step": 1833
    },
    {
      "epoch": 0.6710574460300036,
      "grad_norm": 0.038043420761823654,
      "learning_rate": 3.353028064992615e-05,
      "loss": 0.0016,
      "step": 1834
    },
    {
      "epoch": 0.6714233443102817,
      "grad_norm": 5.697388172149658,
      "learning_rate": 3.352104874446086e-05,
      "loss": 0.2275,
      "step": 1835
    },
    {
      "epoch": 0.6717892425905598,
      "grad_norm": 10.993441581726074,
      "learning_rate": 3.351181683899557e-05,
      "loss": 0.7304,
      "step": 1836
    },
    {
      "epoch": 0.6721551408708379,
      "grad_norm": 0.21685178577899933,
      "learning_rate": 3.350258493353028e-05,
      "loss": 0.0032,
      "step": 1837
    },
    {
      "epoch": 0.6725210391511159,
      "grad_norm": 0.16552934050559998,
      "learning_rate": 3.349335302806499e-05,
      "loss": 0.0035,
      "step": 1838
    },
    {
      "epoch": 0.672886937431394,
      "grad_norm": 0.07791784405708313,
      "learning_rate": 3.34841211225997e-05,
      "loss": 0.0018,
      "step": 1839
    },
    {
      "epoch": 0.6732528357116722,
      "grad_norm": 6.052730560302734,
      "learning_rate": 3.347488921713442e-05,
      "loss": 0.0183,
      "step": 1840
    },
    {
      "epoch": 0.6736187339919503,
      "grad_norm": 0.02339455671608448,
      "learning_rate": 3.346565731166913e-05,
      "loss": 0.001,
      "step": 1841
    },
    {
      "epoch": 0.6739846322722283,
      "grad_norm": 0.029792435467243195,
      "learning_rate": 3.345642540620384e-05,
      "loss": 0.0012,
      "step": 1842
    },
    {
      "epoch": 0.6743505305525064,
      "grad_norm": 6.144677639007568,
      "learning_rate": 3.344719350073856e-05,
      "loss": 0.6154,
      "step": 1843
    },
    {
      "epoch": 0.6747164288327845,
      "grad_norm": 0.327248752117157,
      "learning_rate": 3.343796159527326e-05,
      "loss": 0.0042,
      "step": 1844
    },
    {
      "epoch": 0.6750823271130626,
      "grad_norm": 0.06135210767388344,
      "learning_rate": 3.342872968980798e-05,
      "loss": 0.0015,
      "step": 1845
    },
    {
      "epoch": 0.6754482253933407,
      "grad_norm": 10.373112678527832,
      "learning_rate": 3.341949778434269e-05,
      "loss": 0.1431,
      "step": 1846
    },
    {
      "epoch": 0.6758141236736187,
      "grad_norm": 4.642188549041748,
      "learning_rate": 3.34102658788774e-05,
      "loss": 0.117,
      "step": 1847
    },
    {
      "epoch": 0.6761800219538968,
      "grad_norm": 0.1236332505941391,
      "learning_rate": 3.340103397341212e-05,
      "loss": 0.003,
      "step": 1848
    },
    {
      "epoch": 0.6765459202341749,
      "grad_norm": 0.03157122805714607,
      "learning_rate": 3.339180206794683e-05,
      "loss": 0.0014,
      "step": 1849
    },
    {
      "epoch": 0.676911818514453,
      "grad_norm": 0.031632885336875916,
      "learning_rate": 3.338257016248154e-05,
      "loss": 0.0012,
      "step": 1850
    },
    {
      "epoch": 0.677277716794731,
      "grad_norm": 1.688043475151062,
      "learning_rate": 3.337333825701625e-05,
      "loss": 0.0134,
      "step": 1851
    },
    {
      "epoch": 0.6776436150750091,
      "grad_norm": 0.1112988218665123,
      "learning_rate": 3.336410635155096e-05,
      "loss": 0.0022,
      "step": 1852
    },
    {
      "epoch": 0.6780095133552873,
      "grad_norm": 0.03230787441134453,
      "learning_rate": 3.335487444608567e-05,
      "loss": 0.0013,
      "step": 1853
    },
    {
      "epoch": 0.6783754116355654,
      "grad_norm": 3.9207866191864014,
      "learning_rate": 3.334564254062039e-05,
      "loss": 0.022,
      "step": 1854
    },
    {
      "epoch": 0.6787413099158434,
      "grad_norm": 1.4212805032730103,
      "learning_rate": 3.33364106351551e-05,
      "loss": 0.0188,
      "step": 1855
    },
    {
      "epoch": 0.6791072081961215,
      "grad_norm": 7.130736827850342,
      "learning_rate": 3.332717872968981e-05,
      "loss": 0.1077,
      "step": 1856
    },
    {
      "epoch": 0.6794731064763996,
      "grad_norm": 0.9275796413421631,
      "learning_rate": 3.331794682422452e-05,
      "loss": 0.017,
      "step": 1857
    },
    {
      "epoch": 0.6798390047566777,
      "grad_norm": 0.035910412669181824,
      "learning_rate": 3.330871491875923e-05,
      "loss": 0.0015,
      "step": 1858
    },
    {
      "epoch": 0.6802049030369557,
      "grad_norm": 1.6036497354507446,
      "learning_rate": 3.329948301329395e-05,
      "loss": 0.3448,
      "step": 1859
    },
    {
      "epoch": 0.6805708013172338,
      "grad_norm": 0.04253338277339935,
      "learning_rate": 3.329025110782866e-05,
      "loss": 0.0018,
      "step": 1860
    },
    {
      "epoch": 0.6809366995975119,
      "grad_norm": 5.54117488861084,
      "learning_rate": 3.328101920236337e-05,
      "loss": 0.1207,
      "step": 1861
    },
    {
      "epoch": 0.68130259787779,
      "grad_norm": 0.10457110404968262,
      "learning_rate": 3.3271787296898085e-05,
      "loss": 0.0033,
      "step": 1862
    },
    {
      "epoch": 0.681668496158068,
      "grad_norm": 0.05511416867375374,
      "learning_rate": 3.326255539143279e-05,
      "loss": 0.0025,
      "step": 1863
    },
    {
      "epoch": 0.6820343944383461,
      "grad_norm": 13.38876724243164,
      "learning_rate": 3.3253323485967507e-05,
      "loss": 0.2981,
      "step": 1864
    },
    {
      "epoch": 0.6824002927186242,
      "grad_norm": 0.051368504762649536,
      "learning_rate": 3.324409158050222e-05,
      "loss": 0.0023,
      "step": 1865
    },
    {
      "epoch": 0.6827661909989023,
      "grad_norm": 0.041452161967754364,
      "learning_rate": 3.323485967503693e-05,
      "loss": 0.0019,
      "step": 1866
    },
    {
      "epoch": 0.6831320892791803,
      "grad_norm": 0.2830202281475067,
      "learning_rate": 3.3225627769571645e-05,
      "loss": 0.0079,
      "step": 1867
    },
    {
      "epoch": 0.6834979875594585,
      "grad_norm": 5.837390899658203,
      "learning_rate": 3.3216395864106356e-05,
      "loss": 0.1414,
      "step": 1868
    },
    {
      "epoch": 0.6838638858397366,
      "grad_norm": 1.6007667779922485,
      "learning_rate": 3.3207163958641066e-05,
      "loss": 0.3516,
      "step": 1869
    },
    {
      "epoch": 0.6842297841200147,
      "grad_norm": 0.06447337567806244,
      "learning_rate": 3.319793205317578e-05,
      "loss": 0.0031,
      "step": 1870
    },
    {
      "epoch": 0.6845956824002927,
      "grad_norm": 0.6246155500411987,
      "learning_rate": 3.318870014771049e-05,
      "loss": 0.0088,
      "step": 1871
    },
    {
      "epoch": 0.6849615806805708,
      "grad_norm": 0.06960929930210114,
      "learning_rate": 3.31794682422452e-05,
      "loss": 0.0031,
      "step": 1872
    },
    {
      "epoch": 0.6853274789608489,
      "grad_norm": 0.10577204078435898,
      "learning_rate": 3.3170236336779915e-05,
      "loss": 0.0046,
      "step": 1873
    },
    {
      "epoch": 0.685693377241127,
      "grad_norm": 7.663261890411377,
      "learning_rate": 3.3161004431314626e-05,
      "loss": 0.3741,
      "step": 1874
    },
    {
      "epoch": 0.686059275521405,
      "grad_norm": 0.3591771125793457,
      "learning_rate": 3.3151772525849336e-05,
      "loss": 0.0113,
      "step": 1875
    },
    {
      "epoch": 0.6864251738016831,
      "grad_norm": 0.46971848607063293,
      "learning_rate": 3.314254062038405e-05,
      "loss": 0.0086,
      "step": 1876
    },
    {
      "epoch": 0.6867910720819612,
      "grad_norm": 6.296411991119385,
      "learning_rate": 3.313330871491876e-05,
      "loss": 0.2057,
      "step": 1877
    },
    {
      "epoch": 0.6871569703622393,
      "grad_norm": 0.08612236380577087,
      "learning_rate": 3.3124076809453475e-05,
      "loss": 0.0039,
      "step": 1878
    },
    {
      "epoch": 0.6875228686425173,
      "grad_norm": 4.667773246765137,
      "learning_rate": 3.3114844903988185e-05,
      "loss": 0.2409,
      "step": 1879
    },
    {
      "epoch": 0.6878887669227954,
      "grad_norm": 7.733899116516113,
      "learning_rate": 3.3105612998522896e-05,
      "loss": 0.2762,
      "step": 1880
    },
    {
      "epoch": 0.6882546652030735,
      "grad_norm": 17.74125099182129,
      "learning_rate": 3.309638109305761e-05,
      "loss": 0.1724,
      "step": 1881
    },
    {
      "epoch": 0.6886205634833517,
      "grad_norm": 0.6430237293243408,
      "learning_rate": 3.308714918759232e-05,
      "loss": 0.0127,
      "step": 1882
    },
    {
      "epoch": 0.6889864617636298,
      "grad_norm": 1.9825129508972168,
      "learning_rate": 3.307791728212703e-05,
      "loss": 0.0298,
      "step": 1883
    },
    {
      "epoch": 0.6893523600439078,
      "grad_norm": 0.3301848769187927,
      "learning_rate": 3.3068685376661745e-05,
      "loss": 0.0139,
      "step": 1884
    },
    {
      "epoch": 0.6897182583241859,
      "grad_norm": 0.43627846240997314,
      "learning_rate": 3.3059453471196455e-05,
      "loss": 0.0074,
      "step": 1885
    },
    {
      "epoch": 0.690084156604464,
      "grad_norm": 0.077665314078331,
      "learning_rate": 3.305022156573117e-05,
      "loss": 0.0037,
      "step": 1886
    },
    {
      "epoch": 0.6904500548847421,
      "grad_norm": 0.1685972362756729,
      "learning_rate": 3.304098966026588e-05,
      "loss": 0.0065,
      "step": 1887
    },
    {
      "epoch": 0.6908159531650201,
      "grad_norm": 0.07923941314220428,
      "learning_rate": 3.3031757754800594e-05,
      "loss": 0.0036,
      "step": 1888
    },
    {
      "epoch": 0.6911818514452982,
      "grad_norm": 0.21823859214782715,
      "learning_rate": 3.3022525849335304e-05,
      "loss": 0.0068,
      "step": 1889
    },
    {
      "epoch": 0.6915477497255763,
      "grad_norm": 0.3008313477039337,
      "learning_rate": 3.3013293943870015e-05,
      "loss": 0.0088,
      "step": 1890
    },
    {
      "epoch": 0.6919136480058544,
      "grad_norm": 0.2706022262573242,
      "learning_rate": 3.3004062038404725e-05,
      "loss": 0.0086,
      "step": 1891
    },
    {
      "epoch": 0.6922795462861324,
      "grad_norm": 1.0016307830810547,
      "learning_rate": 3.299483013293944e-05,
      "loss": 0.0135,
      "step": 1892
    },
    {
      "epoch": 0.6926454445664105,
      "grad_norm": 0.08008371293544769,
      "learning_rate": 3.298559822747415e-05,
      "loss": 0.0027,
      "step": 1893
    },
    {
      "epoch": 0.6930113428466886,
      "grad_norm": 1.4992748498916626,
      "learning_rate": 3.2976366322008864e-05,
      "loss": 0.2955,
      "step": 1894
    },
    {
      "epoch": 0.6933772411269667,
      "grad_norm": 0.13983695209026337,
      "learning_rate": 3.2967134416543574e-05,
      "loss": 0.0036,
      "step": 1895
    },
    {
      "epoch": 0.6937431394072447,
      "grad_norm": 0.16777819395065308,
      "learning_rate": 3.2957902511078285e-05,
      "loss": 0.0041,
      "step": 1896
    },
    {
      "epoch": 0.6941090376875229,
      "grad_norm": 0.16482704877853394,
      "learning_rate": 3.2948670605613e-05,
      "loss": 0.0056,
      "step": 1897
    },
    {
      "epoch": 0.694474935967801,
      "grad_norm": 0.08413726836442947,
      "learning_rate": 3.293943870014771e-05,
      "loss": 0.0031,
      "step": 1898
    },
    {
      "epoch": 0.6948408342480791,
      "grad_norm": 2.5302886962890625,
      "learning_rate": 3.293020679468242e-05,
      "loss": 0.0149,
      "step": 1899
    },
    {
      "epoch": 0.6952067325283571,
      "grad_norm": 0.5524428486824036,
      "learning_rate": 3.292097488921714e-05,
      "loss": 0.0071,
      "step": 1900
    },
    {
      "epoch": 0.6955726308086352,
      "grad_norm": 1.5459927320480347,
      "learning_rate": 3.2911742983751844e-05,
      "loss": 0.3248,
      "step": 1901
    },
    {
      "epoch": 0.6959385290889133,
      "grad_norm": 4.556272029876709,
      "learning_rate": 3.2902511078286555e-05,
      "loss": 0.2044,
      "step": 1902
    },
    {
      "epoch": 0.6963044273691914,
      "grad_norm": 0.35628265142440796,
      "learning_rate": 3.289327917282127e-05,
      "loss": 0.0057,
      "step": 1903
    },
    {
      "epoch": 0.6966703256494694,
      "grad_norm": 16.294687271118164,
      "learning_rate": 3.288404726735598e-05,
      "loss": 0.2379,
      "step": 1904
    },
    {
      "epoch": 0.6970362239297475,
      "grad_norm": 0.13818359375,
      "learning_rate": 3.28748153618907e-05,
      "loss": 0.0043,
      "step": 1905
    },
    {
      "epoch": 0.6974021222100256,
      "grad_norm": 21.71449851989746,
      "learning_rate": 3.286558345642541e-05,
      "loss": 0.0878,
      "step": 1906
    },
    {
      "epoch": 0.6977680204903037,
      "grad_norm": 5.0618510246276855,
      "learning_rate": 3.285635155096012e-05,
      "loss": 0.0155,
      "step": 1907
    },
    {
      "epoch": 0.6981339187705817,
      "grad_norm": 0.09689901769161224,
      "learning_rate": 3.284711964549483e-05,
      "loss": 0.0045,
      "step": 1908
    },
    {
      "epoch": 0.6984998170508598,
      "grad_norm": 0.20164629817008972,
      "learning_rate": 3.283788774002954e-05,
      "loss": 0.0065,
      "step": 1909
    },
    {
      "epoch": 0.698865715331138,
      "grad_norm": 0.16885969042778015,
      "learning_rate": 3.282865583456425e-05,
      "loss": 0.0062,
      "step": 1910
    },
    {
      "epoch": 0.6992316136114161,
      "grad_norm": 15.456985473632812,
      "learning_rate": 3.281942392909897e-05,
      "loss": 0.5468,
      "step": 1911
    },
    {
      "epoch": 0.6995975118916941,
      "grad_norm": 3.360994577407837,
      "learning_rate": 3.281019202363368e-05,
      "loss": 0.0184,
      "step": 1912
    },
    {
      "epoch": 0.6999634101719722,
      "grad_norm": 3.9883904457092285,
      "learning_rate": 3.280096011816839e-05,
      "loss": 0.0192,
      "step": 1913
    },
    {
      "epoch": 0.7003293084522503,
      "grad_norm": 0.13741803169250488,
      "learning_rate": 3.27917282127031e-05,
      "loss": 0.0058,
      "step": 1914
    },
    {
      "epoch": 0.7006952067325284,
      "grad_norm": 0.09274721145629883,
      "learning_rate": 3.278249630723781e-05,
      "loss": 0.0043,
      "step": 1915
    },
    {
      "epoch": 0.7010611050128064,
      "grad_norm": 0.4100906252861023,
      "learning_rate": 3.277326440177253e-05,
      "loss": 0.0103,
      "step": 1916
    },
    {
      "epoch": 0.7014270032930845,
      "grad_norm": 0.6367105841636658,
      "learning_rate": 3.276403249630724e-05,
      "loss": 0.008,
      "step": 1917
    },
    {
      "epoch": 0.7017929015733626,
      "grad_norm": 0.1192326694726944,
      "learning_rate": 3.275480059084195e-05,
      "loss": 0.0056,
      "step": 1918
    },
    {
      "epoch": 0.7021587998536407,
      "grad_norm": 0.16362448036670685,
      "learning_rate": 3.274556868537667e-05,
      "loss": 0.0069,
      "step": 1919
    },
    {
      "epoch": 0.7025246981339188,
      "grad_norm": 0.15850292146205902,
      "learning_rate": 3.273633677991137e-05,
      "loss": 0.0056,
      "step": 1920
    },
    {
      "epoch": 0.7028905964141968,
      "grad_norm": 3.5090222358703613,
      "learning_rate": 3.272710487444608e-05,
      "loss": 0.0272,
      "step": 1921
    },
    {
      "epoch": 0.7032564946944749,
      "grad_norm": 0.05330631881952286,
      "learning_rate": 3.27178729689808e-05,
      "loss": 0.0024,
      "step": 1922
    },
    {
      "epoch": 0.703622392974753,
      "grad_norm": 31.537609100341797,
      "learning_rate": 3.270864106351551e-05,
      "loss": 0.3322,
      "step": 1923
    },
    {
      "epoch": 0.7039882912550312,
      "grad_norm": 13.582193374633789,
      "learning_rate": 3.269940915805022e-05,
      "loss": 0.3778,
      "step": 1924
    },
    {
      "epoch": 0.7043541895353091,
      "grad_norm": 0.06720732152462006,
      "learning_rate": 3.269017725258494e-05,
      "loss": 0.0028,
      "step": 1925
    },
    {
      "epoch": 0.7047200878155873,
      "grad_norm": 0.03727424144744873,
      "learning_rate": 3.268094534711965e-05,
      "loss": 0.0017,
      "step": 1926
    },
    {
      "epoch": 0.7050859860958654,
      "grad_norm": 0.143751859664917,
      "learning_rate": 3.267171344165436e-05,
      "loss": 0.0031,
      "step": 1927
    },
    {
      "epoch": 0.7054518843761435,
      "grad_norm": 0.051822636276483536,
      "learning_rate": 3.266248153618907e-05,
      "loss": 0.0024,
      "step": 1928
    },
    {
      "epoch": 0.7058177826564215,
      "grad_norm": 1.8832542896270752,
      "learning_rate": 3.265324963072378e-05,
      "loss": 0.0108,
      "step": 1929
    },
    {
      "epoch": 0.7061836809366996,
      "grad_norm": 0.260347843170166,
      "learning_rate": 3.26440177252585e-05,
      "loss": 0.0068,
      "step": 1930
    },
    {
      "epoch": 0.7065495792169777,
      "grad_norm": 0.08212118595838547,
      "learning_rate": 3.263478581979321e-05,
      "loss": 0.0027,
      "step": 1931
    },
    {
      "epoch": 0.7069154774972558,
      "grad_norm": 0.31406503915786743,
      "learning_rate": 3.262555391432792e-05,
      "loss": 0.0045,
      "step": 1932
    },
    {
      "epoch": 0.7072813757775338,
      "grad_norm": 0.18015927076339722,
      "learning_rate": 3.261632200886263e-05,
      "loss": 0.0045,
      "step": 1933
    },
    {
      "epoch": 0.7076472740578119,
      "grad_norm": 0.06598324328660965,
      "learning_rate": 3.260709010339734e-05,
      "loss": 0.0029,
      "step": 1934
    },
    {
      "epoch": 0.70801317233809,
      "grad_norm": 0.03595101088285446,
      "learning_rate": 3.259785819793206e-05,
      "loss": 0.0016,
      "step": 1935
    },
    {
      "epoch": 0.7083790706183681,
      "grad_norm": 0.025496840476989746,
      "learning_rate": 3.258862629246677e-05,
      "loss": 0.0012,
      "step": 1936
    },
    {
      "epoch": 0.7087449688986461,
      "grad_norm": 12.16235065460205,
      "learning_rate": 3.257939438700148e-05,
      "loss": 0.1798,
      "step": 1937
    },
    {
      "epoch": 0.7091108671789242,
      "grad_norm": 7.500272274017334,
      "learning_rate": 3.2570162481536196e-05,
      "loss": 0.1644,
      "step": 1938
    },
    {
      "epoch": 0.7094767654592024,
      "grad_norm": 13.296486854553223,
      "learning_rate": 3.25609305760709e-05,
      "loss": 0.1566,
      "step": 1939
    },
    {
      "epoch": 0.7098426637394805,
      "grad_norm": 0.07533790916204453,
      "learning_rate": 3.255169867060561e-05,
      "loss": 0.0023,
      "step": 1940
    },
    {
      "epoch": 0.7102085620197585,
      "grad_norm": 0.032131511718034744,
      "learning_rate": 3.254246676514033e-05,
      "loss": 0.0014,
      "step": 1941
    },
    {
      "epoch": 0.7105744603000366,
      "grad_norm": 0.03475900739431381,
      "learning_rate": 3.253323485967504e-05,
      "loss": 0.0016,
      "step": 1942
    },
    {
      "epoch": 0.7109403585803147,
      "grad_norm": 0.2184230536222458,
      "learning_rate": 3.252400295420975e-05,
      "loss": 0.003,
      "step": 1943
    },
    {
      "epoch": 0.7113062568605928,
      "grad_norm": 0.0130386957898736,
      "learning_rate": 3.2514771048744466e-05,
      "loss": 0.0006,
      "step": 1944
    },
    {
      "epoch": 0.7116721551408708,
      "grad_norm": 0.06149248778820038,
      "learning_rate": 3.2505539143279176e-05,
      "loss": 0.0016,
      "step": 1945
    },
    {
      "epoch": 0.7120380534211489,
      "grad_norm": 2.6922824382781982,
      "learning_rate": 3.249630723781389e-05,
      "loss": 0.5391,
      "step": 1946
    },
    {
      "epoch": 0.712403951701427,
      "grad_norm": 0.07403847575187683,
      "learning_rate": 3.24870753323486e-05,
      "loss": 0.0012,
      "step": 1947
    },
    {
      "epoch": 0.7127698499817051,
      "grad_norm": 0.22764749825000763,
      "learning_rate": 3.247784342688331e-05,
      "loss": 0.0021,
      "step": 1948
    },
    {
      "epoch": 0.7131357482619831,
      "grad_norm": 0.034508705139160156,
      "learning_rate": 3.2468611521418025e-05,
      "loss": 0.0015,
      "step": 1949
    },
    {
      "epoch": 0.7135016465422612,
      "grad_norm": 0.020472703501582146,
      "learning_rate": 3.2459379615952736e-05,
      "loss": 0.0009,
      "step": 1950
    },
    {
      "epoch": 0.7138675448225393,
      "grad_norm": 6.200559139251709,
      "learning_rate": 3.2450147710487446e-05,
      "loss": 0.2841,
      "step": 1951
    },
    {
      "epoch": 0.7142334431028174,
      "grad_norm": 0.0918068066239357,
      "learning_rate": 3.244091580502216e-05,
      "loss": 0.0017,
      "step": 1952
    },
    {
      "epoch": 0.7145993413830956,
      "grad_norm": 0.6126900315284729,
      "learning_rate": 3.243168389955687e-05,
      "loss": 0.0058,
      "step": 1953
    },
    {
      "epoch": 0.7149652396633736,
      "grad_norm": 8.778240203857422,
      "learning_rate": 3.242245199409158e-05,
      "loss": 0.0101,
      "step": 1954
    },
    {
      "epoch": 0.7153311379436517,
      "grad_norm": 23.65386390686035,
      "learning_rate": 3.2413220088626295e-05,
      "loss": 0.113,
      "step": 1955
    },
    {
      "epoch": 0.7156970362239298,
      "grad_norm": 0.22287267446517944,
      "learning_rate": 3.2403988183161006e-05,
      "loss": 0.0032,
      "step": 1956
    },
    {
      "epoch": 0.7160629345042079,
      "grad_norm": 0.03347810357809067,
      "learning_rate": 3.239475627769572e-05,
      "loss": 0.0017,
      "step": 1957
    },
    {
      "epoch": 0.7164288327844859,
      "grad_norm": 0.024979230016469955,
      "learning_rate": 3.238552437223043e-05,
      "loss": 0.0013,
      "step": 1958
    },
    {
      "epoch": 0.716794731064764,
      "grad_norm": 0.04101972654461861,
      "learning_rate": 3.237629246676514e-05,
      "loss": 0.0017,
      "step": 1959
    },
    {
      "epoch": 0.7171606293450421,
      "grad_norm": 0.03305108845233917,
      "learning_rate": 3.2367060561299855e-05,
      "loss": 0.0017,
      "step": 1960
    },
    {
      "epoch": 0.7175265276253202,
      "grad_norm": 0.04302297905087471,
      "learning_rate": 3.2357828655834565e-05,
      "loss": 0.0019,
      "step": 1961
    },
    {
      "epoch": 0.7178924259055982,
      "grad_norm": 0.04425369203090668,
      "learning_rate": 3.2348596750369276e-05,
      "loss": 0.0016,
      "step": 1962
    },
    {
      "epoch": 0.7182583241858763,
      "grad_norm": 0.04440101981163025,
      "learning_rate": 3.233936484490399e-05,
      "loss": 0.0017,
      "step": 1963
    },
    {
      "epoch": 0.7186242224661544,
      "grad_norm": 0.03148224949836731,
      "learning_rate": 3.2330132939438704e-05,
      "loss": 0.0015,
      "step": 1964
    },
    {
      "epoch": 0.7189901207464325,
      "grad_norm": 0.055971380323171616,
      "learning_rate": 3.2320901033973414e-05,
      "loss": 0.0023,
      "step": 1965
    },
    {
      "epoch": 0.7193560190267105,
      "grad_norm": 0.026385385543107986,
      "learning_rate": 3.2311669128508125e-05,
      "loss": 0.0012,
      "step": 1966
    },
    {
      "epoch": 0.7197219173069886,
      "grad_norm": 0.021906334906816483,
      "learning_rate": 3.2302437223042835e-05,
      "loss": 0.0011,
      "step": 1967
    },
    {
      "epoch": 0.7200878155872668,
      "grad_norm": 0.2751903533935547,
      "learning_rate": 3.229320531757755e-05,
      "loss": 0.0056,
      "step": 1968
    },
    {
      "epoch": 0.7204537138675449,
      "grad_norm": 0.3750927448272705,
      "learning_rate": 3.228397341211226e-05,
      "loss": 0.0042,
      "step": 1969
    },
    {
      "epoch": 0.7208196121478229,
      "grad_norm": 0.022914990782737732,
      "learning_rate": 3.2274741506646974e-05,
      "loss": 0.0012,
      "step": 1970
    },
    {
      "epoch": 0.721185510428101,
      "grad_norm": 0.03678980842232704,
      "learning_rate": 3.2265509601181684e-05,
      "loss": 0.0015,
      "step": 1971
    },
    {
      "epoch": 0.7215514087083791,
      "grad_norm": 3.6797335147857666,
      "learning_rate": 3.2256277695716395e-05,
      "loss": 0.2589,
      "step": 1972
    },
    {
      "epoch": 0.7219173069886572,
      "grad_norm": 0.9436518549919128,
      "learning_rate": 3.2247045790251105e-05,
      "loss": 0.0054,
      "step": 1973
    },
    {
      "epoch": 0.7222832052689352,
      "grad_norm": 0.03878078609704971,
      "learning_rate": 3.223781388478582e-05,
      "loss": 0.0015,
      "step": 1974
    },
    {
      "epoch": 0.7226491035492133,
      "grad_norm": 5.549458026885986,
      "learning_rate": 3.222858197932053e-05,
      "loss": 0.2549,
      "step": 1975
    },
    {
      "epoch": 0.7230150018294914,
      "grad_norm": 0.15916861593723297,
      "learning_rate": 3.221935007385525e-05,
      "loss": 0.0021,
      "step": 1976
    },
    {
      "epoch": 0.7233809001097695,
      "grad_norm": 0.8011974692344666,
      "learning_rate": 3.2210118168389954e-05,
      "loss": 0.0053,
      "step": 1977
    },
    {
      "epoch": 0.7237467983900475,
      "grad_norm": 6.756175994873047,
      "learning_rate": 3.2200886262924665e-05,
      "loss": 0.2184,
      "step": 1978
    },
    {
      "epoch": 0.7241126966703256,
      "grad_norm": 2.8995180130004883,
      "learning_rate": 3.219165435745938e-05,
      "loss": 0.0099,
      "step": 1979
    },
    {
      "epoch": 0.7244785949506037,
      "grad_norm": 0.022044379264116287,
      "learning_rate": 3.218242245199409e-05,
      "loss": 0.0011,
      "step": 1980
    },
    {
      "epoch": 0.7248444932308818,
      "grad_norm": 0.03290579095482826,
      "learning_rate": 3.2173190546528803e-05,
      "loss": 0.0012,
      "step": 1981
    },
    {
      "epoch": 0.7252103915111598,
      "grad_norm": 0.7356245517730713,
      "learning_rate": 3.216395864106352e-05,
      "loss": 0.0068,
      "step": 1982
    },
    {
      "epoch": 0.725576289791438,
      "grad_norm": 0.15662041306495667,
      "learning_rate": 3.215472673559823e-05,
      "loss": 0.0033,
      "step": 1983
    },
    {
      "epoch": 0.7259421880717161,
      "grad_norm": 0.03458793833851814,
      "learning_rate": 3.2145494830132935e-05,
      "loss": 0.0013,
      "step": 1984
    },
    {
      "epoch": 0.7263080863519942,
      "grad_norm": 27.444448471069336,
      "learning_rate": 3.213626292466765e-05,
      "loss": 0.1087,
      "step": 1985
    },
    {
      "epoch": 0.7266739846322722,
      "grad_norm": 0.019844409078359604,
      "learning_rate": 3.212703101920236e-05,
      "loss": 0.001,
      "step": 1986
    },
    {
      "epoch": 0.7270398829125503,
      "grad_norm": 0.04969099536538124,
      "learning_rate": 3.211779911373708e-05,
      "loss": 0.0017,
      "step": 1987
    },
    {
      "epoch": 0.7274057811928284,
      "grad_norm": 0.021997960284352303,
      "learning_rate": 3.210856720827179e-05,
      "loss": 0.001,
      "step": 1988
    },
    {
      "epoch": 0.7277716794731065,
      "grad_norm": 0.027941176667809486,
      "learning_rate": 3.20993353028065e-05,
      "loss": 0.001,
      "step": 1989
    },
    {
      "epoch": 0.7281375777533846,
      "grad_norm": 0.014207674190402031,
      "learning_rate": 3.209010339734121e-05,
      "loss": 0.0007,
      "step": 1990
    },
    {
      "epoch": 0.7285034760336626,
      "grad_norm": 0.01874164678156376,
      "learning_rate": 3.208087149187592e-05,
      "loss": 0.0009,
      "step": 1991
    },
    {
      "epoch": 0.7288693743139407,
      "grad_norm": 0.01615038700401783,
      "learning_rate": 3.207163958641063e-05,
      "loss": 0.0008,
      "step": 1992
    },
    {
      "epoch": 0.7292352725942188,
      "grad_norm": 0.23944298923015594,
      "learning_rate": 3.206240768094535e-05,
      "loss": 0.0047,
      "step": 1993
    },
    {
      "epoch": 0.7296011708744969,
      "grad_norm": 4.15326452255249,
      "learning_rate": 3.205317577548006e-05,
      "loss": 0.333,
      "step": 1994
    },
    {
      "epoch": 0.7299670691547749,
      "grad_norm": 36.74120330810547,
      "learning_rate": 3.204394387001477e-05,
      "loss": 0.1463,
      "step": 1995
    },
    {
      "epoch": 0.730332967435053,
      "grad_norm": 22.03507423400879,
      "learning_rate": 3.203471196454948e-05,
      "loss": 0.3517,
      "step": 1996
    },
    {
      "epoch": 0.7306988657153312,
      "grad_norm": 0.045136354863643646,
      "learning_rate": 3.202548005908419e-05,
      "loss": 0.0015,
      "step": 1997
    },
    {
      "epoch": 0.7310647639956093,
      "grad_norm": 0.04863480105996132,
      "learning_rate": 3.201624815361891e-05,
      "loss": 0.0016,
      "step": 1998
    },
    {
      "epoch": 0.7314306622758873,
      "grad_norm": 26.01010513305664,
      "learning_rate": 3.200701624815362e-05,
      "loss": 0.1819,
      "step": 1999
    },
    {
      "epoch": 0.7317965605561654,
      "grad_norm": 0.02124817669391632,
      "learning_rate": 3.199778434268833e-05,
      "loss": 0.001,
      "step": 2000
    },
    {
      "epoch": 0.7321624588364435,
      "grad_norm": 0.11133713275194168,
      "learning_rate": 3.198855243722305e-05,
      "loss": 0.0014,
      "step": 2001
    },
    {
      "epoch": 0.7325283571167216,
      "grad_norm": 0.020998582243919373,
      "learning_rate": 3.197932053175776e-05,
      "loss": 0.001,
      "step": 2002
    },
    {
      "epoch": 0.7328942553969996,
      "grad_norm": 0.05650513246655464,
      "learning_rate": 3.197008862629246e-05,
      "loss": 0.0014,
      "step": 2003
    },
    {
      "epoch": 0.7332601536772777,
      "grad_norm": 0.03375597298145294,
      "learning_rate": 3.196085672082718e-05,
      "loss": 0.0014,
      "step": 2004
    },
    {
      "epoch": 0.7336260519575558,
      "grad_norm": 1.0397711992263794,
      "learning_rate": 3.195162481536189e-05,
      "loss": 0.0081,
      "step": 2005
    },
    {
      "epoch": 0.7339919502378339,
      "grad_norm": 0.26841071248054504,
      "learning_rate": 3.194239290989661e-05,
      "loss": 0.0082,
      "step": 2006
    },
    {
      "epoch": 0.7343578485181119,
      "grad_norm": 12.66553783416748,
      "learning_rate": 3.193316100443132e-05,
      "loss": 0.1326,
      "step": 2007
    },
    {
      "epoch": 0.73472374679839,
      "grad_norm": 0.04123378545045853,
      "learning_rate": 3.192392909896603e-05,
      "loss": 0.0012,
      "step": 2008
    },
    {
      "epoch": 0.7350896450786681,
      "grad_norm": 0.10679668933153152,
      "learning_rate": 3.191469719350074e-05,
      "loss": 0.0032,
      "step": 2009
    },
    {
      "epoch": 0.7354555433589463,
      "grad_norm": 1.9712477922439575,
      "learning_rate": 3.190546528803545e-05,
      "loss": 0.2302,
      "step": 2010
    },
    {
      "epoch": 0.7358214416392242,
      "grad_norm": 0.016686230897903442,
      "learning_rate": 3.189623338257016e-05,
      "loss": 0.0008,
      "step": 2011
    },
    {
      "epoch": 0.7361873399195024,
      "grad_norm": 0.05770118907094002,
      "learning_rate": 3.188700147710488e-05,
      "loss": 0.0018,
      "step": 2012
    },
    {
      "epoch": 0.7365532381997805,
      "grad_norm": 0.07264194637537003,
      "learning_rate": 3.187776957163959e-05,
      "loss": 0.0021,
      "step": 2013
    },
    {
      "epoch": 0.7369191364800586,
      "grad_norm": 0.0254527535289526,
      "learning_rate": 3.18685376661743e-05,
      "loss": 0.001,
      "step": 2014
    },
    {
      "epoch": 0.7372850347603366,
      "grad_norm": 0.1577521413564682,
      "learning_rate": 3.185930576070901e-05,
      "loss": 0.0048,
      "step": 2015
    },
    {
      "epoch": 0.7376509330406147,
      "grad_norm": 0.23751254379749298,
      "learning_rate": 3.185007385524372e-05,
      "loss": 0.0066,
      "step": 2016
    },
    {
      "epoch": 0.7380168313208928,
      "grad_norm": 1.8500930070877075,
      "learning_rate": 3.184084194977844e-05,
      "loss": 0.4349,
      "step": 2017
    },
    {
      "epoch": 0.7383827296011709,
      "grad_norm": 0.04236564040184021,
      "learning_rate": 3.183161004431315e-05,
      "loss": 0.0012,
      "step": 2018
    },
    {
      "epoch": 0.7387486278814489,
      "grad_norm": 0.019209980964660645,
      "learning_rate": 3.182237813884786e-05,
      "loss": 0.0009,
      "step": 2019
    },
    {
      "epoch": 0.739114526161727,
      "grad_norm": 0.062324658036231995,
      "learning_rate": 3.1813146233382576e-05,
      "loss": 0.0024,
      "step": 2020
    },
    {
      "epoch": 0.7394804244420051,
      "grad_norm": 0.02229538932442665,
      "learning_rate": 3.1803914327917286e-05,
      "loss": 0.001,
      "step": 2021
    },
    {
      "epoch": 0.7398463227222832,
      "grad_norm": 0.2773287892341614,
      "learning_rate": 3.179468242245199e-05,
      "loss": 0.0024,
      "step": 2022
    },
    {
      "epoch": 0.7402122210025612,
      "grad_norm": 0.12506921589374542,
      "learning_rate": 3.178545051698671e-05,
      "loss": 0.0033,
      "step": 2023
    },
    {
      "epoch": 0.7405781192828393,
      "grad_norm": 2.1528313159942627,
      "learning_rate": 3.177621861152142e-05,
      "loss": 0.477,
      "step": 2024
    },
    {
      "epoch": 0.7409440175631175,
      "grad_norm": 0.050573594868183136,
      "learning_rate": 3.176698670605613e-05,
      "loss": 0.0016,
      "step": 2025
    },
    {
      "epoch": 0.7413099158433956,
      "grad_norm": 0.019277891144156456,
      "learning_rate": 3.1757754800590846e-05,
      "loss": 0.001,
      "step": 2026
    },
    {
      "epoch": 0.7416758141236737,
      "grad_norm": 0.13239124417304993,
      "learning_rate": 3.1748522895125556e-05,
      "loss": 0.0038,
      "step": 2027
    },
    {
      "epoch": 0.7420417124039517,
      "grad_norm": 3.580116033554077,
      "learning_rate": 3.173929098966027e-05,
      "loss": 0.2298,
      "step": 2028
    },
    {
      "epoch": 0.7424076106842298,
      "grad_norm": 0.2313051074743271,
      "learning_rate": 3.173005908419498e-05,
      "loss": 0.008,
      "step": 2029
    },
    {
      "epoch": 0.7427735089645079,
      "grad_norm": 0.10010110586881638,
      "learning_rate": 3.172082717872969e-05,
      "loss": 0.0046,
      "step": 2030
    },
    {
      "epoch": 0.743139407244786,
      "grad_norm": 0.07856214046478271,
      "learning_rate": 3.1711595273264405e-05,
      "loss": 0.0027,
      "step": 2031
    },
    {
      "epoch": 0.743505305525064,
      "grad_norm": 0.025144057348370552,
      "learning_rate": 3.1702363367799116e-05,
      "loss": 0.0013,
      "step": 2032
    },
    {
      "epoch": 0.7438712038053421,
      "grad_norm": 0.04334985464811325,
      "learning_rate": 3.1693131462333827e-05,
      "loss": 0.0022,
      "step": 2033
    },
    {
      "epoch": 0.7442371020856202,
      "grad_norm": 0.39100924134254456,
      "learning_rate": 3.168389955686854e-05,
      "loss": 0.0095,
      "step": 2034
    },
    {
      "epoch": 0.7446030003658983,
      "grad_norm": 0.026215706020593643,
      "learning_rate": 3.167466765140325e-05,
      "loss": 0.0014,
      "step": 2035
    },
    {
      "epoch": 0.7449688986461763,
      "grad_norm": 9.099326133728027,
      "learning_rate": 3.1665435745937965e-05,
      "loss": 0.2704,
      "step": 2036
    },
    {
      "epoch": 0.7453347969264544,
      "grad_norm": 0.055858906358480453,
      "learning_rate": 3.1656203840472676e-05,
      "loss": 0.0023,
      "step": 2037
    },
    {
      "epoch": 0.7457006952067325,
      "grad_norm": 0.028509967029094696,
      "learning_rate": 3.1646971935007386e-05,
      "loss": 0.0015,
      "step": 2038
    },
    {
      "epoch": 0.7460665934870107,
      "grad_norm": 0.6743654012680054,
      "learning_rate": 3.16377400295421e-05,
      "loss": 0.014,
      "step": 2039
    },
    {
      "epoch": 0.7464324917672887,
      "grad_norm": 0.05960041657090187,
      "learning_rate": 3.1628508124076814e-05,
      "loss": 0.0017,
      "step": 2040
    },
    {
      "epoch": 0.7467983900475668,
      "grad_norm": 10.542858123779297,
      "learning_rate": 3.161927621861152e-05,
      "loss": 0.1229,
      "step": 2041
    },
    {
      "epoch": 0.7471642883278449,
      "grad_norm": 0.3738209009170532,
      "learning_rate": 3.1610044313146235e-05,
      "loss": 0.007,
      "step": 2042
    },
    {
      "epoch": 0.747530186608123,
      "grad_norm": 0.035846419632434845,
      "learning_rate": 3.1600812407680946e-05,
      "loss": 0.0015,
      "step": 2043
    },
    {
      "epoch": 0.747896084888401,
      "grad_norm": 0.1488501876592636,
      "learning_rate": 3.1591580502215656e-05,
      "loss": 0.004,
      "step": 2044
    },
    {
      "epoch": 0.7482619831686791,
      "grad_norm": 0.030357541516423225,
      "learning_rate": 3.1582348596750373e-05,
      "loss": 0.0016,
      "step": 2045
    },
    {
      "epoch": 0.7486278814489572,
      "grad_norm": 0.1512530893087387,
      "learning_rate": 3.1573116691285084e-05,
      "loss": 0.0047,
      "step": 2046
    },
    {
      "epoch": 0.7489937797292353,
      "grad_norm": 0.13373538851737976,
      "learning_rate": 3.1563884785819795e-05,
      "loss": 0.0034,
      "step": 2047
    },
    {
      "epoch": 0.7493596780095133,
      "grad_norm": 0.04064589738845825,
      "learning_rate": 3.1554652880354505e-05,
      "loss": 0.0019,
      "step": 2048
    },
    {
      "epoch": 0.7497255762897914,
      "grad_norm": 0.050444334745407104,
      "learning_rate": 3.1545420974889216e-05,
      "loss": 0.0024,
      "step": 2049
    },
    {
      "epoch": 0.7500914745700695,
      "grad_norm": 0.42003491520881653,
      "learning_rate": 3.153618906942393e-05,
      "loss": 0.0041,
      "step": 2050
    },
    {
      "epoch": 0.7504573728503476,
      "grad_norm": 0.025796517729759216,
      "learning_rate": 3.1526957163958644e-05,
      "loss": 0.0012,
      "step": 2051
    },
    {
      "epoch": 0.7508232711306256,
      "grad_norm": 0.08626840263605118,
      "learning_rate": 3.1517725258493354e-05,
      "loss": 0.0014,
      "step": 2052
    },
    {
      "epoch": 0.7511891694109037,
      "grad_norm": 0.028767075389623642,
      "learning_rate": 3.1508493353028065e-05,
      "loss": 0.0013,
      "step": 2053
    },
    {
      "epoch": 0.7515550676911819,
      "grad_norm": 17.902116775512695,
      "learning_rate": 3.1499261447562775e-05,
      "loss": 0.0837,
      "step": 2054
    },
    {
      "epoch": 0.75192096597146,
      "grad_norm": 0.08072703331708908,
      "learning_rate": 3.1490029542097486e-05,
      "loss": 0.0025,
      "step": 2055
    },
    {
      "epoch": 0.752286864251738,
      "grad_norm": 0.028787905350327492,
      "learning_rate": 3.14807976366322e-05,
      "loss": 0.0012,
      "step": 2056
    },
    {
      "epoch": 0.7526527625320161,
      "grad_norm": 25.1708984375,
      "learning_rate": 3.1471565731166914e-05,
      "loss": 0.395,
      "step": 2057
    },
    {
      "epoch": 0.7530186608122942,
      "grad_norm": 7.806828022003174,
      "learning_rate": 3.146233382570163e-05,
      "loss": 0.3271,
      "step": 2058
    },
    {
      "epoch": 0.7533845590925723,
      "grad_norm": 0.12168089300394058,
      "learning_rate": 3.145310192023634e-05,
      "loss": 0.0018,
      "step": 2059
    },
    {
      "epoch": 0.7537504573728504,
      "grad_norm": 11.964946746826172,
      "learning_rate": 3.1443870014771045e-05,
      "loss": 0.4057,
      "step": 2060
    },
    {
      "epoch": 0.7541163556531284,
      "grad_norm": 0.6522296667098999,
      "learning_rate": 3.143463810930576e-05,
      "loss": 0.005,
      "step": 2061
    },
    {
      "epoch": 0.7544822539334065,
      "grad_norm": 0.18270903825759888,
      "learning_rate": 3.142540620384047e-05,
      "loss": 0.0021,
      "step": 2062
    },
    {
      "epoch": 0.7548481522136846,
      "grad_norm": 8.236763954162598,
      "learning_rate": 3.1416174298375184e-05,
      "loss": 0.4028,
      "step": 2063
    },
    {
      "epoch": 0.7552140504939627,
      "grad_norm": 0.10688059777021408,
      "learning_rate": 3.14069423929099e-05,
      "loss": 0.0016,
      "step": 2064
    },
    {
      "epoch": 0.7555799487742407,
      "grad_norm": 0.09669294208288193,
      "learning_rate": 3.139771048744461e-05,
      "loss": 0.0027,
      "step": 2065
    },
    {
      "epoch": 0.7559458470545188,
      "grad_norm": 0.02026095613837242,
      "learning_rate": 3.138847858197932e-05,
      "loss": 0.0011,
      "step": 2066
    },
    {
      "epoch": 0.756311745334797,
      "grad_norm": 2.256866455078125,
      "learning_rate": 3.137924667651403e-05,
      "loss": 0.24,
      "step": 2067
    },
    {
      "epoch": 0.7566776436150751,
      "grad_norm": 3.959991693496704,
      "learning_rate": 3.137001477104874e-05,
      "loss": 0.3481,
      "step": 2068
    },
    {
      "epoch": 0.7570435418953531,
      "grad_norm": 0.029340527951717377,
      "learning_rate": 3.136078286558346e-05,
      "loss": 0.0013,
      "step": 2069
    },
    {
      "epoch": 0.7574094401756312,
      "grad_norm": 0.02157840132713318,
      "learning_rate": 3.135155096011817e-05,
      "loss": 0.0011,
      "step": 2070
    },
    {
      "epoch": 0.7577753384559093,
      "grad_norm": 4.754260540008545,
      "learning_rate": 3.134231905465288e-05,
      "loss": 0.2171,
      "step": 2071
    },
    {
      "epoch": 0.7581412367361874,
      "grad_norm": 0.7771803140640259,
      "learning_rate": 3.133308714918759e-05,
      "loss": 0.0084,
      "step": 2072
    },
    {
      "epoch": 0.7585071350164654,
      "grad_norm": 0.14015904068946838,
      "learning_rate": 3.13238552437223e-05,
      "loss": 0.0043,
      "step": 2073
    },
    {
      "epoch": 0.7588730332967435,
      "grad_norm": 0.030421659350395203,
      "learning_rate": 3.131462333825701e-05,
      "loss": 0.0015,
      "step": 2074
    },
    {
      "epoch": 0.7592389315770216,
      "grad_norm": 0.7462857961654663,
      "learning_rate": 3.130539143279173e-05,
      "loss": 0.0057,
      "step": 2075
    },
    {
      "epoch": 0.7596048298572997,
      "grad_norm": 2.2953689098358154,
      "learning_rate": 3.129615952732644e-05,
      "loss": 0.0178,
      "step": 2076
    },
    {
      "epoch": 0.7599707281375777,
      "grad_norm": 0.2710864841938019,
      "learning_rate": 3.128692762186116e-05,
      "loss": 0.0101,
      "step": 2077
    },
    {
      "epoch": 0.7603366264178558,
      "grad_norm": 0.08016245812177658,
      "learning_rate": 3.127769571639587e-05,
      "loss": 0.0027,
      "step": 2078
    },
    {
      "epoch": 0.7607025246981339,
      "grad_norm": 0.056032486259937286,
      "learning_rate": 3.126846381093057e-05,
      "loss": 0.0018,
      "step": 2079
    },
    {
      "epoch": 0.761068422978412,
      "grad_norm": 0.13119444251060486,
      "learning_rate": 3.125923190546529e-05,
      "loss": 0.0026,
      "step": 2080
    },
    {
      "epoch": 0.76143432125869,
      "grad_norm": 0.1795186549425125,
      "learning_rate": 3.125e-05,
      "loss": 0.0063,
      "step": 2081
    },
    {
      "epoch": 0.7618002195389681,
      "grad_norm": 7.817904949188232,
      "learning_rate": 3.124076809453471e-05,
      "loss": 0.2622,
      "step": 2082
    },
    {
      "epoch": 0.7621661178192463,
      "grad_norm": 4.258350849151611,
      "learning_rate": 3.123153618906943e-05,
      "loss": 0.3994,
      "step": 2083
    },
    {
      "epoch": 0.7625320160995244,
      "grad_norm": 3.422776460647583,
      "learning_rate": 3.122230428360414e-05,
      "loss": 0.3851,
      "step": 2084
    },
    {
      "epoch": 0.7628979143798024,
      "grad_norm": 0.10407117754220963,
      "learning_rate": 3.121307237813885e-05,
      "loss": 0.0033,
      "step": 2085
    },
    {
      "epoch": 0.7632638126600805,
      "grad_norm": 0.0904773622751236,
      "learning_rate": 3.120384047267356e-05,
      "loss": 0.0029,
      "step": 2086
    },
    {
      "epoch": 0.7636297109403586,
      "grad_norm": 0.07519008219242096,
      "learning_rate": 3.119460856720827e-05,
      "loss": 0.0028,
      "step": 2087
    },
    {
      "epoch": 0.7639956092206367,
      "grad_norm": 0.06580175459384918,
      "learning_rate": 3.118537666174299e-05,
      "loss": 0.0025,
      "step": 2088
    },
    {
      "epoch": 0.7643615075009147,
      "grad_norm": 9.909552574157715,
      "learning_rate": 3.11761447562777e-05,
      "loss": 0.0886,
      "step": 2089
    },
    {
      "epoch": 0.7647274057811928,
      "grad_norm": 10.293466567993164,
      "learning_rate": 3.116691285081241e-05,
      "loss": 0.2898,
      "step": 2090
    },
    {
      "epoch": 0.7650933040614709,
      "grad_norm": 0.21789832413196564,
      "learning_rate": 3.115768094534712e-05,
      "loss": 0.0093,
      "step": 2091
    },
    {
      "epoch": 0.765459202341749,
      "grad_norm": 3.2302498817443848,
      "learning_rate": 3.114844903988183e-05,
      "loss": 0.1566,
      "step": 2092
    },
    {
      "epoch": 0.765825100622027,
      "grad_norm": 0.27340808510780334,
      "learning_rate": 3.113921713441654e-05,
      "loss": 0.0135,
      "step": 2093
    },
    {
      "epoch": 0.7661909989023051,
      "grad_norm": 0.5071172118186951,
      "learning_rate": 3.112998522895126e-05,
      "loss": 0.0247,
      "step": 2094
    },
    {
      "epoch": 0.7665568971825832,
      "grad_norm": 8.741157531738281,
      "learning_rate": 3.112075332348597e-05,
      "loss": 0.0588,
      "step": 2095
    },
    {
      "epoch": 0.7669227954628614,
      "grad_norm": 0.11215847730636597,
      "learning_rate": 3.111152141802068e-05,
      "loss": 0.0047,
      "step": 2096
    },
    {
      "epoch": 0.7672886937431395,
      "grad_norm": 0.4617915153503418,
      "learning_rate": 3.1102289512555397e-05,
      "loss": 0.011,
      "step": 2097
    },
    {
      "epoch": 0.7676545920234175,
      "grad_norm": 0.10872840136289597,
      "learning_rate": 3.10930576070901e-05,
      "loss": 0.0055,
      "step": 2098
    },
    {
      "epoch": 0.7680204903036956,
      "grad_norm": 0.08579543232917786,
      "learning_rate": 3.108382570162482e-05,
      "loss": 0.0038,
      "step": 2099
    },
    {
      "epoch": 0.7683863885839737,
      "grad_norm": 0.2880311906337738,
      "learning_rate": 3.107459379615953e-05,
      "loss": 0.0061,
      "step": 2100
    },
    {
      "epoch": 0.7687522868642518,
      "grad_norm": 4.944327354431152,
      "learning_rate": 3.106536189069424e-05,
      "loss": 0.1779,
      "step": 2101
    },
    {
      "epoch": 0.7691181851445298,
      "grad_norm": 2.0201416015625,
      "learning_rate": 3.1056129985228956e-05,
      "loss": 0.0181,
      "step": 2102
    },
    {
      "epoch": 0.7694840834248079,
      "grad_norm": 7.267033576965332,
      "learning_rate": 3.1046898079763667e-05,
      "loss": 0.2111,
      "step": 2103
    },
    {
      "epoch": 0.769849981705086,
      "grad_norm": 7.8508453369140625,
      "learning_rate": 3.103766617429838e-05,
      "loss": 0.183,
      "step": 2104
    },
    {
      "epoch": 0.7702158799853641,
      "grad_norm": 0.11252445727586746,
      "learning_rate": 3.102843426883309e-05,
      "loss": 0.0027,
      "step": 2105
    },
    {
      "epoch": 0.7705817782656421,
      "grad_norm": 0.073197141289711,
      "learning_rate": 3.10192023633678e-05,
      "loss": 0.0027,
      "step": 2106
    },
    {
      "epoch": 0.7709476765459202,
      "grad_norm": 0.0930585265159607,
      "learning_rate": 3.1009970457902516e-05,
      "loss": 0.0025,
      "step": 2107
    },
    {
      "epoch": 0.7713135748261983,
      "grad_norm": 0.2466934323310852,
      "learning_rate": 3.1000738552437226e-05,
      "loss": 0.0036,
      "step": 2108
    },
    {
      "epoch": 0.7716794731064764,
      "grad_norm": 15.804637908935547,
      "learning_rate": 3.099150664697194e-05,
      "loss": 0.3485,
      "step": 2109
    },
    {
      "epoch": 0.7720453713867544,
      "grad_norm": 6.645871162414551,
      "learning_rate": 3.098227474150665e-05,
      "loss": 0.4648,
      "step": 2110
    },
    {
      "epoch": 0.7724112696670326,
      "grad_norm": 0.2447744458913803,
      "learning_rate": 3.097304283604136e-05,
      "loss": 0.0038,
      "step": 2111
    },
    {
      "epoch": 0.7727771679473107,
      "grad_norm": 2.877535104751587,
      "learning_rate": 3.096381093057607e-05,
      "loss": 0.0135,
      "step": 2112
    },
    {
      "epoch": 0.7731430662275888,
      "grad_norm": 0.02302655763924122,
      "learning_rate": 3.0954579025110786e-05,
      "loss": 0.001,
      "step": 2113
    },
    {
      "epoch": 0.7735089645078668,
      "grad_norm": 12.3150634765625,
      "learning_rate": 3.0945347119645496e-05,
      "loss": 0.1492,
      "step": 2114
    },
    {
      "epoch": 0.7738748627881449,
      "grad_norm": 0.22656311094760895,
      "learning_rate": 3.093611521418021e-05,
      "loss": 0.0033,
      "step": 2115
    },
    {
      "epoch": 0.774240761068423,
      "grad_norm": 12.56922435760498,
      "learning_rate": 3.0926883308714924e-05,
      "loss": 0.0541,
      "step": 2116
    },
    {
      "epoch": 0.7746066593487011,
      "grad_norm": 0.08882946521043777,
      "learning_rate": 3.091765140324963e-05,
      "loss": 0.003,
      "step": 2117
    },
    {
      "epoch": 0.7749725576289791,
      "grad_norm": 1.1988260746002197,
      "learning_rate": 3.0908419497784345e-05,
      "loss": 0.0097,
      "step": 2118
    },
    {
      "epoch": 0.7753384559092572,
      "grad_norm": 0.2181743085384369,
      "learning_rate": 3.0899187592319056e-05,
      "loss": 0.0025,
      "step": 2119
    },
    {
      "epoch": 0.7757043541895353,
      "grad_norm": 0.07875942438840866,
      "learning_rate": 3.0889955686853766e-05,
      "loss": 0.0027,
      "step": 2120
    },
    {
      "epoch": 0.7760702524698134,
      "grad_norm": 0.08846661448478699,
      "learning_rate": 3.0880723781388484e-05,
      "loss": 0.002,
      "step": 2121
    },
    {
      "epoch": 0.7764361507500914,
      "grad_norm": 0.026014087721705437,
      "learning_rate": 3.0871491875923194e-05,
      "loss": 0.001,
      "step": 2122
    },
    {
      "epoch": 0.7768020490303695,
      "grad_norm": 0.025688141584396362,
      "learning_rate": 3.0862259970457905e-05,
      "loss": 0.0009,
      "step": 2123
    },
    {
      "epoch": 0.7771679473106476,
      "grad_norm": 0.06893941760063171,
      "learning_rate": 3.0853028064992615e-05,
      "loss": 0.0024,
      "step": 2124
    },
    {
      "epoch": 0.7775338455909258,
      "grad_norm": 0.09416305273771286,
      "learning_rate": 3.0843796159527326e-05,
      "loss": 0.0016,
      "step": 2125
    },
    {
      "epoch": 0.7778997438712038,
      "grad_norm": 0.017492668703198433,
      "learning_rate": 3.0834564254062036e-05,
      "loss": 0.0009,
      "step": 2126
    },
    {
      "epoch": 0.7782656421514819,
      "grad_norm": 0.19894978404045105,
      "learning_rate": 3.0825332348596754e-05,
      "loss": 0.0043,
      "step": 2127
    },
    {
      "epoch": 0.77863154043176,
      "grad_norm": 3.1517860889434814,
      "learning_rate": 3.0816100443131464e-05,
      "loss": 0.1958,
      "step": 2128
    },
    {
      "epoch": 0.7789974387120381,
      "grad_norm": 5.893014430999756,
      "learning_rate": 3.0806868537666175e-05,
      "loss": 0.2161,
      "step": 2129
    },
    {
      "epoch": 0.7793633369923162,
      "grad_norm": 7.040788650512695,
      "learning_rate": 3.0797636632200885e-05,
      "loss": 0.2841,
      "step": 2130
    },
    {
      "epoch": 0.7797292352725942,
      "grad_norm": 0.08413686603307724,
      "learning_rate": 3.0788404726735596e-05,
      "loss": 0.0015,
      "step": 2131
    },
    {
      "epoch": 0.7800951335528723,
      "grad_norm": 0.09119369834661484,
      "learning_rate": 3.077917282127031e-05,
      "loss": 0.0026,
      "step": 2132
    },
    {
      "epoch": 0.7804610318331504,
      "grad_norm": 0.0782436728477478,
      "learning_rate": 3.0769940915805024e-05,
      "loss": 0.0011,
      "step": 2133
    },
    {
      "epoch": 0.7808269301134285,
      "grad_norm": 0.019850581884384155,
      "learning_rate": 3.0760709010339734e-05,
      "loss": 0.0008,
      "step": 2134
    },
    {
      "epoch": 0.7811928283937065,
      "grad_norm": 0.26323580741882324,
      "learning_rate": 3.075147710487445e-05,
      "loss": 0.0033,
      "step": 2135
    },
    {
      "epoch": 0.7815587266739846,
      "grad_norm": 0.0165097638964653,
      "learning_rate": 3.0742245199409155e-05,
      "loss": 0.0006,
      "step": 2136
    },
    {
      "epoch": 0.7819246249542627,
      "grad_norm": 0.30734965205192566,
      "learning_rate": 3.073301329394387e-05,
      "loss": 0.0069,
      "step": 2137
    },
    {
      "epoch": 0.7822905232345408,
      "grad_norm": 1.7070817947387695,
      "learning_rate": 3.072378138847858e-05,
      "loss": 0.0166,
      "step": 2138
    },
    {
      "epoch": 0.7826564215148188,
      "grad_norm": 0.019137656316161156,
      "learning_rate": 3.0714549483013294e-05,
      "loss": 0.0006,
      "step": 2139
    },
    {
      "epoch": 0.783022319795097,
      "grad_norm": 0.014770661480724812,
      "learning_rate": 3.070531757754801e-05,
      "loss": 0.0008,
      "step": 2140
    },
    {
      "epoch": 0.7833882180753751,
      "grad_norm": 0.15452979505062103,
      "learning_rate": 3.069608567208272e-05,
      "loss": 0.003,
      "step": 2141
    },
    {
      "epoch": 0.7837541163556532,
      "grad_norm": 0.034331463277339935,
      "learning_rate": 3.068685376661743e-05,
      "loss": 0.0009,
      "step": 2142
    },
    {
      "epoch": 0.7841200146359312,
      "grad_norm": 0.16502799093723297,
      "learning_rate": 3.067762186115214e-05,
      "loss": 0.0044,
      "step": 2143
    },
    {
      "epoch": 0.7844859129162093,
      "grad_norm": 8.83232593536377,
      "learning_rate": 3.066838995568685e-05,
      "loss": 0.132,
      "step": 2144
    },
    {
      "epoch": 0.7848518111964874,
      "grad_norm": 0.01158144511282444,
      "learning_rate": 3.0659158050221564e-05,
      "loss": 0.0006,
      "step": 2145
    },
    {
      "epoch": 0.7852177094767655,
      "grad_norm": 0.1256393939256668,
      "learning_rate": 3.064992614475628e-05,
      "loss": 0.0024,
      "step": 2146
    },
    {
      "epoch": 0.7855836077570435,
      "grad_norm": 0.8799862265586853,
      "learning_rate": 3.064069423929099e-05,
      "loss": 0.0127,
      "step": 2147
    },
    {
      "epoch": 0.7859495060373216,
      "grad_norm": 10.641087532043457,
      "learning_rate": 3.06314623338257e-05,
      "loss": 0.2832,
      "step": 2148
    },
    {
      "epoch": 0.7863154043175997,
      "grad_norm": 0.013968144543468952,
      "learning_rate": 3.062223042836041e-05,
      "loss": 0.0007,
      "step": 2149
    },
    {
      "epoch": 0.7866813025978778,
      "grad_norm": 2.6053402423858643,
      "learning_rate": 3.0612998522895123e-05,
      "loss": 0.0156,
      "step": 2150
    },
    {
      "epoch": 0.7870472008781558,
      "grad_norm": 0.0665549784898758,
      "learning_rate": 3.060376661742984e-05,
      "loss": 0.0016,
      "step": 2151
    },
    {
      "epoch": 0.7874130991584339,
      "grad_norm": 0.10209513455629349,
      "learning_rate": 3.059453471196455e-05,
      "loss": 0.0016,
      "step": 2152
    },
    {
      "epoch": 0.787778997438712,
      "grad_norm": 0.7218941450119019,
      "learning_rate": 3.058530280649926e-05,
      "loss": 0.0095,
      "step": 2153
    },
    {
      "epoch": 0.7881448957189902,
      "grad_norm": 3.6963043212890625,
      "learning_rate": 3.057607090103398e-05,
      "loss": 0.1939,
      "step": 2154
    },
    {
      "epoch": 0.7885107939992682,
      "grad_norm": 0.11730324476957321,
      "learning_rate": 3.056683899556868e-05,
      "loss": 0.0015,
      "step": 2155
    },
    {
      "epoch": 0.7888766922795463,
      "grad_norm": 0.2950694262981415,
      "learning_rate": 3.0557607090103393e-05,
      "loss": 0.0036,
      "step": 2156
    },
    {
      "epoch": 0.7892425905598244,
      "grad_norm": 0.008954416960477829,
      "learning_rate": 3.054837518463811e-05,
      "loss": 0.0004,
      "step": 2157
    },
    {
      "epoch": 0.7896084888401025,
      "grad_norm": 16.44771957397461,
      "learning_rate": 3.053914327917282e-05,
      "loss": 0.1209,
      "step": 2158
    },
    {
      "epoch": 0.7899743871203805,
      "grad_norm": 0.03964805230498314,
      "learning_rate": 3.052991137370754e-05,
      "loss": 0.0007,
      "step": 2159
    },
    {
      "epoch": 0.7903402854006586,
      "grad_norm": 0.05378180369734764,
      "learning_rate": 3.052067946824225e-05,
      "loss": 0.0013,
      "step": 2160
    },
    {
      "epoch": 0.7907061836809367,
      "grad_norm": 0.1115657165646553,
      "learning_rate": 3.0511447562776956e-05,
      "loss": 0.002,
      "step": 2161
    },
    {
      "epoch": 0.7910720819612148,
      "grad_norm": 0.10567750781774521,
      "learning_rate": 3.0502215657311674e-05,
      "loss": 0.0018,
      "step": 2162
    },
    {
      "epoch": 0.7914379802414928,
      "grad_norm": 0.14717160165309906,
      "learning_rate": 3.049298375184638e-05,
      "loss": 0.0019,
      "step": 2163
    },
    {
      "epoch": 0.7918038785217709,
      "grad_norm": 0.01605193130671978,
      "learning_rate": 3.048375184638109e-05,
      "loss": 0.0004,
      "step": 2164
    },
    {
      "epoch": 0.792169776802049,
      "grad_norm": 0.07372279465198517,
      "learning_rate": 3.047451994091581e-05,
      "loss": 0.0014,
      "step": 2165
    },
    {
      "epoch": 0.7925356750823271,
      "grad_norm": 0.009821723215281963,
      "learning_rate": 3.046528803545052e-05,
      "loss": 0.0004,
      "step": 2166
    },
    {
      "epoch": 0.7929015733626052,
      "grad_norm": 0.11772026866674423,
      "learning_rate": 3.0456056129985233e-05,
      "loss": 0.0027,
      "step": 2167
    },
    {
      "epoch": 0.7932674716428832,
      "grad_norm": 0.1515015959739685,
      "learning_rate": 3.0446824224519944e-05,
      "loss": 0.0024,
      "step": 2168
    },
    {
      "epoch": 0.7936333699231614,
      "grad_norm": 0.01493052951991558,
      "learning_rate": 3.0437592319054654e-05,
      "loss": 0.0007,
      "step": 2169
    },
    {
      "epoch": 0.7939992682034395,
      "grad_norm": 0.25747740268707275,
      "learning_rate": 3.0428360413589368e-05,
      "loss": 0.002,
      "step": 2170
    },
    {
      "epoch": 0.7943651664837176,
      "grad_norm": 0.009936428628861904,
      "learning_rate": 3.041912850812408e-05,
      "loss": 0.0005,
      "step": 2171
    },
    {
      "epoch": 0.7947310647639956,
      "grad_norm": 0.013525445014238358,
      "learning_rate": 3.040989660265879e-05,
      "loss": 0.0006,
      "step": 2172
    },
    {
      "epoch": 0.7950969630442737,
      "grad_norm": 0.07357588410377502,
      "learning_rate": 3.0400664697193503e-05,
      "loss": 0.0007,
      "step": 2173
    },
    {
      "epoch": 0.7954628613245518,
      "grad_norm": 25.10590171813965,
      "learning_rate": 3.0391432791728214e-05,
      "loss": 0.0578,
      "step": 2174
    },
    {
      "epoch": 0.7958287596048299,
      "grad_norm": 0.012385008856654167,
      "learning_rate": 3.0382200886262924e-05,
      "loss": 0.0005,
      "step": 2175
    },
    {
      "epoch": 0.7961946578851079,
      "grad_norm": 33.30012893676758,
      "learning_rate": 3.037296898079764e-05,
      "loss": 0.1055,
      "step": 2176
    },
    {
      "epoch": 0.796560556165386,
      "grad_norm": 0.09485034644603729,
      "learning_rate": 3.036373707533235e-05,
      "loss": 0.0016,
      "step": 2177
    },
    {
      "epoch": 0.7969264544456641,
      "grad_norm": 0.028018590062856674,
      "learning_rate": 3.0354505169867066e-05,
      "loss": 0.0007,
      "step": 2178
    },
    {
      "epoch": 0.7972923527259422,
      "grad_norm": 0.00980694591999054,
      "learning_rate": 3.0345273264401773e-05,
      "loss": 0.0004,
      "step": 2179
    },
    {
      "epoch": 0.7976582510062202,
      "grad_norm": 0.05808039382100105,
      "learning_rate": 3.0336041358936484e-05,
      "loss": 0.0012,
      "step": 2180
    },
    {
      "epoch": 0.7980241492864983,
      "grad_norm": 0.010728336870670319,
      "learning_rate": 3.03268094534712e-05,
      "loss": 0.0005,
      "step": 2181
    },
    {
      "epoch": 0.7983900475667765,
      "grad_norm": 0.2913774251937866,
      "learning_rate": 3.031757754800591e-05,
      "loss": 0.0044,
      "step": 2182
    },
    {
      "epoch": 0.7987559458470546,
      "grad_norm": 6.825941562652588,
      "learning_rate": 3.030834564254062e-05,
      "loss": 0.1327,
      "step": 2183
    },
    {
      "epoch": 0.7991218441273326,
      "grad_norm": 19.08030128479004,
      "learning_rate": 3.0299113737075336e-05,
      "loss": 0.1581,
      "step": 2184
    },
    {
      "epoch": 0.7994877424076107,
      "grad_norm": 0.008691324852406979,
      "learning_rate": 3.0289881831610047e-05,
      "loss": 0.0004,
      "step": 2185
    },
    {
      "epoch": 0.7998536406878888,
      "grad_norm": 0.01127107348293066,
      "learning_rate": 3.0280649926144754e-05,
      "loss": 0.0005,
      "step": 2186
    },
    {
      "epoch": 0.8002195389681669,
      "grad_norm": 0.007250374183058739,
      "learning_rate": 3.027141802067947e-05,
      "loss": 0.0004,
      "step": 2187
    },
    {
      "epoch": 0.8005854372484449,
      "grad_norm": 2.153625726699829,
      "learning_rate": 3.0262186115214182e-05,
      "loss": 0.5435,
      "step": 2188
    },
    {
      "epoch": 0.800951335528723,
      "grad_norm": 14.320625305175781,
      "learning_rate": 3.0252954209748896e-05,
      "loss": 0.1624,
      "step": 2189
    },
    {
      "epoch": 0.8013172338090011,
      "grad_norm": 0.009944872930645943,
      "learning_rate": 3.0243722304283606e-05,
      "loss": 0.0004,
      "step": 2190
    },
    {
      "epoch": 0.8016831320892792,
      "grad_norm": 0.01052294671535492,
      "learning_rate": 3.0234490398818317e-05,
      "loss": 0.0005,
      "step": 2191
    },
    {
      "epoch": 0.8020490303695572,
      "grad_norm": 0.011454523541033268,
      "learning_rate": 3.022525849335303e-05,
      "loss": 0.0006,
      "step": 2192
    },
    {
      "epoch": 0.8024149286498353,
      "grad_norm": 13.805289268493652,
      "learning_rate": 3.021602658788774e-05,
      "loss": 0.1443,
      "step": 2193
    },
    {
      "epoch": 0.8027808269301134,
      "grad_norm": 0.014970376156270504,
      "learning_rate": 3.0206794682422452e-05,
      "loss": 0.0007,
      "step": 2194
    },
    {
      "epoch": 0.8031467252103915,
      "grad_norm": 1.964624285697937,
      "learning_rate": 3.0197562776957166e-05,
      "loss": 0.0099,
      "step": 2195
    },
    {
      "epoch": 0.8035126234906695,
      "grad_norm": 0.021007323637604713,
      "learning_rate": 3.0188330871491876e-05,
      "loss": 0.0008,
      "step": 2196
    },
    {
      "epoch": 0.8038785217709477,
      "grad_norm": 0.02988172322511673,
      "learning_rate": 3.0179098966026587e-05,
      "loss": 0.0007,
      "step": 2197
    },
    {
      "epoch": 0.8042444200512258,
      "grad_norm": 0.04754714295268059,
      "learning_rate": 3.01698670605613e-05,
      "loss": 0.0008,
      "step": 2198
    },
    {
      "epoch": 0.8046103183315039,
      "grad_norm": 0.05278211086988449,
      "learning_rate": 3.016063515509601e-05,
      "loss": 0.0013,
      "step": 2199
    },
    {
      "epoch": 0.8049762166117819,
      "grad_norm": 0.06615059822797775,
      "learning_rate": 3.015140324963073e-05,
      "loss": 0.0011,
      "step": 2200
    },
    {
      "epoch": 0.80534211489206,
      "grad_norm": 8.547704696655273,
      "learning_rate": 3.0142171344165436e-05,
      "loss": 0.1039,
      "step": 2201
    },
    {
      "epoch": 0.8057080131723381,
      "grad_norm": 0.08662967383861542,
      "learning_rate": 3.0132939438700146e-05,
      "loss": 0.0017,
      "step": 2202
    },
    {
      "epoch": 0.8060739114526162,
      "grad_norm": 0.013335993513464928,
      "learning_rate": 3.0123707533234864e-05,
      "loss": 0.0006,
      "step": 2203
    },
    {
      "epoch": 0.8064398097328943,
      "grad_norm": 13.060247421264648,
      "learning_rate": 3.0114475627769574e-05,
      "loss": 0.3474,
      "step": 2204
    },
    {
      "epoch": 0.8068057080131723,
      "grad_norm": 0.016270864754915237,
      "learning_rate": 3.010524372230428e-05,
      "loss": 0.0008,
      "step": 2205
    },
    {
      "epoch": 0.8071716062934504,
      "grad_norm": 7.9611687660217285,
      "learning_rate": 3.0096011816839e-05,
      "loss": 0.436,
      "step": 2206
    },
    {
      "epoch": 0.8075375045737285,
      "grad_norm": 0.020304502919316292,
      "learning_rate": 3.008677991137371e-05,
      "loss": 0.0009,
      "step": 2207
    },
    {
      "epoch": 0.8079034028540066,
      "grad_norm": 0.8861445188522339,
      "learning_rate": 3.0077548005908423e-05,
      "loss": 0.0049,
      "step": 2208
    },
    {
      "epoch": 0.8082693011342846,
      "grad_norm": 8.580401420593262,
      "learning_rate": 3.0068316100443134e-05,
      "loss": 0.1541,
      "step": 2209
    },
    {
      "epoch": 0.8086351994145627,
      "grad_norm": 4.0436553955078125,
      "learning_rate": 3.0059084194977844e-05,
      "loss": 0.021,
      "step": 2210
    },
    {
      "epoch": 0.8090010976948409,
      "grad_norm": 0.12871721386909485,
      "learning_rate": 3.004985228951256e-05,
      "loss": 0.0022,
      "step": 2211
    },
    {
      "epoch": 0.809366995975119,
      "grad_norm": 6.060528755187988,
      "learning_rate": 3.004062038404727e-05,
      "loss": 0.0217,
      "step": 2212
    },
    {
      "epoch": 0.809732894255397,
      "grad_norm": 0.12236885726451874,
      "learning_rate": 3.003138847858198e-05,
      "loss": 0.0022,
      "step": 2213
    },
    {
      "epoch": 0.8100987925356751,
      "grad_norm": 1.7250896692276,
      "learning_rate": 3.0022156573116693e-05,
      "loss": 0.0084,
      "step": 2214
    },
    {
      "epoch": 0.8104646908159532,
      "grad_norm": 0.1835588812828064,
      "learning_rate": 3.0012924667651404e-05,
      "loss": 0.0025,
      "step": 2215
    },
    {
      "epoch": 0.8108305890962313,
      "grad_norm": 0.037250131368637085,
      "learning_rate": 3.0003692762186115e-05,
      "loss": 0.0009,
      "step": 2216
    },
    {
      "epoch": 0.8111964873765093,
      "grad_norm": 0.7085587382316589,
      "learning_rate": 2.999446085672083e-05,
      "loss": 0.0074,
      "step": 2217
    },
    {
      "epoch": 0.8115623856567874,
      "grad_norm": 0.17334501445293427,
      "learning_rate": 2.998522895125554e-05,
      "loss": 0.0025,
      "step": 2218
    },
    {
      "epoch": 0.8119282839370655,
      "grad_norm": 6.599008083343506,
      "learning_rate": 2.9975997045790256e-05,
      "loss": 0.0344,
      "step": 2219
    },
    {
      "epoch": 0.8122941822173436,
      "grad_norm": 0.09909096360206604,
      "learning_rate": 2.9966765140324963e-05,
      "loss": 0.0018,
      "step": 2220
    },
    {
      "epoch": 0.8126600804976216,
      "grad_norm": 4.9623188972473145,
      "learning_rate": 2.9957533234859674e-05,
      "loss": 0.3822,
      "step": 2221
    },
    {
      "epoch": 0.8130259787778997,
      "grad_norm": 0.13683900237083435,
      "learning_rate": 2.994830132939439e-05,
      "loss": 0.0025,
      "step": 2222
    },
    {
      "epoch": 0.8133918770581778,
      "grad_norm": 0.11967572569847107,
      "learning_rate": 2.9939069423929102e-05,
      "loss": 0.0019,
      "step": 2223
    },
    {
      "epoch": 0.813757775338456,
      "grad_norm": 0.17720085382461548,
      "learning_rate": 2.992983751846381e-05,
      "loss": 0.0037,
      "step": 2224
    },
    {
      "epoch": 0.814123673618734,
      "grad_norm": 0.6193429827690125,
      "learning_rate": 2.9920605612998526e-05,
      "loss": 0.0053,
      "step": 2225
    },
    {
      "epoch": 0.814489571899012,
      "grad_norm": 8.979415893554688,
      "learning_rate": 2.9911373707533237e-05,
      "loss": 0.0214,
      "step": 2226
    },
    {
      "epoch": 0.8148554701792902,
      "grad_norm": 0.016746053472161293,
      "learning_rate": 2.9902141802067944e-05,
      "loss": 0.0008,
      "step": 2227
    },
    {
      "epoch": 0.8152213684595683,
      "grad_norm": 0.01985037326812744,
      "learning_rate": 2.989290989660266e-05,
      "loss": 0.0009,
      "step": 2228
    },
    {
      "epoch": 0.8155872667398463,
      "grad_norm": 0.014459765516221523,
      "learning_rate": 2.9883677991137372e-05,
      "loss": 0.0007,
      "step": 2229
    },
    {
      "epoch": 0.8159531650201244,
      "grad_norm": 0.07507485896348953,
      "learning_rate": 2.9874446085672086e-05,
      "loss": 0.0018,
      "step": 2230
    },
    {
      "epoch": 0.8163190633004025,
      "grad_norm": 0.015374776907265186,
      "learning_rate": 2.9865214180206796e-05,
      "loss": 0.0008,
      "step": 2231
    },
    {
      "epoch": 0.8166849615806806,
      "grad_norm": 0.038925912231206894,
      "learning_rate": 2.9855982274741507e-05,
      "loss": 0.0012,
      "step": 2232
    },
    {
      "epoch": 0.8170508598609586,
      "grad_norm": 0.09271692484617233,
      "learning_rate": 2.984675036927622e-05,
      "loss": 0.002,
      "step": 2233
    },
    {
      "epoch": 0.8174167581412367,
      "grad_norm": 0.054618388414382935,
      "learning_rate": 2.983751846381093e-05,
      "loss": 0.0011,
      "step": 2234
    },
    {
      "epoch": 0.8177826564215148,
      "grad_norm": 0.018944522365927696,
      "learning_rate": 2.9828286558345642e-05,
      "loss": 0.0007,
      "step": 2235
    },
    {
      "epoch": 0.8181485547017929,
      "grad_norm": 0.02522527240216732,
      "learning_rate": 2.9819054652880356e-05,
      "loss": 0.0007,
      "step": 2236
    },
    {
      "epoch": 0.818514452982071,
      "grad_norm": 0.046465497463941574,
      "learning_rate": 2.9809822747415067e-05,
      "loss": 0.0012,
      "step": 2237
    },
    {
      "epoch": 0.818880351262349,
      "grad_norm": 0.3906405568122864,
      "learning_rate": 2.9800590841949784e-05,
      "loss": 0.0028,
      "step": 2238
    },
    {
      "epoch": 0.8192462495426271,
      "grad_norm": 0.011170784942805767,
      "learning_rate": 2.979135893648449e-05,
      "loss": 0.0005,
      "step": 2239
    },
    {
      "epoch": 0.8196121478229053,
      "grad_norm": 0.019397318363189697,
      "learning_rate": 2.97821270310192e-05,
      "loss": 0.0008,
      "step": 2240
    },
    {
      "epoch": 0.8199780461031834,
      "grad_norm": 0.0406508594751358,
      "learning_rate": 2.977289512555392e-05,
      "loss": 0.0013,
      "step": 2241
    },
    {
      "epoch": 0.8203439443834614,
      "grad_norm": 12.436427116394043,
      "learning_rate": 2.976366322008863e-05,
      "loss": 0.1158,
      "step": 2242
    },
    {
      "epoch": 0.8207098426637395,
      "grad_norm": 33.44866180419922,
      "learning_rate": 2.9754431314623337e-05,
      "loss": 0.0399,
      "step": 2243
    },
    {
      "epoch": 0.8210757409440176,
      "grad_norm": 0.01845637522637844,
      "learning_rate": 2.9745199409158054e-05,
      "loss": 0.0008,
      "step": 2244
    },
    {
      "epoch": 0.8214416392242957,
      "grad_norm": 0.008441337384283543,
      "learning_rate": 2.9735967503692764e-05,
      "loss": 0.0004,
      "step": 2245
    },
    {
      "epoch": 0.8218075375045737,
      "grad_norm": 0.02250954881310463,
      "learning_rate": 2.972673559822747e-05,
      "loss": 0.0008,
      "step": 2246
    },
    {
      "epoch": 0.8221734357848518,
      "grad_norm": 0.06148984283208847,
      "learning_rate": 2.971750369276219e-05,
      "loss": 0.0014,
      "step": 2247
    },
    {
      "epoch": 0.8225393340651299,
      "grad_norm": 4.941781997680664,
      "learning_rate": 2.97082717872969e-05,
      "loss": 0.0114,
      "step": 2248
    },
    {
      "epoch": 0.822905232345408,
      "grad_norm": 0.015038466081023216,
      "learning_rate": 2.9699039881831613e-05,
      "loss": 0.0007,
      "step": 2249
    },
    {
      "epoch": 0.823271130625686,
      "grad_norm": 0.018513519316911697,
      "learning_rate": 2.9689807976366324e-05,
      "loss": 0.0007,
      "step": 2250
    },
    {
      "epoch": 0.8236370289059641,
      "grad_norm": 10.553889274597168,
      "learning_rate": 2.9680576070901035e-05,
      "loss": 0.1159,
      "step": 2251
    },
    {
      "epoch": 0.8240029271862422,
      "grad_norm": 0.022073732689023018,
      "learning_rate": 2.967134416543575e-05,
      "loss": 0.0008,
      "step": 2252
    },
    {
      "epoch": 0.8243688254665203,
      "grad_norm": 0.01764705590903759,
      "learning_rate": 2.966211225997046e-05,
      "loss": 0.0007,
      "step": 2253
    },
    {
      "epoch": 0.8247347237467983,
      "grad_norm": 0.01258089393377304,
      "learning_rate": 2.965288035450517e-05,
      "loss": 0.0006,
      "step": 2254
    },
    {
      "epoch": 0.8251006220270765,
      "grad_norm": 3.772271156311035,
      "learning_rate": 2.9643648449039884e-05,
      "loss": 0.0094,
      "step": 2255
    },
    {
      "epoch": 0.8254665203073546,
      "grad_norm": 0.18835732340812683,
      "learning_rate": 2.9634416543574594e-05,
      "loss": 0.0018,
      "step": 2256
    },
    {
      "epoch": 0.8258324185876327,
      "grad_norm": 0.03484392911195755,
      "learning_rate": 2.9625184638109305e-05,
      "loss": 0.0007,
      "step": 2257
    },
    {
      "epoch": 0.8261983168679107,
      "grad_norm": 0.10869388282299042,
      "learning_rate": 2.961595273264402e-05,
      "loss": 0.0018,
      "step": 2258
    },
    {
      "epoch": 0.8265642151481888,
      "grad_norm": 4.178879737854004,
      "learning_rate": 2.960672082717873e-05,
      "loss": 0.4233,
      "step": 2259
    },
    {
      "epoch": 0.8269301134284669,
      "grad_norm": 20.41912078857422,
      "learning_rate": 2.9597488921713446e-05,
      "loss": 0.5359,
      "step": 2260
    },
    {
      "epoch": 0.827296011708745,
      "grad_norm": 0.011897003278136253,
      "learning_rate": 2.9588257016248157e-05,
      "loss": 0.0005,
      "step": 2261
    },
    {
      "epoch": 0.827661909989023,
      "grad_norm": 5.7722930908203125,
      "learning_rate": 2.9579025110782864e-05,
      "loss": 0.3614,
      "step": 2262
    },
    {
      "epoch": 0.8280278082693011,
      "grad_norm": 0.03792958706617355,
      "learning_rate": 2.956979320531758e-05,
      "loss": 0.001,
      "step": 2263
    },
    {
      "epoch": 0.8283937065495792,
      "grad_norm": 0.011072899214923382,
      "learning_rate": 2.9560561299852292e-05,
      "loss": 0.0006,
      "step": 2264
    },
    {
      "epoch": 0.8287596048298573,
      "grad_norm": 0.04971572384238243,
      "learning_rate": 2.9551329394387e-05,
      "loss": 0.0013,
      "step": 2265
    },
    {
      "epoch": 0.8291255031101353,
      "grad_norm": 12.903741836547852,
      "learning_rate": 2.9542097488921716e-05,
      "loss": 0.1709,
      "step": 2266
    },
    {
      "epoch": 0.8294914013904134,
      "grad_norm": 15.852808952331543,
      "learning_rate": 2.9532865583456427e-05,
      "loss": 0.0795,
      "step": 2267
    },
    {
      "epoch": 0.8298572996706916,
      "grad_norm": 1.4390485286712646,
      "learning_rate": 2.952363367799114e-05,
      "loss": 0.0124,
      "step": 2268
    },
    {
      "epoch": 0.8302231979509697,
      "grad_norm": 0.18045000731945038,
      "learning_rate": 2.951440177252585e-05,
      "loss": 0.0038,
      "step": 2269
    },
    {
      "epoch": 0.8305890962312477,
      "grad_norm": 3.0153937339782715,
      "learning_rate": 2.9505169867060562e-05,
      "loss": 0.2324,
      "step": 2270
    },
    {
      "epoch": 0.8309549945115258,
      "grad_norm": 0.35312244296073914,
      "learning_rate": 2.9495937961595276e-05,
      "loss": 0.0055,
      "step": 2271
    },
    {
      "epoch": 0.8313208927918039,
      "grad_norm": 0.043996118009090424,
      "learning_rate": 2.9486706056129987e-05,
      "loss": 0.0015,
      "step": 2272
    },
    {
      "epoch": 0.831686791072082,
      "grad_norm": 10.45129680633545,
      "learning_rate": 2.9477474150664697e-05,
      "loss": 0.1001,
      "step": 2273
    },
    {
      "epoch": 0.8320526893523601,
      "grad_norm": 0.06293060630559921,
      "learning_rate": 2.946824224519941e-05,
      "loss": 0.0024,
      "step": 2274
    },
    {
      "epoch": 0.8324185876326381,
      "grad_norm": 11.971413612365723,
      "learning_rate": 2.945901033973412e-05,
      "loss": 0.1488,
      "step": 2275
    },
    {
      "epoch": 0.8327844859129162,
      "grad_norm": 0.018979081884026527,
      "learning_rate": 2.9449778434268832e-05,
      "loss": 0.0008,
      "step": 2276
    },
    {
      "epoch": 0.8331503841931943,
      "grad_norm": 0.03411589190363884,
      "learning_rate": 2.9440546528803546e-05,
      "loss": 0.0012,
      "step": 2277
    },
    {
      "epoch": 0.8335162824734724,
      "grad_norm": 0.03995240852236748,
      "learning_rate": 2.9431314623338257e-05,
      "loss": 0.0015,
      "step": 2278
    },
    {
      "epoch": 0.8338821807537504,
      "grad_norm": 0.07735110074281693,
      "learning_rate": 2.9422082717872974e-05,
      "loss": 0.0022,
      "step": 2279
    },
    {
      "epoch": 0.8342480790340285,
      "grad_norm": 9.519315719604492,
      "learning_rate": 2.9412850812407685e-05,
      "loss": 0.1836,
      "step": 2280
    },
    {
      "epoch": 0.8346139773143066,
      "grad_norm": 0.14812444150447845,
      "learning_rate": 2.940361890694239e-05,
      "loss": 0.002,
      "step": 2281
    },
    {
      "epoch": 0.8349798755945848,
      "grad_norm": 0.0641964003443718,
      "learning_rate": 2.939438700147711e-05,
      "loss": 0.0019,
      "step": 2282
    },
    {
      "epoch": 0.8353457738748628,
      "grad_norm": 0.36481422185897827,
      "learning_rate": 2.938515509601182e-05,
      "loss": 0.0049,
      "step": 2283
    },
    {
      "epoch": 0.8357116721551409,
      "grad_norm": 0.017011608928442,
      "learning_rate": 2.9375923190546527e-05,
      "loss": 0.0008,
      "step": 2284
    },
    {
      "epoch": 0.836077570435419,
      "grad_norm": 0.1494755893945694,
      "learning_rate": 2.9366691285081244e-05,
      "loss": 0.0025,
      "step": 2285
    },
    {
      "epoch": 0.8364434687156971,
      "grad_norm": 0.05257638171315193,
      "learning_rate": 2.9357459379615955e-05,
      "loss": 0.0015,
      "step": 2286
    },
    {
      "epoch": 0.8368093669959751,
      "grad_norm": 0.1205996498465538,
      "learning_rate": 2.9348227474150665e-05,
      "loss": 0.0026,
      "step": 2287
    },
    {
      "epoch": 0.8371752652762532,
      "grad_norm": 0.14593611657619476,
      "learning_rate": 2.933899556868538e-05,
      "loss": 0.004,
      "step": 2288
    },
    {
      "epoch": 0.8375411635565313,
      "grad_norm": 13.34325122833252,
      "learning_rate": 2.932976366322009e-05,
      "loss": 0.1457,
      "step": 2289
    },
    {
      "epoch": 0.8379070618368094,
      "grad_norm": 3.3368616104125977,
      "learning_rate": 2.9320531757754804e-05,
      "loss": 0.2326,
      "step": 2290
    },
    {
      "epoch": 0.8382729601170874,
      "grad_norm": 0.05741732567548752,
      "learning_rate": 2.9311299852289514e-05,
      "loss": 0.002,
      "step": 2291
    },
    {
      "epoch": 0.8386388583973655,
      "grad_norm": 0.08018413931131363,
      "learning_rate": 2.9302067946824225e-05,
      "loss": 0.0022,
      "step": 2292
    },
    {
      "epoch": 0.8390047566776436,
      "grad_norm": 7.298826217651367,
      "learning_rate": 2.929283604135894e-05,
      "loss": 0.0326,
      "step": 2293
    },
    {
      "epoch": 0.8393706549579217,
      "grad_norm": 0.013412373140454292,
      "learning_rate": 2.928360413589365e-05,
      "loss": 0.0006,
      "step": 2294
    },
    {
      "epoch": 0.8397365532381997,
      "grad_norm": 0.032284364104270935,
      "learning_rate": 2.927437223042836e-05,
      "loss": 0.0011,
      "step": 2295
    },
    {
      "epoch": 0.8401024515184778,
      "grad_norm": 13.905638694763184,
      "learning_rate": 2.9265140324963074e-05,
      "loss": 0.2,
      "step": 2296
    },
    {
      "epoch": 0.840468349798756,
      "grad_norm": 0.18865230679512024,
      "learning_rate": 2.9255908419497784e-05,
      "loss": 0.0048,
      "step": 2297
    },
    {
      "epoch": 0.8408342480790341,
      "grad_norm": 0.13258232176303864,
      "learning_rate": 2.9246676514032495e-05,
      "loss": 0.0036,
      "step": 2298
    },
    {
      "epoch": 0.8412001463593121,
      "grad_norm": 5.628713130950928,
      "learning_rate": 2.9237444608567212e-05,
      "loss": 0.0254,
      "step": 2299
    },
    {
      "epoch": 0.8415660446395902,
      "grad_norm": 0.02479073777794838,
      "learning_rate": 2.922821270310192e-05,
      "loss": 0.0008,
      "step": 2300
    },
    {
      "epoch": 0.8419319429198683,
      "grad_norm": 0.10962385684251785,
      "learning_rate": 2.9218980797636637e-05,
      "loss": 0.0021,
      "step": 2301
    },
    {
      "epoch": 0.8422978412001464,
      "grad_norm": 2.80621337890625,
      "learning_rate": 2.9209748892171347e-05,
      "loss": 0.2432,
      "step": 2302
    },
    {
      "epoch": 0.8426637394804244,
      "grad_norm": 0.019096119329333305,
      "learning_rate": 2.9200516986706054e-05,
      "loss": 0.0008,
      "step": 2303
    },
    {
      "epoch": 0.8430296377607025,
      "grad_norm": 1.703490138053894,
      "learning_rate": 2.919128508124077e-05,
      "loss": 0.0118,
      "step": 2304
    },
    {
      "epoch": 0.8433955360409806,
      "grad_norm": 0.08021625131368637,
      "learning_rate": 2.9182053175775482e-05,
      "loss": 0.0026,
      "step": 2305
    },
    {
      "epoch": 0.8437614343212587,
      "grad_norm": 7.40974760055542,
      "learning_rate": 2.9172821270310193e-05,
      "loss": 0.388,
      "step": 2306
    },
    {
      "epoch": 0.8441273326015367,
      "grad_norm": 0.08682569861412048,
      "learning_rate": 2.9163589364844907e-05,
      "loss": 0.0024,
      "step": 2307
    },
    {
      "epoch": 0.8444932308818148,
      "grad_norm": 0.4706585109233856,
      "learning_rate": 2.9154357459379617e-05,
      "loss": 0.0055,
      "step": 2308
    },
    {
      "epoch": 0.8448591291620929,
      "grad_norm": 0.14325034618377686,
      "learning_rate": 2.914512555391433e-05,
      "loss": 0.0034,
      "step": 2309
    },
    {
      "epoch": 0.845225027442371,
      "grad_norm": 3.56807804107666,
      "learning_rate": 2.913589364844904e-05,
      "loss": 0.0206,
      "step": 2310
    },
    {
      "epoch": 0.8455909257226492,
      "grad_norm": 0.09884609282016754,
      "learning_rate": 2.9126661742983752e-05,
      "loss": 0.002,
      "step": 2311
    },
    {
      "epoch": 0.8459568240029272,
      "grad_norm": 0.6897984147071838,
      "learning_rate": 2.9117429837518466e-05,
      "loss": 0.0079,
      "step": 2312
    },
    {
      "epoch": 0.8463227222832053,
      "grad_norm": 0.26429808139801025,
      "learning_rate": 2.9108197932053177e-05,
      "loss": 0.0059,
      "step": 2313
    },
    {
      "epoch": 0.8466886205634834,
      "grad_norm": 0.05191277340054512,
      "learning_rate": 2.9098966026587887e-05,
      "loss": 0.001,
      "step": 2314
    },
    {
      "epoch": 0.8470545188437615,
      "grad_norm": 0.07315676659345627,
      "learning_rate": 2.90897341211226e-05,
      "loss": 0.0017,
      "step": 2315
    },
    {
      "epoch": 0.8474204171240395,
      "grad_norm": 0.029645366594195366,
      "learning_rate": 2.9080502215657312e-05,
      "loss": 0.0008,
      "step": 2316
    },
    {
      "epoch": 0.8477863154043176,
      "grad_norm": 0.047427788376808167,
      "learning_rate": 2.9071270310192022e-05,
      "loss": 0.0011,
      "step": 2317
    },
    {
      "epoch": 0.8481522136845957,
      "grad_norm": 4.2201385498046875,
      "learning_rate": 2.906203840472674e-05,
      "loss": 0.0176,
      "step": 2318
    },
    {
      "epoch": 0.8485181119648738,
      "grad_norm": 0.03314806520938873,
      "learning_rate": 2.9052806499261447e-05,
      "loss": 0.0008,
      "step": 2319
    },
    {
      "epoch": 0.8488840102451518,
      "grad_norm": 0.012494320049881935,
      "learning_rate": 2.9043574593796164e-05,
      "loss": 0.0005,
      "step": 2320
    },
    {
      "epoch": 0.8492499085254299,
      "grad_norm": 3.033682107925415,
      "learning_rate": 2.9034342688330875e-05,
      "loss": 0.2954,
      "step": 2321
    },
    {
      "epoch": 0.849615806805708,
      "grad_norm": 0.11109353601932526,
      "learning_rate": 2.9025110782865582e-05,
      "loss": 0.0021,
      "step": 2322
    },
    {
      "epoch": 0.8499817050859861,
      "grad_norm": 0.06776965409517288,
      "learning_rate": 2.90158788774003e-05,
      "loss": 0.0018,
      "step": 2323
    },
    {
      "epoch": 0.8503476033662641,
      "grad_norm": 0.010376938618719578,
      "learning_rate": 2.900664697193501e-05,
      "loss": 0.0005,
      "step": 2324
    },
    {
      "epoch": 0.8507135016465422,
      "grad_norm": 0.039810001850128174,
      "learning_rate": 2.899741506646972e-05,
      "loss": 0.0009,
      "step": 2325
    },
    {
      "epoch": 0.8510793999268204,
      "grad_norm": 0.3594454526901245,
      "learning_rate": 2.8988183161004434e-05,
      "loss": 0.0032,
      "step": 2326
    },
    {
      "epoch": 0.8514452982070985,
      "grad_norm": 0.23796789348125458,
      "learning_rate": 2.8978951255539145e-05,
      "loss": 0.0024,
      "step": 2327
    },
    {
      "epoch": 0.8518111964873765,
      "grad_norm": 0.011097910813987255,
      "learning_rate": 2.8969719350073855e-05,
      "loss": 0.0005,
      "step": 2328
    },
    {
      "epoch": 0.8521770947676546,
      "grad_norm": 9.558564186096191,
      "learning_rate": 2.896048744460857e-05,
      "loss": 0.0416,
      "step": 2329
    },
    {
      "epoch": 0.8525429930479327,
      "grad_norm": 0.09764181822538376,
      "learning_rate": 2.895125553914328e-05,
      "loss": 0.0018,
      "step": 2330
    },
    {
      "epoch": 0.8529088913282108,
      "grad_norm": 0.0429607629776001,
      "learning_rate": 2.8942023633677994e-05,
      "loss": 0.0011,
      "step": 2331
    },
    {
      "epoch": 0.8532747896084888,
      "grad_norm": 0.07645070552825928,
      "learning_rate": 2.8932791728212704e-05,
      "loss": 0.0018,
      "step": 2332
    },
    {
      "epoch": 0.8536406878887669,
      "grad_norm": 0.08440320193767548,
      "learning_rate": 2.8923559822747415e-05,
      "loss": 0.0021,
      "step": 2333
    },
    {
      "epoch": 0.854006586169045,
      "grad_norm": 0.12306660413742065,
      "learning_rate": 2.891432791728213e-05,
      "loss": 0.0022,
      "step": 2334
    },
    {
      "epoch": 0.8543724844493231,
      "grad_norm": 4.068546295166016,
      "learning_rate": 2.890509601181684e-05,
      "loss": 0.0173,
      "step": 2335
    },
    {
      "epoch": 0.8547383827296011,
      "grad_norm": 0.3068051338195801,
      "learning_rate": 2.889586410635155e-05,
      "loss": 0.0035,
      "step": 2336
    },
    {
      "epoch": 0.8551042810098792,
      "grad_norm": 0.011022981256246567,
      "learning_rate": 2.8886632200886267e-05,
      "loss": 0.0004,
      "step": 2337
    },
    {
      "epoch": 0.8554701792901573,
      "grad_norm": 0.3144238293170929,
      "learning_rate": 2.8877400295420974e-05,
      "loss": 0.0066,
      "step": 2338
    },
    {
      "epoch": 0.8558360775704354,
      "grad_norm": 0.04222669452428818,
      "learning_rate": 2.886816838995569e-05,
      "loss": 0.001,
      "step": 2339
    },
    {
      "epoch": 0.8562019758507134,
      "grad_norm": 0.008244151249527931,
      "learning_rate": 2.8858936484490402e-05,
      "loss": 0.0004,
      "step": 2340
    },
    {
      "epoch": 0.8565678741309916,
      "grad_norm": 3.6072769165039062,
      "learning_rate": 2.884970457902511e-05,
      "loss": 0.2882,
      "step": 2341
    },
    {
      "epoch": 0.8569337724112697,
      "grad_norm": 0.04715666174888611,
      "learning_rate": 2.8840472673559827e-05,
      "loss": 0.001,
      "step": 2342
    },
    {
      "epoch": 0.8572996706915478,
      "grad_norm": 0.1809810847043991,
      "learning_rate": 2.8831240768094537e-05,
      "loss": 0.0043,
      "step": 2343
    },
    {
      "epoch": 0.8576655689718259,
      "grad_norm": 0.27102574706077576,
      "learning_rate": 2.8822008862629248e-05,
      "loss": 0.0055,
      "step": 2344
    },
    {
      "epoch": 0.8580314672521039,
      "grad_norm": 0.1321425884962082,
      "learning_rate": 2.881277695716396e-05,
      "loss": 0.0024,
      "step": 2345
    },
    {
      "epoch": 0.858397365532382,
      "grad_norm": 9.336758613586426,
      "learning_rate": 2.8803545051698672e-05,
      "loss": 0.2584,
      "step": 2346
    },
    {
      "epoch": 0.8587632638126601,
      "grad_norm": 0.26309043169021606,
      "learning_rate": 2.8794313146233383e-05,
      "loss": 0.004,
      "step": 2347
    },
    {
      "epoch": 0.8591291620929382,
      "grad_norm": 0.06777963042259216,
      "learning_rate": 2.8785081240768097e-05,
      "loss": 0.0017,
      "step": 2348
    },
    {
      "epoch": 0.8594950603732162,
      "grad_norm": 0.018203282728791237,
      "learning_rate": 2.8775849335302807e-05,
      "loss": 0.0005,
      "step": 2349
    },
    {
      "epoch": 0.8598609586534943,
      "grad_norm": 0.8760929107666016,
      "learning_rate": 2.876661742983752e-05,
      "loss": 0.0084,
      "step": 2350
    },
    {
      "epoch": 0.8602268569337724,
      "grad_norm": 0.016411621123552322,
      "learning_rate": 2.8757385524372232e-05,
      "loss": 0.0006,
      "step": 2351
    },
    {
      "epoch": 0.8605927552140505,
      "grad_norm": 0.055022966116666794,
      "learning_rate": 2.8748153618906942e-05,
      "loss": 0.0009,
      "step": 2352
    },
    {
      "epoch": 0.8609586534943285,
      "grad_norm": 0.02048979140818119,
      "learning_rate": 2.8738921713441656e-05,
      "loss": 0.0005,
      "step": 2353
    },
    {
      "epoch": 0.8613245517746067,
      "grad_norm": 3.983909845352173,
      "learning_rate": 2.8729689807976367e-05,
      "loss": 0.2696,
      "step": 2354
    },
    {
      "epoch": 0.8616904500548848,
      "grad_norm": 0.028523484244942665,
      "learning_rate": 2.8720457902511077e-05,
      "loss": 0.0005,
      "step": 2355
    },
    {
      "epoch": 0.8620563483351629,
      "grad_norm": 0.0066682081669569016,
      "learning_rate": 2.8711225997045795e-05,
      "loss": 0.0003,
      "step": 2356
    },
    {
      "epoch": 0.8624222466154409,
      "grad_norm": 27.120361328125,
      "learning_rate": 2.8701994091580502e-05,
      "loss": 0.1203,
      "step": 2357
    },
    {
      "epoch": 0.862788144895719,
      "grad_norm": 0.006011583376675844,
      "learning_rate": 2.8692762186115212e-05,
      "loss": 0.0003,
      "step": 2358
    },
    {
      "epoch": 0.8631540431759971,
      "grad_norm": 8.908696174621582,
      "learning_rate": 2.868353028064993e-05,
      "loss": 0.3698,
      "step": 2359
    },
    {
      "epoch": 0.8635199414562752,
      "grad_norm": 0.11809727549552917,
      "learning_rate": 2.8674298375184637e-05,
      "loss": 0.0035,
      "step": 2360
    },
    {
      "epoch": 0.8638858397365532,
      "grad_norm": 0.012679662555456161,
      "learning_rate": 2.8665066469719354e-05,
      "loss": 0.0004,
      "step": 2361
    },
    {
      "epoch": 0.8642517380168313,
      "grad_norm": 0.03528186306357384,
      "learning_rate": 2.8655834564254065e-05,
      "loss": 0.0007,
      "step": 2362
    },
    {
      "epoch": 0.8646176362971094,
      "grad_norm": 0.1782093346118927,
      "learning_rate": 2.8646602658788775e-05,
      "loss": 0.0023,
      "step": 2363
    },
    {
      "epoch": 0.8649835345773875,
      "grad_norm": 19.66499900817871,
      "learning_rate": 2.863737075332349e-05,
      "loss": 0.1272,
      "step": 2364
    },
    {
      "epoch": 0.8653494328576655,
      "grad_norm": 0.006736451759934425,
      "learning_rate": 2.86281388478582e-05,
      "loss": 0.0003,
      "step": 2365
    },
    {
      "epoch": 0.8657153311379436,
      "grad_norm": 0.019063647836446762,
      "learning_rate": 2.861890694239291e-05,
      "loss": 0.0005,
      "step": 2366
    },
    {
      "epoch": 0.8660812294182217,
      "grad_norm": 0.20570670068264008,
      "learning_rate": 2.8609675036927624e-05,
      "loss": 0.0025,
      "step": 2367
    },
    {
      "epoch": 0.8664471276984999,
      "grad_norm": 0.1950778216123581,
      "learning_rate": 2.8600443131462335e-05,
      "loss": 0.0033,
      "step": 2368
    },
    {
      "epoch": 0.8668130259787779,
      "grad_norm": 3.3898096084594727,
      "learning_rate": 2.859121122599705e-05,
      "loss": 0.297,
      "step": 2369
    },
    {
      "epoch": 0.867178924259056,
      "grad_norm": 0.08605406433343887,
      "learning_rate": 2.858197932053176e-05,
      "loss": 0.0027,
      "step": 2370
    },
    {
      "epoch": 0.8675448225393341,
      "grad_norm": 0.19329982995986938,
      "learning_rate": 2.857274741506647e-05,
      "loss": 0.0056,
      "step": 2371
    },
    {
      "epoch": 0.8679107208196122,
      "grad_norm": 14.90194034576416,
      "learning_rate": 2.8563515509601184e-05,
      "loss": 0.1799,
      "step": 2372
    },
    {
      "epoch": 0.8682766190998902,
      "grad_norm": 0.13984830677509308,
      "learning_rate": 2.8554283604135894e-05,
      "loss": 0.0042,
      "step": 2373
    },
    {
      "epoch": 0.8686425173801683,
      "grad_norm": 1.5414798259735107,
      "learning_rate": 2.8545051698670605e-05,
      "loss": 0.2106,
      "step": 2374
    },
    {
      "epoch": 0.8690084156604464,
      "grad_norm": 22.387693405151367,
      "learning_rate": 2.8535819793205322e-05,
      "loss": 0.0516,
      "step": 2375
    },
    {
      "epoch": 0.8693743139407245,
      "grad_norm": 1.7697194814682007,
      "learning_rate": 2.852658788774003e-05,
      "loss": 0.0139,
      "step": 2376
    },
    {
      "epoch": 0.8697402122210025,
      "grad_norm": 0.14903174340724945,
      "learning_rate": 2.851735598227474e-05,
      "loss": 0.0047,
      "step": 2377
    },
    {
      "epoch": 0.8701061105012806,
      "grad_norm": 0.14246955513954163,
      "learning_rate": 2.8508124076809457e-05,
      "loss": 0.0046,
      "step": 2378
    },
    {
      "epoch": 0.8704720087815587,
      "grad_norm": 3.4655942916870117,
      "learning_rate": 2.8498892171344164e-05,
      "loss": 0.5072,
      "step": 2379
    },
    {
      "epoch": 0.8708379070618368,
      "grad_norm": 5.127075672149658,
      "learning_rate": 2.8489660265878882e-05,
      "loss": 0.1563,
      "step": 2380
    },
    {
      "epoch": 0.871203805342115,
      "grad_norm": 0.38901588320732117,
      "learning_rate": 2.8480428360413592e-05,
      "loss": 0.0054,
      "step": 2381
    },
    {
      "epoch": 0.8715697036223929,
      "grad_norm": 0.057362150400877,
      "learning_rate": 2.8471196454948303e-05,
      "loss": 0.0009,
      "step": 2382
    },
    {
      "epoch": 0.871935601902671,
      "grad_norm": 2.8942067623138428,
      "learning_rate": 2.8461964549483017e-05,
      "loss": 0.013,
      "step": 2383
    },
    {
      "epoch": 0.8723015001829492,
      "grad_norm": 0.17383091151714325,
      "learning_rate": 2.8452732644017727e-05,
      "loss": 0.0054,
      "step": 2384
    },
    {
      "epoch": 0.8726673984632273,
      "grad_norm": 0.10648547857999802,
      "learning_rate": 2.8443500738552438e-05,
      "loss": 0.0027,
      "step": 2385
    },
    {
      "epoch": 0.8730332967435053,
      "grad_norm": 0.15953654050827026,
      "learning_rate": 2.8434268833087152e-05,
      "loss": 0.0058,
      "step": 2386
    },
    {
      "epoch": 0.8733991950237834,
      "grad_norm": 0.19960947334766388,
      "learning_rate": 2.8425036927621862e-05,
      "loss": 0.0043,
      "step": 2387
    },
    {
      "epoch": 0.8737650933040615,
      "grad_norm": 0.09301581233739853,
      "learning_rate": 2.8415805022156573e-05,
      "loss": 0.0016,
      "step": 2388
    },
    {
      "epoch": 0.8741309915843396,
      "grad_norm": 0.31374457478523254,
      "learning_rate": 2.8406573116691287e-05,
      "loss": 0.0091,
      "step": 2389
    },
    {
      "epoch": 0.8744968898646176,
      "grad_norm": 1.636184811592102,
      "learning_rate": 2.8397341211225997e-05,
      "loss": 0.2031,
      "step": 2390
    },
    {
      "epoch": 0.8748627881448957,
      "grad_norm": 0.07174541801214218,
      "learning_rate": 2.838810930576071e-05,
      "loss": 0.0008,
      "step": 2391
    },
    {
      "epoch": 0.8752286864251738,
      "grad_norm": 0.31115826964378357,
      "learning_rate": 2.8378877400295422e-05,
      "loss": 0.003,
      "step": 2392
    },
    {
      "epoch": 0.8755945847054519,
      "grad_norm": 0.0568607822060585,
      "learning_rate": 2.8369645494830132e-05,
      "loss": 0.0016,
      "step": 2393
    },
    {
      "epoch": 0.8759604829857299,
      "grad_norm": 0.030345726758241653,
      "learning_rate": 2.836041358936485e-05,
      "loss": 0.001,
      "step": 2394
    },
    {
      "epoch": 0.876326381266008,
      "grad_norm": 0.08085419982671738,
      "learning_rate": 2.8351181683899557e-05,
      "loss": 0.0037,
      "step": 2395
    },
    {
      "epoch": 0.8766922795462861,
      "grad_norm": 0.717153012752533,
      "learning_rate": 2.8341949778434267e-05,
      "loss": 0.0066,
      "step": 2396
    },
    {
      "epoch": 0.8770581778265643,
      "grad_norm": 0.1362871676683426,
      "learning_rate": 2.8332717872968985e-05,
      "loss": 0.0068,
      "step": 2397
    },
    {
      "epoch": 0.8774240761068423,
      "grad_norm": 0.07076391577720642,
      "learning_rate": 2.8323485967503692e-05,
      "loss": 0.0007,
      "step": 2398
    },
    {
      "epoch": 0.8777899743871204,
      "grad_norm": 0.8894248008728027,
      "learning_rate": 2.8314254062038402e-05,
      "loss": 0.0116,
      "step": 2399
    },
    {
      "epoch": 0.8781558726673985,
      "grad_norm": 0.17146944999694824,
      "learning_rate": 2.830502215657312e-05,
      "loss": 0.0049,
      "step": 2400
    },
    {
      "epoch": 0.8785217709476766,
      "grad_norm": 0.2051824927330017,
      "learning_rate": 2.829579025110783e-05,
      "loss": 0.0014,
      "step": 2401
    },
    {
      "epoch": 0.8788876692279546,
      "grad_norm": 0.012371894903481007,
      "learning_rate": 2.8286558345642544e-05,
      "loss": 0.0004,
      "step": 2402
    },
    {
      "epoch": 0.8792535675082327,
      "grad_norm": 0.010924341157078743,
      "learning_rate": 2.8277326440177255e-05,
      "loss": 0.0005,
      "step": 2403
    },
    {
      "epoch": 0.8796194657885108,
      "grad_norm": 0.02683316543698311,
      "learning_rate": 2.8268094534711965e-05,
      "loss": 0.0009,
      "step": 2404
    },
    {
      "epoch": 0.8799853640687889,
      "grad_norm": 0.1066422089934349,
      "learning_rate": 2.825886262924668e-05,
      "loss": 0.0024,
      "step": 2405
    },
    {
      "epoch": 0.8803512623490669,
      "grad_norm": 0.017891431227326393,
      "learning_rate": 2.824963072378139e-05,
      "loss": 0.0006,
      "step": 2406
    },
    {
      "epoch": 0.880717160629345,
      "grad_norm": 0.007211391348391771,
      "learning_rate": 2.82403988183161e-05,
      "loss": 0.0004,
      "step": 2407
    },
    {
      "epoch": 0.8810830589096231,
      "grad_norm": 4.469084739685059,
      "learning_rate": 2.8231166912850814e-05,
      "loss": 0.2507,
      "step": 2408
    },
    {
      "epoch": 0.8814489571899012,
      "grad_norm": 3.5832526683807373,
      "learning_rate": 2.8221935007385525e-05,
      "loss": 0.0128,
      "step": 2409
    },
    {
      "epoch": 0.8818148554701792,
      "grad_norm": 4.577999114990234,
      "learning_rate": 2.821270310192024e-05,
      "loss": 0.2669,
      "step": 2410
    },
    {
      "epoch": 0.8821807537504573,
      "grad_norm": 0.01055457629263401,
      "learning_rate": 2.820347119645495e-05,
      "loss": 0.0005,
      "step": 2411
    },
    {
      "epoch": 0.8825466520307355,
      "grad_norm": 0.018739400431513786,
      "learning_rate": 2.819423929098966e-05,
      "loss": 0.0006,
      "step": 2412
    },
    {
      "epoch": 0.8829125503110136,
      "grad_norm": 2.8237311840057373,
      "learning_rate": 2.8185007385524377e-05,
      "loss": 0.2109,
      "step": 2413
    },
    {
      "epoch": 0.8832784485912917,
      "grad_norm": 0.060895077884197235,
      "learning_rate": 2.8175775480059084e-05,
      "loss": 0.0012,
      "step": 2414
    },
    {
      "epoch": 0.8836443468715697,
      "grad_norm": 14.983053207397461,
      "learning_rate": 2.8166543574593795e-05,
      "loss": 0.2805,
      "step": 2415
    },
    {
      "epoch": 0.8840102451518478,
      "grad_norm": 6.067799091339111,
      "learning_rate": 2.8157311669128512e-05,
      "loss": 0.167,
      "step": 2416
    },
    {
      "epoch": 0.8843761434321259,
      "grad_norm": 0.024524515494704247,
      "learning_rate": 2.814807976366322e-05,
      "loss": 0.0007,
      "step": 2417
    },
    {
      "epoch": 0.884742041712404,
      "grad_norm": 0.009640267118811607,
      "learning_rate": 2.813884785819793e-05,
      "loss": 0.0004,
      "step": 2418
    },
    {
      "epoch": 0.885107939992682,
      "grad_norm": 0.16847209632396698,
      "learning_rate": 2.8129615952732647e-05,
      "loss": 0.0033,
      "step": 2419
    },
    {
      "epoch": 0.8854738382729601,
      "grad_norm": 0.1462797075510025,
      "learning_rate": 2.8120384047267358e-05,
      "loss": 0.001,
      "step": 2420
    },
    {
      "epoch": 0.8858397365532382,
      "grad_norm": 12.551939964294434,
      "learning_rate": 2.8111152141802072e-05,
      "loss": 0.0401,
      "step": 2421
    },
    {
      "epoch": 0.8862056348335163,
      "grad_norm": 7.287128448486328,
      "learning_rate": 2.8101920236336782e-05,
      "loss": 0.0184,
      "step": 2422
    },
    {
      "epoch": 0.8865715331137943,
      "grad_norm": 0.19664889574050903,
      "learning_rate": 2.8092688330871493e-05,
      "loss": 0.0037,
      "step": 2423
    },
    {
      "epoch": 0.8869374313940724,
      "grad_norm": 0.04665127769112587,
      "learning_rate": 2.8083456425406207e-05,
      "loss": 0.0013,
      "step": 2424
    },
    {
      "epoch": 0.8873033296743505,
      "grad_norm": 0.12603791058063507,
      "learning_rate": 2.8074224519940917e-05,
      "loss": 0.0021,
      "step": 2425
    },
    {
      "epoch": 0.8876692279546287,
      "grad_norm": 0.2589413821697235,
      "learning_rate": 2.8064992614475628e-05,
      "loss": 0.0056,
      "step": 2426
    },
    {
      "epoch": 0.8880351262349067,
      "grad_norm": 7.433413028717041,
      "learning_rate": 2.8055760709010342e-05,
      "loss": 0.1994,
      "step": 2427
    },
    {
      "epoch": 0.8884010245151848,
      "grad_norm": 0.008150879293680191,
      "learning_rate": 2.8046528803545052e-05,
      "loss": 0.0004,
      "step": 2428
    },
    {
      "epoch": 0.8887669227954629,
      "grad_norm": 0.15379375219345093,
      "learning_rate": 2.8037296898079763e-05,
      "loss": 0.0013,
      "step": 2429
    },
    {
      "epoch": 0.889132821075741,
      "grad_norm": 0.15200605988502502,
      "learning_rate": 2.8028064992614477e-05,
      "loss": 0.0037,
      "step": 2430
    },
    {
      "epoch": 0.889498719356019,
      "grad_norm": 0.009201060980558395,
      "learning_rate": 2.8018833087149187e-05,
      "loss": 0.0005,
      "step": 2431
    },
    {
      "epoch": 0.8898646176362971,
      "grad_norm": 0.09404323995113373,
      "learning_rate": 2.8009601181683905e-05,
      "loss": 0.0022,
      "step": 2432
    },
    {
      "epoch": 0.8902305159165752,
      "grad_norm": 0.013448581099510193,
      "learning_rate": 2.8000369276218612e-05,
      "loss": 0.0005,
      "step": 2433
    },
    {
      "epoch": 0.8905964141968533,
      "grad_norm": 2.6169352531433105,
      "learning_rate": 2.7991137370753323e-05,
      "loss": 0.2083,
      "step": 2434
    },
    {
      "epoch": 0.8909623124771313,
      "grad_norm": 0.025228960439562798,
      "learning_rate": 2.798190546528804e-05,
      "loss": 0.0007,
      "step": 2435
    },
    {
      "epoch": 0.8913282107574094,
      "grad_norm": 0.0718347430229187,
      "learning_rate": 2.7972673559822747e-05,
      "loss": 0.0019,
      "step": 2436
    },
    {
      "epoch": 0.8916941090376875,
      "grad_norm": 0.011073248460888863,
      "learning_rate": 2.7963441654357458e-05,
      "loss": 0.0005,
      "step": 2437
    },
    {
      "epoch": 0.8920600073179656,
      "grad_norm": 1.1715282201766968,
      "learning_rate": 2.7954209748892175e-05,
      "loss": 0.1694,
      "step": 2438
    },
    {
      "epoch": 0.8924259055982436,
      "grad_norm": 0.6235449314117432,
      "learning_rate": 2.7944977843426885e-05,
      "loss": 0.0087,
      "step": 2439
    },
    {
      "epoch": 0.8927918038785218,
      "grad_norm": 0.014110995456576347,
      "learning_rate": 2.79357459379616e-05,
      "loss": 0.0006,
      "step": 2440
    },
    {
      "epoch": 0.8931577021587999,
      "grad_norm": 1.9529935121536255,
      "learning_rate": 2.792651403249631e-05,
      "loss": 0.1327,
      "step": 2441
    },
    {
      "epoch": 0.893523600439078,
      "grad_norm": 5.241689205169678,
      "learning_rate": 2.791728212703102e-05,
      "loss": 0.366,
      "step": 2442
    },
    {
      "epoch": 0.893889498719356,
      "grad_norm": 0.30248934030532837,
      "learning_rate": 2.7908050221565734e-05,
      "loss": 0.0114,
      "step": 2443
    },
    {
      "epoch": 0.8942553969996341,
      "grad_norm": 3.5531766414642334,
      "learning_rate": 2.7898818316100445e-05,
      "loss": 0.0246,
      "step": 2444
    },
    {
      "epoch": 0.8946212952799122,
      "grad_norm": 0.01836216449737549,
      "learning_rate": 2.7889586410635156e-05,
      "loss": 0.0005,
      "step": 2445
    },
    {
      "epoch": 0.8949871935601903,
      "grad_norm": 0.29527318477630615,
      "learning_rate": 2.788035450516987e-05,
      "loss": 0.0135,
      "step": 2446
    },
    {
      "epoch": 0.8953530918404683,
      "grad_norm": 4.541627883911133,
      "learning_rate": 2.787112259970458e-05,
      "loss": 0.2238,
      "step": 2447
    },
    {
      "epoch": 0.8957189901207464,
      "grad_norm": 0.06520239263772964,
      "learning_rate": 2.786189069423929e-05,
      "loss": 0.0015,
      "step": 2448
    },
    {
      "epoch": 0.8960848884010245,
      "grad_norm": 0.2246903032064438,
      "learning_rate": 2.7852658788774004e-05,
      "loss": 0.0153,
      "step": 2449
    },
    {
      "epoch": 0.8964507866813026,
      "grad_norm": 0.2544086277484894,
      "learning_rate": 2.7843426883308715e-05,
      "loss": 0.0035,
      "step": 2450
    },
    {
      "epoch": 0.8968166849615807,
      "grad_norm": 0.00803438015282154,
      "learning_rate": 2.7834194977843432e-05,
      "loss": 0.0004,
      "step": 2451
    },
    {
      "epoch": 0.8971825832418587,
      "grad_norm": 1.4097932577133179,
      "learning_rate": 2.782496307237814e-05,
      "loss": 0.1032,
      "step": 2452
    },
    {
      "epoch": 0.8975484815221368,
      "grad_norm": 0.026053963229060173,
      "learning_rate": 2.781573116691285e-05,
      "loss": 0.0006,
      "step": 2453
    },
    {
      "epoch": 0.897914379802415,
      "grad_norm": 0.03176390379667282,
      "learning_rate": 2.7806499261447567e-05,
      "loss": 0.0011,
      "step": 2454
    },
    {
      "epoch": 0.8982802780826931,
      "grad_norm": 1.2261488437652588,
      "learning_rate": 2.7797267355982275e-05,
      "loss": 0.0872,
      "step": 2455
    },
    {
      "epoch": 0.8986461763629711,
      "grad_norm": 1.0759289264678955,
      "learning_rate": 2.7788035450516985e-05,
      "loss": 0.0072,
      "step": 2456
    },
    {
      "epoch": 0.8990120746432492,
      "grad_norm": 3.108198642730713,
      "learning_rate": 2.7778803545051702e-05,
      "loss": 0.0616,
      "step": 2457
    },
    {
      "epoch": 0.8993779729235273,
      "grad_norm": 6.767861843109131,
      "learning_rate": 2.7769571639586413e-05,
      "loss": 0.3696,
      "step": 2458
    },
    {
      "epoch": 0.8997438712038054,
      "grad_norm": 2.318535566329956,
      "learning_rate": 2.776033973412112e-05,
      "loss": 0.0319,
      "step": 2459
    },
    {
      "epoch": 0.9001097694840834,
      "grad_norm": 0.13344858586788177,
      "learning_rate": 2.7751107828655837e-05,
      "loss": 0.0034,
      "step": 2460
    },
    {
      "epoch": 0.9004756677643615,
      "grad_norm": 0.19371597468852997,
      "learning_rate": 2.7741875923190548e-05,
      "loss": 0.0044,
      "step": 2461
    },
    {
      "epoch": 0.9008415660446396,
      "grad_norm": 0.036145880818367004,
      "learning_rate": 2.7732644017725262e-05,
      "loss": 0.001,
      "step": 2462
    },
    {
      "epoch": 0.9012074643249177,
      "grad_norm": 1.174586534500122,
      "learning_rate": 2.7723412112259972e-05,
      "loss": 0.0421,
      "step": 2463
    },
    {
      "epoch": 0.9015733626051957,
      "grad_norm": 0.7428464293479919,
      "learning_rate": 2.7714180206794683e-05,
      "loss": 0.0089,
      "step": 2464
    },
    {
      "epoch": 0.9019392608854738,
      "grad_norm": 0.2935730516910553,
      "learning_rate": 2.7704948301329397e-05,
      "loss": 0.0063,
      "step": 2465
    },
    {
      "epoch": 0.9023051591657519,
      "grad_norm": 1.436362385749817,
      "learning_rate": 2.7695716395864108e-05,
      "loss": 0.047,
      "step": 2466
    },
    {
      "epoch": 0.90267105744603,
      "grad_norm": 0.23355573415756226,
      "learning_rate": 2.7686484490398818e-05,
      "loss": 0.0037,
      "step": 2467
    },
    {
      "epoch": 0.903036955726308,
      "grad_norm": 0.03045356459915638,
      "learning_rate": 2.7677252584933532e-05,
      "loss": 0.0009,
      "step": 2468
    },
    {
      "epoch": 0.9034028540065862,
      "grad_norm": 0.020573245361447334,
      "learning_rate": 2.7668020679468243e-05,
      "loss": 0.0005,
      "step": 2469
    },
    {
      "epoch": 0.9037687522868643,
      "grad_norm": 1.5990699529647827,
      "learning_rate": 2.7658788774002953e-05,
      "loss": 0.0049,
      "step": 2470
    },
    {
      "epoch": 0.9041346505671424,
      "grad_norm": 11.232032775878906,
      "learning_rate": 2.7649556868537667e-05,
      "loss": 0.0975,
      "step": 2471
    },
    {
      "epoch": 0.9045005488474204,
      "grad_norm": 0.031614549458026886,
      "learning_rate": 2.7640324963072378e-05,
      "loss": 0.0007,
      "step": 2472
    },
    {
      "epoch": 0.9048664471276985,
      "grad_norm": 3.7049529552459717,
      "learning_rate": 2.7631093057607095e-05,
      "loss": 0.0955,
      "step": 2473
    },
    {
      "epoch": 0.9052323454079766,
      "grad_norm": 0.00898729357868433,
      "learning_rate": 2.7621861152141802e-05,
      "loss": 0.0004,
      "step": 2474
    },
    {
      "epoch": 0.9055982436882547,
      "grad_norm": 0.1839628517627716,
      "learning_rate": 2.7612629246676513e-05,
      "loss": 0.0033,
      "step": 2475
    },
    {
      "epoch": 0.9059641419685327,
      "grad_norm": 2.023604393005371,
      "learning_rate": 2.760339734121123e-05,
      "loss": 0.0308,
      "step": 2476
    },
    {
      "epoch": 0.9063300402488108,
      "grad_norm": 0.010604613460600376,
      "learning_rate": 2.759416543574594e-05,
      "loss": 0.0004,
      "step": 2477
    },
    {
      "epoch": 0.9066959385290889,
      "grad_norm": 0.23932267725467682,
      "learning_rate": 2.7584933530280648e-05,
      "loss": 0.0027,
      "step": 2478
    },
    {
      "epoch": 0.907061836809367,
      "grad_norm": 0.1413584053516388,
      "learning_rate": 2.7575701624815365e-05,
      "loss": 0.0025,
      "step": 2479
    },
    {
      "epoch": 0.907427735089645,
      "grad_norm": 0.011939426884055138,
      "learning_rate": 2.7566469719350076e-05,
      "loss": 0.0005,
      "step": 2480
    },
    {
      "epoch": 0.9077936333699231,
      "grad_norm": 1.0148255825042725,
      "learning_rate": 2.755723781388479e-05,
      "loss": 0.0076,
      "step": 2481
    },
    {
      "epoch": 0.9081595316502012,
      "grad_norm": 0.007628806866705418,
      "learning_rate": 2.75480059084195e-05,
      "loss": 0.0003,
      "step": 2482
    },
    {
      "epoch": 0.9085254299304794,
      "grad_norm": 0.006238657981157303,
      "learning_rate": 2.753877400295421e-05,
      "loss": 0.0003,
      "step": 2483
    },
    {
      "epoch": 0.9088913282107574,
      "grad_norm": 0.005595592316240072,
      "learning_rate": 2.7529542097488925e-05,
      "loss": 0.0003,
      "step": 2484
    },
    {
      "epoch": 0.9092572264910355,
      "grad_norm": 0.006963836494833231,
      "learning_rate": 2.7520310192023635e-05,
      "loss": 0.0003,
      "step": 2485
    },
    {
      "epoch": 0.9096231247713136,
      "grad_norm": 0.0071249669417738914,
      "learning_rate": 2.7511078286558346e-05,
      "loss": 0.0003,
      "step": 2486
    },
    {
      "epoch": 0.9099890230515917,
      "grad_norm": 7.965728759765625,
      "learning_rate": 2.750184638109306e-05,
      "loss": 0.2114,
      "step": 2487
    },
    {
      "epoch": 0.9103549213318698,
      "grad_norm": 0.008594954386353493,
      "learning_rate": 2.749261447562777e-05,
      "loss": 0.0004,
      "step": 2488
    },
    {
      "epoch": 0.9107208196121478,
      "grad_norm": 0.01024838350713253,
      "learning_rate": 2.748338257016248e-05,
      "loss": 0.0004,
      "step": 2489
    },
    {
      "epoch": 0.9110867178924259,
      "grad_norm": 0.07532355934381485,
      "learning_rate": 2.7474150664697195e-05,
      "loss": 0.0007,
      "step": 2490
    },
    {
      "epoch": 0.911452616172704,
      "grad_norm": 0.005415230989456177,
      "learning_rate": 2.7464918759231905e-05,
      "loss": 0.0002,
      "step": 2491
    },
    {
      "epoch": 0.9118185144529821,
      "grad_norm": 2.0201172828674316,
      "learning_rate": 2.7455686853766622e-05,
      "loss": 0.5044,
      "step": 2492
    },
    {
      "epoch": 0.9121844127332601,
      "grad_norm": 0.007929674349725246,
      "learning_rate": 2.744645494830133e-05,
      "loss": 0.0004,
      "step": 2493
    },
    {
      "epoch": 0.9125503110135382,
      "grad_norm": 17.112545013427734,
      "learning_rate": 2.743722304283604e-05,
      "loss": 0.5844,
      "step": 2494
    },
    {
      "epoch": 0.9129162092938163,
      "grad_norm": 0.008557843044400215,
      "learning_rate": 2.7427991137370757e-05,
      "loss": 0.0004,
      "step": 2495
    },
    {
      "epoch": 0.9132821075740944,
      "grad_norm": 0.008999458514153957,
      "learning_rate": 2.7418759231905468e-05,
      "loss": 0.0004,
      "step": 2496
    },
    {
      "epoch": 0.9136480058543724,
      "grad_norm": 35.17911148071289,
      "learning_rate": 2.7409527326440175e-05,
      "loss": 0.6497,
      "step": 2497
    },
    {
      "epoch": 0.9140139041346506,
      "grad_norm": 0.009386838413774967,
      "learning_rate": 2.7400295420974893e-05,
      "loss": 0.0004,
      "step": 2498
    },
    {
      "epoch": 0.9143798024149287,
      "grad_norm": 0.010294799692928791,
      "learning_rate": 2.7391063515509603e-05,
      "loss": 0.0004,
      "step": 2499
    },
    {
      "epoch": 0.9147457006952068,
      "grad_norm": 10.489164352416992,
      "learning_rate": 2.738183161004431e-05,
      "loss": 0.223,
      "step": 2500
    },
    {
      "epoch": 0.9151115989754848,
      "grad_norm": 0.010627656243741512,
      "learning_rate": 2.7372599704579028e-05,
      "loss": 0.0005,
      "step": 2501
    },
    {
      "epoch": 0.9154774972557629,
      "grad_norm": 0.010161353275179863,
      "learning_rate": 2.7363367799113738e-05,
      "loss": 0.0005,
      "step": 2502
    },
    {
      "epoch": 0.915843395536041,
      "grad_norm": 0.02858750894665718,
      "learning_rate": 2.7354135893648452e-05,
      "loss": 0.0008,
      "step": 2503
    },
    {
      "epoch": 0.9162092938163191,
      "grad_norm": 4.589656352996826,
      "learning_rate": 2.7344903988183163e-05,
      "loss": 0.3135,
      "step": 2504
    },
    {
      "epoch": 0.9165751920965971,
      "grad_norm": 0.013493206351995468,
      "learning_rate": 2.7335672082717873e-05,
      "loss": 0.0005,
      "step": 2505
    },
    {
      "epoch": 0.9169410903768752,
      "grad_norm": 0.5237334966659546,
      "learning_rate": 2.7326440177252587e-05,
      "loss": 0.0073,
      "step": 2506
    },
    {
      "epoch": 0.9173069886571533,
      "grad_norm": 0.03015860728919506,
      "learning_rate": 2.7317208271787298e-05,
      "loss": 0.0008,
      "step": 2507
    },
    {
      "epoch": 0.9176728869374314,
      "grad_norm": 3.5779221057891846,
      "learning_rate": 2.7307976366322008e-05,
      "loss": 0.2461,
      "step": 2508
    },
    {
      "epoch": 0.9180387852177094,
      "grad_norm": 0.014928746037185192,
      "learning_rate": 2.7298744460856722e-05,
      "loss": 0.0007,
      "step": 2509
    },
    {
      "epoch": 0.9184046834979875,
      "grad_norm": 2.2055485248565674,
      "learning_rate": 2.7289512555391433e-05,
      "loss": 0.4488,
      "step": 2510
    },
    {
      "epoch": 0.9187705817782656,
      "grad_norm": 0.014232759363949299,
      "learning_rate": 2.728028064992615e-05,
      "loss": 0.0007,
      "step": 2511
    },
    {
      "epoch": 0.9191364800585438,
      "grad_norm": 0.03175671398639679,
      "learning_rate": 2.7271048744460857e-05,
      "loss": 0.0012,
      "step": 2512
    },
    {
      "epoch": 0.9195023783388218,
      "grad_norm": 0.025431813672184944,
      "learning_rate": 2.7261816838995568e-05,
      "loss": 0.0009,
      "step": 2513
    },
    {
      "epoch": 0.9198682766190999,
      "grad_norm": 0.021341735497117043,
      "learning_rate": 2.7252584933530285e-05,
      "loss": 0.0008,
      "step": 2514
    },
    {
      "epoch": 0.920234174899378,
      "grad_norm": 0.019728118553757668,
      "learning_rate": 2.7243353028064996e-05,
      "loss": 0.0008,
      "step": 2515
    },
    {
      "epoch": 0.9206000731796561,
      "grad_norm": 0.11735562235116959,
      "learning_rate": 2.7234121122599703e-05,
      "loss": 0.0034,
      "step": 2516
    },
    {
      "epoch": 0.9209659714599341,
      "grad_norm": 1.2307708263397217,
      "learning_rate": 2.722488921713442e-05,
      "loss": 0.0197,
      "step": 2517
    },
    {
      "epoch": 0.9213318697402122,
      "grad_norm": 16.871898651123047,
      "learning_rate": 2.721565731166913e-05,
      "loss": 0.4032,
      "step": 2518
    },
    {
      "epoch": 0.9216977680204903,
      "grad_norm": 7.887607574462891,
      "learning_rate": 2.7206425406203838e-05,
      "loss": 0.0758,
      "step": 2519
    },
    {
      "epoch": 0.9220636663007684,
      "grad_norm": 1.9904025793075562,
      "learning_rate": 2.7197193500738555e-05,
      "loss": 0.3782,
      "step": 2520
    },
    {
      "epoch": 0.9224295645810465,
      "grad_norm": 0.04256922006607056,
      "learning_rate": 2.7187961595273266e-05,
      "loss": 0.0016,
      "step": 2521
    },
    {
      "epoch": 0.9227954628613245,
      "grad_norm": 1.2438559532165527,
      "learning_rate": 2.717872968980798e-05,
      "loss": 0.0226,
      "step": 2522
    },
    {
      "epoch": 0.9231613611416026,
      "grad_norm": 1.3270453214645386,
      "learning_rate": 2.716949778434269e-05,
      "loss": 0.0118,
      "step": 2523
    },
    {
      "epoch": 0.9235272594218807,
      "grad_norm": 0.06168406829237938,
      "learning_rate": 2.71602658788774e-05,
      "loss": 0.0024,
      "step": 2524
    },
    {
      "epoch": 0.9238931577021589,
      "grad_norm": 0.08483677357435226,
      "learning_rate": 2.7151033973412115e-05,
      "loss": 0.0036,
      "step": 2525
    },
    {
      "epoch": 0.9242590559824369,
      "grad_norm": 1.2694214582443237,
      "learning_rate": 2.7141802067946825e-05,
      "loss": 0.0215,
      "step": 2526
    },
    {
      "epoch": 0.924624954262715,
      "grad_norm": 0.0738210678100586,
      "learning_rate": 2.7132570162481536e-05,
      "loss": 0.0033,
      "step": 2527
    },
    {
      "epoch": 0.9249908525429931,
      "grad_norm": 1.0486245155334473,
      "learning_rate": 2.712333825701625e-05,
      "loss": 0.0144,
      "step": 2528
    },
    {
      "epoch": 0.9253567508232712,
      "grad_norm": 0.15310901403427124,
      "learning_rate": 2.711410635155096e-05,
      "loss": 0.0068,
      "step": 2529
    },
    {
      "epoch": 0.9257226491035492,
      "grad_norm": 2.612610340118408,
      "learning_rate": 2.710487444608567e-05,
      "loss": 0.0294,
      "step": 2530
    },
    {
      "epoch": 0.9260885473838273,
      "grad_norm": 0.15296556055545807,
      "learning_rate": 2.7095642540620385e-05,
      "loss": 0.0073,
      "step": 2531
    },
    {
      "epoch": 0.9264544456641054,
      "grad_norm": 0.08584308624267578,
      "learning_rate": 2.7086410635155095e-05,
      "loss": 0.0036,
      "step": 2532
    },
    {
      "epoch": 0.9268203439443835,
      "grad_norm": 0.0713866576552391,
      "learning_rate": 2.7077178729689813e-05,
      "loss": 0.003,
      "step": 2533
    },
    {
      "epoch": 0.9271862422246615,
      "grad_norm": 0.04400838539004326,
      "learning_rate": 2.7067946824224523e-05,
      "loss": 0.002,
      "step": 2534
    },
    {
      "epoch": 0.9275521405049396,
      "grad_norm": 0.08018136024475098,
      "learning_rate": 2.705871491875923e-05,
      "loss": 0.003,
      "step": 2535
    },
    {
      "epoch": 0.9279180387852177,
      "grad_norm": 0.04168882593512535,
      "learning_rate": 2.7049483013293948e-05,
      "loss": 0.0019,
      "step": 2536
    },
    {
      "epoch": 0.9282839370654958,
      "grad_norm": 37.503150939941406,
      "learning_rate": 2.7040251107828658e-05,
      "loss": 0.1595,
      "step": 2537
    },
    {
      "epoch": 0.9286498353457738,
      "grad_norm": 1.3807260990142822,
      "learning_rate": 2.7031019202363365e-05,
      "loss": 0.0084,
      "step": 2538
    },
    {
      "epoch": 0.9290157336260519,
      "grad_norm": 0.041417092084884644,
      "learning_rate": 2.7021787296898083e-05,
      "loss": 0.0018,
      "step": 2539
    },
    {
      "epoch": 0.92938163190633,
      "grad_norm": 0.02388579584658146,
      "learning_rate": 2.7012555391432793e-05,
      "loss": 0.001,
      "step": 2540
    },
    {
      "epoch": 0.9297475301866082,
      "grad_norm": 0.0985739603638649,
      "learning_rate": 2.7003323485967507e-05,
      "loss": 0.0019,
      "step": 2541
    },
    {
      "epoch": 0.9301134284668862,
      "grad_norm": 0.043983589857816696,
      "learning_rate": 2.6994091580502218e-05,
      "loss": 0.0015,
      "step": 2542
    },
    {
      "epoch": 0.9304793267471643,
      "grad_norm": 0.047176726162433624,
      "learning_rate": 2.6984859675036928e-05,
      "loss": 0.002,
      "step": 2543
    },
    {
      "epoch": 0.9308452250274424,
      "grad_norm": 0.12402182072401047,
      "learning_rate": 2.6975627769571642e-05,
      "loss": 0.0025,
      "step": 2544
    },
    {
      "epoch": 0.9312111233077205,
      "grad_norm": 0.05142500624060631,
      "learning_rate": 2.6966395864106353e-05,
      "loss": 0.0014,
      "step": 2545
    },
    {
      "epoch": 0.9315770215879985,
      "grad_norm": 0.052246008068323135,
      "learning_rate": 2.6957163958641063e-05,
      "loss": 0.0019,
      "step": 2546
    },
    {
      "epoch": 0.9319429198682766,
      "grad_norm": 11.036200523376465,
      "learning_rate": 2.6947932053175777e-05,
      "loss": 0.0322,
      "step": 2547
    },
    {
      "epoch": 0.9323088181485547,
      "grad_norm": 0.022371137514710426,
      "learning_rate": 2.6938700147710488e-05,
      "loss": 0.001,
      "step": 2548
    },
    {
      "epoch": 0.9326747164288328,
      "grad_norm": 0.029248928651213646,
      "learning_rate": 2.6929468242245198e-05,
      "loss": 0.0013,
      "step": 2549
    },
    {
      "epoch": 0.9330406147091108,
      "grad_norm": 3.582366466522217,
      "learning_rate": 2.6920236336779912e-05,
      "loss": 0.3825,
      "step": 2550
    },
    {
      "epoch": 0.9334065129893889,
      "grad_norm": 3.227651596069336,
      "learning_rate": 2.6911004431314623e-05,
      "loss": 0.0092,
      "step": 2551
    },
    {
      "epoch": 0.933772411269667,
      "grad_norm": 0.03365964815020561,
      "learning_rate": 2.690177252584934e-05,
      "loss": 0.0014,
      "step": 2552
    },
    {
      "epoch": 0.9341383095499451,
      "grad_norm": 0.028224747627973557,
      "learning_rate": 2.689254062038405e-05,
      "loss": 0.0012,
      "step": 2553
    },
    {
      "epoch": 0.9345042078302231,
      "grad_norm": 15.368185043334961,
      "learning_rate": 2.6883308714918758e-05,
      "loss": 0.1056,
      "step": 2554
    },
    {
      "epoch": 0.9348701061105013,
      "grad_norm": 0.0895269364118576,
      "learning_rate": 2.6874076809453475e-05,
      "loss": 0.0016,
      "step": 2555
    },
    {
      "epoch": 0.9352360043907794,
      "grad_norm": 0.029711725190281868,
      "learning_rate": 2.6864844903988186e-05,
      "loss": 0.0013,
      "step": 2556
    },
    {
      "epoch": 0.9356019026710575,
      "grad_norm": 0.03922191634774208,
      "learning_rate": 2.6855612998522893e-05,
      "loss": 0.0015,
      "step": 2557
    },
    {
      "epoch": 0.9359678009513356,
      "grad_norm": 4.624883651733398,
      "learning_rate": 2.684638109305761e-05,
      "loss": 0.3772,
      "step": 2558
    },
    {
      "epoch": 0.9363336992316136,
      "grad_norm": 0.06617605686187744,
      "learning_rate": 2.683714918759232e-05,
      "loss": 0.0012,
      "step": 2559
    },
    {
      "epoch": 0.9366995975118917,
      "grad_norm": 11.265473365783691,
      "learning_rate": 2.682791728212703e-05,
      "loss": 0.0562,
      "step": 2560
    },
    {
      "epoch": 0.9370654957921698,
      "grad_norm": 0.03102906048297882,
      "learning_rate": 2.6818685376661745e-05,
      "loss": 0.0008,
      "step": 2561
    },
    {
      "epoch": 0.9374313940724479,
      "grad_norm": 0.22534961998462677,
      "learning_rate": 2.6809453471196456e-05,
      "loss": 0.0021,
      "step": 2562
    },
    {
      "epoch": 0.9377972923527259,
      "grad_norm": 0.02669086866080761,
      "learning_rate": 2.680022156573117e-05,
      "loss": 0.0011,
      "step": 2563
    },
    {
      "epoch": 0.938163190633004,
      "grad_norm": 0.0600670650601387,
      "learning_rate": 2.679098966026588e-05,
      "loss": 0.002,
      "step": 2564
    },
    {
      "epoch": 0.9385290889132821,
      "grad_norm": 0.02471977286040783,
      "learning_rate": 2.678175775480059e-05,
      "loss": 0.0011,
      "step": 2565
    },
    {
      "epoch": 0.9388949871935602,
      "grad_norm": 10.416007995605469,
      "learning_rate": 2.6772525849335305e-05,
      "loss": 0.2937,
      "step": 2566
    },
    {
      "epoch": 0.9392608854738382,
      "grad_norm": 5.03302526473999,
      "learning_rate": 2.6763293943870015e-05,
      "loss": 0.6089,
      "step": 2567
    },
    {
      "epoch": 0.9396267837541163,
      "grad_norm": 0.046395283192396164,
      "learning_rate": 2.6754062038404726e-05,
      "loss": 0.0019,
      "step": 2568
    },
    {
      "epoch": 0.9399926820343945,
      "grad_norm": 0.026579461991786957,
      "learning_rate": 2.674483013293944e-05,
      "loss": 0.0012,
      "step": 2569
    },
    {
      "epoch": 0.9403585803146726,
      "grad_norm": 24.351896286010742,
      "learning_rate": 2.673559822747415e-05,
      "loss": 0.5257,
      "step": 2570
    },
    {
      "epoch": 0.9407244785949506,
      "grad_norm": 0.15977421402931213,
      "learning_rate": 2.672636632200886e-05,
      "loss": 0.0045,
      "step": 2571
    },
    {
      "epoch": 0.9410903768752287,
      "grad_norm": 23.414104461669922,
      "learning_rate": 2.6717134416543578e-05,
      "loss": 0.1098,
      "step": 2572
    },
    {
      "epoch": 0.9414562751555068,
      "grad_norm": 0.08411747962236404,
      "learning_rate": 2.6707902511078285e-05,
      "loss": 0.0024,
      "step": 2573
    },
    {
      "epoch": 0.9418221734357849,
      "grad_norm": 0.046255405992269516,
      "learning_rate": 2.6698670605613003e-05,
      "loss": 0.002,
      "step": 2574
    },
    {
      "epoch": 0.9421880717160629,
      "grad_norm": 16.438709259033203,
      "learning_rate": 2.6689438700147713e-05,
      "loss": 0.1773,
      "step": 2575
    },
    {
      "epoch": 0.942553969996341,
      "grad_norm": 7.59377384185791,
      "learning_rate": 2.668020679468242e-05,
      "loss": 0.0331,
      "step": 2576
    },
    {
      "epoch": 0.9429198682766191,
      "grad_norm": 0.07829441130161285,
      "learning_rate": 2.6670974889217138e-05,
      "loss": 0.0032,
      "step": 2577
    },
    {
      "epoch": 0.9432857665568972,
      "grad_norm": 10.652874946594238,
      "learning_rate": 2.6661742983751848e-05,
      "loss": 0.2149,
      "step": 2578
    },
    {
      "epoch": 0.9436516648371752,
      "grad_norm": 3.1017119884490967,
      "learning_rate": 2.665251107828656e-05,
      "loss": 0.3059,
      "step": 2579
    },
    {
      "epoch": 0.9440175631174533,
      "grad_norm": 9.917399406433105,
      "learning_rate": 2.6643279172821273e-05,
      "loss": 0.1724,
      "step": 2580
    },
    {
      "epoch": 0.9443834613977314,
      "grad_norm": 0.06291015446186066,
      "learning_rate": 2.6634047267355983e-05,
      "loss": 0.0024,
      "step": 2581
    },
    {
      "epoch": 0.9447493596780095,
      "grad_norm": 0.10149252414703369,
      "learning_rate": 2.6624815361890697e-05,
      "loss": 0.0038,
      "step": 2582
    },
    {
      "epoch": 0.9451152579582875,
      "grad_norm": 0.09539443999528885,
      "learning_rate": 2.6615583456425408e-05,
      "loss": 0.0041,
      "step": 2583
    },
    {
      "epoch": 0.9454811562385657,
      "grad_norm": 0.10646620392799377,
      "learning_rate": 2.660635155096012e-05,
      "loss": 0.0043,
      "step": 2584
    },
    {
      "epoch": 0.9458470545188438,
      "grad_norm": 2.29754638671875,
      "learning_rate": 2.6597119645494832e-05,
      "loss": 0.2712,
      "step": 2585
    },
    {
      "epoch": 0.9462129527991219,
      "grad_norm": 0.5174437761306763,
      "learning_rate": 2.6587887740029543e-05,
      "loss": 0.0129,
      "step": 2586
    },
    {
      "epoch": 0.9465788510793999,
      "grad_norm": 0.13202469050884247,
      "learning_rate": 2.6578655834564253e-05,
      "loss": 0.0054,
      "step": 2587
    },
    {
      "epoch": 0.946944749359678,
      "grad_norm": 0.5809246897697449,
      "learning_rate": 2.6569423929098967e-05,
      "loss": 0.019,
      "step": 2588
    },
    {
      "epoch": 0.9473106476399561,
      "grad_norm": 0.14604789018630981,
      "learning_rate": 2.6560192023633678e-05,
      "loss": 0.0046,
      "step": 2589
    },
    {
      "epoch": 0.9476765459202342,
      "grad_norm": 0.3036283254623413,
      "learning_rate": 2.655096011816839e-05,
      "loss": 0.0091,
      "step": 2590
    },
    {
      "epoch": 0.9480424442005122,
      "grad_norm": 15.699737548828125,
      "learning_rate": 2.6541728212703106e-05,
      "loss": 0.072,
      "step": 2591
    },
    {
      "epoch": 0.9484083424807903,
      "grad_norm": 0.14738549292087555,
      "learning_rate": 2.6532496307237813e-05,
      "loss": 0.0051,
      "step": 2592
    },
    {
      "epoch": 0.9487742407610684,
      "grad_norm": 0.6843262910842896,
      "learning_rate": 2.652326440177253e-05,
      "loss": 0.0115,
      "step": 2593
    },
    {
      "epoch": 0.9491401390413465,
      "grad_norm": 1.1390122175216675,
      "learning_rate": 2.651403249630724e-05,
      "loss": 0.0125,
      "step": 2594
    },
    {
      "epoch": 0.9495060373216246,
      "grad_norm": 0.14405173063278198,
      "learning_rate": 2.6504800590841948e-05,
      "loss": 0.0059,
      "step": 2595
    },
    {
      "epoch": 0.9498719356019026,
      "grad_norm": 0.061118338257074356,
      "learning_rate": 2.6495568685376665e-05,
      "loss": 0.0021,
      "step": 2596
    },
    {
      "epoch": 0.9502378338821807,
      "grad_norm": 0.31926611065864563,
      "learning_rate": 2.6486336779911376e-05,
      "loss": 0.0064,
      "step": 2597
    },
    {
      "epoch": 0.9506037321624589,
      "grad_norm": 0.11115774512290955,
      "learning_rate": 2.6477104874446086e-05,
      "loss": 0.0034,
      "step": 2598
    },
    {
      "epoch": 0.950969630442737,
      "grad_norm": 0.04260997846722603,
      "learning_rate": 2.64678729689808e-05,
      "loss": 0.0016,
      "step": 2599
    },
    {
      "epoch": 0.951335528723015,
      "grad_norm": 8.575345993041992,
      "learning_rate": 2.645864106351551e-05,
      "loss": 0.2309,
      "step": 2600
    },
    {
      "epoch": 0.9517014270032931,
      "grad_norm": 0.0440555214881897,
      "learning_rate": 2.644940915805022e-05,
      "loss": 0.0011,
      "step": 2601
    },
    {
      "epoch": 0.9520673252835712,
      "grad_norm": 0.1246153861284256,
      "learning_rate": 2.6440177252584935e-05,
      "loss": 0.0021,
      "step": 2602
    },
    {
      "epoch": 0.9524332235638493,
      "grad_norm": 0.02324896678328514,
      "learning_rate": 2.6430945347119646e-05,
      "loss": 0.0009,
      "step": 2603
    },
    {
      "epoch": 0.9527991218441273,
      "grad_norm": 0.5150194764137268,
      "learning_rate": 2.642171344165436e-05,
      "loss": 0.0055,
      "step": 2604
    },
    {
      "epoch": 0.9531650201244054,
      "grad_norm": 0.024301450699567795,
      "learning_rate": 2.641248153618907e-05,
      "loss": 0.0009,
      "step": 2605
    },
    {
      "epoch": 0.9535309184046835,
      "grad_norm": 0.32154402136802673,
      "learning_rate": 2.640324963072378e-05,
      "loss": 0.0036,
      "step": 2606
    },
    {
      "epoch": 0.9538968166849616,
      "grad_norm": 0.018949532881379128,
      "learning_rate": 2.6394017725258495e-05,
      "loss": 0.0008,
      "step": 2607
    },
    {
      "epoch": 0.9542627149652396,
      "grad_norm": 0.03918946906924248,
      "learning_rate": 2.6384785819793205e-05,
      "loss": 0.0014,
      "step": 2608
    },
    {
      "epoch": 0.9546286132455177,
      "grad_norm": 0.3271595239639282,
      "learning_rate": 2.6375553914327916e-05,
      "loss": 0.0044,
      "step": 2609
    },
    {
      "epoch": 0.9549945115257958,
      "grad_norm": 11.550783157348633,
      "learning_rate": 2.6366322008862633e-05,
      "loss": 0.1462,
      "step": 2610
    },
    {
      "epoch": 0.955360409806074,
      "grad_norm": 0.032109133899211884,
      "learning_rate": 2.635709010339734e-05,
      "loss": 0.0011,
      "step": 2611
    },
    {
      "epoch": 0.955726308086352,
      "grad_norm": 0.07624780386686325,
      "learning_rate": 2.6347858197932058e-05,
      "loss": 0.0019,
      "step": 2612
    },
    {
      "epoch": 0.9560922063666301,
      "grad_norm": 0.015542085282504559,
      "learning_rate": 2.6338626292466768e-05,
      "loss": 0.0006,
      "step": 2613
    },
    {
      "epoch": 0.9564581046469082,
      "grad_norm": 0.03954515606164932,
      "learning_rate": 2.6329394387001475e-05,
      "loss": 0.0011,
      "step": 2614
    },
    {
      "epoch": 0.9568240029271863,
      "grad_norm": 1.8754205703735352,
      "learning_rate": 2.6320162481536193e-05,
      "loss": 0.3733,
      "step": 2615
    },
    {
      "epoch": 0.9571899012074643,
      "grad_norm": 0.06464025378227234,
      "learning_rate": 2.6310930576070903e-05,
      "loss": 0.0009,
      "step": 2616
    },
    {
      "epoch": 0.9575557994877424,
      "grad_norm": 0.021534042432904243,
      "learning_rate": 2.6301698670605614e-05,
      "loss": 0.0008,
      "step": 2617
    },
    {
      "epoch": 0.9579216977680205,
      "grad_norm": 0.02146449312567711,
      "learning_rate": 2.6292466765140328e-05,
      "loss": 0.0009,
      "step": 2618
    },
    {
      "epoch": 0.9582875960482986,
      "grad_norm": 0.05555645376443863,
      "learning_rate": 2.628323485967504e-05,
      "loss": 0.0016,
      "step": 2619
    },
    {
      "epoch": 0.9586534943285766,
      "grad_norm": 0.013226288370788097,
      "learning_rate": 2.627400295420975e-05,
      "loss": 0.0006,
      "step": 2620
    },
    {
      "epoch": 0.9590193926088547,
      "grad_norm": 1.8897143602371216,
      "learning_rate": 2.6264771048744463e-05,
      "loss": 0.3795,
      "step": 2621
    },
    {
      "epoch": 0.9593852908891328,
      "grad_norm": 5.592034816741943,
      "learning_rate": 2.6255539143279173e-05,
      "loss": 0.2433,
      "step": 2622
    },
    {
      "epoch": 0.9597511891694109,
      "grad_norm": 0.020753895863890648,
      "learning_rate": 2.6246307237813887e-05,
      "loss": 0.0009,
      "step": 2623
    },
    {
      "epoch": 0.9601170874496889,
      "grad_norm": 0.06250546127557755,
      "learning_rate": 2.6237075332348598e-05,
      "loss": 0.0021,
      "step": 2624
    },
    {
      "epoch": 0.960482985729967,
      "grad_norm": 0.03132772073149681,
      "learning_rate": 2.622784342688331e-05,
      "loss": 0.0013,
      "step": 2625
    },
    {
      "epoch": 0.9608488840102452,
      "grad_norm": 1.7287977933883667,
      "learning_rate": 2.6218611521418022e-05,
      "loss": 0.3714,
      "step": 2626
    },
    {
      "epoch": 0.9612147822905233,
      "grad_norm": 0.06189589574933052,
      "learning_rate": 2.6209379615952733e-05,
      "loss": 0.0026,
      "step": 2627
    },
    {
      "epoch": 0.9615806805708014,
      "grad_norm": 0.13623768091201782,
      "learning_rate": 2.6200147710487443e-05,
      "loss": 0.0051,
      "step": 2628
    },
    {
      "epoch": 0.9619465788510794,
      "grad_norm": 0.0645032450556755,
      "learning_rate": 2.619091580502216e-05,
      "loss": 0.0028,
      "step": 2629
    },
    {
      "epoch": 0.9623124771313575,
      "grad_norm": 0.13837115466594696,
      "learning_rate": 2.6181683899556868e-05,
      "loss": 0.0055,
      "step": 2630
    },
    {
      "epoch": 0.9626783754116356,
      "grad_norm": 0.07775513827800751,
      "learning_rate": 2.617245199409158e-05,
      "loss": 0.0034,
      "step": 2631
    },
    {
      "epoch": 0.9630442736919137,
      "grad_norm": 0.13342443108558655,
      "learning_rate": 2.6163220088626296e-05,
      "loss": 0.0056,
      "step": 2632
    },
    {
      "epoch": 0.9634101719721917,
      "grad_norm": 22.465566635131836,
      "learning_rate": 2.6153988183161003e-05,
      "loss": 0.4223,
      "step": 2633
    },
    {
      "epoch": 0.9637760702524698,
      "grad_norm": 0.07467301189899445,
      "learning_rate": 2.614475627769572e-05,
      "loss": 0.0032,
      "step": 2634
    },
    {
      "epoch": 0.9641419685327479,
      "grad_norm": 0.08347884565591812,
      "learning_rate": 2.613552437223043e-05,
      "loss": 0.0038,
      "step": 2635
    },
    {
      "epoch": 0.964507866813026,
      "grad_norm": 0.10841310769319534,
      "learning_rate": 2.612629246676514e-05,
      "loss": 0.0048,
      "step": 2636
    },
    {
      "epoch": 0.964873765093304,
      "grad_norm": 0.09116119891405106,
      "learning_rate": 2.6117060561299855e-05,
      "loss": 0.0042,
      "step": 2637
    },
    {
      "epoch": 0.9652396633735821,
      "grad_norm": 0.12764424085617065,
      "learning_rate": 2.6107828655834566e-05,
      "loss": 0.0059,
      "step": 2638
    },
    {
      "epoch": 0.9656055616538602,
      "grad_norm": 0.02046887017786503,
      "learning_rate": 2.6098596750369276e-05,
      "loss": 0.001,
      "step": 2639
    },
    {
      "epoch": 0.9659714599341384,
      "grad_norm": 0.04150936007499695,
      "learning_rate": 2.608936484490399e-05,
      "loss": 0.0019,
      "step": 2640
    },
    {
      "epoch": 0.9663373582144164,
      "grad_norm": 0.10797789692878723,
      "learning_rate": 2.60801329394387e-05,
      "loss": 0.0049,
      "step": 2641
    },
    {
      "epoch": 0.9667032564946945,
      "grad_norm": 0.47909289598464966,
      "learning_rate": 2.6070901033973415e-05,
      "loss": 0.0042,
      "step": 2642
    },
    {
      "epoch": 0.9670691547749726,
      "grad_norm": 3.671593427658081,
      "learning_rate": 2.6061669128508125e-05,
      "loss": 0.2092,
      "step": 2643
    },
    {
      "epoch": 0.9674350530552507,
      "grad_norm": 0.5866138935089111,
      "learning_rate": 2.6052437223042836e-05,
      "loss": 0.006,
      "step": 2644
    },
    {
      "epoch": 0.9678009513355287,
      "grad_norm": 0.09561919420957565,
      "learning_rate": 2.604320531757755e-05,
      "loss": 0.0042,
      "step": 2645
    },
    {
      "epoch": 0.9681668496158068,
      "grad_norm": 0.09881139546632767,
      "learning_rate": 2.603397341211226e-05,
      "loss": 0.0035,
      "step": 2646
    },
    {
      "epoch": 0.9685327478960849,
      "grad_norm": 0.04196678102016449,
      "learning_rate": 2.602474150664697e-05,
      "loss": 0.0018,
      "step": 2647
    },
    {
      "epoch": 0.968898646176363,
      "grad_norm": 0.2800406515598297,
      "learning_rate": 2.601550960118169e-05,
      "loss": 0.0055,
      "step": 2648
    },
    {
      "epoch": 0.969264544456641,
      "grad_norm": 0.06818892061710358,
      "learning_rate": 2.6006277695716395e-05,
      "loss": 0.003,
      "step": 2649
    },
    {
      "epoch": 0.9696304427369191,
      "grad_norm": 22.69625473022461,
      "learning_rate": 2.5997045790251106e-05,
      "loss": 0.1561,
      "step": 2650
    },
    {
      "epoch": 0.9699963410171972,
      "grad_norm": 9.659928321838379,
      "learning_rate": 2.5987813884785823e-05,
      "loss": 0.1701,
      "step": 2651
    },
    {
      "epoch": 0.9703622392974753,
      "grad_norm": 0.06489446759223938,
      "learning_rate": 2.597858197932053e-05,
      "loss": 0.0028,
      "step": 2652
    },
    {
      "epoch": 0.9707281375777533,
      "grad_norm": 0.12958703935146332,
      "learning_rate": 2.5969350073855248e-05,
      "loss": 0.0025,
      "step": 2653
    },
    {
      "epoch": 0.9710940358580314,
      "grad_norm": 3.594942808151245,
      "learning_rate": 2.596011816838996e-05,
      "loss": 0.2187,
      "step": 2654
    },
    {
      "epoch": 0.9714599341383096,
      "grad_norm": 1.227676510810852,
      "learning_rate": 2.595088626292467e-05,
      "loss": 0.0086,
      "step": 2655
    },
    {
      "epoch": 0.9718258324185877,
      "grad_norm": 3.954880952835083,
      "learning_rate": 2.5941654357459383e-05,
      "loss": 0.0168,
      "step": 2656
    },
    {
      "epoch": 0.9721917306988657,
      "grad_norm": 0.10849419981241226,
      "learning_rate": 2.5932422451994093e-05,
      "loss": 0.0028,
      "step": 2657
    },
    {
      "epoch": 0.9725576289791438,
      "grad_norm": 0.23058877885341644,
      "learning_rate": 2.5923190546528804e-05,
      "loss": 0.0044,
      "step": 2658
    },
    {
      "epoch": 0.9729235272594219,
      "grad_norm": 0.040511488914489746,
      "learning_rate": 2.5913958641063518e-05,
      "loss": 0.0017,
      "step": 2659
    },
    {
      "epoch": 0.9732894255397,
      "grad_norm": 26.643451690673828,
      "learning_rate": 2.590472673559823e-05,
      "loss": 0.0933,
      "step": 2660
    },
    {
      "epoch": 0.973655323819978,
      "grad_norm": 0.032034024596214294,
      "learning_rate": 2.589549483013294e-05,
      "loss": 0.0014,
      "step": 2661
    },
    {
      "epoch": 0.9740212221002561,
      "grad_norm": 7.284882068634033,
      "learning_rate": 2.5886262924667653e-05,
      "loss": 0.7792,
      "step": 2662
    },
    {
      "epoch": 0.9743871203805342,
      "grad_norm": 0.01808491349220276,
      "learning_rate": 2.5877031019202364e-05,
      "loss": 0.0008,
      "step": 2663
    },
    {
      "epoch": 0.9747530186608123,
      "grad_norm": 0.14059317111968994,
      "learning_rate": 2.5867799113737077e-05,
      "loss": 0.0042,
      "step": 2664
    },
    {
      "epoch": 0.9751189169410904,
      "grad_norm": 0.18214142322540283,
      "learning_rate": 2.5858567208271788e-05,
      "loss": 0.005,
      "step": 2665
    },
    {
      "epoch": 0.9754848152213684,
      "grad_norm": 0.19955478608608246,
      "learning_rate": 2.58493353028065e-05,
      "loss": 0.0047,
      "step": 2666
    },
    {
      "epoch": 0.9758507135016465,
      "grad_norm": 2.361823081970215,
      "learning_rate": 2.5840103397341216e-05,
      "loss": 0.4068,
      "step": 2667
    },
    {
      "epoch": 0.9762166117819246,
      "grad_norm": 10.694450378417969,
      "learning_rate": 2.5830871491875923e-05,
      "loss": 0.3862,
      "step": 2668
    },
    {
      "epoch": 0.9765825100622028,
      "grad_norm": 0.4670516848564148,
      "learning_rate": 2.5821639586410634e-05,
      "loss": 0.0068,
      "step": 2669
    },
    {
      "epoch": 0.9769484083424808,
      "grad_norm": 0.31839150190353394,
      "learning_rate": 2.581240768094535e-05,
      "loss": 0.0055,
      "step": 2670
    },
    {
      "epoch": 0.9773143066227589,
      "grad_norm": 0.028634384274482727,
      "learning_rate": 2.5803175775480058e-05,
      "loss": 0.0013,
      "step": 2671
    },
    {
      "epoch": 0.977680204903037,
      "grad_norm": 19.71830940246582,
      "learning_rate": 2.579394387001477e-05,
      "loss": 0.1819,
      "step": 2672
    },
    {
      "epoch": 0.9780461031833151,
      "grad_norm": 0.03332719951868057,
      "learning_rate": 2.5784711964549486e-05,
      "loss": 0.0015,
      "step": 2673
    },
    {
      "epoch": 0.9784120014635931,
      "grad_norm": 1.8760509490966797,
      "learning_rate": 2.5775480059084196e-05,
      "loss": 0.0124,
      "step": 2674
    },
    {
      "epoch": 0.9787778997438712,
      "grad_norm": 0.04584743455052376,
      "learning_rate": 2.576624815361891e-05,
      "loss": 0.0019,
      "step": 2675
    },
    {
      "epoch": 0.9791437980241493,
      "grad_norm": 0.056005727499723434,
      "learning_rate": 2.575701624815362e-05,
      "loss": 0.0022,
      "step": 2676
    },
    {
      "epoch": 0.9795096963044274,
      "grad_norm": 2.6575496196746826,
      "learning_rate": 2.574778434268833e-05,
      "loss": 0.2049,
      "step": 2677
    },
    {
      "epoch": 0.9798755945847054,
      "grad_norm": 0.06169693544507027,
      "learning_rate": 2.5738552437223045e-05,
      "loss": 0.002,
      "step": 2678
    },
    {
      "epoch": 0.9802414928649835,
      "grad_norm": 2.7368862628936768,
      "learning_rate": 2.5729320531757756e-05,
      "loss": 0.1795,
      "step": 2679
    },
    {
      "epoch": 0.9806073911452616,
      "grad_norm": 0.05711686238646507,
      "learning_rate": 2.5720088626292467e-05,
      "loss": 0.0024,
      "step": 2680
    },
    {
      "epoch": 0.9809732894255397,
      "grad_norm": 4.828252792358398,
      "learning_rate": 2.571085672082718e-05,
      "loss": 0.025,
      "step": 2681
    },
    {
      "epoch": 0.9813391877058177,
      "grad_norm": 0.04075026139616966,
      "learning_rate": 2.570162481536189e-05,
      "loss": 0.0019,
      "step": 2682
    },
    {
      "epoch": 0.9817050859860958,
      "grad_norm": 0.27417561411857605,
      "learning_rate": 2.5692392909896605e-05,
      "loss": 0.0102,
      "step": 2683
    },
    {
      "epoch": 0.982070984266374,
      "grad_norm": 0.047912128269672394,
      "learning_rate": 2.5683161004431316e-05,
      "loss": 0.0021,
      "step": 2684
    },
    {
      "epoch": 0.9824368825466521,
      "grad_norm": 0.017791245132684708,
      "learning_rate": 2.5673929098966026e-05,
      "loss": 0.0007,
      "step": 2685
    },
    {
      "epoch": 0.9828027808269301,
      "grad_norm": 0.3270077109336853,
      "learning_rate": 2.5664697193500743e-05,
      "loss": 0.0057,
      "step": 2686
    },
    {
      "epoch": 0.9831686791072082,
      "grad_norm": 0.10433890670537949,
      "learning_rate": 2.565546528803545e-05,
      "loss": 0.0021,
      "step": 2687
    },
    {
      "epoch": 0.9835345773874863,
      "grad_norm": 0.0628061592578888,
      "learning_rate": 2.564623338257016e-05,
      "loss": 0.0025,
      "step": 2688
    },
    {
      "epoch": 0.9839004756677644,
      "grad_norm": 0.2038898766040802,
      "learning_rate": 2.563700147710488e-05,
      "loss": 0.0053,
      "step": 2689
    },
    {
      "epoch": 0.9842663739480424,
      "grad_norm": 0.05257059633731842,
      "learning_rate": 2.5627769571639586e-05,
      "loss": 0.002,
      "step": 2690
    },
    {
      "epoch": 0.9846322722283205,
      "grad_norm": 0.6812283396720886,
      "learning_rate": 2.5618537666174296e-05,
      "loss": 0.0044,
      "step": 2691
    },
    {
      "epoch": 0.9849981705085986,
      "grad_norm": 0.03375031054019928,
      "learning_rate": 2.5609305760709013e-05,
      "loss": 0.0014,
      "step": 2692
    },
    {
      "epoch": 0.9853640687888767,
      "grad_norm": 0.030720941722393036,
      "learning_rate": 2.5600073855243724e-05,
      "loss": 0.0009,
      "step": 2693
    },
    {
      "epoch": 0.9857299670691547,
      "grad_norm": 0.04028170183300972,
      "learning_rate": 2.5590841949778438e-05,
      "loss": 0.0015,
      "step": 2694
    },
    {
      "epoch": 0.9860958653494328,
      "grad_norm": 0.15428870916366577,
      "learning_rate": 2.558161004431315e-05,
      "loss": 0.0021,
      "step": 2695
    },
    {
      "epoch": 0.9864617636297109,
      "grad_norm": 0.9232298135757446,
      "learning_rate": 2.557237813884786e-05,
      "loss": 0.015,
      "step": 2696
    },
    {
      "epoch": 0.986827661909989,
      "grad_norm": 0.9436476826667786,
      "learning_rate": 2.5563146233382573e-05,
      "loss": 0.009,
      "step": 2697
    },
    {
      "epoch": 0.9871935601902672,
      "grad_norm": 2.997331380844116,
      "learning_rate": 2.5553914327917284e-05,
      "loss": 0.3471,
      "step": 2698
    },
    {
      "epoch": 0.9875594584705452,
      "grad_norm": 0.30270037055015564,
      "learning_rate": 2.5544682422451994e-05,
      "loss": 0.0067,
      "step": 2699
    },
    {
      "epoch": 0.9879253567508233,
      "grad_norm": 0.03084562160074711,
      "learning_rate": 2.5535450516986708e-05,
      "loss": 0.0012,
      "step": 2700
    },
    {
      "epoch": 0.9882912550311014,
      "grad_norm": 18.255599975585938,
      "learning_rate": 2.552621861152142e-05,
      "loss": 0.3,
      "step": 2701
    },
    {
      "epoch": 0.9886571533113795,
      "grad_norm": 0.15699726343154907,
      "learning_rate": 2.551698670605613e-05,
      "loss": 0.0049,
      "step": 2702
    },
    {
      "epoch": 0.9890230515916575,
      "grad_norm": 0.05757717415690422,
      "learning_rate": 2.5507754800590843e-05,
      "loss": 0.002,
      "step": 2703
    },
    {
      "epoch": 0.9893889498719356,
      "grad_norm": 0.056506991386413574,
      "learning_rate": 2.5498522895125554e-05,
      "loss": 0.0021,
      "step": 2704
    },
    {
      "epoch": 0.9897548481522137,
      "grad_norm": 7.483556270599365,
      "learning_rate": 2.548929098966027e-05,
      "loss": 0.2614,
      "step": 2705
    },
    {
      "epoch": 0.9901207464324918,
      "grad_norm": 2.592970132827759,
      "learning_rate": 2.5480059084194978e-05,
      "loss": 0.2076,
      "step": 2706
    },
    {
      "epoch": 0.9904866447127698,
      "grad_norm": 6.375938892364502,
      "learning_rate": 2.547082717872969e-05,
      "loss": 0.1599,
      "step": 2707
    },
    {
      "epoch": 0.9908525429930479,
      "grad_norm": 0.11985413730144501,
      "learning_rate": 2.5461595273264406e-05,
      "loss": 0.0042,
      "step": 2708
    },
    {
      "epoch": 0.991218441273326,
      "grad_norm": 1.132277250289917,
      "learning_rate": 2.5452363367799113e-05,
      "loss": 0.0079,
      "step": 2709
    },
    {
      "epoch": 0.9915843395536041,
      "grad_norm": 3.641314744949341,
      "learning_rate": 2.5443131462333824e-05,
      "loss": 0.3154,
      "step": 2710
    },
    {
      "epoch": 0.9919502378338821,
      "grad_norm": 0.20364488661289215,
      "learning_rate": 2.543389955686854e-05,
      "loss": 0.0035,
      "step": 2711
    },
    {
      "epoch": 0.9923161361141603,
      "grad_norm": 0.042372800409793854,
      "learning_rate": 2.542466765140325e-05,
      "loss": 0.0018,
      "step": 2712
    },
    {
      "epoch": 0.9926820343944384,
      "grad_norm": 0.19246011972427368,
      "learning_rate": 2.5415435745937966e-05,
      "loss": 0.0024,
      "step": 2713
    },
    {
      "epoch": 0.9930479326747165,
      "grad_norm": 2.6803464889526367,
      "learning_rate": 2.5406203840472676e-05,
      "loss": 0.1488,
      "step": 2714
    },
    {
      "epoch": 0.9934138309549945,
      "grad_norm": 21.203615188598633,
      "learning_rate": 2.5396971935007387e-05,
      "loss": 0.2085,
      "step": 2715
    },
    {
      "epoch": 0.9937797292352726,
      "grad_norm": 1.1919541358947754,
      "learning_rate": 2.53877400295421e-05,
      "loss": 0.0098,
      "step": 2716
    },
    {
      "epoch": 0.9941456275155507,
      "grad_norm": 5.2423295974731445,
      "learning_rate": 2.537850812407681e-05,
      "loss": 0.183,
      "step": 2717
    },
    {
      "epoch": 0.9945115257958288,
      "grad_norm": 0.23891915380954742,
      "learning_rate": 2.536927621861152e-05,
      "loss": 0.0031,
      "step": 2718
    },
    {
      "epoch": 0.9948774240761068,
      "grad_norm": 0.03420918807387352,
      "learning_rate": 2.5360044313146236e-05,
      "loss": 0.0013,
      "step": 2719
    },
    {
      "epoch": 0.9952433223563849,
      "grad_norm": 0.03581967204809189,
      "learning_rate": 2.5350812407680946e-05,
      "loss": 0.0015,
      "step": 2720
    },
    {
      "epoch": 0.995609220636663,
      "grad_norm": 23.206592559814453,
      "learning_rate": 2.5341580502215657e-05,
      "loss": 0.0523,
      "step": 2721
    },
    {
      "epoch": 0.9959751189169411,
      "grad_norm": 3.727323293685913,
      "learning_rate": 2.533234859675037e-05,
      "loss": 0.2297,
      "step": 2722
    },
    {
      "epoch": 0.9963410171972191,
      "grad_norm": 1.2294844388961792,
      "learning_rate": 2.532311669128508e-05,
      "loss": 0.1261,
      "step": 2723
    },
    {
      "epoch": 0.9967069154774972,
      "grad_norm": 7.103829383850098,
      "learning_rate": 2.53138847858198e-05,
      "loss": 0.0466,
      "step": 2724
    },
    {
      "epoch": 0.9970728137577753,
      "grad_norm": 10.326226234436035,
      "learning_rate": 2.5304652880354506e-05,
      "loss": 0.1268,
      "step": 2725
    },
    {
      "epoch": 0.9974387120380535,
      "grad_norm": 0.5198456645011902,
      "learning_rate": 2.5295420974889216e-05,
      "loss": 0.0076,
      "step": 2726
    },
    {
      "epoch": 0.9978046103183315,
      "grad_norm": 0.3136453330516815,
      "learning_rate": 2.5286189069423934e-05,
      "loss": 0.0062,
      "step": 2727
    },
    {
      "epoch": 0.9981705085986096,
      "grad_norm": 0.06677121669054031,
      "learning_rate": 2.527695716395864e-05,
      "loss": 0.0018,
      "step": 2728
    },
    {
      "epoch": 0.9985364068788877,
      "grad_norm": 0.10190611332654953,
      "learning_rate": 2.526772525849335e-05,
      "loss": 0.0028,
      "step": 2729
    },
    {
      "epoch": 0.9989023051591658,
      "grad_norm": 0.05115601047873497,
      "learning_rate": 2.525849335302807e-05,
      "loss": 0.0011,
      "step": 2730
    },
    {
      "epoch": 0.9992682034394438,
      "grad_norm": 0.019759967923164368,
      "learning_rate": 2.524926144756278e-05,
      "loss": 0.0009,
      "step": 2731
    },
    {
      "epoch": 0.9996341017197219,
      "grad_norm": 0.05653541907668114,
      "learning_rate": 2.5240029542097486e-05,
      "loss": 0.0018,
      "step": 2732
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.011047453619539738,
      "learning_rate": 2.5230797636632204e-05,
      "loss": 0.0005,
      "step": 2733
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9897076986414162,
      "eval_loss": 0.03610680252313614,
      "eval_runtime": 46.8555,
      "eval_samples_per_second": 103.68,
      "eval_steps_per_second": 6.488,
      "step": 2733
    },
    {
      "epoch": 1.000365898280278,
      "grad_norm": 0.08082527667284012,
      "learning_rate": 2.5221565731166914e-05,
      "loss": 0.002,
      "step": 2734
    },
    {
      "epoch": 1.0007317965605562,
      "grad_norm": 0.8398702144622803,
      "learning_rate": 2.5212333825701628e-05,
      "loss": 0.0163,
      "step": 2735
    },
    {
      "epoch": 1.0010976948408343,
      "grad_norm": 0.015966927632689476,
      "learning_rate": 2.520310192023634e-05,
      "loss": 0.0007,
      "step": 2736
    },
    {
      "epoch": 1.0014635931211124,
      "grad_norm": 0.0830625519156456,
      "learning_rate": 2.519387001477105e-05,
      "loss": 0.002,
      "step": 2737
    },
    {
      "epoch": 1.0018294914013903,
      "grad_norm": 0.4810104966163635,
      "learning_rate": 2.5184638109305763e-05,
      "loss": 0.0063,
      "step": 2738
    },
    {
      "epoch": 1.0021953896816684,
      "grad_norm": 0.5944202542304993,
      "learning_rate": 2.5175406203840474e-05,
      "loss": 0.0051,
      "step": 2739
    },
    {
      "epoch": 1.0025612879619465,
      "grad_norm": 0.01938275620341301,
      "learning_rate": 2.5166174298375184e-05,
      "loss": 0.0007,
      "step": 2740
    },
    {
      "epoch": 1.0029271862422247,
      "grad_norm": 0.019745314493775368,
      "learning_rate": 2.5156942392909898e-05,
      "loss": 0.0007,
      "step": 2741
    },
    {
      "epoch": 1.0032930845225028,
      "grad_norm": 0.1678568720817566,
      "learning_rate": 2.514771048744461e-05,
      "loss": 0.0024,
      "step": 2742
    },
    {
      "epoch": 1.0036589828027809,
      "grad_norm": 0.10100062191486359,
      "learning_rate": 2.5138478581979326e-05,
      "loss": 0.0018,
      "step": 2743
    },
    {
      "epoch": 1.004024881083059,
      "grad_norm": 0.036672282963991165,
      "learning_rate": 2.5129246676514033e-05,
      "loss": 0.0012,
      "step": 2744
    },
    {
      "epoch": 1.004390779363337,
      "grad_norm": 11.295361518859863,
      "learning_rate": 2.5120014771048744e-05,
      "loss": 0.0331,
      "step": 2745
    },
    {
      "epoch": 1.004756677643615,
      "grad_norm": 0.018862564116716385,
      "learning_rate": 2.511078286558346e-05,
      "loss": 0.0008,
      "step": 2746
    },
    {
      "epoch": 1.005122575923893,
      "grad_norm": 0.051881760358810425,
      "learning_rate": 2.5101550960118168e-05,
      "loss": 0.0013,
      "step": 2747
    },
    {
      "epoch": 1.0054884742041712,
      "grad_norm": 0.07491950690746307,
      "learning_rate": 2.509231905465288e-05,
      "loss": 0.0015,
      "step": 2748
    },
    {
      "epoch": 1.0058543724844493,
      "grad_norm": 0.012169167399406433,
      "learning_rate": 2.5083087149187596e-05,
      "loss": 0.0005,
      "step": 2749
    },
    {
      "epoch": 1.0062202707647274,
      "grad_norm": 0.22182968258857727,
      "learning_rate": 2.5073855243722307e-05,
      "loss": 0.0031,
      "step": 2750
    },
    {
      "epoch": 1.0065861690450055,
      "grad_norm": 0.07552787661552429,
      "learning_rate": 2.5064623338257014e-05,
      "loss": 0.0012,
      "step": 2751
    },
    {
      "epoch": 1.0069520673252836,
      "grad_norm": 5.86866569519043,
      "learning_rate": 2.505539143279173e-05,
      "loss": 0.0936,
      "step": 2752
    },
    {
      "epoch": 1.0073179656055617,
      "grad_norm": 0.01261125784367323,
      "learning_rate": 2.504615952732644e-05,
      "loss": 0.0006,
      "step": 2753
    },
    {
      "epoch": 1.0076838638858396,
      "grad_norm": 0.007101427298039198,
      "learning_rate": 2.5036927621861156e-05,
      "loss": 0.0003,
      "step": 2754
    },
    {
      "epoch": 1.0080497621661177,
      "grad_norm": 0.036480892449617386,
      "learning_rate": 2.5027695716395866e-05,
      "loss": 0.0007,
      "step": 2755
    },
    {
      "epoch": 1.0084156604463959,
      "grad_norm": 0.015146817080676556,
      "learning_rate": 2.5018463810930577e-05,
      "loss": 0.0006,
      "step": 2756
    },
    {
      "epoch": 1.008781558726674,
      "grad_norm": 0.033164381980895996,
      "learning_rate": 2.500923190546529e-05,
      "loss": 0.0008,
      "step": 2757
    },
    {
      "epoch": 1.009147457006952,
      "grad_norm": 1.0949952602386475,
      "learning_rate": 2.5e-05,
      "loss": 0.0155,
      "step": 2758
    },
    {
      "epoch": 1.0095133552872302,
      "grad_norm": 0.012886625714600086,
      "learning_rate": 2.4990768094534715e-05,
      "loss": 0.0006,
      "step": 2759
    },
    {
      "epoch": 1.0098792535675083,
      "grad_norm": 0.9573665261268616,
      "learning_rate": 2.4981536189069422e-05,
      "loss": 0.0051,
      "step": 2760
    },
    {
      "epoch": 1.0102451518477864,
      "grad_norm": 0.008906185626983643,
      "learning_rate": 2.4972304283604136e-05,
      "loss": 0.0004,
      "step": 2761
    },
    {
      "epoch": 1.0106110501280643,
      "grad_norm": 0.015067358501255512,
      "learning_rate": 2.496307237813885e-05,
      "loss": 0.0005,
      "step": 2762
    },
    {
      "epoch": 1.0109769484083424,
      "grad_norm": 4.709773063659668,
      "learning_rate": 2.495384047267356e-05,
      "loss": 0.3114,
      "step": 2763
    },
    {
      "epoch": 1.0113428466886205,
      "grad_norm": 0.013595347292721272,
      "learning_rate": 2.494460856720827e-05,
      "loss": 0.0006,
      "step": 2764
    },
    {
      "epoch": 1.0117087449688986,
      "grad_norm": 7.174707889556885,
      "learning_rate": 2.4935376661742985e-05,
      "loss": 0.4242,
      "step": 2765
    },
    {
      "epoch": 1.0120746432491767,
      "grad_norm": 0.025550326332449913,
      "learning_rate": 2.4926144756277696e-05,
      "loss": 0.0007,
      "step": 2766
    },
    {
      "epoch": 1.0124405415294548,
      "grad_norm": 0.013800282031297684,
      "learning_rate": 2.491691285081241e-05,
      "loss": 0.0006,
      "step": 2767
    },
    {
      "epoch": 1.012806439809733,
      "grad_norm": 0.028058933094143867,
      "learning_rate": 2.490768094534712e-05,
      "loss": 0.0007,
      "step": 2768
    },
    {
      "epoch": 1.013172338090011,
      "grad_norm": 0.017024481669068336,
      "learning_rate": 2.4898449039881834e-05,
      "loss": 0.0007,
      "step": 2769
    },
    {
      "epoch": 1.0135382363702892,
      "grad_norm": 0.010079368017613888,
      "learning_rate": 2.4889217134416545e-05,
      "loss": 0.0004,
      "step": 2770
    },
    {
      "epoch": 1.013904134650567,
      "grad_norm": 0.012860339134931564,
      "learning_rate": 2.487998522895126e-05,
      "loss": 0.0005,
      "step": 2771
    },
    {
      "epoch": 1.0142700329308452,
      "grad_norm": 0.02001948468387127,
      "learning_rate": 2.487075332348597e-05,
      "loss": 0.0007,
      "step": 2772
    },
    {
      "epoch": 1.0146359312111233,
      "grad_norm": 0.036402612924575806,
      "learning_rate": 2.486152141802068e-05,
      "loss": 0.0006,
      "step": 2773
    },
    {
      "epoch": 1.0150018294914014,
      "grad_norm": 0.28369125723838806,
      "learning_rate": 2.4852289512555394e-05,
      "loss": 0.0039,
      "step": 2774
    },
    {
      "epoch": 1.0153677277716795,
      "grad_norm": 0.028123287484049797,
      "learning_rate": 2.4843057607090104e-05,
      "loss": 0.0008,
      "step": 2775
    },
    {
      "epoch": 1.0157336260519576,
      "grad_norm": 0.06601066142320633,
      "learning_rate": 2.4833825701624815e-05,
      "loss": 0.0015,
      "step": 2776
    },
    {
      "epoch": 1.0160995243322357,
      "grad_norm": 0.019214052706956863,
      "learning_rate": 2.482459379615953e-05,
      "loss": 0.0007,
      "step": 2777
    },
    {
      "epoch": 1.0164654226125138,
      "grad_norm": 0.0840875655412674,
      "learning_rate": 2.4815361890694243e-05,
      "loss": 0.0011,
      "step": 2778
    },
    {
      "epoch": 1.0168313208927917,
      "grad_norm": 0.05446089804172516,
      "learning_rate": 2.480612998522895e-05,
      "loss": 0.0011,
      "step": 2779
    },
    {
      "epoch": 1.0171972191730698,
      "grad_norm": 0.010811123996973038,
      "learning_rate": 2.4796898079763664e-05,
      "loss": 0.0005,
      "step": 2780
    },
    {
      "epoch": 1.017563117453348,
      "grad_norm": 0.011407673358917236,
      "learning_rate": 2.4787666174298378e-05,
      "loss": 0.0004,
      "step": 2781
    },
    {
      "epoch": 1.017929015733626,
      "grad_norm": 1.3037885427474976,
      "learning_rate": 2.4778434268833088e-05,
      "loss": 0.0102,
      "step": 2782
    },
    {
      "epoch": 1.0182949140139042,
      "grad_norm": 0.014581265859305859,
      "learning_rate": 2.47692023633678e-05,
      "loss": 0.0006,
      "step": 2783
    },
    {
      "epoch": 1.0186608122941823,
      "grad_norm": 0.01095176488161087,
      "learning_rate": 2.4759970457902513e-05,
      "loss": 0.0005,
      "step": 2784
    },
    {
      "epoch": 1.0190267105744604,
      "grad_norm": 0.015049893409013748,
      "learning_rate": 2.4750738552437223e-05,
      "loss": 0.0006,
      "step": 2785
    },
    {
      "epoch": 1.0193926088547385,
      "grad_norm": 2.1174614429473877,
      "learning_rate": 2.4741506646971937e-05,
      "loss": 0.0168,
      "step": 2786
    },
    {
      "epoch": 1.0197585071350164,
      "grad_norm": 0.0429539792239666,
      "learning_rate": 2.4732274741506648e-05,
      "loss": 0.0011,
      "step": 2787
    },
    {
      "epoch": 1.0201244054152945,
      "grad_norm": 0.024778880178928375,
      "learning_rate": 2.4723042836041362e-05,
      "loss": 0.0007,
      "step": 2788
    },
    {
      "epoch": 1.0204903036955726,
      "grad_norm": 6.050967216491699,
      "learning_rate": 2.4713810930576072e-05,
      "loss": 0.0601,
      "step": 2789
    },
    {
      "epoch": 1.0208562019758507,
      "grad_norm": 0.4151097238063812,
      "learning_rate": 2.4704579025110783e-05,
      "loss": 0.0044,
      "step": 2790
    },
    {
      "epoch": 1.0212221002561288,
      "grad_norm": 0.014708014205098152,
      "learning_rate": 2.4695347119645497e-05,
      "loss": 0.0006,
      "step": 2791
    },
    {
      "epoch": 1.021587998536407,
      "grad_norm": 0.1672939509153366,
      "learning_rate": 2.4686115214180207e-05,
      "loss": 0.0022,
      "step": 2792
    },
    {
      "epoch": 1.021953896816685,
      "grad_norm": 6.963943958282471,
      "learning_rate": 2.467688330871492e-05,
      "loss": 0.0125,
      "step": 2793
    },
    {
      "epoch": 1.0223197950969631,
      "grad_norm": 0.011313047260046005,
      "learning_rate": 2.4667651403249632e-05,
      "loss": 0.0005,
      "step": 2794
    },
    {
      "epoch": 1.022685693377241,
      "grad_norm": 0.010489699430763721,
      "learning_rate": 2.4658419497784342e-05,
      "loss": 0.0003,
      "step": 2795
    },
    {
      "epoch": 1.0230515916575191,
      "grad_norm": 22.583091735839844,
      "learning_rate": 2.4649187592319056e-05,
      "loss": 0.165,
      "step": 2796
    },
    {
      "epoch": 1.0234174899377972,
      "grad_norm": 9.215481758117676,
      "learning_rate": 2.463995568685377e-05,
      "loss": 0.036,
      "step": 2797
    },
    {
      "epoch": 1.0237833882180754,
      "grad_norm": 0.007602694444358349,
      "learning_rate": 2.4630723781388477e-05,
      "loss": 0.0003,
      "step": 2798
    },
    {
      "epoch": 1.0241492864983535,
      "grad_norm": 0.020928367972373962,
      "learning_rate": 2.462149187592319e-05,
      "loss": 0.0007,
      "step": 2799
    },
    {
      "epoch": 1.0245151847786316,
      "grad_norm": 0.015208023600280285,
      "learning_rate": 2.4612259970457905e-05,
      "loss": 0.0005,
      "step": 2800
    },
    {
      "epoch": 1.0248810830589097,
      "grad_norm": 0.011537192389369011,
      "learning_rate": 2.4603028064992616e-05,
      "loss": 0.0005,
      "step": 2801
    },
    {
      "epoch": 1.0252469813391878,
      "grad_norm": 0.007692322134971619,
      "learning_rate": 2.4593796159527326e-05,
      "loss": 0.0003,
      "step": 2802
    },
    {
      "epoch": 1.025612879619466,
      "grad_norm": 0.048709686845541,
      "learning_rate": 2.458456425406204e-05,
      "loss": 0.001,
      "step": 2803
    },
    {
      "epoch": 1.0259787778997438,
      "grad_norm": 0.00933071132749319,
      "learning_rate": 2.457533234859675e-05,
      "loss": 0.0004,
      "step": 2804
    },
    {
      "epoch": 1.026344676180022,
      "grad_norm": 0.005618406925350428,
      "learning_rate": 2.456610044313146e-05,
      "loss": 0.0003,
      "step": 2805
    },
    {
      "epoch": 1.0267105744603,
      "grad_norm": 0.009875643998384476,
      "learning_rate": 2.4556868537666175e-05,
      "loss": 0.0004,
      "step": 2806
    },
    {
      "epoch": 1.0270764727405781,
      "grad_norm": 0.0055536567233502865,
      "learning_rate": 2.454763663220089e-05,
      "loss": 0.0003,
      "step": 2807
    },
    {
      "epoch": 1.0274423710208562,
      "grad_norm": 0.015558036975562572,
      "learning_rate": 2.45384047267356e-05,
      "loss": 0.0006,
      "step": 2808
    },
    {
      "epoch": 1.0278082693011343,
      "grad_norm": 0.010067100636661053,
      "learning_rate": 2.452917282127031e-05,
      "loss": 0.0004,
      "step": 2809
    },
    {
      "epoch": 1.0281741675814124,
      "grad_norm": 0.019061608240008354,
      "learning_rate": 2.4519940915805024e-05,
      "loss": 0.0005,
      "step": 2810
    },
    {
      "epoch": 1.0285400658616906,
      "grad_norm": 0.021458622068166733,
      "learning_rate": 2.4510709010339735e-05,
      "loss": 0.0004,
      "step": 2811
    },
    {
      "epoch": 1.0289059641419684,
      "grad_norm": 0.008132196962833405,
      "learning_rate": 2.450147710487445e-05,
      "loss": 0.0004,
      "step": 2812
    },
    {
      "epoch": 1.0292718624222466,
      "grad_norm": 0.007447022013366222,
      "learning_rate": 2.449224519940916e-05,
      "loss": 0.0003,
      "step": 2813
    },
    {
      "epoch": 1.0296377607025247,
      "grad_norm": 0.033418986946344376,
      "learning_rate": 2.448301329394387e-05,
      "loss": 0.0006,
      "step": 2814
    },
    {
      "epoch": 1.0300036589828028,
      "grad_norm": 0.005853499285876751,
      "learning_rate": 2.4473781388478584e-05,
      "loss": 0.0003,
      "step": 2815
    },
    {
      "epoch": 1.0303695572630809,
      "grad_norm": 0.005615739151835442,
      "learning_rate": 2.4464549483013294e-05,
      "loss": 0.0003,
      "step": 2816
    },
    {
      "epoch": 1.030735455543359,
      "grad_norm": 0.033910781145095825,
      "learning_rate": 2.4455317577548005e-05,
      "loss": 0.0006,
      "step": 2817
    },
    {
      "epoch": 1.031101353823637,
      "grad_norm": 0.011294005438685417,
      "learning_rate": 2.444608567208272e-05,
      "loss": 0.0004,
      "step": 2818
    },
    {
      "epoch": 1.0314672521039152,
      "grad_norm": 0.009336862713098526,
      "learning_rate": 2.4436853766617433e-05,
      "loss": 0.0004,
      "step": 2819
    },
    {
      "epoch": 1.031833150384193,
      "grad_norm": 0.979201078414917,
      "learning_rate": 2.4427621861152143e-05,
      "loss": 0.0056,
      "step": 2820
    },
    {
      "epoch": 1.0321990486644712,
      "grad_norm": 0.0047829169780015945,
      "learning_rate": 2.4418389955686854e-05,
      "loss": 0.0002,
      "step": 2821
    },
    {
      "epoch": 1.0325649469447493,
      "grad_norm": 0.009327632375061512,
      "learning_rate": 2.4409158050221568e-05,
      "loss": 0.0004,
      "step": 2822
    },
    {
      "epoch": 1.0329308452250274,
      "grad_norm": 0.011036950163543224,
      "learning_rate": 2.439992614475628e-05,
      "loss": 0.0004,
      "step": 2823
    },
    {
      "epoch": 1.0332967435053055,
      "grad_norm": 0.01911260187625885,
      "learning_rate": 2.439069423929099e-05,
      "loss": 0.0005,
      "step": 2824
    },
    {
      "epoch": 1.0336626417855836,
      "grad_norm": 8.878238677978516,
      "learning_rate": 2.4381462333825703e-05,
      "loss": 0.2222,
      "step": 2825
    },
    {
      "epoch": 1.0340285400658618,
      "grad_norm": 0.15310177206993103,
      "learning_rate": 2.4372230428360417e-05,
      "loss": 0.0008,
      "step": 2826
    },
    {
      "epoch": 1.0343944383461399,
      "grad_norm": 5.251101016998291,
      "learning_rate": 2.4362998522895127e-05,
      "loss": 0.1977,
      "step": 2827
    },
    {
      "epoch": 1.0347603366264178,
      "grad_norm": 0.015653613954782486,
      "learning_rate": 2.4353766617429838e-05,
      "loss": 0.0004,
      "step": 2828
    },
    {
      "epoch": 1.0351262349066959,
      "grad_norm": 0.22587384283542633,
      "learning_rate": 2.4344534711964552e-05,
      "loss": 0.0031,
      "step": 2829
    },
    {
      "epoch": 1.035492133186974,
      "grad_norm": 0.08038066327571869,
      "learning_rate": 2.4335302806499262e-05,
      "loss": 0.0008,
      "step": 2830
    },
    {
      "epoch": 1.035858031467252,
      "grad_norm": 0.006741483695805073,
      "learning_rate": 2.4326070901033973e-05,
      "loss": 0.0003,
      "step": 2831
    },
    {
      "epoch": 1.0362239297475302,
      "grad_norm": 10.018200874328613,
      "learning_rate": 2.4316838995568687e-05,
      "loss": 0.4292,
      "step": 2832
    },
    {
      "epoch": 1.0365898280278083,
      "grad_norm": 0.08071452379226685,
      "learning_rate": 2.4307607090103397e-05,
      "loss": 0.0011,
      "step": 2833
    },
    {
      "epoch": 1.0369557263080864,
      "grad_norm": 15.924749374389648,
      "learning_rate": 2.429837518463811e-05,
      "loss": 0.1977,
      "step": 2834
    },
    {
      "epoch": 1.0373216245883645,
      "grad_norm": 0.009641203097999096,
      "learning_rate": 2.4289143279172822e-05,
      "loss": 0.0003,
      "step": 2835
    },
    {
      "epoch": 1.0376875228686426,
      "grad_norm": 27.263856887817383,
      "learning_rate": 2.4279911373707532e-05,
      "loss": 0.3193,
      "step": 2836
    },
    {
      "epoch": 1.0380534211489205,
      "grad_norm": 0.23043079674243927,
      "learning_rate": 2.4270679468242246e-05,
      "loss": 0.0034,
      "step": 2837
    },
    {
      "epoch": 1.0384193194291986,
      "grad_norm": 12.759759902954102,
      "learning_rate": 2.426144756277696e-05,
      "loss": 0.0809,
      "step": 2838
    },
    {
      "epoch": 1.0387852177094767,
      "grad_norm": 0.0052197822369635105,
      "learning_rate": 2.425221565731167e-05,
      "loss": 0.0002,
      "step": 2839
    },
    {
      "epoch": 1.0391511159897548,
      "grad_norm": 0.007017247844487429,
      "learning_rate": 2.424298375184638e-05,
      "loss": 0.0003,
      "step": 2840
    },
    {
      "epoch": 1.039517014270033,
      "grad_norm": 0.007474356330931187,
      "learning_rate": 2.4233751846381095e-05,
      "loss": 0.0003,
      "step": 2841
    },
    {
      "epoch": 1.039882912550311,
      "grad_norm": 0.12507842481136322,
      "learning_rate": 2.4224519940915806e-05,
      "loss": 0.001,
      "step": 2842
    },
    {
      "epoch": 1.0402488108305892,
      "grad_norm": 9.174826622009277,
      "learning_rate": 2.4215288035450516e-05,
      "loss": 0.0164,
      "step": 2843
    },
    {
      "epoch": 1.0406147091108673,
      "grad_norm": 0.00745109049603343,
      "learning_rate": 2.420605612998523e-05,
      "loss": 0.0003,
      "step": 2844
    },
    {
      "epoch": 1.0409806073911452,
      "grad_norm": 0.011495605111122131,
      "learning_rate": 2.4196824224519944e-05,
      "loss": 0.0003,
      "step": 2845
    },
    {
      "epoch": 1.0413465056714233,
      "grad_norm": 3.4946606159210205,
      "learning_rate": 2.418759231905465e-05,
      "loss": 0.0117,
      "step": 2846
    },
    {
      "epoch": 1.0417124039517014,
      "grad_norm": 0.007262317929416895,
      "learning_rate": 2.4178360413589365e-05,
      "loss": 0.0003,
      "step": 2847
    },
    {
      "epoch": 1.0420783022319795,
      "grad_norm": 0.00984103512018919,
      "learning_rate": 2.416912850812408e-05,
      "loss": 0.0004,
      "step": 2848
    },
    {
      "epoch": 1.0424442005122576,
      "grad_norm": 0.008908335119485855,
      "learning_rate": 2.415989660265879e-05,
      "loss": 0.0004,
      "step": 2849
    },
    {
      "epoch": 1.0428100987925357,
      "grad_norm": 0.00895612221211195,
      "learning_rate": 2.41506646971935e-05,
      "loss": 0.0003,
      "step": 2850
    },
    {
      "epoch": 1.0431759970728138,
      "grad_norm": 0.004742453806102276,
      "learning_rate": 2.4141432791728214e-05,
      "loss": 0.0002,
      "step": 2851
    },
    {
      "epoch": 1.043541895353092,
      "grad_norm": 6.4714484214782715,
      "learning_rate": 2.4132200886262925e-05,
      "loss": 0.04,
      "step": 2852
    },
    {
      "epoch": 1.0439077936333698,
      "grad_norm": 0.024991340935230255,
      "learning_rate": 2.412296898079764e-05,
      "loss": 0.0005,
      "step": 2853
    },
    {
      "epoch": 1.044273691913648,
      "grad_norm": 0.006916295737028122,
      "learning_rate": 2.411373707533235e-05,
      "loss": 0.0003,
      "step": 2854
    },
    {
      "epoch": 1.044639590193926,
      "grad_norm": 0.006792658474296331,
      "learning_rate": 2.410450516986706e-05,
      "loss": 0.0003,
      "step": 2855
    },
    {
      "epoch": 1.0450054884742042,
      "grad_norm": 0.008394352160394192,
      "learning_rate": 2.4095273264401774e-05,
      "loss": 0.0003,
      "step": 2856
    },
    {
      "epoch": 1.0453713867544823,
      "grad_norm": 1.9613031148910522,
      "learning_rate": 2.4086041358936488e-05,
      "loss": 0.473,
      "step": 2857
    },
    {
      "epoch": 1.0457372850347604,
      "grad_norm": 0.0052049350924789906,
      "learning_rate": 2.40768094534712e-05,
      "loss": 0.0002,
      "step": 2858
    },
    {
      "epoch": 1.0461031833150385,
      "grad_norm": 0.00585103128105402,
      "learning_rate": 2.406757754800591e-05,
      "loss": 0.0003,
      "step": 2859
    },
    {
      "epoch": 1.0464690815953166,
      "grad_norm": 0.33803707361221313,
      "learning_rate": 2.4058345642540623e-05,
      "loss": 0.0021,
      "step": 2860
    },
    {
      "epoch": 1.0468349798755945,
      "grad_norm": 0.01671196147799492,
      "learning_rate": 2.4049113737075333e-05,
      "loss": 0.0007,
      "step": 2861
    },
    {
      "epoch": 1.0472008781558726,
      "grad_norm": 0.10107366740703583,
      "learning_rate": 2.4039881831610044e-05,
      "loss": 0.0016,
      "step": 2862
    },
    {
      "epoch": 1.0475667764361507,
      "grad_norm": 0.056211426854133606,
      "learning_rate": 2.4030649926144758e-05,
      "loss": 0.001,
      "step": 2863
    },
    {
      "epoch": 1.0479326747164288,
      "grad_norm": 0.025993840768933296,
      "learning_rate": 2.4021418020679472e-05,
      "loss": 0.0006,
      "step": 2864
    },
    {
      "epoch": 1.048298572996707,
      "grad_norm": 0.008937486447393894,
      "learning_rate": 2.401218611521418e-05,
      "loss": 0.0004,
      "step": 2865
    },
    {
      "epoch": 1.048664471276985,
      "grad_norm": 0.013743315823376179,
      "learning_rate": 2.4002954209748893e-05,
      "loss": 0.0006,
      "step": 2866
    },
    {
      "epoch": 1.0490303695572631,
      "grad_norm": 0.013949545100331306,
      "learning_rate": 2.3993722304283607e-05,
      "loss": 0.0006,
      "step": 2867
    },
    {
      "epoch": 1.0493962678375413,
      "grad_norm": 0.008974815718829632,
      "learning_rate": 2.3984490398818317e-05,
      "loss": 0.0004,
      "step": 2868
    },
    {
      "epoch": 1.0497621661178191,
      "grad_norm": 0.030996210873126984,
      "learning_rate": 2.3975258493353028e-05,
      "loss": 0.0007,
      "step": 2869
    },
    {
      "epoch": 1.0501280643980972,
      "grad_norm": 0.01285515632480383,
      "learning_rate": 2.3966026587887742e-05,
      "loss": 0.0005,
      "step": 2870
    },
    {
      "epoch": 1.0504939626783754,
      "grad_norm": 0.014583093114197254,
      "learning_rate": 2.3956794682422452e-05,
      "loss": 0.0006,
      "step": 2871
    },
    {
      "epoch": 1.0508598609586535,
      "grad_norm": 39.1047477722168,
      "learning_rate": 2.3947562776957166e-05,
      "loss": 0.0677,
      "step": 2872
    },
    {
      "epoch": 1.0512257592389316,
      "grad_norm": 0.02143367938697338,
      "learning_rate": 2.3938330871491877e-05,
      "loss": 0.0008,
      "step": 2873
    },
    {
      "epoch": 1.0515916575192097,
      "grad_norm": 0.016595570370554924,
      "learning_rate": 2.3929098966026588e-05,
      "loss": 0.0006,
      "step": 2874
    },
    {
      "epoch": 1.0519575557994878,
      "grad_norm": 2.115847587585449,
      "learning_rate": 2.39198670605613e-05,
      "loss": 0.0072,
      "step": 2875
    },
    {
      "epoch": 1.052323454079766,
      "grad_norm": 0.01625988446176052,
      "learning_rate": 2.3910635155096012e-05,
      "loss": 0.0007,
      "step": 2876
    },
    {
      "epoch": 1.0526893523600438,
      "grad_norm": 0.021268539130687714,
      "learning_rate": 2.3901403249630726e-05,
      "loss": 0.0006,
      "step": 2877
    },
    {
      "epoch": 1.053055250640322,
      "grad_norm": 0.0217280313372612,
      "learning_rate": 2.3892171344165436e-05,
      "loss": 0.0008,
      "step": 2878
    },
    {
      "epoch": 1.0534211489206,
      "grad_norm": 6.653130531311035,
      "learning_rate": 2.388293943870015e-05,
      "loss": 0.026,
      "step": 2879
    },
    {
      "epoch": 1.0537870472008781,
      "grad_norm": 0.012279422953724861,
      "learning_rate": 2.387370753323486e-05,
      "loss": 0.0005,
      "step": 2880
    },
    {
      "epoch": 1.0541529454811562,
      "grad_norm": 0.012521792203187943,
      "learning_rate": 2.386447562776957e-05,
      "loss": 0.0005,
      "step": 2881
    },
    {
      "epoch": 1.0545188437614343,
      "grad_norm": 0.016624923795461655,
      "learning_rate": 2.3855243722304285e-05,
      "loss": 0.0007,
      "step": 2882
    },
    {
      "epoch": 1.0548847420417125,
      "grad_norm": 0.29372310638427734,
      "learning_rate": 2.3846011816839e-05,
      "loss": 0.003,
      "step": 2883
    },
    {
      "epoch": 1.0552506403219906,
      "grad_norm": 0.021887777373194695,
      "learning_rate": 2.3836779911373707e-05,
      "loss": 0.0009,
      "step": 2884
    },
    {
      "epoch": 1.0556165386022687,
      "grad_norm": 0.014837740920484066,
      "learning_rate": 2.382754800590842e-05,
      "loss": 0.0006,
      "step": 2885
    },
    {
      "epoch": 1.0559824368825466,
      "grad_norm": 0.015066131018102169,
      "learning_rate": 2.3818316100443134e-05,
      "loss": 0.0006,
      "step": 2886
    },
    {
      "epoch": 1.0563483351628247,
      "grad_norm": 0.005373795982450247,
      "learning_rate": 2.3809084194977845e-05,
      "loss": 0.0002,
      "step": 2887
    },
    {
      "epoch": 1.0567142334431028,
      "grad_norm": 0.014918684028089046,
      "learning_rate": 2.3799852289512556e-05,
      "loss": 0.0006,
      "step": 2888
    },
    {
      "epoch": 1.057080131723381,
      "grad_norm": 0.02860303781926632,
      "learning_rate": 2.379062038404727e-05,
      "loss": 0.001,
      "step": 2889
    },
    {
      "epoch": 1.057446030003659,
      "grad_norm": 0.01210333127528429,
      "learning_rate": 2.378138847858198e-05,
      "loss": 0.0005,
      "step": 2890
    },
    {
      "epoch": 1.057811928283937,
      "grad_norm": 0.014008075930178165,
      "learning_rate": 2.377215657311669e-05,
      "loss": 0.0006,
      "step": 2891
    },
    {
      "epoch": 1.0581778265642152,
      "grad_norm": 1.1322237253189087,
      "learning_rate": 2.3762924667651405e-05,
      "loss": 0.0073,
      "step": 2892
    },
    {
      "epoch": 1.0585437248444933,
      "grad_norm": 0.012929214164614677,
      "learning_rate": 2.3753692762186115e-05,
      "loss": 0.0006,
      "step": 2893
    },
    {
      "epoch": 1.0589096231247712,
      "grad_norm": 0.00777131924405694,
      "learning_rate": 2.374446085672083e-05,
      "loss": 0.0003,
      "step": 2894
    },
    {
      "epoch": 1.0592755214050493,
      "grad_norm": 0.42729246616363525,
      "learning_rate": 2.373522895125554e-05,
      "loss": 0.0034,
      "step": 2895
    },
    {
      "epoch": 1.0596414196853274,
      "grad_norm": 0.0063050794415175915,
      "learning_rate": 2.3725997045790253e-05,
      "loss": 0.0003,
      "step": 2896
    },
    {
      "epoch": 1.0600073179656055,
      "grad_norm": 0.011966973543167114,
      "learning_rate": 2.3716765140324964e-05,
      "loss": 0.0005,
      "step": 2897
    },
    {
      "epoch": 1.0603732162458837,
      "grad_norm": 0.01991327665746212,
      "learning_rate": 2.3707533234859678e-05,
      "loss": 0.0005,
      "step": 2898
    },
    {
      "epoch": 1.0607391145261618,
      "grad_norm": 0.009026345796883106,
      "learning_rate": 2.369830132939439e-05,
      "loss": 0.0004,
      "step": 2899
    },
    {
      "epoch": 1.0611050128064399,
      "grad_norm": 0.06006752327084541,
      "learning_rate": 2.36890694239291e-05,
      "loss": 0.0009,
      "step": 2900
    },
    {
      "epoch": 1.061470911086718,
      "grad_norm": 0.02576424553990364,
      "learning_rate": 2.3679837518463813e-05,
      "loss": 0.0005,
      "step": 2901
    },
    {
      "epoch": 1.0618368093669959,
      "grad_norm": 0.01132575049996376,
      "learning_rate": 2.3670605612998527e-05,
      "loss": 0.0005,
      "step": 2902
    },
    {
      "epoch": 1.062202707647274,
      "grad_norm": 0.04211842641234398,
      "learning_rate": 2.3661373707533234e-05,
      "loss": 0.0008,
      "step": 2903
    },
    {
      "epoch": 1.062568605927552,
      "grad_norm": 0.03917732089757919,
      "learning_rate": 2.3652141802067948e-05,
      "loss": 0.0008,
      "step": 2904
    },
    {
      "epoch": 1.0629345042078302,
      "grad_norm": 0.005875925067812204,
      "learning_rate": 2.3642909896602662e-05,
      "loss": 0.0003,
      "step": 2905
    },
    {
      "epoch": 1.0633004024881083,
      "grad_norm": 0.008164709433913231,
      "learning_rate": 2.363367799113737e-05,
      "loss": 0.0004,
      "step": 2906
    },
    {
      "epoch": 1.0636663007683864,
      "grad_norm": 13.201425552368164,
      "learning_rate": 2.3624446085672083e-05,
      "loss": 0.1557,
      "step": 2907
    },
    {
      "epoch": 1.0640321990486645,
      "grad_norm": 0.025470459833741188,
      "learning_rate": 2.3615214180206797e-05,
      "loss": 0.0006,
      "step": 2908
    },
    {
      "epoch": 1.0643980973289426,
      "grad_norm": 1.9917594194412231,
      "learning_rate": 2.3605982274741508e-05,
      "loss": 0.5006,
      "step": 2909
    },
    {
      "epoch": 1.0647639956092205,
      "grad_norm": 0.007372874300926924,
      "learning_rate": 2.3596750369276218e-05,
      "loss": 0.0003,
      "step": 2910
    },
    {
      "epoch": 1.0651298938894986,
      "grad_norm": 0.016981983557343483,
      "learning_rate": 2.3587518463810932e-05,
      "loss": 0.0005,
      "step": 2911
    },
    {
      "epoch": 1.0654957921697767,
      "grad_norm": 0.12204470485448837,
      "learning_rate": 2.3578286558345643e-05,
      "loss": 0.0013,
      "step": 2912
    },
    {
      "epoch": 1.0658616904500549,
      "grad_norm": 0.011132786981761456,
      "learning_rate": 2.3569054652880357e-05,
      "loss": 0.0005,
      "step": 2913
    },
    {
      "epoch": 1.066227588730333,
      "grad_norm": 0.021876458078622818,
      "learning_rate": 2.3559822747415067e-05,
      "loss": 0.0007,
      "step": 2914
    },
    {
      "epoch": 1.066593487010611,
      "grad_norm": 0.009426278993487358,
      "learning_rate": 2.355059084194978e-05,
      "loss": 0.0004,
      "step": 2915
    },
    {
      "epoch": 1.0669593852908892,
      "grad_norm": 14.507377624511719,
      "learning_rate": 2.354135893648449e-05,
      "loss": 0.1917,
      "step": 2916
    },
    {
      "epoch": 1.0673252835711673,
      "grad_norm": 0.015073688700795174,
      "learning_rate": 2.3532127031019202e-05,
      "loss": 0.0006,
      "step": 2917
    },
    {
      "epoch": 1.0676911818514454,
      "grad_norm": 0.014668206684291363,
      "learning_rate": 2.3522895125553916e-05,
      "loss": 0.0006,
      "step": 2918
    },
    {
      "epoch": 1.0680570801317233,
      "grad_norm": 6.153843879699707,
      "learning_rate": 2.3513663220088627e-05,
      "loss": 0.2361,
      "step": 2919
    },
    {
      "epoch": 1.0684229784120014,
      "grad_norm": 0.9889324903488159,
      "learning_rate": 2.350443131462334e-05,
      "loss": 0.0093,
      "step": 2920
    },
    {
      "epoch": 1.0687888766922795,
      "grad_norm": 0.01992800645530224,
      "learning_rate": 2.349519940915805e-05,
      "loss": 0.0008,
      "step": 2921
    },
    {
      "epoch": 1.0691547749725576,
      "grad_norm": 0.053348418325185776,
      "learning_rate": 2.348596750369276e-05,
      "loss": 0.0009,
      "step": 2922
    },
    {
      "epoch": 1.0695206732528357,
      "grad_norm": 0.2926748991012573,
      "learning_rate": 2.3476735598227476e-05,
      "loss": 0.0018,
      "step": 2923
    },
    {
      "epoch": 1.0698865715331138,
      "grad_norm": 0.01931336522102356,
      "learning_rate": 2.346750369276219e-05,
      "loss": 0.0008,
      "step": 2924
    },
    {
      "epoch": 1.070252469813392,
      "grad_norm": 0.027931908145546913,
      "learning_rate": 2.3458271787296897e-05,
      "loss": 0.0011,
      "step": 2925
    },
    {
      "epoch": 1.07061836809367,
      "grad_norm": 0.023586565628647804,
      "learning_rate": 2.344903988183161e-05,
      "loss": 0.0009,
      "step": 2926
    },
    {
      "epoch": 1.070984266373948,
      "grad_norm": 0.00982736237347126,
      "learning_rate": 2.3439807976366325e-05,
      "loss": 0.0004,
      "step": 2927
    },
    {
      "epoch": 1.071350164654226,
      "grad_norm": 0.016268063336610794,
      "learning_rate": 2.3430576070901035e-05,
      "loss": 0.0006,
      "step": 2928
    },
    {
      "epoch": 1.0717160629345042,
      "grad_norm": 0.017533745616674423,
      "learning_rate": 2.3421344165435746e-05,
      "loss": 0.0007,
      "step": 2929
    },
    {
      "epoch": 1.0720819612147823,
      "grad_norm": 6.293336391448975,
      "learning_rate": 2.341211225997046e-05,
      "loss": 0.4844,
      "step": 2930
    },
    {
      "epoch": 1.0724478594950604,
      "grad_norm": 1.2797914743423462,
      "learning_rate": 2.340288035450517e-05,
      "loss": 0.0142,
      "step": 2931
    },
    {
      "epoch": 1.0728137577753385,
      "grad_norm": 0.08089164644479752,
      "learning_rate": 2.339364844903988e-05,
      "loss": 0.0011,
      "step": 2932
    },
    {
      "epoch": 1.0731796560556166,
      "grad_norm": 0.018480738624930382,
      "learning_rate": 2.3384416543574595e-05,
      "loss": 0.0007,
      "step": 2933
    },
    {
      "epoch": 1.0735455543358947,
      "grad_norm": 0.010367212817072868,
      "learning_rate": 2.337518463810931e-05,
      "loss": 0.0005,
      "step": 2934
    },
    {
      "epoch": 1.0739114526161726,
      "grad_norm": 0.01418833527714014,
      "learning_rate": 2.336595273264402e-05,
      "loss": 0.0006,
      "step": 2935
    },
    {
      "epoch": 1.0742773508964507,
      "grad_norm": 0.017649943009018898,
      "learning_rate": 2.335672082717873e-05,
      "loss": 0.0008,
      "step": 2936
    },
    {
      "epoch": 1.0746432491767288,
      "grad_norm": 0.6315052509307861,
      "learning_rate": 2.3347488921713444e-05,
      "loss": 0.0038,
      "step": 2937
    },
    {
      "epoch": 1.075009147457007,
      "grad_norm": 0.01677023246884346,
      "learning_rate": 2.3338257016248154e-05,
      "loss": 0.0007,
      "step": 2938
    },
    {
      "epoch": 1.075375045737285,
      "grad_norm": 0.8695914149284363,
      "learning_rate": 2.3329025110782868e-05,
      "loss": 0.0043,
      "step": 2939
    },
    {
      "epoch": 1.0757409440175631,
      "grad_norm": 0.02772458828985691,
      "learning_rate": 2.331979320531758e-05,
      "loss": 0.0008,
      "step": 2940
    },
    {
      "epoch": 1.0761068422978413,
      "grad_norm": 0.028052445501089096,
      "learning_rate": 2.331056129985229e-05,
      "loss": 0.001,
      "step": 2941
    },
    {
      "epoch": 1.0764727405781194,
      "grad_norm": 0.08270027488470078,
      "learning_rate": 2.3301329394387003e-05,
      "loss": 0.0011,
      "step": 2942
    },
    {
      "epoch": 1.0768386388583973,
      "grad_norm": 0.15207186341285706,
      "learning_rate": 2.3292097488921717e-05,
      "loss": 0.0012,
      "step": 2943
    },
    {
      "epoch": 1.0772045371386754,
      "grad_norm": 0.01791827753186226,
      "learning_rate": 2.3282865583456424e-05,
      "loss": 0.0007,
      "step": 2944
    },
    {
      "epoch": 1.0775704354189535,
      "grad_norm": 0.014745253138244152,
      "learning_rate": 2.3273633677991138e-05,
      "loss": 0.0005,
      "step": 2945
    },
    {
      "epoch": 1.0779363336992316,
      "grad_norm": 0.13499760627746582,
      "learning_rate": 2.3264401772525852e-05,
      "loss": 0.0013,
      "step": 2946
    },
    {
      "epoch": 1.0783022319795097,
      "grad_norm": 0.012736164033412933,
      "learning_rate": 2.3255169867060563e-05,
      "loss": 0.0006,
      "step": 2947
    },
    {
      "epoch": 1.0786681302597878,
      "grad_norm": 6.83962345123291,
      "learning_rate": 2.3245937961595273e-05,
      "loss": 0.3725,
      "step": 2948
    },
    {
      "epoch": 1.079034028540066,
      "grad_norm": 5.972532749176025,
      "learning_rate": 2.3236706056129987e-05,
      "loss": 0.133,
      "step": 2949
    },
    {
      "epoch": 1.079399926820344,
      "grad_norm": 0.01595529541373253,
      "learning_rate": 2.3227474150664698e-05,
      "loss": 0.0006,
      "step": 2950
    },
    {
      "epoch": 1.0797658251006221,
      "grad_norm": 2.856369972229004,
      "learning_rate": 2.3218242245199408e-05,
      "loss": 0.0115,
      "step": 2951
    },
    {
      "epoch": 1.0801317233809,
      "grad_norm": 13.93835163116455,
      "learning_rate": 2.3209010339734122e-05,
      "loss": 0.0361,
      "step": 2952
    },
    {
      "epoch": 1.0804976216611781,
      "grad_norm": 0.016563212499022484,
      "learning_rate": 2.3199778434268836e-05,
      "loss": 0.0007,
      "step": 2953
    },
    {
      "epoch": 1.0808635199414562,
      "grad_norm": 0.012709799222648144,
      "learning_rate": 2.3190546528803547e-05,
      "loss": 0.0006,
      "step": 2954
    },
    {
      "epoch": 1.0812294182217344,
      "grad_norm": 31.79897117614746,
      "learning_rate": 2.3181314623338257e-05,
      "loss": 0.067,
      "step": 2955
    },
    {
      "epoch": 1.0815953165020125,
      "grad_norm": 0.0402151457965374,
      "learning_rate": 2.317208271787297e-05,
      "loss": 0.0009,
      "step": 2956
    },
    {
      "epoch": 1.0819612147822906,
      "grad_norm": 0.017955685034394264,
      "learning_rate": 2.316285081240768e-05,
      "loss": 0.0006,
      "step": 2957
    },
    {
      "epoch": 1.0823271130625687,
      "grad_norm": 0.010181659832596779,
      "learning_rate": 2.3153618906942396e-05,
      "loss": 0.0005,
      "step": 2958
    },
    {
      "epoch": 1.0826930113428468,
      "grad_norm": 0.06656995415687561,
      "learning_rate": 2.3144387001477106e-05,
      "loss": 0.001,
      "step": 2959
    },
    {
      "epoch": 1.0830589096231247,
      "grad_norm": 0.01657700538635254,
      "learning_rate": 2.3135155096011817e-05,
      "loss": 0.0007,
      "step": 2960
    },
    {
      "epoch": 1.0834248079034028,
      "grad_norm": 0.015653757378458977,
      "learning_rate": 2.312592319054653e-05,
      "loss": 0.0007,
      "step": 2961
    },
    {
      "epoch": 1.083790706183681,
      "grad_norm": 0.045003682374954224,
      "learning_rate": 2.311669128508124e-05,
      "loss": 0.0012,
      "step": 2962
    },
    {
      "epoch": 1.084156604463959,
      "grad_norm": 0.013579069636762142,
      "learning_rate": 2.3107459379615952e-05,
      "loss": 0.0005,
      "step": 2963
    },
    {
      "epoch": 1.0845225027442371,
      "grad_norm": 0.06305903941392899,
      "learning_rate": 2.3098227474150666e-05,
      "loss": 0.001,
      "step": 2964
    },
    {
      "epoch": 1.0848884010245152,
      "grad_norm": 0.014161727391183376,
      "learning_rate": 2.308899556868538e-05,
      "loss": 0.0006,
      "step": 2965
    },
    {
      "epoch": 1.0852542993047933,
      "grad_norm": 0.01309218816459179,
      "learning_rate": 2.307976366322009e-05,
      "loss": 0.0006,
      "step": 2966
    },
    {
      "epoch": 1.0856201975850714,
      "grad_norm": 0.01727750152349472,
      "learning_rate": 2.30705317577548e-05,
      "loss": 0.0007,
      "step": 2967
    },
    {
      "epoch": 1.0859860958653493,
      "grad_norm": 0.03421613574028015,
      "learning_rate": 2.3061299852289515e-05,
      "loss": 0.0007,
      "step": 2968
    },
    {
      "epoch": 1.0863519941456274,
      "grad_norm": 0.02496761456131935,
      "learning_rate": 2.3052067946824225e-05,
      "loss": 0.0007,
      "step": 2969
    },
    {
      "epoch": 1.0867178924259056,
      "grad_norm": 0.015244511887431145,
      "learning_rate": 2.3042836041358936e-05,
      "loss": 0.0006,
      "step": 2970
    },
    {
      "epoch": 1.0870837907061837,
      "grad_norm": 0.03327210247516632,
      "learning_rate": 2.303360413589365e-05,
      "loss": 0.0007,
      "step": 2971
    },
    {
      "epoch": 1.0874496889864618,
      "grad_norm": 0.012945636175572872,
      "learning_rate": 2.3024372230428364e-05,
      "loss": 0.0005,
      "step": 2972
    },
    {
      "epoch": 1.0878155872667399,
      "grad_norm": 0.012276324443519115,
      "learning_rate": 2.3015140324963074e-05,
      "loss": 0.0005,
      "step": 2973
    },
    {
      "epoch": 1.088181485547018,
      "grad_norm": 0.0113680986687541,
      "learning_rate": 2.3005908419497785e-05,
      "loss": 0.0005,
      "step": 2974
    },
    {
      "epoch": 1.088547383827296,
      "grad_norm": 0.012398471124470234,
      "learning_rate": 2.29966765140325e-05,
      "loss": 0.0005,
      "step": 2975
    },
    {
      "epoch": 1.088913282107574,
      "grad_norm": 0.01655825413763523,
      "learning_rate": 2.298744460856721e-05,
      "loss": 0.0006,
      "step": 2976
    },
    {
      "epoch": 1.089279180387852,
      "grad_norm": 28.908626556396484,
      "learning_rate": 2.297821270310192e-05,
      "loss": 0.0656,
      "step": 2977
    },
    {
      "epoch": 1.0896450786681302,
      "grad_norm": 0.009063655510544777,
      "learning_rate": 2.2968980797636634e-05,
      "loss": 0.0004,
      "step": 2978
    },
    {
      "epoch": 1.0900109769484083,
      "grad_norm": 1.7160165309906006,
      "learning_rate": 2.2959748892171344e-05,
      "loss": 0.0089,
      "step": 2979
    },
    {
      "epoch": 1.0903768752286864,
      "grad_norm": 0.22451218962669373,
      "learning_rate": 2.2950516986706058e-05,
      "loss": 0.0015,
      "step": 2980
    },
    {
      "epoch": 1.0907427735089645,
      "grad_norm": 23.89752197265625,
      "learning_rate": 2.294128508124077e-05,
      "loss": 0.5274,
      "step": 2981
    },
    {
      "epoch": 1.0911086717892426,
      "grad_norm": 0.2451171576976776,
      "learning_rate": 2.293205317577548e-05,
      "loss": 0.0017,
      "step": 2982
    },
    {
      "epoch": 1.0914745700695208,
      "grad_norm": 5.443453788757324,
      "learning_rate": 2.2922821270310193e-05,
      "loss": 0.2413,
      "step": 2983
    },
    {
      "epoch": 1.0918404683497989,
      "grad_norm": 0.29296207427978516,
      "learning_rate": 2.2913589364844907e-05,
      "loss": 0.0048,
      "step": 2984
    },
    {
      "epoch": 1.0922063666300768,
      "grad_norm": 0.008304160088300705,
      "learning_rate": 2.2904357459379618e-05,
      "loss": 0.0004,
      "step": 2985
    },
    {
      "epoch": 1.0925722649103549,
      "grad_norm": 0.03000006452202797,
      "learning_rate": 2.2895125553914328e-05,
      "loss": 0.0005,
      "step": 2986
    },
    {
      "epoch": 1.092938163190633,
      "grad_norm": 0.010177670046687126,
      "learning_rate": 2.2885893648449042e-05,
      "loss": 0.0004,
      "step": 2987
    },
    {
      "epoch": 1.093304061470911,
      "grad_norm": 0.009244498796761036,
      "learning_rate": 2.2876661742983753e-05,
      "loss": 0.0004,
      "step": 2988
    },
    {
      "epoch": 1.0936699597511892,
      "grad_norm": 0.007619443349540234,
      "learning_rate": 2.2867429837518463e-05,
      "loss": 0.0003,
      "step": 2989
    },
    {
      "epoch": 1.0940358580314673,
      "grad_norm": 0.008940479718148708,
      "learning_rate": 2.2858197932053177e-05,
      "loss": 0.0004,
      "step": 2990
    },
    {
      "epoch": 1.0944017563117454,
      "grad_norm": 0.11216933280229568,
      "learning_rate": 2.284896602658789e-05,
      "loss": 0.0023,
      "step": 2991
    },
    {
      "epoch": 1.0947676545920235,
      "grad_norm": 0.013176364824175835,
      "learning_rate": 2.28397341211226e-05,
      "loss": 0.0005,
      "step": 2992
    },
    {
      "epoch": 1.0951335528723014,
      "grad_norm": 0.01564912125468254,
      "learning_rate": 2.2830502215657312e-05,
      "loss": 0.0007,
      "step": 2993
    },
    {
      "epoch": 1.0954994511525795,
      "grad_norm": 0.009666398167610168,
      "learning_rate": 2.2821270310192026e-05,
      "loss": 0.0004,
      "step": 2994
    },
    {
      "epoch": 1.0958653494328576,
      "grad_norm": 3.502549648284912,
      "learning_rate": 2.2812038404726737e-05,
      "loss": 0.1999,
      "step": 2995
    },
    {
      "epoch": 1.0962312477131357,
      "grad_norm": 23.580364227294922,
      "learning_rate": 2.2802806499261447e-05,
      "loss": 0.0976,
      "step": 2996
    },
    {
      "epoch": 1.0965971459934138,
      "grad_norm": 0.006059134379029274,
      "learning_rate": 2.279357459379616e-05,
      "loss": 0.0003,
      "step": 2997
    },
    {
      "epoch": 1.096963044273692,
      "grad_norm": 0.06165052205324173,
      "learning_rate": 2.2784342688330872e-05,
      "loss": 0.0007,
      "step": 2998
    },
    {
      "epoch": 1.09732894255397,
      "grad_norm": 0.008054126985371113,
      "learning_rate": 2.2775110782865586e-05,
      "loss": 0.0003,
      "step": 2999
    },
    {
      "epoch": 1.0976948408342482,
      "grad_norm": 0.23892438411712646,
      "learning_rate": 2.2765878877400296e-05,
      "loss": 0.0023,
      "step": 3000
    },
    {
      "epoch": 1.098060739114526,
      "grad_norm": 7.436591625213623,
      "learning_rate": 2.2756646971935007e-05,
      "loss": 0.3634,
      "step": 3001
    },
    {
      "epoch": 1.0984266373948042,
      "grad_norm": 0.03491814434528351,
      "learning_rate": 2.274741506646972e-05,
      "loss": 0.0005,
      "step": 3002
    },
    {
      "epoch": 1.0987925356750823,
      "grad_norm": 0.013051160611212254,
      "learning_rate": 2.273818316100443e-05,
      "loss": 0.0004,
      "step": 3003
    },
    {
      "epoch": 1.0991584339553604,
      "grad_norm": 4.160589694976807,
      "learning_rate": 2.2728951255539145e-05,
      "loss": 0.0118,
      "step": 3004
    },
    {
      "epoch": 1.0995243322356385,
      "grad_norm": 0.053739096969366074,
      "learning_rate": 2.2719719350073856e-05,
      "loss": 0.0007,
      "step": 3005
    },
    {
      "epoch": 1.0998902305159166,
      "grad_norm": 0.3295167088508606,
      "learning_rate": 2.271048744460857e-05,
      "loss": 0.0036,
      "step": 3006
    },
    {
      "epoch": 1.1002561287961947,
      "grad_norm": 0.009628892876207829,
      "learning_rate": 2.270125553914328e-05,
      "loss": 0.0004,
      "step": 3007
    },
    {
      "epoch": 1.1006220270764728,
      "grad_norm": 1.2677468061447144,
      "learning_rate": 2.269202363367799e-05,
      "loss": 0.0163,
      "step": 3008
    },
    {
      "epoch": 1.1009879253567507,
      "grad_norm": 0.0056749670766294,
      "learning_rate": 2.2682791728212705e-05,
      "loss": 0.0003,
      "step": 3009
    },
    {
      "epoch": 1.1013538236370288,
      "grad_norm": 0.10440031439065933,
      "learning_rate": 2.267355982274742e-05,
      "loss": 0.0013,
      "step": 3010
    },
    {
      "epoch": 1.101719721917307,
      "grad_norm": 0.011501152999699116,
      "learning_rate": 2.2664327917282126e-05,
      "loss": 0.0005,
      "step": 3011
    },
    {
      "epoch": 1.102085620197585,
      "grad_norm": 8.851265907287598,
      "learning_rate": 2.265509601181684e-05,
      "loss": 0.2079,
      "step": 3012
    },
    {
      "epoch": 1.1024515184778632,
      "grad_norm": 0.01641157642006874,
      "learning_rate": 2.2645864106351554e-05,
      "loss": 0.0006,
      "step": 3013
    },
    {
      "epoch": 1.1028174167581413,
      "grad_norm": 0.011132481507956982,
      "learning_rate": 2.2636632200886264e-05,
      "loss": 0.0004,
      "step": 3014
    },
    {
      "epoch": 1.1031833150384194,
      "grad_norm": 9.384973526000977,
      "learning_rate": 2.2627400295420975e-05,
      "loss": 0.3635,
      "step": 3015
    },
    {
      "epoch": 1.1035492133186975,
      "grad_norm": 0.014887782745063305,
      "learning_rate": 2.261816838995569e-05,
      "loss": 0.0005,
      "step": 3016
    },
    {
      "epoch": 1.1039151115989756,
      "grad_norm": 0.019779982045292854,
      "learning_rate": 2.26089364844904e-05,
      "loss": 0.0005,
      "step": 3017
    },
    {
      "epoch": 1.1042810098792535,
      "grad_norm": 0.01554072555154562,
      "learning_rate": 2.259970457902511e-05,
      "loss": 0.0005,
      "step": 3018
    },
    {
      "epoch": 1.1046469081595316,
      "grad_norm": 0.012088476680219173,
      "learning_rate": 2.2590472673559824e-05,
      "loss": 0.0005,
      "step": 3019
    },
    {
      "epoch": 1.1050128064398097,
      "grad_norm": 3.083653688430786,
      "learning_rate": 2.2581240768094534e-05,
      "loss": 0.2286,
      "step": 3020
    },
    {
      "epoch": 1.1053787047200878,
      "grad_norm": 0.00987911131232977,
      "learning_rate": 2.2572008862629248e-05,
      "loss": 0.0004,
      "step": 3021
    },
    {
      "epoch": 1.105744603000366,
      "grad_norm": 1.8729026317596436,
      "learning_rate": 2.256277695716396e-05,
      "loss": 0.438,
      "step": 3022
    },
    {
      "epoch": 1.106110501280644,
      "grad_norm": 0.009433216415345669,
      "learning_rate": 2.2553545051698673e-05,
      "loss": 0.0004,
      "step": 3023
    },
    {
      "epoch": 1.1064763995609221,
      "grad_norm": 9.996583938598633,
      "learning_rate": 2.2544313146233383e-05,
      "loss": 0.122,
      "step": 3024
    },
    {
      "epoch": 1.1068422978412,
      "grad_norm": 0.024046119302511215,
      "learning_rate": 2.2535081240768097e-05,
      "loss": 0.0006,
      "step": 3025
    },
    {
      "epoch": 1.1072081961214781,
      "grad_norm": 12.609514236450195,
      "learning_rate": 2.2525849335302808e-05,
      "loss": 0.0256,
      "step": 3026
    },
    {
      "epoch": 1.1075740944017562,
      "grad_norm": 0.3171626329421997,
      "learning_rate": 2.251661742983752e-05,
      "loss": 0.0063,
      "step": 3027
    },
    {
      "epoch": 1.1079399926820344,
      "grad_norm": 0.05020453408360481,
      "learning_rate": 2.2507385524372232e-05,
      "loss": 0.0011,
      "step": 3028
    },
    {
      "epoch": 1.1083058909623125,
      "grad_norm": 0.12803754210472107,
      "learning_rate": 2.2498153618906946e-05,
      "loss": 0.0016,
      "step": 3029
    },
    {
      "epoch": 1.1086717892425906,
      "grad_norm": 0.015050179325044155,
      "learning_rate": 2.2488921713441653e-05,
      "loss": 0.0006,
      "step": 3030
    },
    {
      "epoch": 1.1090376875228687,
      "grad_norm": 9.373754501342773,
      "learning_rate": 2.2479689807976367e-05,
      "loss": 0.1288,
      "step": 3031
    },
    {
      "epoch": 1.1094035858031468,
      "grad_norm": 0.010504404082894325,
      "learning_rate": 2.247045790251108e-05,
      "loss": 0.0004,
      "step": 3032
    },
    {
      "epoch": 1.109769484083425,
      "grad_norm": 0.2605797052383423,
      "learning_rate": 2.246122599704579e-05,
      "loss": 0.0058,
      "step": 3033
    },
    {
      "epoch": 1.1101353823637028,
      "grad_norm": 0.10210049152374268,
      "learning_rate": 2.2451994091580502e-05,
      "loss": 0.0019,
      "step": 3034
    },
    {
      "epoch": 1.110501280643981,
      "grad_norm": 0.1287134438753128,
      "learning_rate": 2.2442762186115216e-05,
      "loss": 0.0017,
      "step": 3035
    },
    {
      "epoch": 1.110867178924259,
      "grad_norm": 0.07610952109098434,
      "learning_rate": 2.2433530280649927e-05,
      "loss": 0.0011,
      "step": 3036
    },
    {
      "epoch": 1.1112330772045371,
      "grad_norm": 4.8396315574646,
      "learning_rate": 2.2424298375184637e-05,
      "loss": 0.027,
      "step": 3037
    },
    {
      "epoch": 1.1115989754848152,
      "grad_norm": 0.2755325138568878,
      "learning_rate": 2.241506646971935e-05,
      "loss": 0.0033,
      "step": 3038
    },
    {
      "epoch": 1.1119648737650933,
      "grad_norm": 7.024994850158691,
      "learning_rate": 2.2405834564254062e-05,
      "loss": 0.1622,
      "step": 3039
    },
    {
      "epoch": 1.1123307720453715,
      "grad_norm": 0.09376496076583862,
      "learning_rate": 2.2396602658788776e-05,
      "loss": 0.001,
      "step": 3040
    },
    {
      "epoch": 1.1126966703256496,
      "grad_norm": 0.04831380024552345,
      "learning_rate": 2.2387370753323486e-05,
      "loss": 0.0011,
      "step": 3041
    },
    {
      "epoch": 1.1130625686059274,
      "grad_norm": 0.021889958530664444,
      "learning_rate": 2.23781388478582e-05,
      "loss": 0.0007,
      "step": 3042
    },
    {
      "epoch": 1.1134284668862056,
      "grad_norm": 0.3613564372062683,
      "learning_rate": 2.236890694239291e-05,
      "loss": 0.0036,
      "step": 3043
    },
    {
      "epoch": 1.1137943651664837,
      "grad_norm": 0.016797401010990143,
      "learning_rate": 2.2359675036927625e-05,
      "loss": 0.0006,
      "step": 3044
    },
    {
      "epoch": 1.1141602634467618,
      "grad_norm": 0.015917547047138214,
      "learning_rate": 2.2350443131462335e-05,
      "loss": 0.0006,
      "step": 3045
    },
    {
      "epoch": 1.1145261617270399,
      "grad_norm": 0.24754951894283295,
      "learning_rate": 2.2341211225997046e-05,
      "loss": 0.0034,
      "step": 3046
    },
    {
      "epoch": 1.114892060007318,
      "grad_norm": 0.3076060116291046,
      "learning_rate": 2.233197932053176e-05,
      "loss": 0.0022,
      "step": 3047
    },
    {
      "epoch": 1.115257958287596,
      "grad_norm": 0.009954329580068588,
      "learning_rate": 2.232274741506647e-05,
      "loss": 0.0005,
      "step": 3048
    },
    {
      "epoch": 1.1156238565678742,
      "grad_norm": 0.015208226628601551,
      "learning_rate": 2.231351550960118e-05,
      "loss": 0.0006,
      "step": 3049
    },
    {
      "epoch": 1.1159897548481523,
      "grad_norm": 0.4202074408531189,
      "learning_rate": 2.2304283604135895e-05,
      "loss": 0.008,
      "step": 3050
    },
    {
      "epoch": 1.1163556531284302,
      "grad_norm": 1.173966407775879,
      "learning_rate": 2.229505169867061e-05,
      "loss": 0.0103,
      "step": 3051
    },
    {
      "epoch": 1.1167215514087083,
      "grad_norm": 3.8232383728027344,
      "learning_rate": 2.2285819793205316e-05,
      "loss": 0.1679,
      "step": 3052
    },
    {
      "epoch": 1.1170874496889864,
      "grad_norm": 0.04528050124645233,
      "learning_rate": 2.227658788774003e-05,
      "loss": 0.0008,
      "step": 3053
    },
    {
      "epoch": 1.1174533479692645,
      "grad_norm": 6.29957389831543,
      "learning_rate": 2.2267355982274744e-05,
      "loss": 0.2403,
      "step": 3054
    },
    {
      "epoch": 1.1178192462495427,
      "grad_norm": 4.217491149902344,
      "learning_rate": 2.2258124076809454e-05,
      "loss": 0.2569,
      "step": 3055
    },
    {
      "epoch": 1.1181851445298208,
      "grad_norm": 0.23507311940193176,
      "learning_rate": 2.2248892171344165e-05,
      "loss": 0.0022,
      "step": 3056
    },
    {
      "epoch": 1.1185510428100989,
      "grad_norm": 0.20950791239738464,
      "learning_rate": 2.223966026587888e-05,
      "loss": 0.0018,
      "step": 3057
    },
    {
      "epoch": 1.1189169410903768,
      "grad_norm": 0.017259104177355766,
      "learning_rate": 2.223042836041359e-05,
      "loss": 0.0007,
      "step": 3058
    },
    {
      "epoch": 1.1192828393706549,
      "grad_norm": 0.01038543414324522,
      "learning_rate": 2.2221196454948303e-05,
      "loss": 0.0005,
      "step": 3059
    },
    {
      "epoch": 1.119648737650933,
      "grad_norm": 0.14535725116729736,
      "learning_rate": 2.2211964549483014e-05,
      "loss": 0.0013,
      "step": 3060
    },
    {
      "epoch": 1.120014635931211,
      "grad_norm": 0.012281753122806549,
      "learning_rate": 2.2202732644017728e-05,
      "loss": 0.0005,
      "step": 3061
    },
    {
      "epoch": 1.1203805342114892,
      "grad_norm": 0.016543641686439514,
      "learning_rate": 2.219350073855244e-05,
      "loss": 0.0007,
      "step": 3062
    },
    {
      "epoch": 1.1207464324917673,
      "grad_norm": 0.050940562039613724,
      "learning_rate": 2.218426883308715e-05,
      "loss": 0.0009,
      "step": 3063
    },
    {
      "epoch": 1.1211123307720454,
      "grad_norm": 0.029818616807460785,
      "learning_rate": 2.2175036927621863e-05,
      "loss": 0.0006,
      "step": 3064
    },
    {
      "epoch": 1.1214782290523235,
      "grad_norm": 0.25089409947395325,
      "learning_rate": 2.2165805022156573e-05,
      "loss": 0.0026,
      "step": 3065
    },
    {
      "epoch": 1.1218441273326016,
      "grad_norm": 0.014243151061236858,
      "learning_rate": 2.2156573116691287e-05,
      "loss": 0.0004,
      "step": 3066
    },
    {
      "epoch": 1.1222100256128795,
      "grad_norm": 0.28030040860176086,
      "learning_rate": 2.2147341211225998e-05,
      "loss": 0.0047,
      "step": 3067
    },
    {
      "epoch": 1.1225759238931576,
      "grad_norm": 0.013604438863694668,
      "learning_rate": 2.213810930576071e-05,
      "loss": 0.0005,
      "step": 3068
    },
    {
      "epoch": 1.1229418221734357,
      "grad_norm": 0.03591897711157799,
      "learning_rate": 2.2128877400295422e-05,
      "loss": 0.0009,
      "step": 3069
    },
    {
      "epoch": 1.1233077204537139,
      "grad_norm": 0.025469325482845306,
      "learning_rate": 2.2119645494830136e-05,
      "loss": 0.001,
      "step": 3070
    },
    {
      "epoch": 1.123673618733992,
      "grad_norm": 0.24081894755363464,
      "learning_rate": 2.2110413589364844e-05,
      "loss": 0.004,
      "step": 3071
    },
    {
      "epoch": 1.12403951701427,
      "grad_norm": 0.013010436668992043,
      "learning_rate": 2.2101181683899557e-05,
      "loss": 0.0006,
      "step": 3072
    },
    {
      "epoch": 1.1244054152945482,
      "grad_norm": 0.014039091765880585,
      "learning_rate": 2.209194977843427e-05,
      "loss": 0.0005,
      "step": 3073
    },
    {
      "epoch": 1.1247713135748263,
      "grad_norm": 0.013271324336528778,
      "learning_rate": 2.2082717872968982e-05,
      "loss": 0.0005,
      "step": 3074
    },
    {
      "epoch": 1.1251372118551042,
      "grad_norm": 0.013376806862652302,
      "learning_rate": 2.2073485967503692e-05,
      "loss": 0.0005,
      "step": 3075
    },
    {
      "epoch": 1.1255031101353823,
      "grad_norm": 0.14847509562969208,
      "learning_rate": 2.2064254062038406e-05,
      "loss": 0.0017,
      "step": 3076
    },
    {
      "epoch": 1.1258690084156604,
      "grad_norm": 0.15025761723518372,
      "learning_rate": 2.2055022156573117e-05,
      "loss": 0.0013,
      "step": 3077
    },
    {
      "epoch": 1.1262349066959385,
      "grad_norm": 0.010811627842485905,
      "learning_rate": 2.2045790251107828e-05,
      "loss": 0.0005,
      "step": 3078
    },
    {
      "epoch": 1.1266008049762166,
      "grad_norm": 0.35469338297843933,
      "learning_rate": 2.203655834564254e-05,
      "loss": 0.0032,
      "step": 3079
    },
    {
      "epoch": 1.1269667032564947,
      "grad_norm": 0.008379826322197914,
      "learning_rate": 2.2027326440177255e-05,
      "loss": 0.0004,
      "step": 3080
    },
    {
      "epoch": 1.1273326015367728,
      "grad_norm": 0.016223985701799393,
      "learning_rate": 2.2018094534711966e-05,
      "loss": 0.0006,
      "step": 3081
    },
    {
      "epoch": 1.127698499817051,
      "grad_norm": 0.10751519352197647,
      "learning_rate": 2.2008862629246676e-05,
      "loss": 0.0014,
      "step": 3082
    },
    {
      "epoch": 1.128064398097329,
      "grad_norm": 0.008932581171393394,
      "learning_rate": 2.199963072378139e-05,
      "loss": 0.0004,
      "step": 3083
    },
    {
      "epoch": 1.128430296377607,
      "grad_norm": 0.29334044456481934,
      "learning_rate": 2.19903988183161e-05,
      "loss": 0.0026,
      "step": 3084
    },
    {
      "epoch": 1.128796194657885,
      "grad_norm": 0.007366046775132418,
      "learning_rate": 2.1981166912850815e-05,
      "loss": 0.0003,
      "step": 3085
    },
    {
      "epoch": 1.1291620929381632,
      "grad_norm": 0.010298652574419975,
      "learning_rate": 2.1971935007385525e-05,
      "loss": 0.0004,
      "step": 3086
    },
    {
      "epoch": 1.1295279912184413,
      "grad_norm": 0.009781657718122005,
      "learning_rate": 2.1962703101920236e-05,
      "loss": 0.0004,
      "step": 3087
    },
    {
      "epoch": 1.1298938894987194,
      "grad_norm": 0.10570962727069855,
      "learning_rate": 2.195347119645495e-05,
      "loss": 0.0013,
      "step": 3088
    },
    {
      "epoch": 1.1302597877789975,
      "grad_norm": 0.17154525220394135,
      "learning_rate": 2.1944239290989664e-05,
      "loss": 0.001,
      "step": 3089
    },
    {
      "epoch": 1.1306256860592756,
      "grad_norm": 1.3409245014190674,
      "learning_rate": 2.193500738552437e-05,
      "loss": 0.0103,
      "step": 3090
    },
    {
      "epoch": 1.1309915843395535,
      "grad_norm": 0.009363027289509773,
      "learning_rate": 2.1925775480059085e-05,
      "loss": 0.0004,
      "step": 3091
    },
    {
      "epoch": 1.1313574826198316,
      "grad_norm": 0.01619800552725792,
      "learning_rate": 2.19165435745938e-05,
      "loss": 0.0006,
      "step": 3092
    },
    {
      "epoch": 1.1317233809001097,
      "grad_norm": 9.236496925354004,
      "learning_rate": 2.190731166912851e-05,
      "loss": 0.055,
      "step": 3093
    },
    {
      "epoch": 1.1320892791803878,
      "grad_norm": 0.006561451591551304,
      "learning_rate": 2.189807976366322e-05,
      "loss": 0.0003,
      "step": 3094
    },
    {
      "epoch": 1.132455177460666,
      "grad_norm": 0.009446714073419571,
      "learning_rate": 2.1888847858197934e-05,
      "loss": 0.0004,
      "step": 3095
    },
    {
      "epoch": 1.132821075740944,
      "grad_norm": 0.01212611235678196,
      "learning_rate": 2.1879615952732645e-05,
      "loss": 0.0004,
      "step": 3096
    },
    {
      "epoch": 1.1331869740212221,
      "grad_norm": 0.00928456149995327,
      "learning_rate": 2.1870384047267355e-05,
      "loss": 0.0004,
      "step": 3097
    },
    {
      "epoch": 1.1335528723015003,
      "grad_norm": 0.008931029587984085,
      "learning_rate": 2.186115214180207e-05,
      "loss": 0.0004,
      "step": 3098
    },
    {
      "epoch": 1.1339187705817784,
      "grad_norm": 7.803125858306885,
      "learning_rate": 2.1851920236336783e-05,
      "loss": 0.3832,
      "step": 3099
    },
    {
      "epoch": 1.1342846688620563,
      "grad_norm": 0.5098270177841187,
      "learning_rate": 2.1842688330871493e-05,
      "loss": 0.0041,
      "step": 3100
    },
    {
      "epoch": 1.1346505671423344,
      "grad_norm": 0.014646833762526512,
      "learning_rate": 2.1833456425406204e-05,
      "loss": 0.0004,
      "step": 3101
    },
    {
      "epoch": 1.1350164654226125,
      "grad_norm": 0.22690288722515106,
      "learning_rate": 2.1824224519940918e-05,
      "loss": 0.0013,
      "step": 3102
    },
    {
      "epoch": 1.1353823637028906,
      "grad_norm": 0.07715766131877899,
      "learning_rate": 2.181499261447563e-05,
      "loss": 0.0009,
      "step": 3103
    },
    {
      "epoch": 1.1357482619831687,
      "grad_norm": 18.501203536987305,
      "learning_rate": 2.180576070901034e-05,
      "loss": 0.5693,
      "step": 3104
    },
    {
      "epoch": 1.1361141602634468,
      "grad_norm": 0.08476994931697845,
      "learning_rate": 2.1796528803545053e-05,
      "loss": 0.0009,
      "step": 3105
    },
    {
      "epoch": 1.136480058543725,
      "grad_norm": 0.017515581101179123,
      "learning_rate": 2.1787296898079764e-05,
      "loss": 0.0004,
      "step": 3106
    },
    {
      "epoch": 1.136845956824003,
      "grad_norm": 0.013849022798240185,
      "learning_rate": 2.1778064992614477e-05,
      "loss": 0.0005,
      "step": 3107
    },
    {
      "epoch": 1.137211855104281,
      "grad_norm": 0.08891665190458298,
      "learning_rate": 2.1768833087149188e-05,
      "loss": 0.0009,
      "step": 3108
    },
    {
      "epoch": 1.137577753384559,
      "grad_norm": 0.008240005932748318,
      "learning_rate": 2.17596011816839e-05,
      "loss": 0.0003,
      "step": 3109
    },
    {
      "epoch": 1.1379436516648371,
      "grad_norm": 6.037247180938721,
      "learning_rate": 2.1750369276218613e-05,
      "loss": 0.1426,
      "step": 3110
    },
    {
      "epoch": 1.1383095499451152,
      "grad_norm": 0.044649988412857056,
      "learning_rate": 2.1741137370753326e-05,
      "loss": 0.0008,
      "step": 3111
    },
    {
      "epoch": 1.1386754482253933,
      "grad_norm": 0.025776563212275505,
      "learning_rate": 2.1731905465288037e-05,
      "loss": 0.0006,
      "step": 3112
    },
    {
      "epoch": 1.1390413465056715,
      "grad_norm": 0.011815650388598442,
      "learning_rate": 2.1722673559822748e-05,
      "loss": 0.0004,
      "step": 3113
    },
    {
      "epoch": 1.1394072447859496,
      "grad_norm": 9.358113288879395,
      "learning_rate": 2.171344165435746e-05,
      "loss": 0.1932,
      "step": 3114
    },
    {
      "epoch": 1.1397731430662277,
      "grad_norm": 9.221000671386719,
      "learning_rate": 2.1704209748892172e-05,
      "loss": 0.168,
      "step": 3115
    },
    {
      "epoch": 1.1401390413465058,
      "grad_norm": 0.01394871436059475,
      "learning_rate": 2.1694977843426883e-05,
      "loss": 0.0005,
      "step": 3116
    },
    {
      "epoch": 1.1405049396267837,
      "grad_norm": 0.12319813668727875,
      "learning_rate": 2.1685745937961597e-05,
      "loss": 0.0007,
      "step": 3117
    },
    {
      "epoch": 1.1408708379070618,
      "grad_norm": 13.573166847229004,
      "learning_rate": 2.167651403249631e-05,
      "loss": 0.083,
      "step": 3118
    },
    {
      "epoch": 1.14123673618734,
      "grad_norm": 0.010240583680570126,
      "learning_rate": 2.1667282127031018e-05,
      "loss": 0.0003,
      "step": 3119
    },
    {
      "epoch": 1.141602634467618,
      "grad_norm": 0.6969809532165527,
      "learning_rate": 2.165805022156573e-05,
      "loss": 0.0056,
      "step": 3120
    },
    {
      "epoch": 1.1419685327478961,
      "grad_norm": 0.00904578436166048,
      "learning_rate": 2.1648818316100445e-05,
      "loss": 0.0004,
      "step": 3121
    },
    {
      "epoch": 1.1423344310281742,
      "grad_norm": 0.5914815068244934,
      "learning_rate": 2.1639586410635156e-05,
      "loss": 0.0036,
      "step": 3122
    },
    {
      "epoch": 1.1427003293084523,
      "grad_norm": 0.009337070398032665,
      "learning_rate": 2.1630354505169867e-05,
      "loss": 0.0004,
      "step": 3123
    },
    {
      "epoch": 1.1430662275887302,
      "grad_norm": 0.009361001662909985,
      "learning_rate": 2.162112259970458e-05,
      "loss": 0.0004,
      "step": 3124
    },
    {
      "epoch": 1.1434321258690083,
      "grad_norm": 0.010017944499850273,
      "learning_rate": 2.161189069423929e-05,
      "loss": 0.0004,
      "step": 3125
    },
    {
      "epoch": 1.1437980241492864,
      "grad_norm": 0.007662912830710411,
      "learning_rate": 2.1602658788774005e-05,
      "loss": 0.0003,
      "step": 3126
    },
    {
      "epoch": 1.1441639224295646,
      "grad_norm": 0.009647880680859089,
      "learning_rate": 2.1593426883308716e-05,
      "loss": 0.0004,
      "step": 3127
    },
    {
      "epoch": 1.1445298207098427,
      "grad_norm": 0.07521982491016388,
      "learning_rate": 2.1584194977843426e-05,
      "loss": 0.001,
      "step": 3128
    },
    {
      "epoch": 1.1448957189901208,
      "grad_norm": 0.01289602555334568,
      "learning_rate": 2.157496307237814e-05,
      "loss": 0.0004,
      "step": 3129
    },
    {
      "epoch": 1.1452616172703989,
      "grad_norm": 0.008924438618123531,
      "learning_rate": 2.1565731166912854e-05,
      "loss": 0.0004,
      "step": 3130
    },
    {
      "epoch": 1.145627515550677,
      "grad_norm": 0.009568078443408012,
      "learning_rate": 2.1556499261447565e-05,
      "loss": 0.0004,
      "step": 3131
    },
    {
      "epoch": 1.145993413830955,
      "grad_norm": 1.174254298210144,
      "learning_rate": 2.1547267355982275e-05,
      "loss": 0.0093,
      "step": 3132
    },
    {
      "epoch": 1.146359312111233,
      "grad_norm": 2.111025810241699,
      "learning_rate": 2.153803545051699e-05,
      "loss": 0.0228,
      "step": 3133
    },
    {
      "epoch": 1.146725210391511,
      "grad_norm": 0.04659072309732437,
      "learning_rate": 2.15288035450517e-05,
      "loss": 0.0008,
      "step": 3134
    },
    {
      "epoch": 1.1470911086717892,
      "grad_norm": 0.010614963248372078,
      "learning_rate": 2.151957163958641e-05,
      "loss": 0.0004,
      "step": 3135
    },
    {
      "epoch": 1.1474570069520673,
      "grad_norm": 0.01315068919211626,
      "learning_rate": 2.1510339734121124e-05,
      "loss": 0.0004,
      "step": 3136
    },
    {
      "epoch": 1.1478229052323454,
      "grad_norm": 0.40193572640419006,
      "learning_rate": 2.1501107828655838e-05,
      "loss": 0.0028,
      "step": 3137
    },
    {
      "epoch": 1.1481888035126235,
      "grad_norm": 0.04548206925392151,
      "learning_rate": 2.1491875923190545e-05,
      "loss": 0.0007,
      "step": 3138
    },
    {
      "epoch": 1.1485547017929016,
      "grad_norm": 0.011353031732141972,
      "learning_rate": 2.148264401772526e-05,
      "loss": 0.0002,
      "step": 3139
    },
    {
      "epoch": 1.1489206000731795,
      "grad_norm": 0.011754250153899193,
      "learning_rate": 2.1473412112259973e-05,
      "loss": 0.0004,
      "step": 3140
    },
    {
      "epoch": 1.1492864983534576,
      "grad_norm": 0.009210828691720963,
      "learning_rate": 2.1464180206794684e-05,
      "loss": 0.0003,
      "step": 3141
    },
    {
      "epoch": 1.1496523966337358,
      "grad_norm": 0.5060833096504211,
      "learning_rate": 2.1454948301329394e-05,
      "loss": 0.0022,
      "step": 3142
    },
    {
      "epoch": 1.1500182949140139,
      "grad_norm": 0.012434794567525387,
      "learning_rate": 2.1445716395864108e-05,
      "loss": 0.0005,
      "step": 3143
    },
    {
      "epoch": 1.150384193194292,
      "grad_norm": 0.650405764579773,
      "learning_rate": 2.143648449039882e-05,
      "loss": 0.0042,
      "step": 3144
    },
    {
      "epoch": 1.15075009147457,
      "grad_norm": 0.005522820167243481,
      "learning_rate": 2.1427252584933533e-05,
      "loss": 0.0002,
      "step": 3145
    },
    {
      "epoch": 1.1511159897548482,
      "grad_norm": 0.04679917171597481,
      "learning_rate": 2.1418020679468243e-05,
      "loss": 0.0006,
      "step": 3146
    },
    {
      "epoch": 1.1514818880351263,
      "grad_norm": 0.011098410934209824,
      "learning_rate": 2.1408788774002954e-05,
      "loss": 0.0004,
      "step": 3147
    },
    {
      "epoch": 1.1518477863154044,
      "grad_norm": 0.4534562826156616,
      "learning_rate": 2.1399556868537668e-05,
      "loss": 0.0024,
      "step": 3148
    },
    {
      "epoch": 1.1522136845956825,
      "grad_norm": 0.028850579634308815,
      "learning_rate": 2.1390324963072378e-05,
      "loss": 0.0006,
      "step": 3149
    },
    {
      "epoch": 1.1525795828759604,
      "grad_norm": 0.007756910752505064,
      "learning_rate": 2.1381093057607092e-05,
      "loss": 0.0003,
      "step": 3150
    },
    {
      "epoch": 1.1529454811562385,
      "grad_norm": 0.0073447562754154205,
      "learning_rate": 2.1371861152141803e-05,
      "loss": 0.0003,
      "step": 3151
    },
    {
      "epoch": 1.1533113794365166,
      "grad_norm": 2.4712586402893066,
      "learning_rate": 2.1362629246676517e-05,
      "loss": 0.6254,
      "step": 3152
    },
    {
      "epoch": 1.1536772777167947,
      "grad_norm": 0.0067286184057593346,
      "learning_rate": 2.1353397341211227e-05,
      "loss": 0.0003,
      "step": 3153
    },
    {
      "epoch": 1.1540431759970728,
      "grad_norm": 0.1720251590013504,
      "learning_rate": 2.1344165435745938e-05,
      "loss": 0.0012,
      "step": 3154
    },
    {
      "epoch": 1.154409074277351,
      "grad_norm": 0.008273164741694927,
      "learning_rate": 2.133493353028065e-05,
      "loss": 0.0004,
      "step": 3155
    },
    {
      "epoch": 1.154774972557629,
      "grad_norm": 0.056431569159030914,
      "learning_rate": 2.1325701624815366e-05,
      "loss": 0.0005,
      "step": 3156
    },
    {
      "epoch": 1.155140870837907,
      "grad_norm": 0.5431950688362122,
      "learning_rate": 2.1316469719350073e-05,
      "loss": 0.0018,
      "step": 3157
    },
    {
      "epoch": 1.155506769118185,
      "grad_norm": 0.007493387907743454,
      "learning_rate": 2.1307237813884787e-05,
      "loss": 0.0003,
      "step": 3158
    },
    {
      "epoch": 1.1558726673984632,
      "grad_norm": 0.007222231477499008,
      "learning_rate": 2.12980059084195e-05,
      "loss": 0.0003,
      "step": 3159
    },
    {
      "epoch": 1.1562385656787413,
      "grad_norm": 0.011149332858622074,
      "learning_rate": 2.128877400295421e-05,
      "loss": 0.0005,
      "step": 3160
    },
    {
      "epoch": 1.1566044639590194,
      "grad_norm": 0.006505921948701143,
      "learning_rate": 2.127954209748892e-05,
      "loss": 0.0003,
      "step": 3161
    },
    {
      "epoch": 1.1569703622392975,
      "grad_norm": 0.00957836676388979,
      "learning_rate": 2.1270310192023636e-05,
      "loss": 0.0005,
      "step": 3162
    },
    {
      "epoch": 1.1573362605195756,
      "grad_norm": 0.011533444747328758,
      "learning_rate": 2.1261078286558346e-05,
      "loss": 0.0005,
      "step": 3163
    },
    {
      "epoch": 1.1577021587998537,
      "grad_norm": 0.008497724309563637,
      "learning_rate": 2.1251846381093057e-05,
      "loss": 0.0004,
      "step": 3164
    },
    {
      "epoch": 1.1580680570801318,
      "grad_norm": 0.008255177177488804,
      "learning_rate": 2.124261447562777e-05,
      "loss": 0.0004,
      "step": 3165
    },
    {
      "epoch": 1.1584339553604097,
      "grad_norm": 0.7140600681304932,
      "learning_rate": 2.123338257016248e-05,
      "loss": 0.0027,
      "step": 3166
    },
    {
      "epoch": 1.1587998536406878,
      "grad_norm": 0.010497407987713814,
      "learning_rate": 2.1224150664697195e-05,
      "loss": 0.0005,
      "step": 3167
    },
    {
      "epoch": 1.159165751920966,
      "grad_norm": 0.009691012091934681,
      "learning_rate": 2.1214918759231906e-05,
      "loss": 0.0004,
      "step": 3168
    },
    {
      "epoch": 1.159531650201244,
      "grad_norm": 0.00948450155556202,
      "learning_rate": 2.120568685376662e-05,
      "loss": 0.0005,
      "step": 3169
    },
    {
      "epoch": 1.1598975484815222,
      "grad_norm": 0.011519008316099644,
      "learning_rate": 2.119645494830133e-05,
      "loss": 0.0004,
      "step": 3170
    },
    {
      "epoch": 1.1602634467618003,
      "grad_norm": 0.01117816660553217,
      "learning_rate": 2.1187223042836044e-05,
      "loss": 0.0005,
      "step": 3171
    },
    {
      "epoch": 1.1606293450420784,
      "grad_norm": 0.0953235998749733,
      "learning_rate": 2.1177991137370755e-05,
      "loss": 0.0009,
      "step": 3172
    },
    {
      "epoch": 1.1609952433223563,
      "grad_norm": 0.0389227494597435,
      "learning_rate": 2.1168759231905465e-05,
      "loss": 0.0006,
      "step": 3173
    },
    {
      "epoch": 1.1613611416026344,
      "grad_norm": 0.011005885899066925,
      "learning_rate": 2.115952732644018e-05,
      "loss": 0.0005,
      "step": 3174
    },
    {
      "epoch": 1.1617270398829125,
      "grad_norm": 0.013279328122735023,
      "learning_rate": 2.1150295420974893e-05,
      "loss": 0.0005,
      "step": 3175
    },
    {
      "epoch": 1.1620929381631906,
      "grad_norm": 0.008540036156773567,
      "learning_rate": 2.11410635155096e-05,
      "loss": 0.0004,
      "step": 3176
    },
    {
      "epoch": 1.1624588364434687,
      "grad_norm": 0.008555000647902489,
      "learning_rate": 2.1131831610044314e-05,
      "loss": 0.0004,
      "step": 3177
    },
    {
      "epoch": 1.1628247347237468,
      "grad_norm": 19.728214263916016,
      "learning_rate": 2.1122599704579028e-05,
      "loss": 0.1495,
      "step": 3178
    },
    {
      "epoch": 1.163190633004025,
      "grad_norm": 0.009353105910122395,
      "learning_rate": 2.1113367799113735e-05,
      "loss": 0.0004,
      "step": 3179
    },
    {
      "epoch": 1.163556531284303,
      "grad_norm": 0.011264121159911156,
      "learning_rate": 2.110413589364845e-05,
      "loss": 0.0005,
      "step": 3180
    },
    {
      "epoch": 1.1639224295645811,
      "grad_norm": 0.010501482523977757,
      "learning_rate": 2.1094903988183163e-05,
      "loss": 0.0005,
      "step": 3181
    },
    {
      "epoch": 1.1642883278448592,
      "grad_norm": 0.009029158391058445,
      "learning_rate": 2.1085672082717874e-05,
      "loss": 0.0004,
      "step": 3182
    },
    {
      "epoch": 1.1646542261251371,
      "grad_norm": 0.0085203368216753,
      "learning_rate": 2.1076440177252584e-05,
      "loss": 0.0004,
      "step": 3183
    },
    {
      "epoch": 1.1650201244054152,
      "grad_norm": 0.008904622867703438,
      "learning_rate": 2.1067208271787298e-05,
      "loss": 0.0004,
      "step": 3184
    },
    {
      "epoch": 1.1653860226856934,
      "grad_norm": 0.022098267450928688,
      "learning_rate": 2.105797636632201e-05,
      "loss": 0.0005,
      "step": 3185
    },
    {
      "epoch": 1.1657519209659715,
      "grad_norm": 0.009469600394368172,
      "learning_rate": 2.1048744460856723e-05,
      "loss": 0.0005,
      "step": 3186
    },
    {
      "epoch": 1.1661178192462496,
      "grad_norm": 0.06475409865379333,
      "learning_rate": 2.1039512555391433e-05,
      "loss": 0.0008,
      "step": 3187
    },
    {
      "epoch": 1.1664837175265277,
      "grad_norm": 0.007019869517534971,
      "learning_rate": 2.1030280649926147e-05,
      "loss": 0.0003,
      "step": 3188
    },
    {
      "epoch": 1.1668496158068058,
      "grad_norm": 0.009865746833384037,
      "learning_rate": 2.1021048744460858e-05,
      "loss": 0.0005,
      "step": 3189
    },
    {
      "epoch": 1.1672155140870837,
      "grad_norm": 0.011877442710101604,
      "learning_rate": 2.101181683899557e-05,
      "loss": 0.0005,
      "step": 3190
    },
    {
      "epoch": 1.1675814123673618,
      "grad_norm": 0.009673867374658585,
      "learning_rate": 2.1002584933530282e-05,
      "loss": 0.0005,
      "step": 3191
    },
    {
      "epoch": 1.16794731064764,
      "grad_norm": 0.014952362515032291,
      "learning_rate": 2.0993353028064993e-05,
      "loss": 0.0005,
      "step": 3192
    },
    {
      "epoch": 1.168313208927918,
      "grad_norm": 0.05155900865793228,
      "learning_rate": 2.0984121122599707e-05,
      "loss": 0.0007,
      "step": 3193
    },
    {
      "epoch": 1.1686791072081961,
      "grad_norm": 0.05399926006793976,
      "learning_rate": 2.0974889217134417e-05,
      "loss": 0.0007,
      "step": 3194
    },
    {
      "epoch": 1.1690450054884742,
      "grad_norm": 0.007429550401866436,
      "learning_rate": 2.0965657311669128e-05,
      "loss": 0.0003,
      "step": 3195
    },
    {
      "epoch": 1.1694109037687523,
      "grad_norm": 0.013430245220661163,
      "learning_rate": 2.0956425406203842e-05,
      "loss": 0.0005,
      "step": 3196
    },
    {
      "epoch": 1.1697768020490305,
      "grad_norm": 0.007592185866087675,
      "learning_rate": 2.0947193500738556e-05,
      "loss": 0.0004,
      "step": 3197
    },
    {
      "epoch": 1.1701427003293086,
      "grad_norm": 0.007715534418821335,
      "learning_rate": 2.0937961595273263e-05,
      "loss": 0.0004,
      "step": 3198
    },
    {
      "epoch": 1.1705085986095864,
      "grad_norm": 0.006302173249423504,
      "learning_rate": 2.0928729689807977e-05,
      "loss": 0.0003,
      "step": 3199
    },
    {
      "epoch": 1.1708744968898646,
      "grad_norm": 0.007500824984163046,
      "learning_rate": 2.091949778434269e-05,
      "loss": 0.0004,
      "step": 3200
    },
    {
      "epoch": 1.1712403951701427,
      "grad_norm": 0.007322429679334164,
      "learning_rate": 2.09102658788774e-05,
      "loss": 0.0003,
      "step": 3201
    },
    {
      "epoch": 1.1716062934504208,
      "grad_norm": 0.041867129504680634,
      "learning_rate": 2.0901033973412112e-05,
      "loss": 0.0005,
      "step": 3202
    },
    {
      "epoch": 1.1719721917306989,
      "grad_norm": 0.007346983999013901,
      "learning_rate": 2.0891802067946826e-05,
      "loss": 0.0003,
      "step": 3203
    },
    {
      "epoch": 1.172338090010977,
      "grad_norm": 0.03645465150475502,
      "learning_rate": 2.0882570162481536e-05,
      "loss": 0.0005,
      "step": 3204
    },
    {
      "epoch": 1.172703988291255,
      "grad_norm": 10.90464973449707,
      "learning_rate": 2.0873338257016247e-05,
      "loss": 0.4244,
      "step": 3205
    },
    {
      "epoch": 1.173069886571533,
      "grad_norm": 0.7849429249763489,
      "learning_rate": 2.086410635155096e-05,
      "loss": 0.0025,
      "step": 3206
    },
    {
      "epoch": 1.173435784851811,
      "grad_norm": 8.517992973327637,
      "learning_rate": 2.0854874446085675e-05,
      "loss": 0.2479,
      "step": 3207
    },
    {
      "epoch": 1.1738016831320892,
      "grad_norm": 0.005845978856086731,
      "learning_rate": 2.0845642540620385e-05,
      "loss": 0.0003,
      "step": 3208
    },
    {
      "epoch": 1.1741675814123673,
      "grad_norm": 0.007766477297991514,
      "learning_rate": 2.0836410635155096e-05,
      "loss": 0.0004,
      "step": 3209
    },
    {
      "epoch": 1.1745334796926454,
      "grad_norm": 6.920149326324463,
      "learning_rate": 2.082717872968981e-05,
      "loss": 0.4723,
      "step": 3210
    },
    {
      "epoch": 1.1748993779729235,
      "grad_norm": 0.006912372540682554,
      "learning_rate": 2.081794682422452e-05,
      "loss": 0.0003,
      "step": 3211
    },
    {
      "epoch": 1.1752652762532017,
      "grad_norm": 0.07798849791288376,
      "learning_rate": 2.0808714918759234e-05,
      "loss": 0.001,
      "step": 3212
    },
    {
      "epoch": 1.1756311745334798,
      "grad_norm": 0.005390833131968975,
      "learning_rate": 2.0799483013293945e-05,
      "loss": 0.0002,
      "step": 3213
    },
    {
      "epoch": 1.1759970728137579,
      "grad_norm": 0.007722930982708931,
      "learning_rate": 2.0790251107828655e-05,
      "loss": 0.0004,
      "step": 3214
    },
    {
      "epoch": 1.176362971094036,
      "grad_norm": 0.24463096261024475,
      "learning_rate": 2.078101920236337e-05,
      "loss": 0.0027,
      "step": 3215
    },
    {
      "epoch": 1.1767288693743139,
      "grad_norm": 0.15389735996723175,
      "learning_rate": 2.0771787296898083e-05,
      "loss": 0.0012,
      "step": 3216
    },
    {
      "epoch": 1.177094767654592,
      "grad_norm": 0.05114171653985977,
      "learning_rate": 2.076255539143279e-05,
      "loss": 0.0007,
      "step": 3217
    },
    {
      "epoch": 1.17746066593487,
      "grad_norm": 0.04437921568751335,
      "learning_rate": 2.0753323485967504e-05,
      "loss": 0.0008,
      "step": 3218
    },
    {
      "epoch": 1.1778265642151482,
      "grad_norm": 0.47756555676460266,
      "learning_rate": 2.0744091580502218e-05,
      "loss": 0.0071,
      "step": 3219
    },
    {
      "epoch": 1.1781924624954263,
      "grad_norm": 0.07751242071390152,
      "learning_rate": 2.073485967503693e-05,
      "loss": 0.001,
      "step": 3220
    },
    {
      "epoch": 1.1785583607757044,
      "grad_norm": 0.03064548224210739,
      "learning_rate": 2.072562776957164e-05,
      "loss": 0.0008,
      "step": 3221
    },
    {
      "epoch": 1.1789242590559825,
      "grad_norm": 0.6445357799530029,
      "learning_rate": 2.0716395864106353e-05,
      "loss": 0.0076,
      "step": 3222
    },
    {
      "epoch": 1.1792901573362604,
      "grad_norm": 0.014941399917006493,
      "learning_rate": 2.0707163958641064e-05,
      "loss": 0.0004,
      "step": 3223
    },
    {
      "epoch": 1.1796560556165385,
      "grad_norm": 0.010794408619403839,
      "learning_rate": 2.0697932053175774e-05,
      "loss": 0.0004,
      "step": 3224
    },
    {
      "epoch": 1.1800219538968166,
      "grad_norm": 0.03984038531780243,
      "learning_rate": 2.0688700147710488e-05,
      "loss": 0.0008,
      "step": 3225
    },
    {
      "epoch": 1.1803878521770947,
      "grad_norm": 0.011664878576993942,
      "learning_rate": 2.0679468242245202e-05,
      "loss": 0.0005,
      "step": 3226
    },
    {
      "epoch": 1.1807537504573729,
      "grad_norm": 0.017896432429552078,
      "learning_rate": 2.0670236336779913e-05,
      "loss": 0.0005,
      "step": 3227
    },
    {
      "epoch": 1.181119648737651,
      "grad_norm": 0.00978259276598692,
      "learning_rate": 2.0661004431314623e-05,
      "loss": 0.0004,
      "step": 3228
    },
    {
      "epoch": 1.181485547017929,
      "grad_norm": 0.14287631213665009,
      "learning_rate": 2.0651772525849337e-05,
      "loss": 0.001,
      "step": 3229
    },
    {
      "epoch": 1.1818514452982072,
      "grad_norm": 0.4529663324356079,
      "learning_rate": 2.0642540620384048e-05,
      "loss": 0.003,
      "step": 3230
    },
    {
      "epoch": 1.1822173435784853,
      "grad_norm": 0.009674769826233387,
      "learning_rate": 2.0633308714918762e-05,
      "loss": 0.0003,
      "step": 3231
    },
    {
      "epoch": 1.1825832418587632,
      "grad_norm": 0.01108067948371172,
      "learning_rate": 2.0624076809453472e-05,
      "loss": 0.0005,
      "step": 3232
    },
    {
      "epoch": 1.1829491401390413,
      "grad_norm": 0.010737285949289799,
      "learning_rate": 2.0614844903988183e-05,
      "loss": 0.0005,
      "step": 3233
    },
    {
      "epoch": 1.1833150384193194,
      "grad_norm": 0.014941631816327572,
      "learning_rate": 2.0605612998522897e-05,
      "loss": 0.0006,
      "step": 3234
    },
    {
      "epoch": 1.1836809366995975,
      "grad_norm": 0.19738194346427917,
      "learning_rate": 2.0596381093057607e-05,
      "loss": 0.002,
      "step": 3235
    },
    {
      "epoch": 1.1840468349798756,
      "grad_norm": 15.576763153076172,
      "learning_rate": 2.0587149187592318e-05,
      "loss": 0.2976,
      "step": 3236
    },
    {
      "epoch": 1.1844127332601537,
      "grad_norm": 0.0069581312127411366,
      "learning_rate": 2.0577917282127032e-05,
      "loss": 0.0003,
      "step": 3237
    },
    {
      "epoch": 1.1847786315404318,
      "grad_norm": 0.20858226716518402,
      "learning_rate": 2.0568685376661746e-05,
      "loss": 0.0017,
      "step": 3238
    },
    {
      "epoch": 1.1851445298207097,
      "grad_norm": 0.01593598537147045,
      "learning_rate": 2.0559453471196456e-05,
      "loss": 0.0004,
      "step": 3239
    },
    {
      "epoch": 1.1855104281009878,
      "grad_norm": 0.01779775694012642,
      "learning_rate": 2.0550221565731167e-05,
      "loss": 0.0004,
      "step": 3240
    },
    {
      "epoch": 1.185876326381266,
      "grad_norm": 6.671947956085205,
      "learning_rate": 2.054098966026588e-05,
      "loss": 0.4019,
      "step": 3241
    },
    {
      "epoch": 1.186242224661544,
      "grad_norm": 0.15906205773353577,
      "learning_rate": 2.053175775480059e-05,
      "loss": 0.0017,
      "step": 3242
    },
    {
      "epoch": 1.1866081229418222,
      "grad_norm": 0.011083226650953293,
      "learning_rate": 2.0522525849335302e-05,
      "loss": 0.0004,
      "step": 3243
    },
    {
      "epoch": 1.1869740212221003,
      "grad_norm": 0.16909854114055634,
      "learning_rate": 2.0513293943870016e-05,
      "loss": 0.0023,
      "step": 3244
    },
    {
      "epoch": 1.1873399195023784,
      "grad_norm": 0.010023013688623905,
      "learning_rate": 2.050406203840473e-05,
      "loss": 0.0004,
      "step": 3245
    },
    {
      "epoch": 1.1877058177826565,
      "grad_norm": 6.190181255340576,
      "learning_rate": 2.049483013293944e-05,
      "loss": 0.2808,
      "step": 3246
    },
    {
      "epoch": 1.1880717160629346,
      "grad_norm": 0.026222536340355873,
      "learning_rate": 2.048559822747415e-05,
      "loss": 0.0007,
      "step": 3247
    },
    {
      "epoch": 1.1884376143432127,
      "grad_norm": 0.017622314393520355,
      "learning_rate": 2.0476366322008865e-05,
      "loss": 0.0004,
      "step": 3248
    },
    {
      "epoch": 1.1888035126234906,
      "grad_norm": 23.178333282470703,
      "learning_rate": 2.0467134416543575e-05,
      "loss": 0.7,
      "step": 3249
    },
    {
      "epoch": 1.1891694109037687,
      "grad_norm": 0.04297082871198654,
      "learning_rate": 2.0457902511078286e-05,
      "loss": 0.0008,
      "step": 3250
    },
    {
      "epoch": 1.1895353091840468,
      "grad_norm": 52.97172927856445,
      "learning_rate": 2.0448670605613e-05,
      "loss": 0.0693,
      "step": 3251
    },
    {
      "epoch": 1.189901207464325,
      "grad_norm": 0.017277900129556656,
      "learning_rate": 2.043943870014771e-05,
      "loss": 0.0005,
      "step": 3252
    },
    {
      "epoch": 1.190267105744603,
      "grad_norm": 1.4267460107803345,
      "learning_rate": 2.0430206794682424e-05,
      "loss": 0.0089,
      "step": 3253
    },
    {
      "epoch": 1.1906330040248811,
      "grad_norm": 0.07972031086683273,
      "learning_rate": 2.0420974889217135e-05,
      "loss": 0.0017,
      "step": 3254
    },
    {
      "epoch": 1.1909989023051593,
      "grad_norm": 0.03154296055436134,
      "learning_rate": 2.0411742983751845e-05,
      "loss": 0.0007,
      "step": 3255
    },
    {
      "epoch": 1.1913648005854371,
      "grad_norm": 0.12069658190011978,
      "learning_rate": 2.040251107828656e-05,
      "loss": 0.0015,
      "step": 3256
    },
    {
      "epoch": 1.1917306988657153,
      "grad_norm": 0.02671537920832634,
      "learning_rate": 2.0393279172821273e-05,
      "loss": 0.0007,
      "step": 3257
    },
    {
      "epoch": 1.1920965971459934,
      "grad_norm": 0.04235648736357689,
      "learning_rate": 2.0384047267355984e-05,
      "loss": 0.0011,
      "step": 3258
    },
    {
      "epoch": 1.1924624954262715,
      "grad_norm": 0.010338540188968182,
      "learning_rate": 2.0374815361890694e-05,
      "loss": 0.0004,
      "step": 3259
    },
    {
      "epoch": 1.1928283937065496,
      "grad_norm": 0.016851652413606644,
      "learning_rate": 2.036558345642541e-05,
      "loss": 0.0006,
      "step": 3260
    },
    {
      "epoch": 1.1931942919868277,
      "grad_norm": 0.019670160487294197,
      "learning_rate": 2.035635155096012e-05,
      "loss": 0.0006,
      "step": 3261
    },
    {
      "epoch": 1.1935601902671058,
      "grad_norm": 0.057452958077192307,
      "learning_rate": 2.034711964549483e-05,
      "loss": 0.001,
      "step": 3262
    },
    {
      "epoch": 1.193926088547384,
      "grad_norm": 0.01560608297586441,
      "learning_rate": 2.0337887740029543e-05,
      "loss": 0.0006,
      "step": 3263
    },
    {
      "epoch": 1.194291986827662,
      "grad_norm": 0.08079899102449417,
      "learning_rate": 2.0328655834564257e-05,
      "loss": 0.0018,
      "step": 3264
    },
    {
      "epoch": 1.19465788510794,
      "grad_norm": 0.1601235419511795,
      "learning_rate": 2.0319423929098964e-05,
      "loss": 0.0012,
      "step": 3265
    },
    {
      "epoch": 1.195023783388218,
      "grad_norm": 0.013516063801944256,
      "learning_rate": 2.031019202363368e-05,
      "loss": 0.0004,
      "step": 3266
    },
    {
      "epoch": 1.1953896816684961,
      "grad_norm": 0.055684275925159454,
      "learning_rate": 2.0300960118168392e-05,
      "loss": 0.0013,
      "step": 3267
    },
    {
      "epoch": 1.1957555799487742,
      "grad_norm": 0.005900131072849035,
      "learning_rate": 2.0291728212703103e-05,
      "loss": 0.0003,
      "step": 3268
    },
    {
      "epoch": 1.1961214782290523,
      "grad_norm": 0.010602157562971115,
      "learning_rate": 2.0282496307237813e-05,
      "loss": 0.0004,
      "step": 3269
    },
    {
      "epoch": 1.1964873765093305,
      "grad_norm": 0.0648721233010292,
      "learning_rate": 2.0273264401772527e-05,
      "loss": 0.001,
      "step": 3270
    },
    {
      "epoch": 1.1968532747896086,
      "grad_norm": 0.02919216826558113,
      "learning_rate": 2.0264032496307238e-05,
      "loss": 0.0006,
      "step": 3271
    },
    {
      "epoch": 1.1972191730698865,
      "grad_norm": 0.03531704843044281,
      "learning_rate": 2.0254800590841952e-05,
      "loss": 0.0007,
      "step": 3272
    },
    {
      "epoch": 1.1975850713501646,
      "grad_norm": 0.041586145758628845,
      "learning_rate": 2.0245568685376662e-05,
      "loss": 0.0009,
      "step": 3273
    },
    {
      "epoch": 1.1979509696304427,
      "grad_norm": 0.018494807183742523,
      "learning_rate": 2.0236336779911373e-05,
      "loss": 0.0005,
      "step": 3274
    },
    {
      "epoch": 1.1983168679107208,
      "grad_norm": 0.005714723840355873,
      "learning_rate": 2.0227104874446087e-05,
      "loss": 0.0003,
      "step": 3275
    },
    {
      "epoch": 1.198682766190999,
      "grad_norm": 0.0748540610074997,
      "learning_rate": 2.02178729689808e-05,
      "loss": 0.0008,
      "step": 3276
    },
    {
      "epoch": 1.199048664471277,
      "grad_norm": 0.009562901221215725,
      "learning_rate": 2.020864106351551e-05,
      "loss": 0.0004,
      "step": 3277
    },
    {
      "epoch": 1.1994145627515551,
      "grad_norm": 0.006245824042707682,
      "learning_rate": 2.0199409158050222e-05,
      "loss": 0.0003,
      "step": 3278
    },
    {
      "epoch": 1.1997804610318332,
      "grad_norm": 0.008272073231637478,
      "learning_rate": 2.0190177252584936e-05,
      "loss": 0.0003,
      "step": 3279
    },
    {
      "epoch": 1.2001463593121113,
      "grad_norm": 0.005777985323220491,
      "learning_rate": 2.0180945347119646e-05,
      "loss": 0.0003,
      "step": 3280
    },
    {
      "epoch": 1.2005122575923892,
      "grad_norm": 0.020243516191840172,
      "learning_rate": 2.0171713441654357e-05,
      "loss": 0.0003,
      "step": 3281
    },
    {
      "epoch": 1.2008781558726673,
      "grad_norm": 0.0094853350892663,
      "learning_rate": 2.016248153618907e-05,
      "loss": 0.0003,
      "step": 3282
    },
    {
      "epoch": 1.2012440541529454,
      "grad_norm": 0.006705585401505232,
      "learning_rate": 2.0153249630723785e-05,
      "loss": 0.0003,
      "step": 3283
    },
    {
      "epoch": 1.2016099524332235,
      "grad_norm": 0.008366566151380539,
      "learning_rate": 2.0144017725258492e-05,
      "loss": 0.0003,
      "step": 3284
    },
    {
      "epoch": 1.2019758507135017,
      "grad_norm": 0.005836303345859051,
      "learning_rate": 2.0134785819793206e-05,
      "loss": 0.0003,
      "step": 3285
    },
    {
      "epoch": 1.2023417489937798,
      "grad_norm": 0.013155228458344936,
      "learning_rate": 2.012555391432792e-05,
      "loss": 0.0003,
      "step": 3286
    },
    {
      "epoch": 1.2027076472740579,
      "grad_norm": 0.1771305948495865,
      "learning_rate": 2.011632200886263e-05,
      "loss": 0.0034,
      "step": 3287
    },
    {
      "epoch": 1.203073545554336,
      "grad_norm": 0.003961509559303522,
      "learning_rate": 2.010709010339734e-05,
      "loss": 0.0002,
      "step": 3288
    },
    {
      "epoch": 1.2034394438346139,
      "grad_norm": 0.0389350950717926,
      "learning_rate": 2.0097858197932055e-05,
      "loss": 0.0006,
      "step": 3289
    },
    {
      "epoch": 1.203805342114892,
      "grad_norm": 0.008587593212723732,
      "learning_rate": 2.0088626292466765e-05,
      "loss": 0.0003,
      "step": 3290
    },
    {
      "epoch": 1.20417124039517,
      "grad_norm": 0.03372571989893913,
      "learning_rate": 2.0079394387001476e-05,
      "loss": 0.0007,
      "step": 3291
    },
    {
      "epoch": 1.2045371386754482,
      "grad_norm": 0.00641021179035306,
      "learning_rate": 2.007016248153619e-05,
      "loss": 0.0002,
      "step": 3292
    },
    {
      "epoch": 1.2049030369557263,
      "grad_norm": 4.839038372039795,
      "learning_rate": 2.00609305760709e-05,
      "loss": 0.2215,
      "step": 3293
    },
    {
      "epoch": 1.2052689352360044,
      "grad_norm": 0.01819152943789959,
      "learning_rate": 2.0051698670605614e-05,
      "loss": 0.0003,
      "step": 3294
    },
    {
      "epoch": 1.2056348335162825,
      "grad_norm": 13.066176414489746,
      "learning_rate": 2.0042466765140325e-05,
      "loss": 0.2682,
      "step": 3295
    },
    {
      "epoch": 1.2060007317965606,
      "grad_norm": 0.03264129161834717,
      "learning_rate": 2.003323485967504e-05,
      "loss": 0.0005,
      "step": 3296
    },
    {
      "epoch": 1.2063666300768388,
      "grad_norm": 0.005679728463292122,
      "learning_rate": 2.002400295420975e-05,
      "loss": 0.0002,
      "step": 3297
    },
    {
      "epoch": 1.2067325283571166,
      "grad_norm": 0.009226368740200996,
      "learning_rate": 2.0014771048744463e-05,
      "loss": 0.0003,
      "step": 3298
    },
    {
      "epoch": 1.2070984266373948,
      "grad_norm": 0.004619322717189789,
      "learning_rate": 2.0005539143279174e-05,
      "loss": 0.0002,
      "step": 3299
    },
    {
      "epoch": 1.2074643249176729,
      "grad_norm": 0.05935478210449219,
      "learning_rate": 1.9996307237813885e-05,
      "loss": 0.0005,
      "step": 3300
    },
    {
      "epoch": 1.207830223197951,
      "grad_norm": 1.648099422454834,
      "learning_rate": 1.99870753323486e-05,
      "loss": 0.0143,
      "step": 3301
    },
    {
      "epoch": 1.208196121478229,
      "grad_norm": 14.508955001831055,
      "learning_rate": 1.9977843426883312e-05,
      "loss": 0.1644,
      "step": 3302
    },
    {
      "epoch": 1.2085620197585072,
      "grad_norm": 0.047127071768045425,
      "learning_rate": 1.996861152141802e-05,
      "loss": 0.0006,
      "step": 3303
    },
    {
      "epoch": 1.2089279180387853,
      "grad_norm": 0.0822020173072815,
      "learning_rate": 1.9959379615952733e-05,
      "loss": 0.0013,
      "step": 3304
    },
    {
      "epoch": 1.2092938163190632,
      "grad_norm": 0.05884671211242676,
      "learning_rate": 1.9950147710487447e-05,
      "loss": 0.0007,
      "step": 3305
    },
    {
      "epoch": 1.2096597145993413,
      "grad_norm": 0.40928515791893005,
      "learning_rate": 1.9940915805022155e-05,
      "loss": 0.0031,
      "step": 3306
    },
    {
      "epoch": 1.2100256128796194,
      "grad_norm": 0.017835844308137894,
      "learning_rate": 1.993168389955687e-05,
      "loss": 0.0003,
      "step": 3307
    },
    {
      "epoch": 1.2103915111598975,
      "grad_norm": 0.007122342474758625,
      "learning_rate": 1.9922451994091582e-05,
      "loss": 0.0003,
      "step": 3308
    },
    {
      "epoch": 1.2107574094401756,
      "grad_norm": 0.004490976221859455,
      "learning_rate": 1.9913220088626293e-05,
      "loss": 0.0002,
      "step": 3309
    },
    {
      "epoch": 1.2111233077204537,
      "grad_norm": 0.06297315657138824,
      "learning_rate": 1.9903988183161004e-05,
      "loss": 0.0006,
      "step": 3310
    },
    {
      "epoch": 1.2114892060007318,
      "grad_norm": 0.07492397725582123,
      "learning_rate": 1.9894756277695717e-05,
      "loss": 0.0008,
      "step": 3311
    },
    {
      "epoch": 1.21185510428101,
      "grad_norm": 11.40316104888916,
      "learning_rate": 1.9885524372230428e-05,
      "loss": 0.64,
      "step": 3312
    },
    {
      "epoch": 1.212221002561288,
      "grad_norm": 0.00578479515388608,
      "learning_rate": 1.9876292466765142e-05,
      "loss": 0.0003,
      "step": 3313
    },
    {
      "epoch": 1.212586900841566,
      "grad_norm": 0.30362898111343384,
      "learning_rate": 1.9867060561299853e-05,
      "loss": 0.0043,
      "step": 3314
    },
    {
      "epoch": 1.212952799121844,
      "grad_norm": 0.12036679685115814,
      "learning_rate": 1.9857828655834566e-05,
      "loss": 0.001,
      "step": 3315
    },
    {
      "epoch": 1.2133186974021222,
      "grad_norm": 5.8765716552734375,
      "learning_rate": 1.9848596750369277e-05,
      "loss": 0.2093,
      "step": 3316
    },
    {
      "epoch": 1.2136845956824003,
      "grad_norm": 19.887123107910156,
      "learning_rate": 1.983936484490399e-05,
      "loss": 0.1985,
      "step": 3317
    },
    {
      "epoch": 1.2140504939626784,
      "grad_norm": 0.12871411442756653,
      "learning_rate": 1.98301329394387e-05,
      "loss": 0.0022,
      "step": 3318
    },
    {
      "epoch": 1.2144163922429565,
      "grad_norm": 0.09431731700897217,
      "learning_rate": 1.9820901033973412e-05,
      "loss": 0.0009,
      "step": 3319
    },
    {
      "epoch": 1.2147822905232346,
      "grad_norm": 0.16585227847099304,
      "learning_rate": 1.9811669128508126e-05,
      "loss": 0.0018,
      "step": 3320
    },
    {
      "epoch": 1.2151481888035127,
      "grad_norm": 7.6236701011657715,
      "learning_rate": 1.9802437223042837e-05,
      "loss": 0.1083,
      "step": 3321
    },
    {
      "epoch": 1.2155140870837906,
      "grad_norm": 0.2550187408924103,
      "learning_rate": 1.9793205317577547e-05,
      "loss": 0.0024,
      "step": 3322
    },
    {
      "epoch": 1.2158799853640687,
      "grad_norm": 0.013299000449478626,
      "learning_rate": 1.978397341211226e-05,
      "loss": 0.0003,
      "step": 3323
    },
    {
      "epoch": 1.2162458836443468,
      "grad_norm": 0.009596642106771469,
      "learning_rate": 1.9774741506646975e-05,
      "loss": 0.0004,
      "step": 3324
    },
    {
      "epoch": 1.216611781924625,
      "grad_norm": 0.013926897197961807,
      "learning_rate": 1.9765509601181682e-05,
      "loss": 0.0003,
      "step": 3325
    },
    {
      "epoch": 1.216977680204903,
      "grad_norm": 0.010980174876749516,
      "learning_rate": 1.9756277695716396e-05,
      "loss": 0.0003,
      "step": 3326
    },
    {
      "epoch": 1.2173435784851812,
      "grad_norm": 0.17668847739696503,
      "learning_rate": 1.974704579025111e-05,
      "loss": 0.0016,
      "step": 3327
    },
    {
      "epoch": 1.2177094767654593,
      "grad_norm": 0.12056700885295868,
      "learning_rate": 1.973781388478582e-05,
      "loss": 0.0017,
      "step": 3328
    },
    {
      "epoch": 1.2180753750457374,
      "grad_norm": 6.2979912757873535,
      "learning_rate": 1.972858197932053e-05,
      "loss": 0.2763,
      "step": 3329
    },
    {
      "epoch": 1.2184412733260155,
      "grad_norm": 0.05705837160348892,
      "learning_rate": 1.9719350073855245e-05,
      "loss": 0.0007,
      "step": 3330
    },
    {
      "epoch": 1.2188071716062934,
      "grad_norm": 3.499901294708252,
      "learning_rate": 1.9710118168389956e-05,
      "loss": 0.1724,
      "step": 3331
    },
    {
      "epoch": 1.2191730698865715,
      "grad_norm": 0.0068640438839793205,
      "learning_rate": 1.970088626292467e-05,
      "loss": 0.0003,
      "step": 3332
    },
    {
      "epoch": 1.2195389681668496,
      "grad_norm": 0.005087016616016626,
      "learning_rate": 1.969165435745938e-05,
      "loss": 0.0002,
      "step": 3333
    },
    {
      "epoch": 1.2199048664471277,
      "grad_norm": 0.34548524022102356,
      "learning_rate": 1.9682422451994094e-05,
      "loss": 0.0036,
      "step": 3334
    },
    {
      "epoch": 1.2202707647274058,
      "grad_norm": 0.20049351453781128,
      "learning_rate": 1.9673190546528805e-05,
      "loss": 0.004,
      "step": 3335
    },
    {
      "epoch": 1.220636663007684,
      "grad_norm": 0.009419719688594341,
      "learning_rate": 1.9663958641063515e-05,
      "loss": 0.0003,
      "step": 3336
    },
    {
      "epoch": 1.221002561287962,
      "grad_norm": 0.005073810927569866,
      "learning_rate": 1.965472673559823e-05,
      "loss": 0.0002,
      "step": 3337
    },
    {
      "epoch": 1.22136845956824,
      "grad_norm": 2.0467889308929443,
      "learning_rate": 1.964549483013294e-05,
      "loss": 0.5163,
      "step": 3338
    },
    {
      "epoch": 1.221734357848518,
      "grad_norm": 0.008345366455614567,
      "learning_rate": 1.9636262924667654e-05,
      "loss": 0.0004,
      "step": 3339
    },
    {
      "epoch": 1.2221002561287961,
      "grad_norm": 0.05324974283576012,
      "learning_rate": 1.9627031019202364e-05,
      "loss": 0.0006,
      "step": 3340
    },
    {
      "epoch": 1.2224661544090742,
      "grad_norm": 6.352705955505371,
      "learning_rate": 1.9617799113737075e-05,
      "loss": 0.2026,
      "step": 3341
    },
    {
      "epoch": 1.2228320526893524,
      "grad_norm": 0.29622936248779297,
      "learning_rate": 1.960856720827179e-05,
      "loss": 0.0025,
      "step": 3342
    },
    {
      "epoch": 1.2231979509696305,
      "grad_norm": 0.007297398988157511,
      "learning_rate": 1.9599335302806502e-05,
      "loss": 0.0003,
      "step": 3343
    },
    {
      "epoch": 1.2235638492499086,
      "grad_norm": 0.16392262279987335,
      "learning_rate": 1.959010339734121e-05,
      "loss": 0.0019,
      "step": 3344
    },
    {
      "epoch": 1.2239297475301867,
      "grad_norm": 0.03343691676855087,
      "learning_rate": 1.9580871491875924e-05,
      "loss": 0.0006,
      "step": 3345
    },
    {
      "epoch": 1.2242956458104648,
      "grad_norm": 0.35877591371536255,
      "learning_rate": 1.9571639586410638e-05,
      "loss": 0.0034,
      "step": 3346
    },
    {
      "epoch": 1.2246615440907427,
      "grad_norm": 0.04476356506347656,
      "learning_rate": 1.9562407680945348e-05,
      "loss": 0.0004,
      "step": 3347
    },
    {
      "epoch": 1.2250274423710208,
      "grad_norm": 0.1950269639492035,
      "learning_rate": 1.955317577548006e-05,
      "loss": 0.0022,
      "step": 3348
    },
    {
      "epoch": 1.225393340651299,
      "grad_norm": 0.026113523170351982,
      "learning_rate": 1.9543943870014773e-05,
      "loss": 0.0006,
      "step": 3349
    },
    {
      "epoch": 1.225759238931577,
      "grad_norm": 0.3314107060432434,
      "learning_rate": 1.9534711964549483e-05,
      "loss": 0.0037,
      "step": 3350
    },
    {
      "epoch": 1.2261251372118551,
      "grad_norm": 0.06757180392742157,
      "learning_rate": 1.9525480059084194e-05,
      "loss": 0.0008,
      "step": 3351
    },
    {
      "epoch": 1.2264910354921332,
      "grad_norm": 0.011280201375484467,
      "learning_rate": 1.9516248153618908e-05,
      "loss": 0.0004,
      "step": 3352
    },
    {
      "epoch": 1.2268569337724113,
      "grad_norm": 0.32605722546577454,
      "learning_rate": 1.950701624815362e-05,
      "loss": 0.0025,
      "step": 3353
    },
    {
      "epoch": 1.2272228320526892,
      "grad_norm": 0.3718966245651245,
      "learning_rate": 1.9497784342688332e-05,
      "loss": 0.0044,
      "step": 3354
    },
    {
      "epoch": 1.2275887303329673,
      "grad_norm": 0.007467193994671106,
      "learning_rate": 1.9488552437223043e-05,
      "loss": 0.0004,
      "step": 3355
    },
    {
      "epoch": 1.2279546286132454,
      "grad_norm": 0.006677944213151932,
      "learning_rate": 1.9479320531757757e-05,
      "loss": 0.0003,
      "step": 3356
    },
    {
      "epoch": 1.2283205268935236,
      "grad_norm": 0.010543424636125565,
      "learning_rate": 1.9470088626292467e-05,
      "loss": 0.0004,
      "step": 3357
    },
    {
      "epoch": 1.2286864251738017,
      "grad_norm": 0.023497823625802994,
      "learning_rate": 1.946085672082718e-05,
      "loss": 0.0005,
      "step": 3358
    },
    {
      "epoch": 1.2290523234540798,
      "grad_norm": 9.428997993469238,
      "learning_rate": 1.945162481536189e-05,
      "loss": 0.1619,
      "step": 3359
    },
    {
      "epoch": 1.2294182217343579,
      "grad_norm": 0.5450267195701599,
      "learning_rate": 1.9442392909896602e-05,
      "loss": 0.0059,
      "step": 3360
    },
    {
      "epoch": 1.229784120014636,
      "grad_norm": 0.40316012501716614,
      "learning_rate": 1.9433161004431316e-05,
      "loss": 0.0047,
      "step": 3361
    },
    {
      "epoch": 1.230150018294914,
      "grad_norm": 0.008476764895021915,
      "learning_rate": 1.942392909896603e-05,
      "loss": 0.0003,
      "step": 3362
    },
    {
      "epoch": 1.2305159165751922,
      "grad_norm": 0.7629579305648804,
      "learning_rate": 1.9414697193500737e-05,
      "loss": 0.0044,
      "step": 3363
    },
    {
      "epoch": 1.23088181485547,
      "grad_norm": 0.011874595656991005,
      "learning_rate": 1.940546528803545e-05,
      "loss": 0.0005,
      "step": 3364
    },
    {
      "epoch": 1.2312477131357482,
      "grad_norm": 0.015556234866380692,
      "learning_rate": 1.9396233382570165e-05,
      "loss": 0.0006,
      "step": 3365
    },
    {
      "epoch": 1.2316136114160263,
      "grad_norm": 0.14947612583637238,
      "learning_rate": 1.9387001477104876e-05,
      "loss": 0.0014,
      "step": 3366
    },
    {
      "epoch": 1.2319795096963044,
      "grad_norm": 13.863313674926758,
      "learning_rate": 1.9377769571639586e-05,
      "loss": 0.3018,
      "step": 3367
    },
    {
      "epoch": 1.2323454079765825,
      "grad_norm": 0.015280655585229397,
      "learning_rate": 1.93685376661743e-05,
      "loss": 0.0003,
      "step": 3368
    },
    {
      "epoch": 1.2327113062568607,
      "grad_norm": 0.01146431639790535,
      "learning_rate": 1.935930576070901e-05,
      "loss": 0.0005,
      "step": 3369
    },
    {
      "epoch": 1.2330772045371388,
      "grad_norm": 0.04405948892235756,
      "learning_rate": 1.935007385524372e-05,
      "loss": 0.0006,
      "step": 3370
    },
    {
      "epoch": 1.2334431028174166,
      "grad_norm": 0.006565211806446314,
      "learning_rate": 1.9340841949778435e-05,
      "loss": 0.0003,
      "step": 3371
    },
    {
      "epoch": 1.2338090010976948,
      "grad_norm": 0.013909848406910896,
      "learning_rate": 1.933161004431315e-05,
      "loss": 0.0004,
      "step": 3372
    },
    {
      "epoch": 1.2341748993779729,
      "grad_norm": 0.01591118425130844,
      "learning_rate": 1.932237813884786e-05,
      "loss": 0.0003,
      "step": 3373
    },
    {
      "epoch": 1.234540797658251,
      "grad_norm": 0.041993845254182816,
      "learning_rate": 1.931314623338257e-05,
      "loss": 0.0006,
      "step": 3374
    },
    {
      "epoch": 1.234906695938529,
      "grad_norm": 0.0622442364692688,
      "learning_rate": 1.9303914327917284e-05,
      "loss": 0.0007,
      "step": 3375
    },
    {
      "epoch": 1.2352725942188072,
      "grad_norm": 0.12136901170015335,
      "learning_rate": 1.9294682422451995e-05,
      "loss": 0.0012,
      "step": 3376
    },
    {
      "epoch": 1.2356384924990853,
      "grad_norm": 0.35528257489204407,
      "learning_rate": 1.928545051698671e-05,
      "loss": 0.0027,
      "step": 3377
    },
    {
      "epoch": 1.2360043907793634,
      "grad_norm": 0.00546242156997323,
      "learning_rate": 1.927621861152142e-05,
      "loss": 0.0003,
      "step": 3378
    },
    {
      "epoch": 1.2363702890596415,
      "grad_norm": 0.004649352747946978,
      "learning_rate": 1.926698670605613e-05,
      "loss": 0.0002,
      "step": 3379
    },
    {
      "epoch": 1.2367361873399194,
      "grad_norm": 0.02890283241868019,
      "learning_rate": 1.9257754800590844e-05,
      "loss": 0.0005,
      "step": 3380
    },
    {
      "epoch": 1.2371020856201975,
      "grad_norm": 0.5095432996749878,
      "learning_rate": 1.9248522895125554e-05,
      "loss": 0.0074,
      "step": 3381
    },
    {
      "epoch": 1.2374679839004756,
      "grad_norm": 0.009391454979777336,
      "learning_rate": 1.9239290989660265e-05,
      "loss": 0.0003,
      "step": 3382
    },
    {
      "epoch": 1.2378338821807537,
      "grad_norm": 0.02793816290795803,
      "learning_rate": 1.923005908419498e-05,
      "loss": 0.0005,
      "step": 3383
    },
    {
      "epoch": 1.2381997804610319,
      "grad_norm": 0.01486060582101345,
      "learning_rate": 1.9220827178729693e-05,
      "loss": 0.0003,
      "step": 3384
    },
    {
      "epoch": 1.23856567874131,
      "grad_norm": 2.1611948013305664,
      "learning_rate": 1.9211595273264403e-05,
      "loss": 0.0128,
      "step": 3385
    },
    {
      "epoch": 1.238931577021588,
      "grad_norm": 6.380728721618652,
      "learning_rate": 1.9202363367799114e-05,
      "loss": 0.3326,
      "step": 3386
    },
    {
      "epoch": 1.239297475301866,
      "grad_norm": 0.0054658930748701096,
      "learning_rate": 1.9193131462333828e-05,
      "loss": 0.0002,
      "step": 3387
    },
    {
      "epoch": 1.239663373582144,
      "grad_norm": 0.008976873010396957,
      "learning_rate": 1.9183899556868538e-05,
      "loss": 0.0004,
      "step": 3388
    },
    {
      "epoch": 1.2400292718624222,
      "grad_norm": 0.0056688375771045685,
      "learning_rate": 1.917466765140325e-05,
      "loss": 0.0002,
      "step": 3389
    },
    {
      "epoch": 1.2403951701427003,
      "grad_norm": 0.005987746641039848,
      "learning_rate": 1.9165435745937963e-05,
      "loss": 0.0003,
      "step": 3390
    },
    {
      "epoch": 1.2407610684229784,
      "grad_norm": 0.009038273245096207,
      "learning_rate": 1.9156203840472677e-05,
      "loss": 0.0004,
      "step": 3391
    },
    {
      "epoch": 1.2411269667032565,
      "grad_norm": 0.012654747813940048,
      "learning_rate": 1.9146971935007384e-05,
      "loss": 0.0005,
      "step": 3392
    },
    {
      "epoch": 1.2414928649835346,
      "grad_norm": 0.00800019409507513,
      "learning_rate": 1.9137740029542098e-05,
      "loss": 0.0003,
      "step": 3393
    },
    {
      "epoch": 1.2418587632638127,
      "grad_norm": 0.008285420946776867,
      "learning_rate": 1.912850812407681e-05,
      "loss": 0.0003,
      "step": 3394
    },
    {
      "epoch": 1.2422246615440908,
      "grad_norm": 0.04764702543616295,
      "learning_rate": 1.9119276218611522e-05,
      "loss": 0.0007,
      "step": 3395
    },
    {
      "epoch": 1.242590559824369,
      "grad_norm": 0.08097922801971436,
      "learning_rate": 1.9110044313146233e-05,
      "loss": 0.0009,
      "step": 3396
    },
    {
      "epoch": 1.2429564581046468,
      "grad_norm": 0.007577800191938877,
      "learning_rate": 1.9100812407680947e-05,
      "loss": 0.0003,
      "step": 3397
    },
    {
      "epoch": 1.243322356384925,
      "grad_norm": 0.011743912473320961,
      "learning_rate": 1.9091580502215657e-05,
      "loss": 0.0004,
      "step": 3398
    },
    {
      "epoch": 1.243688254665203,
      "grad_norm": 0.006048922892659903,
      "learning_rate": 1.908234859675037e-05,
      "loss": 0.0003,
      "step": 3399
    },
    {
      "epoch": 1.2440541529454812,
      "grad_norm": 18.209314346313477,
      "learning_rate": 1.9073116691285082e-05,
      "loss": 0.0319,
      "step": 3400
    },
    {
      "epoch": 1.2444200512257593,
      "grad_norm": 0.0068343812599778175,
      "learning_rate": 1.9063884785819792e-05,
      "loss": 0.0003,
      "step": 3401
    },
    {
      "epoch": 1.2447859495060374,
      "grad_norm": 0.00521602900698781,
      "learning_rate": 1.9054652880354506e-05,
      "loss": 0.0002,
      "step": 3402
    },
    {
      "epoch": 1.2451518477863155,
      "grad_norm": 0.010436944663524628,
      "learning_rate": 1.904542097488922e-05,
      "loss": 0.0003,
      "step": 3403
    },
    {
      "epoch": 1.2455177460665934,
      "grad_norm": 0.025367649272084236,
      "learning_rate": 1.903618906942393e-05,
      "loss": 0.0005,
      "step": 3404
    },
    {
      "epoch": 1.2458836443468715,
      "grad_norm": 0.006268155761063099,
      "learning_rate": 1.902695716395864e-05,
      "loss": 0.0003,
      "step": 3405
    },
    {
      "epoch": 1.2462495426271496,
      "grad_norm": 0.015662137418985367,
      "learning_rate": 1.9017725258493355e-05,
      "loss": 0.0005,
      "step": 3406
    },
    {
      "epoch": 1.2466154409074277,
      "grad_norm": 0.006914705038070679,
      "learning_rate": 1.9008493353028066e-05,
      "loss": 0.0003,
      "step": 3407
    },
    {
      "epoch": 1.2469813391877058,
      "grad_norm": 0.007304540369659662,
      "learning_rate": 1.8999261447562776e-05,
      "loss": 0.0003,
      "step": 3408
    },
    {
      "epoch": 1.247347237467984,
      "grad_norm": 0.31365278363227844,
      "learning_rate": 1.899002954209749e-05,
      "loss": 0.0028,
      "step": 3409
    },
    {
      "epoch": 1.247713135748262,
      "grad_norm": 0.006279589142650366,
      "learning_rate": 1.8980797636632204e-05,
      "loss": 0.0003,
      "step": 3410
    },
    {
      "epoch": 1.2480790340285401,
      "grad_norm": 0.18357452750205994,
      "learning_rate": 1.897156573116691e-05,
      "loss": 0.0016,
      "step": 3411
    },
    {
      "epoch": 1.2484449323088183,
      "grad_norm": 0.3271850049495697,
      "learning_rate": 1.8962333825701625e-05,
      "loss": 0.0038,
      "step": 3412
    },
    {
      "epoch": 1.2488108305890961,
      "grad_norm": 0.004667554050683975,
      "learning_rate": 1.895310192023634e-05,
      "loss": 0.0002,
      "step": 3413
    },
    {
      "epoch": 1.2491767288693743,
      "grad_norm": 0.07789824157953262,
      "learning_rate": 1.894387001477105e-05,
      "loss": 0.0006,
      "step": 3414
    },
    {
      "epoch": 1.2495426271496524,
      "grad_norm": 0.004536328837275505,
      "learning_rate": 1.893463810930576e-05,
      "loss": 0.0002,
      "step": 3415
    },
    {
      "epoch": 1.2499085254299305,
      "grad_norm": 0.04080643132328987,
      "learning_rate": 1.8925406203840474e-05,
      "loss": 0.0005,
      "step": 3416
    },
    {
      "epoch": 1.2502744237102086,
      "grad_norm": 0.00609021307900548,
      "learning_rate": 1.8916174298375185e-05,
      "loss": 0.0003,
      "step": 3417
    },
    {
      "epoch": 1.2506403219904867,
      "grad_norm": 1.5706346035003662,
      "learning_rate": 1.89069423929099e-05,
      "loss": 0.0085,
      "step": 3418
    },
    {
      "epoch": 1.2510062202707648,
      "grad_norm": 0.010936247184872627,
      "learning_rate": 1.889771048744461e-05,
      "loss": 0.0003,
      "step": 3419
    },
    {
      "epoch": 1.2513721185510427,
      "grad_norm": 0.007529372815042734,
      "learning_rate": 1.888847858197932e-05,
      "loss": 0.0003,
      "step": 3420
    },
    {
      "epoch": 1.2517380168313208,
      "grad_norm": 2.2607085704803467,
      "learning_rate": 1.8879246676514034e-05,
      "loss": 0.5669,
      "step": 3421
    },
    {
      "epoch": 1.252103915111599,
      "grad_norm": 19.13199806213379,
      "learning_rate": 1.8870014771048744e-05,
      "loss": 0.0312,
      "step": 3422
    },
    {
      "epoch": 1.252469813391877,
      "grad_norm": 0.09298786520957947,
      "learning_rate": 1.8860782865583458e-05,
      "loss": 0.0008,
      "step": 3423
    },
    {
      "epoch": 1.2528357116721551,
      "grad_norm": 7.419679164886475,
      "learning_rate": 1.885155096011817e-05,
      "loss": 0.404,
      "step": 3424
    },
    {
      "epoch": 1.2532016099524332,
      "grad_norm": 0.051304563879966736,
      "learning_rate": 1.8842319054652883e-05,
      "loss": 0.0006,
      "step": 3425
    },
    {
      "epoch": 1.2535675082327113,
      "grad_norm": 9.652484893798828,
      "learning_rate": 1.8833087149187593e-05,
      "loss": 0.068,
      "step": 3426
    },
    {
      "epoch": 1.2539334065129895,
      "grad_norm": 0.00872541218996048,
      "learning_rate": 1.8823855243722304e-05,
      "loss": 0.0004,
      "step": 3427
    },
    {
      "epoch": 1.2542993047932676,
      "grad_norm": 0.010955732315778732,
      "learning_rate": 1.8814623338257018e-05,
      "loss": 0.0005,
      "step": 3428
    },
    {
      "epoch": 1.2546652030735457,
      "grad_norm": 0.01693805120885372,
      "learning_rate": 1.880539143279173e-05,
      "loss": 0.0007,
      "step": 3429
    },
    {
      "epoch": 1.2550311013538236,
      "grad_norm": 0.026541857048869133,
      "learning_rate": 1.879615952732644e-05,
      "loss": 0.0009,
      "step": 3430
    },
    {
      "epoch": 1.2553969996341017,
      "grad_norm": 0.14205367863178253,
      "learning_rate": 1.8786927621861153e-05,
      "loss": 0.0019,
      "step": 3431
    },
    {
      "epoch": 1.2557628979143798,
      "grad_norm": 1.991899013519287,
      "learning_rate": 1.8777695716395867e-05,
      "loss": 0.171,
      "step": 3432
    },
    {
      "epoch": 1.256128796194658,
      "grad_norm": 0.02684234082698822,
      "learning_rate": 1.8768463810930577e-05,
      "loss": 0.001,
      "step": 3433
    },
    {
      "epoch": 1.256494694474936,
      "grad_norm": 0.01413586363196373,
      "learning_rate": 1.8759231905465288e-05,
      "loss": 0.0006,
      "step": 3434
    },
    {
      "epoch": 1.2568605927552141,
      "grad_norm": 0.11246520280838013,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.0029,
      "step": 3435
    },
    {
      "epoch": 1.257226491035492,
      "grad_norm": 10.80744457244873,
      "learning_rate": 1.8740768094534712e-05,
      "loss": 0.2598,
      "step": 3436
    },
    {
      "epoch": 1.25759238931577,
      "grad_norm": 0.07287672907114029,
      "learning_rate": 1.8731536189069423e-05,
      "loss": 0.0015,
      "step": 3437
    },
    {
      "epoch": 1.2579582875960482,
      "grad_norm": 0.02327137254178524,
      "learning_rate": 1.8722304283604137e-05,
      "loss": 0.0009,
      "step": 3438
    },
    {
      "epoch": 1.2583241858763263,
      "grad_norm": 0.02252202481031418,
      "learning_rate": 1.8713072378138847e-05,
      "loss": 0.001,
      "step": 3439
    },
    {
      "epoch": 1.2586900841566044,
      "grad_norm": 13.391948699951172,
      "learning_rate": 1.870384047267356e-05,
      "loss": 0.2391,
      "step": 3440
    },
    {
      "epoch": 1.2590559824368825,
      "grad_norm": 0.012792774476110935,
      "learning_rate": 1.8694608567208272e-05,
      "loss": 0.0006,
      "step": 3441
    },
    {
      "epoch": 1.2594218807171607,
      "grad_norm": 0.03385165333747864,
      "learning_rate": 1.8685376661742986e-05,
      "loss": 0.0011,
      "step": 3442
    },
    {
      "epoch": 1.2597877789974388,
      "grad_norm": 17.49083709716797,
      "learning_rate": 1.8676144756277696e-05,
      "loss": 0.6394,
      "step": 3443
    },
    {
      "epoch": 1.2601536772777169,
      "grad_norm": 0.05770367383956909,
      "learning_rate": 1.866691285081241e-05,
      "loss": 0.0017,
      "step": 3444
    },
    {
      "epoch": 1.260519575557995,
      "grad_norm": 0.02252480573952198,
      "learning_rate": 1.865768094534712e-05,
      "loss": 0.0008,
      "step": 3445
    },
    {
      "epoch": 1.260885473838273,
      "grad_norm": 0.024428123608231544,
      "learning_rate": 1.864844903988183e-05,
      "loss": 0.0009,
      "step": 3446
    },
    {
      "epoch": 1.261251372118551,
      "grad_norm": 0.7865980863571167,
      "learning_rate": 1.8639217134416545e-05,
      "loss": 0.0103,
      "step": 3447
    },
    {
      "epoch": 1.261617270398829,
      "grad_norm": 2.8709001541137695,
      "learning_rate": 1.862998522895126e-05,
      "loss": 0.0142,
      "step": 3448
    },
    {
      "epoch": 1.2619831686791072,
      "grad_norm": 0.015553935430943966,
      "learning_rate": 1.8620753323485966e-05,
      "loss": 0.0007,
      "step": 3449
    },
    {
      "epoch": 1.2623490669593853,
      "grad_norm": 0.7061854004859924,
      "learning_rate": 1.861152141802068e-05,
      "loss": 0.0054,
      "step": 3450
    },
    {
      "epoch": 1.2627149652396634,
      "grad_norm": 0.09691692143678665,
      "learning_rate": 1.8602289512555394e-05,
      "loss": 0.001,
      "step": 3451
    },
    {
      "epoch": 1.2630808635199415,
      "grad_norm": 1.3091353178024292,
      "learning_rate": 1.85930576070901e-05,
      "loss": 0.0074,
      "step": 3452
    },
    {
      "epoch": 1.2634467618002194,
      "grad_norm": 0.03056967817246914,
      "learning_rate": 1.8583825701624815e-05,
      "loss": 0.0009,
      "step": 3453
    },
    {
      "epoch": 1.2638126600804975,
      "grad_norm": 0.570536732673645,
      "learning_rate": 1.857459379615953e-05,
      "loss": 0.0071,
      "step": 3454
    },
    {
      "epoch": 1.2641785583607756,
      "grad_norm": 0.8481528162956238,
      "learning_rate": 1.856536189069424e-05,
      "loss": 0.0079,
      "step": 3455
    },
    {
      "epoch": 1.2645444566410537,
      "grad_norm": 0.009111919440329075,
      "learning_rate": 1.855612998522895e-05,
      "loss": 0.0004,
      "step": 3456
    },
    {
      "epoch": 1.2649103549213319,
      "grad_norm": 0.21579761803150177,
      "learning_rate": 1.8546898079763664e-05,
      "loss": 0.0034,
      "step": 3457
    },
    {
      "epoch": 1.26527625320161,
      "grad_norm": 0.052668776363134384,
      "learning_rate": 1.8537666174298375e-05,
      "loss": 0.0005,
      "step": 3458
    },
    {
      "epoch": 1.265642151481888,
      "grad_norm": 6.340452671051025,
      "learning_rate": 1.852843426883309e-05,
      "loss": 0.1408,
      "step": 3459
    },
    {
      "epoch": 1.2660080497621662,
      "grad_norm": 10.27468204498291,
      "learning_rate": 1.85192023633678e-05,
      "loss": 0.031,
      "step": 3460
    },
    {
      "epoch": 1.2663739480424443,
      "grad_norm": 2.811997652053833,
      "learning_rate": 1.8509970457902513e-05,
      "loss": 0.0107,
      "step": 3461
    },
    {
      "epoch": 1.2667398463227224,
      "grad_norm": 0.4510214924812317,
      "learning_rate": 1.8500738552437224e-05,
      "loss": 0.003,
      "step": 3462
    },
    {
      "epoch": 1.2671057446030003,
      "grad_norm": 6.161521911621094,
      "learning_rate": 1.8491506646971938e-05,
      "loss": 0.1974,
      "step": 3463
    },
    {
      "epoch": 1.2674716428832784,
      "grad_norm": 0.007707761600613594,
      "learning_rate": 1.848227474150665e-05,
      "loss": 0.0004,
      "step": 3464
    },
    {
      "epoch": 1.2678375411635565,
      "grad_norm": 0.14329692721366882,
      "learning_rate": 1.847304283604136e-05,
      "loss": 0.0014,
      "step": 3465
    },
    {
      "epoch": 1.2682034394438346,
      "grad_norm": 0.008963217958807945,
      "learning_rate": 1.8463810930576073e-05,
      "loss": 0.0004,
      "step": 3466
    },
    {
      "epoch": 1.2685693377241127,
      "grad_norm": 0.0162354689091444,
      "learning_rate": 1.8454579025110783e-05,
      "loss": 0.0005,
      "step": 3467
    },
    {
      "epoch": 1.2689352360043908,
      "grad_norm": 20.20879364013672,
      "learning_rate": 1.8445347119645494e-05,
      "loss": 0.5106,
      "step": 3468
    },
    {
      "epoch": 1.2693011342846687,
      "grad_norm": 0.1922743022441864,
      "learning_rate": 1.8436115214180208e-05,
      "loss": 0.0013,
      "step": 3469
    },
    {
      "epoch": 1.2696670325649468,
      "grad_norm": 0.010959564708173275,
      "learning_rate": 1.8426883308714922e-05,
      "loss": 0.0005,
      "step": 3470
    },
    {
      "epoch": 1.270032930845225,
      "grad_norm": 22.19922637939453,
      "learning_rate": 1.841765140324963e-05,
      "loss": 0.0416,
      "step": 3471
    },
    {
      "epoch": 1.270398829125503,
      "grad_norm": 9.52852725982666,
      "learning_rate": 1.8408419497784343e-05,
      "loss": 0.1084,
      "step": 3472
    },
    {
      "epoch": 1.2707647274057812,
      "grad_norm": 37.76287078857422,
      "learning_rate": 1.8399187592319057e-05,
      "loss": 0.4007,
      "step": 3473
    },
    {
      "epoch": 1.2711306256860593,
      "grad_norm": 0.2540423572063446,
      "learning_rate": 1.8389955686853767e-05,
      "loss": 0.0013,
      "step": 3474
    },
    {
      "epoch": 1.2714965239663374,
      "grad_norm": 0.010287422686815262,
      "learning_rate": 1.8380723781388478e-05,
      "loss": 0.0005,
      "step": 3475
    },
    {
      "epoch": 1.2718624222466155,
      "grad_norm": 0.015284906141459942,
      "learning_rate": 1.8371491875923192e-05,
      "loss": 0.0004,
      "step": 3476
    },
    {
      "epoch": 1.2722283205268936,
      "grad_norm": 0.04164614528417587,
      "learning_rate": 1.8362259970457902e-05,
      "loss": 0.0008,
      "step": 3477
    },
    {
      "epoch": 1.2725942188071717,
      "grad_norm": 0.2587161362171173,
      "learning_rate": 1.8353028064992616e-05,
      "loss": 0.0022,
      "step": 3478
    },
    {
      "epoch": 1.2729601170874498,
      "grad_norm": 0.5241527557373047,
      "learning_rate": 1.8343796159527327e-05,
      "loss": 0.0039,
      "step": 3479
    },
    {
      "epoch": 1.2733260153677277,
      "grad_norm": 0.16427870094776154,
      "learning_rate": 1.833456425406204e-05,
      "loss": 0.0011,
      "step": 3480
    },
    {
      "epoch": 1.2736919136480058,
      "grad_norm": 0.00997576117515564,
      "learning_rate": 1.832533234859675e-05,
      "loss": 0.0004,
      "step": 3481
    },
    {
      "epoch": 1.274057811928284,
      "grad_norm": 0.009136571548879147,
      "learning_rate": 1.8316100443131462e-05,
      "loss": 0.0004,
      "step": 3482
    },
    {
      "epoch": 1.274423710208562,
      "grad_norm": 0.0071525899693369865,
      "learning_rate": 1.8306868537666176e-05,
      "loss": 0.0003,
      "step": 3483
    },
    {
      "epoch": 1.2747896084888402,
      "grad_norm": 0.017272254452109337,
      "learning_rate": 1.8297636632200886e-05,
      "loss": 0.0004,
      "step": 3484
    },
    {
      "epoch": 1.2751555067691183,
      "grad_norm": 0.6895308494567871,
      "learning_rate": 1.82884047267356e-05,
      "loss": 0.0066,
      "step": 3485
    },
    {
      "epoch": 1.2755214050493962,
      "grad_norm": 0.13321486115455627,
      "learning_rate": 1.827917282127031e-05,
      "loss": 0.0007,
      "step": 3486
    },
    {
      "epoch": 1.2758873033296743,
      "grad_norm": 0.007818316109478474,
      "learning_rate": 1.826994091580502e-05,
      "loss": 0.0004,
      "step": 3487
    },
    {
      "epoch": 1.2762532016099524,
      "grad_norm": 0.10545669496059418,
      "learning_rate": 1.8260709010339735e-05,
      "loss": 0.0005,
      "step": 3488
    },
    {
      "epoch": 1.2766190998902305,
      "grad_norm": 0.007263761479407549,
      "learning_rate": 1.825147710487445e-05,
      "loss": 0.0003,
      "step": 3489
    },
    {
      "epoch": 1.2769849981705086,
      "grad_norm": 0.15870404243469238,
      "learning_rate": 1.8242245199409156e-05,
      "loss": 0.0009,
      "step": 3490
    },
    {
      "epoch": 1.2773508964507867,
      "grad_norm": 0.007761103101074696,
      "learning_rate": 1.823301329394387e-05,
      "loss": 0.0003,
      "step": 3491
    },
    {
      "epoch": 1.2777167947310648,
      "grad_norm": 6.272299289703369,
      "learning_rate": 1.8223781388478584e-05,
      "loss": 0.4869,
      "step": 3492
    },
    {
      "epoch": 1.278082693011343,
      "grad_norm": 0.006881404668092728,
      "learning_rate": 1.8214549483013295e-05,
      "loss": 0.0003,
      "step": 3493
    },
    {
      "epoch": 1.278448591291621,
      "grad_norm": 0.03941008076071739,
      "learning_rate": 1.8205317577548005e-05,
      "loss": 0.0008,
      "step": 3494
    },
    {
      "epoch": 1.2788144895718991,
      "grad_norm": 0.006615237798541784,
      "learning_rate": 1.819608567208272e-05,
      "loss": 0.0003,
      "step": 3495
    },
    {
      "epoch": 1.279180387852177,
      "grad_norm": 0.060966938734054565,
      "learning_rate": 1.818685376661743e-05,
      "loss": 0.0005,
      "step": 3496
    },
    {
      "epoch": 1.2795462861324551,
      "grad_norm": 0.016094299033284187,
      "learning_rate": 1.817762186115214e-05,
      "loss": 0.0004,
      "step": 3497
    },
    {
      "epoch": 1.2799121844127332,
      "grad_norm": 15.225496292114258,
      "learning_rate": 1.8168389955686854e-05,
      "loss": 0.2349,
      "step": 3498
    },
    {
      "epoch": 1.2802780826930114,
      "grad_norm": 0.006420321296900511,
      "learning_rate": 1.815915805022157e-05,
      "loss": 0.0003,
      "step": 3499
    },
    {
      "epoch": 1.2806439809732895,
      "grad_norm": 0.009896103292703629,
      "learning_rate": 1.814992614475628e-05,
      "loss": 0.0004,
      "step": 3500
    },
    {
      "epoch": 1.2810098792535676,
      "grad_norm": 0.009295687079429626,
      "learning_rate": 1.814069423929099e-05,
      "loss": 0.0004,
      "step": 3501
    },
    {
      "epoch": 1.2813757775338455,
      "grad_norm": 0.1731029599905014,
      "learning_rate": 1.8131462333825703e-05,
      "loss": 0.002,
      "step": 3502
    },
    {
      "epoch": 1.2817416758141236,
      "grad_norm": 0.007807739544659853,
      "learning_rate": 1.8122230428360414e-05,
      "loss": 0.0004,
      "step": 3503
    },
    {
      "epoch": 1.2821075740944017,
      "grad_norm": 0.04685850813984871,
      "learning_rate": 1.8112998522895128e-05,
      "loss": 0.0005,
      "step": 3504
    },
    {
      "epoch": 1.2824734723746798,
      "grad_norm": 0.006781814154237509,
      "learning_rate": 1.810376661742984e-05,
      "loss": 0.0003,
      "step": 3505
    },
    {
      "epoch": 1.282839370654958,
      "grad_norm": 0.029277648776769638,
      "learning_rate": 1.809453471196455e-05,
      "loss": 0.0004,
      "step": 3506
    },
    {
      "epoch": 1.283205268935236,
      "grad_norm": 1.2168620824813843,
      "learning_rate": 1.8085302806499263e-05,
      "loss": 0.005,
      "step": 3507
    },
    {
      "epoch": 1.2835711672155141,
      "grad_norm": 0.05980167165398598,
      "learning_rate": 1.8076070901033973e-05,
      "loss": 0.0005,
      "step": 3508
    },
    {
      "epoch": 1.2839370654957922,
      "grad_norm": 0.009838027879595757,
      "learning_rate": 1.8066838995568684e-05,
      "loss": 0.0004,
      "step": 3509
    },
    {
      "epoch": 1.2843029637760703,
      "grad_norm": 0.013022365048527718,
      "learning_rate": 1.8057607090103398e-05,
      "loss": 0.0004,
      "step": 3510
    },
    {
      "epoch": 1.2846688620563484,
      "grad_norm": 0.008852077648043633,
      "learning_rate": 1.8048375184638112e-05,
      "loss": 0.0004,
      "step": 3511
    },
    {
      "epoch": 1.2850347603366263,
      "grad_norm": 0.008825463242828846,
      "learning_rate": 1.8039143279172822e-05,
      "loss": 0.0004,
      "step": 3512
    },
    {
      "epoch": 1.2854006586169044,
      "grad_norm": 0.00889803096652031,
      "learning_rate": 1.8029911373707533e-05,
      "loss": 0.0004,
      "step": 3513
    },
    {
      "epoch": 1.2857665568971826,
      "grad_norm": 0.006406840402632952,
      "learning_rate": 1.8020679468242247e-05,
      "loss": 0.0003,
      "step": 3514
    },
    {
      "epoch": 1.2861324551774607,
      "grad_norm": 0.1593201607465744,
      "learning_rate": 1.8011447562776957e-05,
      "loss": 0.0014,
      "step": 3515
    },
    {
      "epoch": 1.2864983534577388,
      "grad_norm": 0.008662647567689419,
      "learning_rate": 1.8002215657311668e-05,
      "loss": 0.0004,
      "step": 3516
    },
    {
      "epoch": 1.2868642517380169,
      "grad_norm": 0.007426423020660877,
      "learning_rate": 1.7992983751846382e-05,
      "loss": 0.0003,
      "step": 3517
    },
    {
      "epoch": 1.287230150018295,
      "grad_norm": 0.009404859505593777,
      "learning_rate": 1.7983751846381096e-05,
      "loss": 0.0004,
      "step": 3518
    },
    {
      "epoch": 1.2875960482985729,
      "grad_norm": 0.007359879557043314,
      "learning_rate": 1.7974519940915806e-05,
      "loss": 0.0003,
      "step": 3519
    },
    {
      "epoch": 1.287961946578851,
      "grad_norm": 0.006194309331476688,
      "learning_rate": 1.7965288035450517e-05,
      "loss": 0.0002,
      "step": 3520
    },
    {
      "epoch": 1.288327844859129,
      "grad_norm": 0.264801561832428,
      "learning_rate": 1.795605612998523e-05,
      "loss": 0.0011,
      "step": 3521
    },
    {
      "epoch": 1.2886937431394072,
      "grad_norm": 0.006784538272768259,
      "learning_rate": 1.794682422451994e-05,
      "loss": 0.0003,
      "step": 3522
    },
    {
      "epoch": 1.2890596414196853,
      "grad_norm": 0.008029811084270477,
      "learning_rate": 1.7937592319054652e-05,
      "loss": 0.0004,
      "step": 3523
    },
    {
      "epoch": 1.2894255396999634,
      "grad_norm": 0.008191043511033058,
      "learning_rate": 1.7928360413589366e-05,
      "loss": 0.0003,
      "step": 3524
    },
    {
      "epoch": 1.2897914379802415,
      "grad_norm": 0.08365490287542343,
      "learning_rate": 1.7919128508124077e-05,
      "loss": 0.0012,
      "step": 3525
    },
    {
      "epoch": 1.2901573362605196,
      "grad_norm": 0.007555261719971895,
      "learning_rate": 1.790989660265879e-05,
      "loss": 0.0003,
      "step": 3526
    },
    {
      "epoch": 1.2905232345407978,
      "grad_norm": 0.006324216723442078,
      "learning_rate": 1.79006646971935e-05,
      "loss": 0.0003,
      "step": 3527
    },
    {
      "epoch": 1.2908891328210759,
      "grad_norm": 0.03324977308511734,
      "learning_rate": 1.789143279172821e-05,
      "loss": 0.0006,
      "step": 3528
    },
    {
      "epoch": 1.2912550311013538,
      "grad_norm": 0.00550543749704957,
      "learning_rate": 1.7882200886262925e-05,
      "loss": 0.0002,
      "step": 3529
    },
    {
      "epoch": 1.2916209293816319,
      "grad_norm": 0.005932601168751717,
      "learning_rate": 1.787296898079764e-05,
      "loss": 0.0003,
      "step": 3530
    },
    {
      "epoch": 1.29198682766191,
      "grad_norm": 0.015467837452888489,
      "learning_rate": 1.786373707533235e-05,
      "loss": 0.0003,
      "step": 3531
    },
    {
      "epoch": 1.292352725942188,
      "grad_norm": 3.483289957046509,
      "learning_rate": 1.785450516986706e-05,
      "loss": 0.4658,
      "step": 3532
    },
    {
      "epoch": 1.2927186242224662,
      "grad_norm": 0.007057019975036383,
      "learning_rate": 1.7845273264401774e-05,
      "loss": 0.0003,
      "step": 3533
    },
    {
      "epoch": 1.2930845225027443,
      "grad_norm": 0.008183229714632034,
      "learning_rate": 1.7836041358936485e-05,
      "loss": 0.0004,
      "step": 3534
    },
    {
      "epoch": 1.2934504207830222,
      "grad_norm": 0.010247322730720043,
      "learning_rate": 1.7826809453471196e-05,
      "loss": 0.0004,
      "step": 3535
    },
    {
      "epoch": 1.2938163190633003,
      "grad_norm": 0.008473553694784641,
      "learning_rate": 1.781757754800591e-05,
      "loss": 0.0004,
      "step": 3536
    },
    {
      "epoch": 1.2941822173435784,
      "grad_norm": 0.006307257805019617,
      "learning_rate": 1.7808345642540623e-05,
      "loss": 0.0003,
      "step": 3537
    },
    {
      "epoch": 1.2945481156238565,
      "grad_norm": 0.006397646386176348,
      "learning_rate": 1.779911373707533e-05,
      "loss": 0.0003,
      "step": 3538
    },
    {
      "epoch": 1.2949140139041346,
      "grad_norm": 0.006925150752067566,
      "learning_rate": 1.7789881831610045e-05,
      "loss": 0.0003,
      "step": 3539
    },
    {
      "epoch": 1.2952799121844127,
      "grad_norm": 0.007193938363343477,
      "learning_rate": 1.778064992614476e-05,
      "loss": 0.0003,
      "step": 3540
    },
    {
      "epoch": 1.2956458104646909,
      "grad_norm": 0.3553125262260437,
      "learning_rate": 1.777141802067947e-05,
      "loss": 0.0021,
      "step": 3541
    },
    {
      "epoch": 1.296011708744969,
      "grad_norm": 0.006816533859819174,
      "learning_rate": 1.776218611521418e-05,
      "loss": 0.0003,
      "step": 3542
    },
    {
      "epoch": 1.296377607025247,
      "grad_norm": 0.009601273573935032,
      "learning_rate": 1.7752954209748894e-05,
      "loss": 0.0004,
      "step": 3543
    },
    {
      "epoch": 1.2967435053055252,
      "grad_norm": 0.006589301396161318,
      "learning_rate": 1.7743722304283604e-05,
      "loss": 0.0003,
      "step": 3544
    },
    {
      "epoch": 1.297109403585803,
      "grad_norm": 0.014678377658128738,
      "learning_rate": 1.7734490398818318e-05,
      "loss": 0.0006,
      "step": 3545
    },
    {
      "epoch": 1.2974753018660812,
      "grad_norm": 0.015555121004581451,
      "learning_rate": 1.772525849335303e-05,
      "loss": 0.0004,
      "step": 3546
    },
    {
      "epoch": 1.2978412001463593,
      "grad_norm": 0.007872229442000389,
      "learning_rate": 1.771602658788774e-05,
      "loss": 0.0004,
      "step": 3547
    },
    {
      "epoch": 1.2982070984266374,
      "grad_norm": 0.008108869194984436,
      "learning_rate": 1.7706794682422453e-05,
      "loss": 0.0004,
      "step": 3548
    },
    {
      "epoch": 1.2985729967069155,
      "grad_norm": 0.010157916694879532,
      "learning_rate": 1.7697562776957167e-05,
      "loss": 0.0004,
      "step": 3549
    },
    {
      "epoch": 1.2989388949871936,
      "grad_norm": 7.256031513214111,
      "learning_rate": 1.7688330871491878e-05,
      "loss": 0.1373,
      "step": 3550
    },
    {
      "epoch": 1.2993047932674717,
      "grad_norm": 0.009361966513097286,
      "learning_rate": 1.7679098966026588e-05,
      "loss": 0.0004,
      "step": 3551
    },
    {
      "epoch": 1.2996706915477496,
      "grad_norm": 0.007833853363990784,
      "learning_rate": 1.7669867060561302e-05,
      "loss": 0.0004,
      "step": 3552
    },
    {
      "epoch": 1.3000365898280277,
      "grad_norm": 0.007173149846494198,
      "learning_rate": 1.7660635155096013e-05,
      "loss": 0.0003,
      "step": 3553
    },
    {
      "epoch": 1.3004024881083058,
      "grad_norm": 0.009191369637846947,
      "learning_rate": 1.7651403249630723e-05,
      "loss": 0.0004,
      "step": 3554
    },
    {
      "epoch": 1.300768386388584,
      "grad_norm": 0.00806412659585476,
      "learning_rate": 1.7642171344165437e-05,
      "loss": 0.0004,
      "step": 3555
    },
    {
      "epoch": 1.301134284668862,
      "grad_norm": 0.00825346540659666,
      "learning_rate": 1.763293943870015e-05,
      "loss": 0.0003,
      "step": 3556
    },
    {
      "epoch": 1.3015001829491402,
      "grad_norm": 33.34310531616211,
      "learning_rate": 1.7623707533234858e-05,
      "loss": 0.2975,
      "step": 3557
    },
    {
      "epoch": 1.3018660812294183,
      "grad_norm": 0.019979335367679596,
      "learning_rate": 1.7614475627769572e-05,
      "loss": 0.0005,
      "step": 3558
    },
    {
      "epoch": 1.3022319795096964,
      "grad_norm": 0.011214091442525387,
      "learning_rate": 1.7605243722304286e-05,
      "loss": 0.0005,
      "step": 3559
    },
    {
      "epoch": 1.3025978777899745,
      "grad_norm": 0.011703052558004856,
      "learning_rate": 1.7596011816838997e-05,
      "loss": 0.0005,
      "step": 3560
    },
    {
      "epoch": 1.3029637760702526,
      "grad_norm": 0.24954891204833984,
      "learning_rate": 1.7586779911373707e-05,
      "loss": 0.0072,
      "step": 3561
    },
    {
      "epoch": 1.3033296743505305,
      "grad_norm": 5.27633810043335,
      "learning_rate": 1.757754800590842e-05,
      "loss": 0.4346,
      "step": 3562
    },
    {
      "epoch": 1.3036955726308086,
      "grad_norm": 0.04693859815597534,
      "learning_rate": 1.756831610044313e-05,
      "loss": 0.0005,
      "step": 3563
    },
    {
      "epoch": 1.3040614709110867,
      "grad_norm": 8.805634498596191,
      "learning_rate": 1.7559084194977846e-05,
      "loss": 0.2086,
      "step": 3564
    },
    {
      "epoch": 1.3044273691913648,
      "grad_norm": 38.169944763183594,
      "learning_rate": 1.7549852289512556e-05,
      "loss": 0.0662,
      "step": 3565
    },
    {
      "epoch": 1.304793267471643,
      "grad_norm": 0.006351001560688019,
      "learning_rate": 1.7540620384047267e-05,
      "loss": 0.0003,
      "step": 3566
    },
    {
      "epoch": 1.305159165751921,
      "grad_norm": 0.016460182145237923,
      "learning_rate": 1.753138847858198e-05,
      "loss": 0.0007,
      "step": 3567
    },
    {
      "epoch": 1.305525064032199,
      "grad_norm": 0.011846724897623062,
      "learning_rate": 1.752215657311669e-05,
      "loss": 0.0005,
      "step": 3568
    },
    {
      "epoch": 1.305890962312477,
      "grad_norm": 0.027364002540707588,
      "learning_rate": 1.7512924667651405e-05,
      "loss": 0.0007,
      "step": 3569
    },
    {
      "epoch": 1.3062568605927551,
      "grad_norm": 5.9839067459106445,
      "learning_rate": 1.7503692762186116e-05,
      "loss": 0.2985,
      "step": 3570
    },
    {
      "epoch": 1.3066227588730333,
      "grad_norm": 0.025131937116384506,
      "learning_rate": 1.749446085672083e-05,
      "loss": 0.0007,
      "step": 3571
    },
    {
      "epoch": 1.3069886571533114,
      "grad_norm": 0.03521532192826271,
      "learning_rate": 1.748522895125554e-05,
      "loss": 0.001,
      "step": 3572
    },
    {
      "epoch": 1.3073545554335895,
      "grad_norm": 0.026811633259058,
      "learning_rate": 1.747599704579025e-05,
      "loss": 0.001,
      "step": 3573
    },
    {
      "epoch": 1.3077204537138676,
      "grad_norm": 0.03177221491932869,
      "learning_rate": 1.7466765140324965e-05,
      "loss": 0.0009,
      "step": 3574
    },
    {
      "epoch": 1.3080863519941457,
      "grad_norm": 0.06493864953517914,
      "learning_rate": 1.745753323485968e-05,
      "loss": 0.0011,
      "step": 3575
    },
    {
      "epoch": 1.3084522502744238,
      "grad_norm": 0.03323405236005783,
      "learning_rate": 1.7448301329394386e-05,
      "loss": 0.0009,
      "step": 3576
    },
    {
      "epoch": 1.308818148554702,
      "grad_norm": 0.02096899226307869,
      "learning_rate": 1.74390694239291e-05,
      "loss": 0.0007,
      "step": 3577
    },
    {
      "epoch": 1.3091840468349798,
      "grad_norm": 0.02789313532412052,
      "learning_rate": 1.7429837518463814e-05,
      "loss": 0.0008,
      "step": 3578
    },
    {
      "epoch": 1.309549945115258,
      "grad_norm": 0.013829246163368225,
      "learning_rate": 1.7420605612998524e-05,
      "loss": 0.0005,
      "step": 3579
    },
    {
      "epoch": 1.309915843395536,
      "grad_norm": 0.015086740255355835,
      "learning_rate": 1.7411373707533235e-05,
      "loss": 0.0006,
      "step": 3580
    },
    {
      "epoch": 1.3102817416758141,
      "grad_norm": 0.018077921122312546,
      "learning_rate": 1.740214180206795e-05,
      "loss": 0.0007,
      "step": 3581
    },
    {
      "epoch": 1.3106476399560922,
      "grad_norm": 5.738001823425293,
      "learning_rate": 1.739290989660266e-05,
      "loss": 0.1619,
      "step": 3582
    },
    {
      "epoch": 1.3110135382363703,
      "grad_norm": 0.02069113776087761,
      "learning_rate": 1.738367799113737e-05,
      "loss": 0.0007,
      "step": 3583
    },
    {
      "epoch": 1.3113794365166482,
      "grad_norm": 0.009018084034323692,
      "learning_rate": 1.7374446085672084e-05,
      "loss": 0.0004,
      "step": 3584
    },
    {
      "epoch": 1.3117453347969263,
      "grad_norm": 0.01895199716091156,
      "learning_rate": 1.7365214180206794e-05,
      "loss": 0.0006,
      "step": 3585
    },
    {
      "epoch": 1.3121112330772045,
      "grad_norm": 9.471710205078125,
      "learning_rate": 1.7355982274741508e-05,
      "loss": 0.1989,
      "step": 3586
    },
    {
      "epoch": 1.3124771313574826,
      "grad_norm": 0.010507349856197834,
      "learning_rate": 1.734675036927622e-05,
      "loss": 0.0004,
      "step": 3587
    },
    {
      "epoch": 1.3128430296377607,
      "grad_norm": 0.04389691725373268,
      "learning_rate": 1.7337518463810933e-05,
      "loss": 0.0008,
      "step": 3588
    },
    {
      "epoch": 1.3132089279180388,
      "grad_norm": 0.06791482120752335,
      "learning_rate": 1.7328286558345643e-05,
      "loss": 0.0009,
      "step": 3589
    },
    {
      "epoch": 1.313574826198317,
      "grad_norm": 0.01296323724091053,
      "learning_rate": 1.7319054652880357e-05,
      "loss": 0.0005,
      "step": 3590
    },
    {
      "epoch": 1.313940724478595,
      "grad_norm": 0.016118459403514862,
      "learning_rate": 1.7309822747415068e-05,
      "loss": 0.0005,
      "step": 3591
    },
    {
      "epoch": 1.314306622758873,
      "grad_norm": 0.021454671397805214,
      "learning_rate": 1.7300590841949778e-05,
      "loss": 0.0006,
      "step": 3592
    },
    {
      "epoch": 1.3146725210391512,
      "grad_norm": 0.41365957260131836,
      "learning_rate": 1.7291358936484492e-05,
      "loss": 0.0079,
      "step": 3593
    },
    {
      "epoch": 1.3150384193194293,
      "grad_norm": 0.009138185530900955,
      "learning_rate": 1.7282127031019203e-05,
      "loss": 0.0004,
      "step": 3594
    },
    {
      "epoch": 1.3154043175997072,
      "grad_norm": 0.22918668389320374,
      "learning_rate": 1.7272895125553913e-05,
      "loss": 0.002,
      "step": 3595
    },
    {
      "epoch": 1.3157702158799853,
      "grad_norm": 0.05030486360192299,
      "learning_rate": 1.7263663220088627e-05,
      "loss": 0.0011,
      "step": 3596
    },
    {
      "epoch": 1.3161361141602634,
      "grad_norm": 0.02172028087079525,
      "learning_rate": 1.725443131462334e-05,
      "loss": 0.0007,
      "step": 3597
    },
    {
      "epoch": 1.3165020124405415,
      "grad_norm": 0.05011909082531929,
      "learning_rate": 1.7245199409158048e-05,
      "loss": 0.0008,
      "step": 3598
    },
    {
      "epoch": 1.3168679107208197,
      "grad_norm": 0.049771856516599655,
      "learning_rate": 1.7235967503692762e-05,
      "loss": 0.0008,
      "step": 3599
    },
    {
      "epoch": 1.3172338090010978,
      "grad_norm": 0.006991399452090263,
      "learning_rate": 1.7226735598227476e-05,
      "loss": 0.0003,
      "step": 3600
    },
    {
      "epoch": 1.3175997072813757,
      "grad_norm": 0.022415094077587128,
      "learning_rate": 1.7217503692762187e-05,
      "loss": 0.0006,
      "step": 3601
    },
    {
      "epoch": 1.3179656055616538,
      "grad_norm": 0.02169918082654476,
      "learning_rate": 1.7208271787296897e-05,
      "loss": 0.0005,
      "step": 3602
    },
    {
      "epoch": 1.3183315038419319,
      "grad_norm": 1.6611833572387695,
      "learning_rate": 1.719903988183161e-05,
      "loss": 0.0113,
      "step": 3603
    },
    {
      "epoch": 1.31869740212221,
      "grad_norm": 0.005926684942096472,
      "learning_rate": 1.7189807976366322e-05,
      "loss": 0.0003,
      "step": 3604
    },
    {
      "epoch": 1.319063300402488,
      "grad_norm": 0.0046302094124257565,
      "learning_rate": 1.7180576070901036e-05,
      "loss": 0.0002,
      "step": 3605
    },
    {
      "epoch": 1.3194291986827662,
      "grad_norm": 17.48542594909668,
      "learning_rate": 1.7171344165435746e-05,
      "loss": 0.6043,
      "step": 3606
    },
    {
      "epoch": 1.3197950969630443,
      "grad_norm": 0.006609418895095587,
      "learning_rate": 1.716211225997046e-05,
      "loss": 0.0003,
      "step": 3607
    },
    {
      "epoch": 1.3201609952433224,
      "grad_norm": 0.016508780419826508,
      "learning_rate": 1.715288035450517e-05,
      "loss": 0.0005,
      "step": 3608
    },
    {
      "epoch": 1.3205268935236005,
      "grad_norm": 0.007513934746384621,
      "learning_rate": 1.714364844903988e-05,
      "loss": 0.0003,
      "step": 3609
    },
    {
      "epoch": 1.3208927918038786,
      "grad_norm": 0.014295338653028011,
      "learning_rate": 1.7134416543574595e-05,
      "loss": 0.0004,
      "step": 3610
    },
    {
      "epoch": 1.3212586900841565,
      "grad_norm": 0.5484269857406616,
      "learning_rate": 1.7125184638109306e-05,
      "loss": 0.0054,
      "step": 3611
    },
    {
      "epoch": 1.3216245883644346,
      "grad_norm": 0.0635870099067688,
      "learning_rate": 1.711595273264402e-05,
      "loss": 0.0006,
      "step": 3612
    },
    {
      "epoch": 1.3219904866447127,
      "grad_norm": 0.013677765615284443,
      "learning_rate": 1.710672082717873e-05,
      "loss": 0.0003,
      "step": 3613
    },
    {
      "epoch": 1.3223563849249909,
      "grad_norm": 0.7513426542282104,
      "learning_rate": 1.709748892171344e-05,
      "loss": 0.0083,
      "step": 3614
    },
    {
      "epoch": 1.322722283205269,
      "grad_norm": 0.006864702794700861,
      "learning_rate": 1.7088257016248155e-05,
      "loss": 0.0003,
      "step": 3615
    },
    {
      "epoch": 1.323088181485547,
      "grad_norm": 0.01322631910443306,
      "learning_rate": 1.707902511078287e-05,
      "loss": 0.0005,
      "step": 3616
    },
    {
      "epoch": 1.323454079765825,
      "grad_norm": 0.005235433112829924,
      "learning_rate": 1.7069793205317576e-05,
      "loss": 0.0002,
      "step": 3617
    },
    {
      "epoch": 1.323819978046103,
      "grad_norm": 0.06602129340171814,
      "learning_rate": 1.706056129985229e-05,
      "loss": 0.0006,
      "step": 3618
    },
    {
      "epoch": 1.3241858763263812,
      "grad_norm": 0.033843375742435455,
      "learning_rate": 1.7051329394387004e-05,
      "loss": 0.0005,
      "step": 3619
    },
    {
      "epoch": 1.3245517746066593,
      "grad_norm": 0.004936803597956896,
      "learning_rate": 1.7042097488921714e-05,
      "loss": 0.0002,
      "step": 3620
    },
    {
      "epoch": 1.3249176728869374,
      "grad_norm": 0.005549116525799036,
      "learning_rate": 1.7032865583456425e-05,
      "loss": 0.0002,
      "step": 3621
    },
    {
      "epoch": 1.3252835711672155,
      "grad_norm": 0.014956871047616005,
      "learning_rate": 1.702363367799114e-05,
      "loss": 0.0005,
      "step": 3622
    },
    {
      "epoch": 1.3256494694474936,
      "grad_norm": 0.2262580692768097,
      "learning_rate": 1.701440177252585e-05,
      "loss": 0.0034,
      "step": 3623
    },
    {
      "epoch": 1.3260153677277717,
      "grad_norm": 0.005703535862267017,
      "learning_rate": 1.700516986706056e-05,
      "loss": 0.0003,
      "step": 3624
    },
    {
      "epoch": 1.3263812660080498,
      "grad_norm": 0.006644431035965681,
      "learning_rate": 1.6995937961595274e-05,
      "loss": 0.0002,
      "step": 3625
    },
    {
      "epoch": 1.326747164288328,
      "grad_norm": 0.00726970424875617,
      "learning_rate": 1.6986706056129988e-05,
      "loss": 0.0003,
      "step": 3626
    },
    {
      "epoch": 1.327113062568606,
      "grad_norm": 4.50407075881958,
      "learning_rate": 1.6977474150664698e-05,
      "loss": 0.1881,
      "step": 3627
    },
    {
      "epoch": 1.327478960848884,
      "grad_norm": 0.005735277198255062,
      "learning_rate": 1.696824224519941e-05,
      "loss": 0.0003,
      "step": 3628
    },
    {
      "epoch": 1.327844859129162,
      "grad_norm": 0.018872441723942757,
      "learning_rate": 1.6959010339734123e-05,
      "loss": 0.0005,
      "step": 3629
    },
    {
      "epoch": 1.3282107574094402,
      "grad_norm": 0.006840793881565332,
      "learning_rate": 1.6949778434268833e-05,
      "loss": 0.0003,
      "step": 3630
    },
    {
      "epoch": 1.3285766556897183,
      "grad_norm": 0.006299990229308605,
      "learning_rate": 1.6940546528803547e-05,
      "loss": 0.0003,
      "step": 3631
    },
    {
      "epoch": 1.3289425539699964,
      "grad_norm": 0.05048524960875511,
      "learning_rate": 1.6931314623338258e-05,
      "loss": 0.0005,
      "step": 3632
    },
    {
      "epoch": 1.3293084522502745,
      "grad_norm": 0.013322624377906322,
      "learning_rate": 1.6922082717872968e-05,
      "loss": 0.0003,
      "step": 3633
    },
    {
      "epoch": 1.3296743505305524,
      "grad_norm": 0.008361618034541607,
      "learning_rate": 1.6912850812407682e-05,
      "loss": 0.0003,
      "step": 3634
    },
    {
      "epoch": 1.3300402488108305,
      "grad_norm": 0.12233593314886093,
      "learning_rate": 1.6903618906942396e-05,
      "loss": 0.0013,
      "step": 3635
    },
    {
      "epoch": 1.3304061470911086,
      "grad_norm": 0.007603877689689398,
      "learning_rate": 1.6894387001477103e-05,
      "loss": 0.0003,
      "step": 3636
    },
    {
      "epoch": 1.3307720453713867,
      "grad_norm": 0.036120228469371796,
      "learning_rate": 1.6885155096011817e-05,
      "loss": 0.0006,
      "step": 3637
    },
    {
      "epoch": 1.3311379436516648,
      "grad_norm": 0.0072745224460959435,
      "learning_rate": 1.687592319054653e-05,
      "loss": 0.0003,
      "step": 3638
    },
    {
      "epoch": 1.331503841931943,
      "grad_norm": 0.03076029196381569,
      "learning_rate": 1.6866691285081242e-05,
      "loss": 0.0005,
      "step": 3639
    },
    {
      "epoch": 1.331869740212221,
      "grad_norm": 0.013098797760903835,
      "learning_rate": 1.6857459379615952e-05,
      "loss": 0.0003,
      "step": 3640
    },
    {
      "epoch": 1.3322356384924992,
      "grad_norm": 0.3608027696609497,
      "learning_rate": 1.6848227474150666e-05,
      "loss": 0.002,
      "step": 3641
    },
    {
      "epoch": 1.3326015367727773,
      "grad_norm": 0.028674524277448654,
      "learning_rate": 1.6838995568685377e-05,
      "loss": 0.0005,
      "step": 3642
    },
    {
      "epoch": 1.3329674350530554,
      "grad_norm": 0.007417668122798204,
      "learning_rate": 1.6829763663220087e-05,
      "loss": 0.0003,
      "step": 3643
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.00747141195461154,
      "learning_rate": 1.68205317577548e-05,
      "loss": 0.0003,
      "step": 3644
    },
    {
      "epoch": 1.3336992316136114,
      "grad_norm": 0.17886972427368164,
      "learning_rate": 1.6811299852289515e-05,
      "loss": 0.0036,
      "step": 3645
    },
    {
      "epoch": 1.3340651298938895,
      "grad_norm": 0.11214978992938995,
      "learning_rate": 1.6802067946824226e-05,
      "loss": 0.0011,
      "step": 3646
    },
    {
      "epoch": 1.3344310281741676,
      "grad_norm": 0.005222878884524107,
      "learning_rate": 1.6792836041358936e-05,
      "loss": 0.0002,
      "step": 3647
    },
    {
      "epoch": 1.3347969264544457,
      "grad_norm": 25.92600440979004,
      "learning_rate": 1.678360413589365e-05,
      "loss": 0.0292,
      "step": 3648
    },
    {
      "epoch": 1.3351628247347238,
      "grad_norm": 0.0042549255304038525,
      "learning_rate": 1.677437223042836e-05,
      "loss": 0.0002,
      "step": 3649
    },
    {
      "epoch": 1.3355287230150017,
      "grad_norm": 0.009412266314029694,
      "learning_rate": 1.6765140324963075e-05,
      "loss": 0.0003,
      "step": 3650
    },
    {
      "epoch": 1.3358946212952798,
      "grad_norm": 0.005509866867214441,
      "learning_rate": 1.6755908419497785e-05,
      "loss": 0.0002,
      "step": 3651
    },
    {
      "epoch": 1.336260519575558,
      "grad_norm": 0.005265035666525364,
      "learning_rate": 1.6746676514032496e-05,
      "loss": 0.0002,
      "step": 3652
    },
    {
      "epoch": 1.336626417855836,
      "grad_norm": 0.25969669222831726,
      "learning_rate": 1.673744460856721e-05,
      "loss": 0.0032,
      "step": 3653
    },
    {
      "epoch": 1.3369923161361141,
      "grad_norm": 0.007728701923042536,
      "learning_rate": 1.672821270310192e-05,
      "loss": 0.0003,
      "step": 3654
    },
    {
      "epoch": 1.3373582144163922,
      "grad_norm": 0.007709487807005644,
      "learning_rate": 1.671898079763663e-05,
      "loss": 0.0003,
      "step": 3655
    },
    {
      "epoch": 1.3377241126966704,
      "grad_norm": 10.964926719665527,
      "learning_rate": 1.6709748892171345e-05,
      "loss": 0.2677,
      "step": 3656
    },
    {
      "epoch": 1.3380900109769485,
      "grad_norm": 0.00506269047036767,
      "learning_rate": 1.670051698670606e-05,
      "loss": 0.0002,
      "step": 3657
    },
    {
      "epoch": 1.3384559092572266,
      "grad_norm": 0.13231152296066284,
      "learning_rate": 1.669128508124077e-05,
      "loss": 0.0017,
      "step": 3658
    },
    {
      "epoch": 1.3388218075375047,
      "grad_norm": 0.007170175202190876,
      "learning_rate": 1.668205317577548e-05,
      "loss": 0.0003,
      "step": 3659
    },
    {
      "epoch": 1.3391877058177828,
      "grad_norm": 0.237347811460495,
      "learning_rate": 1.6672821270310194e-05,
      "loss": 0.0023,
      "step": 3660
    },
    {
      "epoch": 1.3395536040980607,
      "grad_norm": 0.006403347011655569,
      "learning_rate": 1.6663589364844904e-05,
      "loss": 0.0003,
      "step": 3661
    },
    {
      "epoch": 1.3399195023783388,
      "grad_norm": 0.006289857905358076,
      "learning_rate": 1.6654357459379615e-05,
      "loss": 0.0003,
      "step": 3662
    },
    {
      "epoch": 1.340285400658617,
      "grad_norm": 0.009823044762015343,
      "learning_rate": 1.664512555391433e-05,
      "loss": 0.0003,
      "step": 3663
    },
    {
      "epoch": 1.340651298938895,
      "grad_norm": 0.06387443840503693,
      "learning_rate": 1.6635893648449043e-05,
      "loss": 0.0006,
      "step": 3664
    },
    {
      "epoch": 1.3410171972191731,
      "grad_norm": 0.0037189407739788294,
      "learning_rate": 1.6626661742983753e-05,
      "loss": 0.0002,
      "step": 3665
    },
    {
      "epoch": 1.3413830954994512,
      "grad_norm": 0.09408076107501984,
      "learning_rate": 1.6617429837518464e-05,
      "loss": 0.0007,
      "step": 3666
    },
    {
      "epoch": 1.3417489937797291,
      "grad_norm": 6.68268346786499,
      "learning_rate": 1.6608197932053178e-05,
      "loss": 0.1621,
      "step": 3667
    },
    {
      "epoch": 1.3421148920600072,
      "grad_norm": 0.0050530992448329926,
      "learning_rate": 1.659896602658789e-05,
      "loss": 0.0002,
      "step": 3668
    },
    {
      "epoch": 1.3424807903402853,
      "grad_norm": 0.004666874185204506,
      "learning_rate": 1.65897341211226e-05,
      "loss": 0.0002,
      "step": 3669
    },
    {
      "epoch": 1.3428466886205634,
      "grad_norm": 0.00456647016108036,
      "learning_rate": 1.6580502215657313e-05,
      "loss": 0.0002,
      "step": 3670
    },
    {
      "epoch": 1.3432125869008416,
      "grad_norm": 0.007313751615583897,
      "learning_rate": 1.6571270310192023e-05,
      "loss": 0.0003,
      "step": 3671
    },
    {
      "epoch": 1.3435784851811197,
      "grad_norm": 0.15699104964733124,
      "learning_rate": 1.6562038404726737e-05,
      "loss": 0.0036,
      "step": 3672
    },
    {
      "epoch": 1.3439443834613978,
      "grad_norm": 0.005014872644096613,
      "learning_rate": 1.6552806499261448e-05,
      "loss": 0.0002,
      "step": 3673
    },
    {
      "epoch": 1.3443102817416759,
      "grad_norm": 0.003975377418100834,
      "learning_rate": 1.654357459379616e-05,
      "loss": 0.0002,
      "step": 3674
    },
    {
      "epoch": 1.344676180021954,
      "grad_norm": 0.03313998878002167,
      "learning_rate": 1.6534342688330872e-05,
      "loss": 0.0004,
      "step": 3675
    },
    {
      "epoch": 1.345042078302232,
      "grad_norm": 0.1966564953327179,
      "learning_rate": 1.6525110782865586e-05,
      "loss": 0.0018,
      "step": 3676
    },
    {
      "epoch": 1.34540797658251,
      "grad_norm": 0.021360328420996666,
      "learning_rate": 1.6515878877400297e-05,
      "loss": 0.0003,
      "step": 3677
    },
    {
      "epoch": 1.345773874862788,
      "grad_norm": 0.13655181229114532,
      "learning_rate": 1.6506646971935007e-05,
      "loss": 0.0009,
      "step": 3678
    },
    {
      "epoch": 1.3461397731430662,
      "grad_norm": 0.05982007086277008,
      "learning_rate": 1.649741506646972e-05,
      "loss": 0.0005,
      "step": 3679
    },
    {
      "epoch": 1.3465056714233443,
      "grad_norm": 0.0058397697284817696,
      "learning_rate": 1.6488183161004432e-05,
      "loss": 0.0002,
      "step": 3680
    },
    {
      "epoch": 1.3468715697036224,
      "grad_norm": 0.006599153857678175,
      "learning_rate": 1.6478951255539142e-05,
      "loss": 0.0003,
      "step": 3681
    },
    {
      "epoch": 1.3472374679839005,
      "grad_norm": 0.008071835152804852,
      "learning_rate": 1.6469719350073856e-05,
      "loss": 0.0002,
      "step": 3682
    },
    {
      "epoch": 1.3476033662641784,
      "grad_norm": 0.01907734014093876,
      "learning_rate": 1.646048744460857e-05,
      "loss": 0.0003,
      "step": 3683
    },
    {
      "epoch": 1.3479692645444565,
      "grad_norm": 0.006706871557980776,
      "learning_rate": 1.6451255539143277e-05,
      "loss": 0.0002,
      "step": 3684
    },
    {
      "epoch": 1.3483351628247346,
      "grad_norm": 0.015992341563105583,
      "learning_rate": 1.644202363367799e-05,
      "loss": 0.0003,
      "step": 3685
    },
    {
      "epoch": 1.3487010611050128,
      "grad_norm": 0.014316271059215069,
      "learning_rate": 1.6432791728212705e-05,
      "loss": 0.0003,
      "step": 3686
    },
    {
      "epoch": 1.3490669593852909,
      "grad_norm": 0.060182683169841766,
      "learning_rate": 1.6423559822747416e-05,
      "loss": 0.0005,
      "step": 3687
    },
    {
      "epoch": 1.349432857665569,
      "grad_norm": 0.00698451604694128,
      "learning_rate": 1.6414327917282126e-05,
      "loss": 0.0002,
      "step": 3688
    },
    {
      "epoch": 1.349798755945847,
      "grad_norm": 0.007220246363431215,
      "learning_rate": 1.640509601181684e-05,
      "loss": 0.0003,
      "step": 3689
    },
    {
      "epoch": 1.3501646542261252,
      "grad_norm": 0.005065953824669123,
      "learning_rate": 1.639586410635155e-05,
      "loss": 0.0002,
      "step": 3690
    },
    {
      "epoch": 1.3505305525064033,
      "grad_norm": 0.011637505143880844,
      "learning_rate": 1.6386632200886265e-05,
      "loss": 0.0003,
      "step": 3691
    },
    {
      "epoch": 1.3508964507866814,
      "grad_norm": 2.670304775238037,
      "learning_rate": 1.6377400295420975e-05,
      "loss": 0.0088,
      "step": 3692
    },
    {
      "epoch": 1.3512623490669595,
      "grad_norm": 0.029218371957540512,
      "learning_rate": 1.6368168389955686e-05,
      "loss": 0.0004,
      "step": 3693
    },
    {
      "epoch": 1.3516282473472374,
      "grad_norm": 0.006115978583693504,
      "learning_rate": 1.63589364844904e-05,
      "loss": 0.0003,
      "step": 3694
    },
    {
      "epoch": 1.3519941456275155,
      "grad_norm": 0.004188124556094408,
      "learning_rate": 1.634970457902511e-05,
      "loss": 0.0002,
      "step": 3695
    },
    {
      "epoch": 1.3523600439077936,
      "grad_norm": 0.05505220219492912,
      "learning_rate": 1.6340472673559824e-05,
      "loss": 0.0005,
      "step": 3696
    },
    {
      "epoch": 1.3527259421880717,
      "grad_norm": 0.16120000183582306,
      "learning_rate": 1.6331240768094535e-05,
      "loss": 0.0012,
      "step": 3697
    },
    {
      "epoch": 1.3530918404683498,
      "grad_norm": 0.028079133480787277,
      "learning_rate": 1.632200886262925e-05,
      "loss": 0.0003,
      "step": 3698
    },
    {
      "epoch": 1.353457738748628,
      "grad_norm": 0.06114259734749794,
      "learning_rate": 1.631277695716396e-05,
      "loss": 0.0002,
      "step": 3699
    },
    {
      "epoch": 1.3538236370289058,
      "grad_norm": 0.0057572126388549805,
      "learning_rate": 1.630354505169867e-05,
      "loss": 0.0003,
      "step": 3700
    },
    {
      "epoch": 1.354189535309184,
      "grad_norm": 0.34030401706695557,
      "learning_rate": 1.6294313146233384e-05,
      "loss": 0.0019,
      "step": 3701
    },
    {
      "epoch": 1.354555433589462,
      "grad_norm": 0.004957401659339666,
      "learning_rate": 1.6285081240768098e-05,
      "loss": 0.0002,
      "step": 3702
    },
    {
      "epoch": 1.3549213318697402,
      "grad_norm": 0.005580356810241938,
      "learning_rate": 1.6275849335302805e-05,
      "loss": 0.0002,
      "step": 3703
    },
    {
      "epoch": 1.3552872301500183,
      "grad_norm": 0.030897844582796097,
      "learning_rate": 1.626661742983752e-05,
      "loss": 0.0005,
      "step": 3704
    },
    {
      "epoch": 1.3556531284302964,
      "grad_norm": 0.49688583612442017,
      "learning_rate": 1.6257385524372233e-05,
      "loss": 0.0024,
      "step": 3705
    },
    {
      "epoch": 1.3560190267105745,
      "grad_norm": 0.017789160832762718,
      "learning_rate": 1.6248153618906943e-05,
      "loss": 0.0003,
      "step": 3706
    },
    {
      "epoch": 1.3563849249908526,
      "grad_norm": 0.004918659571558237,
      "learning_rate": 1.6238921713441654e-05,
      "loss": 0.0002,
      "step": 3707
    },
    {
      "epoch": 1.3567508232711307,
      "grad_norm": 0.00450843246653676,
      "learning_rate": 1.6229689807976368e-05,
      "loss": 0.0002,
      "step": 3708
    },
    {
      "epoch": 1.3571167215514088,
      "grad_norm": 13.423137664794922,
      "learning_rate": 1.622045790251108e-05,
      "loss": 0.0376,
      "step": 3709
    },
    {
      "epoch": 1.3574826198316867,
      "grad_norm": 0.042357493191957474,
      "learning_rate": 1.621122599704579e-05,
      "loss": 0.0004,
      "step": 3710
    },
    {
      "epoch": 1.3578485181119648,
      "grad_norm": 0.004952091258019209,
      "learning_rate": 1.6201994091580503e-05,
      "loss": 0.0002,
      "step": 3711
    },
    {
      "epoch": 1.358214416392243,
      "grad_norm": 0.005087613593786955,
      "learning_rate": 1.6192762186115213e-05,
      "loss": 0.0002,
      "step": 3712
    },
    {
      "epoch": 1.358580314672521,
      "grad_norm": 0.003379714209586382,
      "learning_rate": 1.6183530280649927e-05,
      "loss": 0.0002,
      "step": 3713
    },
    {
      "epoch": 1.3589462129527992,
      "grad_norm": 0.02007175236940384,
      "learning_rate": 1.6174298375184638e-05,
      "loss": 0.0002,
      "step": 3714
    },
    {
      "epoch": 1.3593121112330773,
      "grad_norm": 2.2341105937957764,
      "learning_rate": 1.6165066469719352e-05,
      "loss": 0.5687,
      "step": 3715
    },
    {
      "epoch": 1.3596780095133552,
      "grad_norm": 0.12571851909160614,
      "learning_rate": 1.6155834564254062e-05,
      "loss": 0.0006,
      "step": 3716
    },
    {
      "epoch": 1.3600439077936333,
      "grad_norm": 22.689409255981445,
      "learning_rate": 1.6146602658788776e-05,
      "loss": 0.2692,
      "step": 3717
    },
    {
      "epoch": 1.3604098060739114,
      "grad_norm": 0.07616471499204636,
      "learning_rate": 1.6137370753323487e-05,
      "loss": 0.0008,
      "step": 3718
    },
    {
      "epoch": 1.3607757043541895,
      "grad_norm": 0.8473544120788574,
      "learning_rate": 1.6128138847858197e-05,
      "loss": 0.0029,
      "step": 3719
    },
    {
      "epoch": 1.3611416026344676,
      "grad_norm": 0.03126285597681999,
      "learning_rate": 1.611890694239291e-05,
      "loss": 0.0004,
      "step": 3720
    },
    {
      "epoch": 1.3615075009147457,
      "grad_norm": 0.010552596300840378,
      "learning_rate": 1.6109675036927625e-05,
      "loss": 0.0003,
      "step": 3721
    },
    {
      "epoch": 1.3618733991950238,
      "grad_norm": 0.0051748971454799175,
      "learning_rate": 1.6100443131462333e-05,
      "loss": 0.0002,
      "step": 3722
    },
    {
      "epoch": 1.362239297475302,
      "grad_norm": 0.01210105326026678,
      "learning_rate": 1.6091211225997046e-05,
      "loss": 0.0003,
      "step": 3723
    },
    {
      "epoch": 1.36260519575558,
      "grad_norm": 0.1748838871717453,
      "learning_rate": 1.608197932053176e-05,
      "loss": 0.0017,
      "step": 3724
    },
    {
      "epoch": 1.3629710940358581,
      "grad_norm": 0.0053984057158231735,
      "learning_rate": 1.6072747415066468e-05,
      "loss": 0.0003,
      "step": 3725
    },
    {
      "epoch": 1.363336992316136,
      "grad_norm": 0.006555323489010334,
      "learning_rate": 1.606351550960118e-05,
      "loss": 0.0003,
      "step": 3726
    },
    {
      "epoch": 1.3637028905964141,
      "grad_norm": 0.007885373197495937,
      "learning_rate": 1.6054283604135895e-05,
      "loss": 0.0003,
      "step": 3727
    },
    {
      "epoch": 1.3640687888766923,
      "grad_norm": 0.005410488694906235,
      "learning_rate": 1.6045051698670606e-05,
      "loss": 0.0003,
      "step": 3728
    },
    {
      "epoch": 1.3644346871569704,
      "grad_norm": 0.007302268873900175,
      "learning_rate": 1.6035819793205317e-05,
      "loss": 0.0003,
      "step": 3729
    },
    {
      "epoch": 1.3648005854372485,
      "grad_norm": 0.007070745341479778,
      "learning_rate": 1.602658788774003e-05,
      "loss": 0.0003,
      "step": 3730
    },
    {
      "epoch": 1.3651664837175266,
      "grad_norm": 0.0074715279042720795,
      "learning_rate": 1.601735598227474e-05,
      "loss": 0.0004,
      "step": 3731
    },
    {
      "epoch": 1.3655323819978047,
      "grad_norm": 0.12118847668170929,
      "learning_rate": 1.6008124076809455e-05,
      "loss": 0.0008,
      "step": 3732
    },
    {
      "epoch": 1.3658982802780826,
      "grad_norm": 0.006275932304561138,
      "learning_rate": 1.5998892171344165e-05,
      "loss": 0.0003,
      "step": 3733
    },
    {
      "epoch": 1.3662641785583607,
      "grad_norm": 0.009224691428244114,
      "learning_rate": 1.598966026587888e-05,
      "loss": 0.0004,
      "step": 3734
    },
    {
      "epoch": 1.3666300768386388,
      "grad_norm": 0.26964980363845825,
      "learning_rate": 1.598042836041359e-05,
      "loss": 0.0021,
      "step": 3735
    },
    {
      "epoch": 1.366995975118917,
      "grad_norm": 0.009065292775630951,
      "learning_rate": 1.5971196454948304e-05,
      "loss": 0.0003,
      "step": 3736
    },
    {
      "epoch": 1.367361873399195,
      "grad_norm": 0.005955661181360483,
      "learning_rate": 1.5961964549483014e-05,
      "loss": 0.0003,
      "step": 3737
    },
    {
      "epoch": 1.3677277716794731,
      "grad_norm": 0.17529402673244476,
      "learning_rate": 1.5952732644017725e-05,
      "loss": 0.0011,
      "step": 3738
    },
    {
      "epoch": 1.3680936699597512,
      "grad_norm": 0.019971614703536034,
      "learning_rate": 1.594350073855244e-05,
      "loss": 0.0006,
      "step": 3739
    },
    {
      "epoch": 1.3684595682400293,
      "grad_norm": 0.007587276864796877,
      "learning_rate": 1.593426883308715e-05,
      "loss": 0.0004,
      "step": 3740
    },
    {
      "epoch": 1.3688254665203075,
      "grad_norm": 0.028377192094922066,
      "learning_rate": 1.592503692762186e-05,
      "loss": 0.0005,
      "step": 3741
    },
    {
      "epoch": 1.3691913648005856,
      "grad_norm": 0.007643676828593016,
      "learning_rate": 1.5915805022156574e-05,
      "loss": 0.0003,
      "step": 3742
    },
    {
      "epoch": 1.3695572630808635,
      "grad_norm": 0.0070914821699261665,
      "learning_rate": 1.5906573116691288e-05,
      "loss": 0.0003,
      "step": 3743
    },
    {
      "epoch": 1.3699231613611416,
      "grad_norm": 0.009186800569295883,
      "learning_rate": 1.5897341211225995e-05,
      "loss": 0.0003,
      "step": 3744
    },
    {
      "epoch": 1.3702890596414197,
      "grad_norm": 8.197916030883789,
      "learning_rate": 1.588810930576071e-05,
      "loss": 0.0182,
      "step": 3745
    },
    {
      "epoch": 1.3706549579216978,
      "grad_norm": 0.007637581788003445,
      "learning_rate": 1.5878877400295423e-05,
      "loss": 0.0003,
      "step": 3746
    },
    {
      "epoch": 1.371020856201976,
      "grad_norm": 0.17423586547374725,
      "learning_rate": 1.5869645494830134e-05,
      "loss": 0.0011,
      "step": 3747
    },
    {
      "epoch": 1.371386754482254,
      "grad_norm": 0.007397469598799944,
      "learning_rate": 1.5860413589364844e-05,
      "loss": 0.0004,
      "step": 3748
    },
    {
      "epoch": 1.3717526527625319,
      "grad_norm": 0.008932887576520443,
      "learning_rate": 1.5851181683899558e-05,
      "loss": 0.0003,
      "step": 3749
    },
    {
      "epoch": 1.37211855104281,
      "grad_norm": 0.7936921715736389,
      "learning_rate": 1.584194977843427e-05,
      "loss": 0.0008,
      "step": 3750
    },
    {
      "epoch": 1.372484449323088,
      "grad_norm": 0.005778637249022722,
      "learning_rate": 1.5832717872968982e-05,
      "loss": 0.0003,
      "step": 3751
    },
    {
      "epoch": 1.3728503476033662,
      "grad_norm": 0.007308226078748703,
      "learning_rate": 1.5823485967503693e-05,
      "loss": 0.0003,
      "step": 3752
    },
    {
      "epoch": 1.3732162458836443,
      "grad_norm": 0.006799628492444754,
      "learning_rate": 1.5814254062038407e-05,
      "loss": 0.0003,
      "step": 3753
    },
    {
      "epoch": 1.3735821441639224,
      "grad_norm": 1.1362006664276123,
      "learning_rate": 1.5805022156573118e-05,
      "loss": 0.0042,
      "step": 3754
    },
    {
      "epoch": 1.3739480424442005,
      "grad_norm": 0.005821292754262686,
      "learning_rate": 1.5795790251107828e-05,
      "loss": 0.0003,
      "step": 3755
    },
    {
      "epoch": 1.3743139407244787,
      "grad_norm": 0.0067887078039348125,
      "learning_rate": 1.5786558345642542e-05,
      "loss": 0.0003,
      "step": 3756
    },
    {
      "epoch": 1.3746798390047568,
      "grad_norm": 0.004977044183760881,
      "learning_rate": 1.5777326440177253e-05,
      "loss": 0.0002,
      "step": 3757
    },
    {
      "epoch": 1.3750457372850349,
      "grad_norm": 0.007679630536586046,
      "learning_rate": 1.5768094534711966e-05,
      "loss": 0.0003,
      "step": 3758
    },
    {
      "epoch": 1.3754116355653128,
      "grad_norm": 0.24392937123775482,
      "learning_rate": 1.5758862629246677e-05,
      "loss": 0.0028,
      "step": 3759
    },
    {
      "epoch": 1.3757775338455909,
      "grad_norm": 0.005176721606403589,
      "learning_rate": 1.5749630723781388e-05,
      "loss": 0.0002,
      "step": 3760
    },
    {
      "epoch": 1.376143432125869,
      "grad_norm": 0.022417718544602394,
      "learning_rate": 1.57403988183161e-05,
      "loss": 0.0003,
      "step": 3761
    },
    {
      "epoch": 1.376509330406147,
      "grad_norm": 0.006582105532288551,
      "learning_rate": 1.5731166912850815e-05,
      "loss": 0.0003,
      "step": 3762
    },
    {
      "epoch": 1.3768752286864252,
      "grad_norm": 0.004567407071590424,
      "learning_rate": 1.5721935007385523e-05,
      "loss": 0.0002,
      "step": 3763
    },
    {
      "epoch": 1.3772411269667033,
      "grad_norm": 0.006988114211708307,
      "learning_rate": 1.5712703101920237e-05,
      "loss": 0.0003,
      "step": 3764
    },
    {
      "epoch": 1.3776070252469814,
      "grad_norm": 0.006625080481171608,
      "learning_rate": 1.570347119645495e-05,
      "loss": 0.0003,
      "step": 3765
    },
    {
      "epoch": 1.3779729235272593,
      "grad_norm": 0.00680402060970664,
      "learning_rate": 1.569423929098966e-05,
      "loss": 0.0003,
      "step": 3766
    },
    {
      "epoch": 1.3783388218075374,
      "grad_norm": 0.005630633793771267,
      "learning_rate": 1.568500738552437e-05,
      "loss": 0.0003,
      "step": 3767
    },
    {
      "epoch": 1.3787047200878155,
      "grad_norm": 0.0485728494822979,
      "learning_rate": 1.5675775480059086e-05,
      "loss": 0.0007,
      "step": 3768
    },
    {
      "epoch": 1.3790706183680936,
      "grad_norm": 0.006591048091650009,
      "learning_rate": 1.5666543574593796e-05,
      "loss": 0.0003,
      "step": 3769
    },
    {
      "epoch": 1.3794365166483717,
      "grad_norm": 0.005626852624118328,
      "learning_rate": 1.5657311669128507e-05,
      "loss": 0.0003,
      "step": 3770
    },
    {
      "epoch": 1.3798024149286499,
      "grad_norm": 0.004623140208423138,
      "learning_rate": 1.564807976366322e-05,
      "loss": 0.0002,
      "step": 3771
    },
    {
      "epoch": 1.380168313208928,
      "grad_norm": 0.0070443167351186275,
      "learning_rate": 1.5638847858197935e-05,
      "loss": 0.0003,
      "step": 3772
    },
    {
      "epoch": 1.380534211489206,
      "grad_norm": 0.04013427719473839,
      "learning_rate": 1.5629615952732645e-05,
      "loss": 0.0005,
      "step": 3773
    },
    {
      "epoch": 1.3809001097694842,
      "grad_norm": 0.006354098208248615,
      "learning_rate": 1.5620384047267356e-05,
      "loss": 0.0003,
      "step": 3774
    },
    {
      "epoch": 1.3812660080497623,
      "grad_norm": 0.00390565674751997,
      "learning_rate": 1.561115214180207e-05,
      "loss": 0.0002,
      "step": 3775
    },
    {
      "epoch": 1.3816319063300402,
      "grad_norm": 0.006038084626197815,
      "learning_rate": 1.560192023633678e-05,
      "loss": 0.0002,
      "step": 3776
    },
    {
      "epoch": 1.3819978046103183,
      "grad_norm": 0.016241276636719704,
      "learning_rate": 1.5592688330871494e-05,
      "loss": 0.0003,
      "step": 3777
    },
    {
      "epoch": 1.3823637028905964,
      "grad_norm": 0.005580638535320759,
      "learning_rate": 1.5583456425406205e-05,
      "loss": 0.0003,
      "step": 3778
    },
    {
      "epoch": 1.3827296011708745,
      "grad_norm": 0.0063253724947571754,
      "learning_rate": 1.5574224519940915e-05,
      "loss": 0.0003,
      "step": 3779
    },
    {
      "epoch": 1.3830954994511526,
      "grad_norm": 0.021840941160917282,
      "learning_rate": 1.556499261447563e-05,
      "loss": 0.0004,
      "step": 3780
    },
    {
      "epoch": 1.3834613977314307,
      "grad_norm": 2.8308651447296143,
      "learning_rate": 1.555576070901034e-05,
      "loss": 0.0092,
      "step": 3781
    },
    {
      "epoch": 1.3838272960117086,
      "grad_norm": 0.004690852481871843,
      "learning_rate": 1.554652880354505e-05,
      "loss": 0.0002,
      "step": 3782
    },
    {
      "epoch": 1.3841931942919867,
      "grad_norm": 0.16932134330272675,
      "learning_rate": 1.5537296898079764e-05,
      "loss": 0.0012,
      "step": 3783
    },
    {
      "epoch": 1.3845590925722648,
      "grad_norm": 0.005234407726675272,
      "learning_rate": 1.5528064992614478e-05,
      "loss": 0.0002,
      "step": 3784
    },
    {
      "epoch": 1.384924990852543,
      "grad_norm": 0.005006752908229828,
      "learning_rate": 1.551883308714919e-05,
      "loss": 0.0002,
      "step": 3785
    },
    {
      "epoch": 1.385290889132821,
      "grad_norm": 0.004540469963103533,
      "learning_rate": 1.55096011816839e-05,
      "loss": 0.0002,
      "step": 3786
    },
    {
      "epoch": 1.3856567874130992,
      "grad_norm": 16.2827091217041,
      "learning_rate": 1.5500369276218613e-05,
      "loss": 0.3808,
      "step": 3787
    },
    {
      "epoch": 1.3860226856933773,
      "grad_norm": 0.0058243153616786,
      "learning_rate": 1.5491137370753324e-05,
      "loss": 0.0002,
      "step": 3788
    },
    {
      "epoch": 1.3863885839736554,
      "grad_norm": 0.005292855203151703,
      "learning_rate": 1.5481905465288034e-05,
      "loss": 0.0002,
      "step": 3789
    },
    {
      "epoch": 1.3867544822539335,
      "grad_norm": 2.551772356033325,
      "learning_rate": 1.5472673559822748e-05,
      "loss": 0.5408,
      "step": 3790
    },
    {
      "epoch": 1.3871203805342116,
      "grad_norm": 0.006318506319075823,
      "learning_rate": 1.5463441654357462e-05,
      "loss": 0.0003,
      "step": 3791
    },
    {
      "epoch": 1.3874862788144895,
      "grad_norm": 0.005264488980174065,
      "learning_rate": 1.5454209748892173e-05,
      "loss": 0.0002,
      "step": 3792
    },
    {
      "epoch": 1.3878521770947676,
      "grad_norm": 0.006943033076822758,
      "learning_rate": 1.5444977843426883e-05,
      "loss": 0.0003,
      "step": 3793
    },
    {
      "epoch": 1.3882180753750457,
      "grad_norm": 0.005414528306573629,
      "learning_rate": 1.5435745937961597e-05,
      "loss": 0.0002,
      "step": 3794
    },
    {
      "epoch": 1.3885839736553238,
      "grad_norm": 0.004496132489293814,
      "learning_rate": 1.5426514032496308e-05,
      "loss": 0.0002,
      "step": 3795
    },
    {
      "epoch": 1.388949871935602,
      "grad_norm": 0.005981886759400368,
      "learning_rate": 1.5417282127031018e-05,
      "loss": 0.0003,
      "step": 3796
    },
    {
      "epoch": 1.38931577021588,
      "grad_norm": 0.03364774212241173,
      "learning_rate": 1.5408050221565732e-05,
      "loss": 0.0006,
      "step": 3797
    },
    {
      "epoch": 1.3896816684961582,
      "grad_norm": 0.005769865121692419,
      "learning_rate": 1.5398818316100443e-05,
      "loss": 0.0003,
      "step": 3798
    },
    {
      "epoch": 1.390047566776436,
      "grad_norm": 0.01633148454129696,
      "learning_rate": 1.5389586410635157e-05,
      "loss": 0.0004,
      "step": 3799
    },
    {
      "epoch": 1.3904134650567141,
      "grad_norm": 0.006274579092860222,
      "learning_rate": 1.5380354505169867e-05,
      "loss": 0.0003,
      "step": 3800
    },
    {
      "epoch": 1.3907793633369923,
      "grad_norm": 27.19484519958496,
      "learning_rate": 1.5371122599704578e-05,
      "loss": 0.172,
      "step": 3801
    },
    {
      "epoch": 1.3911452616172704,
      "grad_norm": 6.513629913330078,
      "learning_rate": 1.536189069423929e-05,
      "loss": 0.408,
      "step": 3802
    },
    {
      "epoch": 1.3915111598975485,
      "grad_norm": 0.005693602841347456,
      "learning_rate": 1.5352658788774006e-05,
      "loss": 0.0002,
      "step": 3803
    },
    {
      "epoch": 1.3918770581778266,
      "grad_norm": 0.006742436904460192,
      "learning_rate": 1.5343426883308716e-05,
      "loss": 0.0003,
      "step": 3804
    },
    {
      "epoch": 1.3922429564581047,
      "grad_norm": 0.020945768803358078,
      "learning_rate": 1.5334194977843427e-05,
      "loss": 0.0004,
      "step": 3805
    },
    {
      "epoch": 1.3926088547383828,
      "grad_norm": 0.007943275384604931,
      "learning_rate": 1.532496307237814e-05,
      "loss": 0.0004,
      "step": 3806
    },
    {
      "epoch": 1.392974753018661,
      "grad_norm": 0.035509735345840454,
      "learning_rate": 1.531573116691285e-05,
      "loss": 0.0006,
      "step": 3807
    },
    {
      "epoch": 1.393340651298939,
      "grad_norm": 1.1231228113174438,
      "learning_rate": 1.5306499261447562e-05,
      "loss": 0.0047,
      "step": 3808
    },
    {
      "epoch": 1.393706549579217,
      "grad_norm": 6.3579020500183105,
      "learning_rate": 1.5297267355982276e-05,
      "loss": 0.3492,
      "step": 3809
    },
    {
      "epoch": 1.394072447859495,
      "grad_norm": 26.566085815429688,
      "learning_rate": 1.528803545051699e-05,
      "loss": 0.0212,
      "step": 3810
    },
    {
      "epoch": 1.3944383461397731,
      "grad_norm": 0.014601755887269974,
      "learning_rate": 1.5278803545051697e-05,
      "loss": 0.0005,
      "step": 3811
    },
    {
      "epoch": 1.3948042444200512,
      "grad_norm": 6.992506504058838,
      "learning_rate": 1.526957163958641e-05,
      "loss": 0.4301,
      "step": 3812
    },
    {
      "epoch": 1.3951701427003294,
      "grad_norm": 0.039034951478242874,
      "learning_rate": 1.5260339734121125e-05,
      "loss": 0.0011,
      "step": 3813
    },
    {
      "epoch": 1.3955360409806075,
      "grad_norm": 0.05138540640473366,
      "learning_rate": 1.5251107828655837e-05,
      "loss": 0.0012,
      "step": 3814
    },
    {
      "epoch": 1.3959019392608853,
      "grad_norm": 49.0382194519043,
      "learning_rate": 1.5241875923190546e-05,
      "loss": 0.3297,
      "step": 3815
    },
    {
      "epoch": 1.3962678375411635,
      "grad_norm": 0.021175779402256012,
      "learning_rate": 1.523264401772526e-05,
      "loss": 0.0007,
      "step": 3816
    },
    {
      "epoch": 1.3966337358214416,
      "grad_norm": 0.02313562110066414,
      "learning_rate": 1.5223412112259972e-05,
      "loss": 0.0007,
      "step": 3817
    },
    {
      "epoch": 1.3969996341017197,
      "grad_norm": 0.10324865579605103,
      "learning_rate": 1.5214180206794684e-05,
      "loss": 0.0028,
      "step": 3818
    },
    {
      "epoch": 1.3973655323819978,
      "grad_norm": 0.5099996328353882,
      "learning_rate": 1.5204948301329395e-05,
      "loss": 0.0029,
      "step": 3819
    },
    {
      "epoch": 1.397731430662276,
      "grad_norm": 0.009773820638656616,
      "learning_rate": 1.5195716395864107e-05,
      "loss": 0.0004,
      "step": 3820
    },
    {
      "epoch": 1.398097328942554,
      "grad_norm": 0.008974878117442131,
      "learning_rate": 1.518648449039882e-05,
      "loss": 0.0004,
      "step": 3821
    },
    {
      "epoch": 1.3984632272228321,
      "grad_norm": 0.04118890315294266,
      "learning_rate": 1.5177252584933533e-05,
      "loss": 0.0011,
      "step": 3822
    },
    {
      "epoch": 1.3988291255031102,
      "grad_norm": 0.06073432043194771,
      "learning_rate": 1.5168020679468242e-05,
      "loss": 0.0013,
      "step": 3823
    },
    {
      "epoch": 1.3991950237833883,
      "grad_norm": 0.02034696377813816,
      "learning_rate": 1.5158788774002954e-05,
      "loss": 0.0008,
      "step": 3824
    },
    {
      "epoch": 1.3995609220636662,
      "grad_norm": 0.02358930930495262,
      "learning_rate": 1.5149556868537668e-05,
      "loss": 0.0008,
      "step": 3825
    },
    {
      "epoch": 1.3999268203439443,
      "grad_norm": 0.025259025394916534,
      "learning_rate": 1.5140324963072377e-05,
      "loss": 0.0008,
      "step": 3826
    },
    {
      "epoch": 1.4002927186242224,
      "grad_norm": 0.010136164724826813,
      "learning_rate": 1.5131093057607091e-05,
      "loss": 0.0005,
      "step": 3827
    },
    {
      "epoch": 1.4006586169045006,
      "grad_norm": 0.039775896817445755,
      "learning_rate": 1.5121861152141803e-05,
      "loss": 0.0009,
      "step": 3828
    },
    {
      "epoch": 1.4010245151847787,
      "grad_norm": 0.010397523641586304,
      "learning_rate": 1.5112629246676515e-05,
      "loss": 0.0005,
      "step": 3829
    },
    {
      "epoch": 1.4013904134650568,
      "grad_norm": 0.022087134420871735,
      "learning_rate": 1.5103397341211226e-05,
      "loss": 0.0008,
      "step": 3830
    },
    {
      "epoch": 1.4017563117453347,
      "grad_norm": 0.012525924481451511,
      "learning_rate": 1.5094165435745938e-05,
      "loss": 0.0005,
      "step": 3831
    },
    {
      "epoch": 1.4021222100256128,
      "grad_norm": 3.0649847984313965,
      "learning_rate": 1.508493353028065e-05,
      "loss": 0.1812,
      "step": 3832
    },
    {
      "epoch": 1.4024881083058909,
      "grad_norm": 0.013756590895354748,
      "learning_rate": 1.5075701624815364e-05,
      "loss": 0.0005,
      "step": 3833
    },
    {
      "epoch": 1.402854006586169,
      "grad_norm": 0.12914545834064484,
      "learning_rate": 1.5066469719350073e-05,
      "loss": 0.0009,
      "step": 3834
    },
    {
      "epoch": 1.403219904866447,
      "grad_norm": 0.3075130581855774,
      "learning_rate": 1.5057237813884787e-05,
      "loss": 0.004,
      "step": 3835
    },
    {
      "epoch": 1.4035858031467252,
      "grad_norm": 0.019789380952715874,
      "learning_rate": 1.50480059084195e-05,
      "loss": 0.0006,
      "step": 3836
    },
    {
      "epoch": 1.4039517014270033,
      "grad_norm": 15.746247291564941,
      "learning_rate": 1.5038774002954212e-05,
      "loss": 0.2902,
      "step": 3837
    },
    {
      "epoch": 1.4043175997072814,
      "grad_norm": 3.4492921829223633,
      "learning_rate": 1.5029542097488922e-05,
      "loss": 0.4049,
      "step": 3838
    },
    {
      "epoch": 1.4046834979875595,
      "grad_norm": 0.2446790635585785,
      "learning_rate": 1.5020310192023634e-05,
      "loss": 0.0022,
      "step": 3839
    },
    {
      "epoch": 1.4050493962678376,
      "grad_norm": 0.009664432145655155,
      "learning_rate": 1.5011078286558347e-05,
      "loss": 0.0004,
      "step": 3840
    },
    {
      "epoch": 1.4054152945481158,
      "grad_norm": 0.014726339839398861,
      "learning_rate": 1.5001846381093057e-05,
      "loss": 0.0006,
      "step": 3841
    },
    {
      "epoch": 1.4057811928283936,
      "grad_norm": 0.022106396034359932,
      "learning_rate": 1.499261447562777e-05,
      "loss": 0.0008,
      "step": 3842
    },
    {
      "epoch": 1.4061470911086718,
      "grad_norm": 0.02539929375052452,
      "learning_rate": 1.4983382570162482e-05,
      "loss": 0.0009,
      "step": 3843
    },
    {
      "epoch": 1.4065129893889499,
      "grad_norm": 0.01971755176782608,
      "learning_rate": 1.4974150664697196e-05,
      "loss": 0.0007,
      "step": 3844
    },
    {
      "epoch": 1.406878887669228,
      "grad_norm": 0.02997218631207943,
      "learning_rate": 1.4964918759231905e-05,
      "loss": 0.0011,
      "step": 3845
    },
    {
      "epoch": 1.407244785949506,
      "grad_norm": 0.035904865711927414,
      "learning_rate": 1.4955686853766618e-05,
      "loss": 0.0009,
      "step": 3846
    },
    {
      "epoch": 1.4076106842297842,
      "grad_norm": 0.03405081108212471,
      "learning_rate": 1.494645494830133e-05,
      "loss": 0.0011,
      "step": 3847
    },
    {
      "epoch": 1.407976582510062,
      "grad_norm": 0.03441885858774185,
      "learning_rate": 1.4937223042836043e-05,
      "loss": 0.0011,
      "step": 3848
    },
    {
      "epoch": 1.4083424807903402,
      "grad_norm": 2.915046215057373,
      "learning_rate": 1.4927991137370754e-05,
      "loss": 0.0112,
      "step": 3849
    },
    {
      "epoch": 1.4087083790706183,
      "grad_norm": 0.017727680504322052,
      "learning_rate": 1.4918759231905466e-05,
      "loss": 0.0007,
      "step": 3850
    },
    {
      "epoch": 1.4090742773508964,
      "grad_norm": 0.025893747806549072,
      "learning_rate": 1.4909527326440178e-05,
      "loss": 0.0009,
      "step": 3851
    },
    {
      "epoch": 1.4094401756311745,
      "grad_norm": 0.04560904577374458,
      "learning_rate": 1.4900295420974892e-05,
      "loss": 0.0016,
      "step": 3852
    },
    {
      "epoch": 1.4098060739114526,
      "grad_norm": 0.029031552374362946,
      "learning_rate": 1.48910635155096e-05,
      "loss": 0.0011,
      "step": 3853
    },
    {
      "epoch": 1.4101719721917307,
      "grad_norm": 0.05269286781549454,
      "learning_rate": 1.4881831610044315e-05,
      "loss": 0.0015,
      "step": 3854
    },
    {
      "epoch": 1.4105378704720088,
      "grad_norm": 0.17514470219612122,
      "learning_rate": 1.4872599704579027e-05,
      "loss": 0.0041,
      "step": 3855
    },
    {
      "epoch": 1.410903768752287,
      "grad_norm": 0.012507021427154541,
      "learning_rate": 1.4863367799113736e-05,
      "loss": 0.0005,
      "step": 3856
    },
    {
      "epoch": 1.411269667032565,
      "grad_norm": 0.04436388611793518,
      "learning_rate": 1.485413589364845e-05,
      "loss": 0.0013,
      "step": 3857
    },
    {
      "epoch": 1.411635565312843,
      "grad_norm": 0.044265247881412506,
      "learning_rate": 1.4844903988183162e-05,
      "loss": 0.0012,
      "step": 3858
    },
    {
      "epoch": 1.412001463593121,
      "grad_norm": 0.01839429885149002,
      "learning_rate": 1.4835672082717874e-05,
      "loss": 0.0006,
      "step": 3859
    },
    {
      "epoch": 1.4123673618733992,
      "grad_norm": 0.09330692142248154,
      "learning_rate": 1.4826440177252585e-05,
      "loss": 0.0028,
      "step": 3860
    },
    {
      "epoch": 1.4127332601536773,
      "grad_norm": 0.48324763774871826,
      "learning_rate": 1.4817208271787297e-05,
      "loss": 0.0029,
      "step": 3861
    },
    {
      "epoch": 1.4130991584339554,
      "grad_norm": 0.008775205351412296,
      "learning_rate": 1.480797636632201e-05,
      "loss": 0.0004,
      "step": 3862
    },
    {
      "epoch": 1.4134650567142335,
      "grad_norm": 0.016668224707245827,
      "learning_rate": 1.4798744460856723e-05,
      "loss": 0.0006,
      "step": 3863
    },
    {
      "epoch": 1.4138309549945114,
      "grad_norm": 0.013614518567919731,
      "learning_rate": 1.4789512555391432e-05,
      "loss": 0.0006,
      "step": 3864
    },
    {
      "epoch": 1.4141968532747895,
      "grad_norm": 0.012033658102154732,
      "learning_rate": 1.4780280649926146e-05,
      "loss": 0.0005,
      "step": 3865
    },
    {
      "epoch": 1.4145627515550676,
      "grad_norm": 0.0070667024701833725,
      "learning_rate": 1.4771048744460858e-05,
      "loss": 0.0003,
      "step": 3866
    },
    {
      "epoch": 1.4149286498353457,
      "grad_norm": 4.960713863372803,
      "learning_rate": 1.476181683899557e-05,
      "loss": 0.3026,
      "step": 3867
    },
    {
      "epoch": 1.4152945481156238,
      "grad_norm": 0.027299050241708755,
      "learning_rate": 1.4752584933530281e-05,
      "loss": 0.0008,
      "step": 3868
    },
    {
      "epoch": 1.415660446395902,
      "grad_norm": 0.11490294337272644,
      "learning_rate": 1.4743353028064993e-05,
      "loss": 0.0038,
      "step": 3869
    },
    {
      "epoch": 1.41602634467618,
      "grad_norm": 0.14292749762535095,
      "learning_rate": 1.4734121122599706e-05,
      "loss": 0.001,
      "step": 3870
    },
    {
      "epoch": 1.4163922429564582,
      "grad_norm": 0.018945399671792984,
      "learning_rate": 1.4724889217134416e-05,
      "loss": 0.0006,
      "step": 3871
    },
    {
      "epoch": 1.4167581412367363,
      "grad_norm": 0.022599516436457634,
      "learning_rate": 1.4715657311669128e-05,
      "loss": 0.0007,
      "step": 3872
    },
    {
      "epoch": 1.4171240395170144,
      "grad_norm": 0.01092821080237627,
      "learning_rate": 1.4706425406203842e-05,
      "loss": 0.0004,
      "step": 3873
    },
    {
      "epoch": 1.4174899377972925,
      "grad_norm": 0.00815325416624546,
      "learning_rate": 1.4697193500738555e-05,
      "loss": 0.0004,
      "step": 3874
    },
    {
      "epoch": 1.4178558360775704,
      "grad_norm": 0.02450287714600563,
      "learning_rate": 1.4687961595273263e-05,
      "loss": 0.0008,
      "step": 3875
    },
    {
      "epoch": 1.4182217343578485,
      "grad_norm": 0.0635528713464737,
      "learning_rate": 1.4678729689807977e-05,
      "loss": 0.0015,
      "step": 3876
    },
    {
      "epoch": 1.4185876326381266,
      "grad_norm": 1.1641638278961182,
      "learning_rate": 1.466949778434269e-05,
      "loss": 0.0055,
      "step": 3877
    },
    {
      "epoch": 1.4189535309184047,
      "grad_norm": 2.085700035095215,
      "learning_rate": 1.4660265878877402e-05,
      "loss": 0.2339,
      "step": 3878
    },
    {
      "epoch": 1.4193194291986828,
      "grad_norm": 0.02031099982559681,
      "learning_rate": 1.4651033973412112e-05,
      "loss": 0.0007,
      "step": 3879
    },
    {
      "epoch": 1.419685327478961,
      "grad_norm": 0.03187205642461777,
      "learning_rate": 1.4641802067946825e-05,
      "loss": 0.0009,
      "step": 3880
    },
    {
      "epoch": 1.4200512257592388,
      "grad_norm": 0.011147622019052505,
      "learning_rate": 1.4632570162481537e-05,
      "loss": 0.0004,
      "step": 3881
    },
    {
      "epoch": 1.420417124039517,
      "grad_norm": 0.006791383493691683,
      "learning_rate": 1.4623338257016247e-05,
      "loss": 0.0003,
      "step": 3882
    },
    {
      "epoch": 1.420783022319795,
      "grad_norm": 0.21201494336128235,
      "learning_rate": 1.461410635155096e-05,
      "loss": 0.0037,
      "step": 3883
    },
    {
      "epoch": 1.4211489206000731,
      "grad_norm": 0.3642657399177551,
      "learning_rate": 1.4604874446085674e-05,
      "loss": 0.0034,
      "step": 3884
    },
    {
      "epoch": 1.4215148188803512,
      "grad_norm": 0.008497160859405994,
      "learning_rate": 1.4595642540620386e-05,
      "loss": 0.0003,
      "step": 3885
    },
    {
      "epoch": 1.4218807171606294,
      "grad_norm": 0.006668700836598873,
      "learning_rate": 1.4586410635155096e-05,
      "loss": 0.0003,
      "step": 3886
    },
    {
      "epoch": 1.4222466154409075,
      "grad_norm": 0.0062116472981870174,
      "learning_rate": 1.4577178729689809e-05,
      "loss": 0.0003,
      "step": 3887
    },
    {
      "epoch": 1.4226125137211856,
      "grad_norm": 0.007947543635964394,
      "learning_rate": 1.456794682422452e-05,
      "loss": 0.0003,
      "step": 3888
    },
    {
      "epoch": 1.4229784120014637,
      "grad_norm": 0.07692423462867737,
      "learning_rate": 1.4558714918759233e-05,
      "loss": 0.002,
      "step": 3889
    },
    {
      "epoch": 1.4233443102817418,
      "grad_norm": 0.01307628769427538,
      "learning_rate": 1.4549483013293944e-05,
      "loss": 0.0004,
      "step": 3890
    },
    {
      "epoch": 1.4237102085620197,
      "grad_norm": 0.028877485543489456,
      "learning_rate": 1.4540251107828656e-05,
      "loss": 0.0007,
      "step": 3891
    },
    {
      "epoch": 1.4240761068422978,
      "grad_norm": 0.008642594330012798,
      "learning_rate": 1.453101920236337e-05,
      "loss": 0.0003,
      "step": 3892
    },
    {
      "epoch": 1.424442005122576,
      "grad_norm": 0.18771176040172577,
      "learning_rate": 1.4521787296898082e-05,
      "loss": 0.0035,
      "step": 3893
    },
    {
      "epoch": 1.424807903402854,
      "grad_norm": 0.005379003938287497,
      "learning_rate": 1.4512555391432791e-05,
      "loss": 0.0002,
      "step": 3894
    },
    {
      "epoch": 1.4251738016831321,
      "grad_norm": 0.007763344328850508,
      "learning_rate": 1.4503323485967505e-05,
      "loss": 0.0004,
      "step": 3895
    },
    {
      "epoch": 1.4255396999634102,
      "grad_norm": 0.10414665192365646,
      "learning_rate": 1.4494091580502217e-05,
      "loss": 0.0029,
      "step": 3896
    },
    {
      "epoch": 1.4259055982436881,
      "grad_norm": 0.005085620563477278,
      "learning_rate": 1.4484859675036928e-05,
      "loss": 0.0002,
      "step": 3897
    },
    {
      "epoch": 1.4262714965239662,
      "grad_norm": 0.007809446658939123,
      "learning_rate": 1.447562776957164e-05,
      "loss": 0.0003,
      "step": 3898
    },
    {
      "epoch": 1.4266373948042443,
      "grad_norm": 0.08002348989248276,
      "learning_rate": 1.4466395864106352e-05,
      "loss": 0.0008,
      "step": 3899
    },
    {
      "epoch": 1.4270032930845225,
      "grad_norm": 5.288859844207764,
      "learning_rate": 1.4457163958641064e-05,
      "loss": 0.2735,
      "step": 3900
    },
    {
      "epoch": 1.4273691913648006,
      "grad_norm": 0.04041401296854019,
      "learning_rate": 1.4447932053175775e-05,
      "loss": 0.0008,
      "step": 3901
    },
    {
      "epoch": 1.4277350896450787,
      "grad_norm": 0.09373247623443604,
      "learning_rate": 1.4438700147710487e-05,
      "loss": 0.0009,
      "step": 3902
    },
    {
      "epoch": 1.4281009879253568,
      "grad_norm": 0.017604978755116463,
      "learning_rate": 1.4429468242245201e-05,
      "loss": 0.0004,
      "step": 3903
    },
    {
      "epoch": 1.428466886205635,
      "grad_norm": 0.017239172011613846,
      "learning_rate": 1.4420236336779913e-05,
      "loss": 0.0003,
      "step": 3904
    },
    {
      "epoch": 1.428832784485913,
      "grad_norm": 0.12726810574531555,
      "learning_rate": 1.4411004431314624e-05,
      "loss": 0.0009,
      "step": 3905
    },
    {
      "epoch": 1.429198682766191,
      "grad_norm": 0.005110755097121,
      "learning_rate": 1.4401772525849336e-05,
      "loss": 0.0002,
      "step": 3906
    },
    {
      "epoch": 1.4295645810464692,
      "grad_norm": 5.081525802612305,
      "learning_rate": 1.4392540620384048e-05,
      "loss": 0.0064,
      "step": 3907
    },
    {
      "epoch": 1.429930479326747,
      "grad_norm": 0.009699770249426365,
      "learning_rate": 1.438330871491876e-05,
      "loss": 0.0003,
      "step": 3908
    },
    {
      "epoch": 1.4302963776070252,
      "grad_norm": 0.010034791193902493,
      "learning_rate": 1.4374076809453471e-05,
      "loss": 0.0004,
      "step": 3909
    },
    {
      "epoch": 1.4306622758873033,
      "grad_norm": 0.011184345930814743,
      "learning_rate": 1.4364844903988183e-05,
      "loss": 0.0004,
      "step": 3910
    },
    {
      "epoch": 1.4310281741675814,
      "grad_norm": 0.005327730439603329,
      "learning_rate": 1.4355612998522897e-05,
      "loss": 0.0002,
      "step": 3911
    },
    {
      "epoch": 1.4313940724478595,
      "grad_norm": 0.1421309858560562,
      "learning_rate": 1.4346381093057606e-05,
      "loss": 0.0028,
      "step": 3912
    },
    {
      "epoch": 1.4317599707281377,
      "grad_norm": 0.006638459395617247,
      "learning_rate": 1.4337149187592318e-05,
      "loss": 0.0003,
      "step": 3913
    },
    {
      "epoch": 1.4321258690084155,
      "grad_norm": 0.043224629014730453,
      "learning_rate": 1.4327917282127032e-05,
      "loss": 0.0006,
      "step": 3914
    },
    {
      "epoch": 1.4324917672886937,
      "grad_norm": 0.17543911933898926,
      "learning_rate": 1.4318685376661745e-05,
      "loss": 0.0054,
      "step": 3915
    },
    {
      "epoch": 1.4328576655689718,
      "grad_norm": 0.009843871928751469,
      "learning_rate": 1.4309453471196455e-05,
      "loss": 0.0003,
      "step": 3916
    },
    {
      "epoch": 1.4332235638492499,
      "grad_norm": 0.008856077678501606,
      "learning_rate": 1.4300221565731167e-05,
      "loss": 0.0003,
      "step": 3917
    },
    {
      "epoch": 1.433589462129528,
      "grad_norm": 0.09332273155450821,
      "learning_rate": 1.429098966026588e-05,
      "loss": 0.0026,
      "step": 3918
    },
    {
      "epoch": 1.433955360409806,
      "grad_norm": 0.008499838411808014,
      "learning_rate": 1.4281757754800592e-05,
      "loss": 0.0003,
      "step": 3919
    },
    {
      "epoch": 1.4343212586900842,
      "grad_norm": 0.012390444055199623,
      "learning_rate": 1.4272525849335302e-05,
      "loss": 0.0004,
      "step": 3920
    },
    {
      "epoch": 1.4346871569703623,
      "grad_norm": 0.1268148422241211,
      "learning_rate": 1.4263293943870015e-05,
      "loss": 0.0026,
      "step": 3921
    },
    {
      "epoch": 1.4350530552506404,
      "grad_norm": 9.805891036987305,
      "learning_rate": 1.4254062038404729e-05,
      "loss": 0.0144,
      "step": 3922
    },
    {
      "epoch": 1.4354189535309185,
      "grad_norm": 0.09058788418769836,
      "learning_rate": 1.4244830132939441e-05,
      "loss": 0.0026,
      "step": 3923
    },
    {
      "epoch": 1.4357848518111964,
      "grad_norm": 0.12705588340759277,
      "learning_rate": 1.4235598227474151e-05,
      "loss": 0.0027,
      "step": 3924
    },
    {
      "epoch": 1.4361507500914745,
      "grad_norm": 0.013544915243983269,
      "learning_rate": 1.4226366322008864e-05,
      "loss": 0.0002,
      "step": 3925
    },
    {
      "epoch": 1.4365166483717526,
      "grad_norm": 0.021797772496938705,
      "learning_rate": 1.4217134416543576e-05,
      "loss": 0.0005,
      "step": 3926
    },
    {
      "epoch": 1.4368825466520307,
      "grad_norm": 0.08350890874862671,
      "learning_rate": 1.4207902511078286e-05,
      "loss": 0.0022,
      "step": 3927
    },
    {
      "epoch": 1.4372484449323089,
      "grad_norm": 0.018075330182909966,
      "learning_rate": 1.4198670605612999e-05,
      "loss": 0.0004,
      "step": 3928
    },
    {
      "epoch": 1.437614343212587,
      "grad_norm": 0.007522772066295147,
      "learning_rate": 1.4189438700147711e-05,
      "loss": 0.0002,
      "step": 3929
    },
    {
      "epoch": 1.4379802414928649,
      "grad_norm": 0.0055830031633377075,
      "learning_rate": 1.4180206794682425e-05,
      "loss": 0.0002,
      "step": 3930
    },
    {
      "epoch": 1.438346139773143,
      "grad_norm": 0.0584937147796154,
      "learning_rate": 1.4170974889217134e-05,
      "loss": 0.0014,
      "step": 3931
    },
    {
      "epoch": 1.438712038053421,
      "grad_norm": 0.014518242329359055,
      "learning_rate": 1.4161742983751846e-05,
      "loss": 0.0004,
      "step": 3932
    },
    {
      "epoch": 1.4390779363336992,
      "grad_norm": 0.006070802453905344,
      "learning_rate": 1.415251107828656e-05,
      "loss": 0.0003,
      "step": 3933
    },
    {
      "epoch": 1.4394438346139773,
      "grad_norm": 3.5416369438171387,
      "learning_rate": 1.4143279172821272e-05,
      "loss": 0.1998,
      "step": 3934
    },
    {
      "epoch": 1.4398097328942554,
      "grad_norm": 0.004318627994507551,
      "learning_rate": 1.4134047267355983e-05,
      "loss": 0.0002,
      "step": 3935
    },
    {
      "epoch": 1.4401756311745335,
      "grad_norm": 0.09875593334436417,
      "learning_rate": 1.4124815361890695e-05,
      "loss": 0.0011,
      "step": 3936
    },
    {
      "epoch": 1.4405415294548116,
      "grad_norm": 0.005922273732721806,
      "learning_rate": 1.4115583456425407e-05,
      "loss": 0.0003,
      "step": 3937
    },
    {
      "epoch": 1.4409074277350897,
      "grad_norm": 0.0666550025343895,
      "learning_rate": 1.410635155096012e-05,
      "loss": 0.0008,
      "step": 3938
    },
    {
      "epoch": 1.4412733260153678,
      "grad_norm": 0.009299950674176216,
      "learning_rate": 1.409711964549483e-05,
      "loss": 0.0002,
      "step": 3939
    },
    {
      "epoch": 1.441639224295646,
      "grad_norm": 0.0056312112137675285,
      "learning_rate": 1.4087887740029542e-05,
      "loss": 0.0002,
      "step": 3940
    },
    {
      "epoch": 1.4420051225759238,
      "grad_norm": 0.008318420499563217,
      "learning_rate": 1.4078655834564256e-05,
      "loss": 0.0003,
      "step": 3941
    },
    {
      "epoch": 1.442371020856202,
      "grad_norm": 0.09523820877075195,
      "learning_rate": 1.4069423929098965e-05,
      "loss": 0.0007,
      "step": 3942
    },
    {
      "epoch": 1.44273691913648,
      "grad_norm": 0.027628686279058456,
      "learning_rate": 1.4060192023633679e-05,
      "loss": 0.0004,
      "step": 3943
    },
    {
      "epoch": 1.4431028174167582,
      "grad_norm": 0.009160085581243038,
      "learning_rate": 1.4050960118168391e-05,
      "loss": 0.0003,
      "step": 3944
    },
    {
      "epoch": 1.4434687156970363,
      "grad_norm": 0.004625244531780481,
      "learning_rate": 1.4041728212703103e-05,
      "loss": 0.0002,
      "step": 3945
    },
    {
      "epoch": 1.4438346139773144,
      "grad_norm": 0.0067979879677295685,
      "learning_rate": 1.4032496307237814e-05,
      "loss": 0.0003,
      "step": 3946
    },
    {
      "epoch": 1.4442005122575923,
      "grad_norm": 0.006252759136259556,
      "learning_rate": 1.4023264401772526e-05,
      "loss": 0.0003,
      "step": 3947
    },
    {
      "epoch": 1.4445664105378704,
      "grad_norm": 0.10563553124666214,
      "learning_rate": 1.4014032496307238e-05,
      "loss": 0.001,
      "step": 3948
    },
    {
      "epoch": 1.4449323088181485,
      "grad_norm": 0.008291146717965603,
      "learning_rate": 1.4004800590841952e-05,
      "loss": 0.0003,
      "step": 3949
    },
    {
      "epoch": 1.4452982070984266,
      "grad_norm": 0.009138152003288269,
      "learning_rate": 1.3995568685376661e-05,
      "loss": 0.0003,
      "step": 3950
    },
    {
      "epoch": 1.4456641053787047,
      "grad_norm": 0.004630940966308117,
      "learning_rate": 1.3986336779911374e-05,
      "loss": 0.0002,
      "step": 3951
    },
    {
      "epoch": 1.4460300036589828,
      "grad_norm": 0.005487330257892609,
      "learning_rate": 1.3977104874446087e-05,
      "loss": 0.0002,
      "step": 3952
    },
    {
      "epoch": 1.446395901939261,
      "grad_norm": 3.29585599899292,
      "learning_rate": 1.39678729689808e-05,
      "loss": 0.2022,
      "step": 3953
    },
    {
      "epoch": 1.446761800219539,
      "grad_norm": 0.004926501773297787,
      "learning_rate": 1.395864106351551e-05,
      "loss": 0.0002,
      "step": 3954
    },
    {
      "epoch": 1.4471276984998171,
      "grad_norm": 0.027538960799574852,
      "learning_rate": 1.3949409158050222e-05,
      "loss": 0.0003,
      "step": 3955
    },
    {
      "epoch": 1.4474935967800953,
      "grad_norm": 13.181702613830566,
      "learning_rate": 1.3940177252584935e-05,
      "loss": 0.4016,
      "step": 3956
    },
    {
      "epoch": 1.4478594950603731,
      "grad_norm": 0.007388906553387642,
      "learning_rate": 1.3930945347119645e-05,
      "loss": 0.0003,
      "step": 3957
    },
    {
      "epoch": 1.4482253933406513,
      "grad_norm": 0.17886269092559814,
      "learning_rate": 1.3921713441654358e-05,
      "loss": 0.0026,
      "step": 3958
    },
    {
      "epoch": 1.4485912916209294,
      "grad_norm": 0.2125595510005951,
      "learning_rate": 1.391248153618907e-05,
      "loss": 0.005,
      "step": 3959
    },
    {
      "epoch": 1.4489571899012075,
      "grad_norm": 21.701858520507812,
      "learning_rate": 1.3903249630723784e-05,
      "loss": 0.4975,
      "step": 3960
    },
    {
      "epoch": 1.4493230881814856,
      "grad_norm": 0.05838427692651749,
      "learning_rate": 1.3894017725258493e-05,
      "loss": 0.0006,
      "step": 3961
    },
    {
      "epoch": 1.4496889864617637,
      "grad_norm": 0.07852599769830704,
      "learning_rate": 1.3884785819793206e-05,
      "loss": 0.0007,
      "step": 3962
    },
    {
      "epoch": 1.4500548847420416,
      "grad_norm": 0.02482706308364868,
      "learning_rate": 1.3875553914327919e-05,
      "loss": 0.0003,
      "step": 3963
    },
    {
      "epoch": 1.4504207830223197,
      "grad_norm": 0.24890142679214478,
      "learning_rate": 1.3866322008862631e-05,
      "loss": 0.0011,
      "step": 3964
    },
    {
      "epoch": 1.4507866813025978,
      "grad_norm": 0.006503795273602009,
      "learning_rate": 1.3857090103397342e-05,
      "loss": 0.0003,
      "step": 3965
    },
    {
      "epoch": 1.451152579582876,
      "grad_norm": 0.17022040486335754,
      "learning_rate": 1.3847858197932054e-05,
      "loss": 0.003,
      "step": 3966
    },
    {
      "epoch": 1.451518477863154,
      "grad_norm": 15.521991729736328,
      "learning_rate": 1.3838626292466766e-05,
      "loss": 0.4358,
      "step": 3967
    },
    {
      "epoch": 1.4518843761434321,
      "grad_norm": 0.037761591374874115,
      "learning_rate": 1.3829394387001477e-05,
      "loss": 0.0005,
      "step": 3968
    },
    {
      "epoch": 1.4522502744237102,
      "grad_norm": 0.006072793621569872,
      "learning_rate": 1.3820162481536189e-05,
      "loss": 0.0003,
      "step": 3969
    },
    {
      "epoch": 1.4526161727039884,
      "grad_norm": 0.1782481074333191,
      "learning_rate": 1.3810930576070901e-05,
      "loss": 0.0023,
      "step": 3970
    },
    {
      "epoch": 1.4529820709842665,
      "grad_norm": 0.00788299273699522,
      "learning_rate": 1.3801698670605615e-05,
      "loss": 0.0003,
      "step": 3971
    },
    {
      "epoch": 1.4533479692645446,
      "grad_norm": 0.004364347085356712,
      "learning_rate": 1.3792466765140324e-05,
      "loss": 0.0002,
      "step": 3972
    },
    {
      "epoch": 1.4537138675448225,
      "grad_norm": 0.007228373549878597,
      "learning_rate": 1.3783234859675038e-05,
      "loss": 0.0003,
      "step": 3973
    },
    {
      "epoch": 1.4540797658251006,
      "grad_norm": 4.751836776733398,
      "learning_rate": 1.377400295420975e-05,
      "loss": 0.1846,
      "step": 3974
    },
    {
      "epoch": 1.4544456641053787,
      "grad_norm": 0.005238227546215057,
      "learning_rate": 1.3764771048744462e-05,
      "loss": 0.0002,
      "step": 3975
    },
    {
      "epoch": 1.4548115623856568,
      "grad_norm": 0.012375573627650738,
      "learning_rate": 1.3755539143279173e-05,
      "loss": 0.0004,
      "step": 3976
    },
    {
      "epoch": 1.455177460665935,
      "grad_norm": 0.00591261824592948,
      "learning_rate": 1.3746307237813885e-05,
      "loss": 0.0003,
      "step": 3977
    },
    {
      "epoch": 1.455543358946213,
      "grad_norm": 2.392815589904785,
      "learning_rate": 1.3737075332348597e-05,
      "loss": 0.188,
      "step": 3978
    },
    {
      "epoch": 1.4559092572264911,
      "grad_norm": 0.4500861167907715,
      "learning_rate": 1.3727843426883311e-05,
      "loss": 0.0066,
      "step": 3979
    },
    {
      "epoch": 1.456275155506769,
      "grad_norm": 0.010431904345750809,
      "learning_rate": 1.371861152141802e-05,
      "loss": 0.0003,
      "step": 3980
    },
    {
      "epoch": 1.4566410537870471,
      "grad_norm": 0.004857638385146856,
      "learning_rate": 1.3709379615952734e-05,
      "loss": 0.0002,
      "step": 3981
    },
    {
      "epoch": 1.4570069520673252,
      "grad_norm": 0.16156207025051117,
      "learning_rate": 1.3700147710487446e-05,
      "loss": 0.0016,
      "step": 3982
    },
    {
      "epoch": 1.4573728503476033,
      "grad_norm": 21.990325927734375,
      "learning_rate": 1.3690915805022155e-05,
      "loss": 0.175,
      "step": 3983
    },
    {
      "epoch": 1.4577387486278814,
      "grad_norm": 7.955433368682861,
      "learning_rate": 1.3681683899556869e-05,
      "loss": 0.1382,
      "step": 3984
    },
    {
      "epoch": 1.4581046469081596,
      "grad_norm": 0.1394226998090744,
      "learning_rate": 1.3672451994091581e-05,
      "loss": 0.0045,
      "step": 3985
    },
    {
      "epoch": 1.4584705451884377,
      "grad_norm": 0.21643424034118652,
      "learning_rate": 1.3663220088626294e-05,
      "loss": 0.006,
      "step": 3986
    },
    {
      "epoch": 1.4588364434687158,
      "grad_norm": 0.01599455624818802,
      "learning_rate": 1.3653988183161004e-05,
      "loss": 0.0003,
      "step": 3987
    },
    {
      "epoch": 1.4592023417489939,
      "grad_norm": 0.00713999243453145,
      "learning_rate": 1.3644756277695716e-05,
      "loss": 0.0003,
      "step": 3988
    },
    {
      "epoch": 1.459568240029272,
      "grad_norm": 0.22308097779750824,
      "learning_rate": 1.3635524372230429e-05,
      "loss": 0.0022,
      "step": 3989
    },
    {
      "epoch": 1.4599341383095499,
      "grad_norm": 0.005570400971919298,
      "learning_rate": 1.3626292466765143e-05,
      "loss": 0.0002,
      "step": 3990
    },
    {
      "epoch": 1.460300036589828,
      "grad_norm": 0.4013859033584595,
      "learning_rate": 1.3617060561299851e-05,
      "loss": 0.0148,
      "step": 3991
    },
    {
      "epoch": 1.460665934870106,
      "grad_norm": 0.04499814286828041,
      "learning_rate": 1.3607828655834565e-05,
      "loss": 0.0005,
      "step": 3992
    },
    {
      "epoch": 1.4610318331503842,
      "grad_norm": 0.024039169773459435,
      "learning_rate": 1.3598596750369278e-05,
      "loss": 0.0004,
      "step": 3993
    },
    {
      "epoch": 1.4613977314306623,
      "grad_norm": 0.006567932665348053,
      "learning_rate": 1.358936484490399e-05,
      "loss": 0.0003,
      "step": 3994
    },
    {
      "epoch": 1.4617636297109404,
      "grad_norm": 0.009509388357400894,
      "learning_rate": 1.35801329394387e-05,
      "loss": 0.0003,
      "step": 3995
    },
    {
      "epoch": 1.4621295279912183,
      "grad_norm": 0.005102129653096199,
      "learning_rate": 1.3570901033973413e-05,
      "loss": 0.0002,
      "step": 3996
    },
    {
      "epoch": 1.4624954262714964,
      "grad_norm": 0.5159423351287842,
      "learning_rate": 1.3561669128508125e-05,
      "loss": 0.0031,
      "step": 3997
    },
    {
      "epoch": 1.4628613245517745,
      "grad_norm": 0.009157221764326096,
      "learning_rate": 1.3552437223042835e-05,
      "loss": 0.0002,
      "step": 3998
    },
    {
      "epoch": 1.4632272228320526,
      "grad_norm": 1.4796454906463623,
      "learning_rate": 1.3543205317577548e-05,
      "loss": 0.1563,
      "step": 3999
    },
    {
      "epoch": 1.4635931211123308,
      "grad_norm": 1.552032709121704,
      "learning_rate": 1.3533973412112262e-05,
      "loss": 0.0088,
      "step": 4000
    },
    {
      "epoch": 1.4639590193926089,
      "grad_norm": 0.0035703314933925867,
      "learning_rate": 1.3524741506646974e-05,
      "loss": 0.0002,
      "step": 4001
    },
    {
      "epoch": 1.464324917672887,
      "grad_norm": 0.11024852097034454,
      "learning_rate": 1.3515509601181683e-05,
      "loss": 0.0006,
      "step": 4002
    },
    {
      "epoch": 1.464690815953165,
      "grad_norm": 0.20563316345214844,
      "learning_rate": 1.3506277695716397e-05,
      "loss": 0.0011,
      "step": 4003
    },
    {
      "epoch": 1.4650567142334432,
      "grad_norm": 0.261508971452713,
      "learning_rate": 1.3497045790251109e-05,
      "loss": 0.0036,
      "step": 4004
    },
    {
      "epoch": 1.4654226125137213,
      "grad_norm": 1.436331868171692,
      "learning_rate": 1.3487813884785821e-05,
      "loss": 0.1385,
      "step": 4005
    },
    {
      "epoch": 1.4657885107939992,
      "grad_norm": 0.3746028244495392,
      "learning_rate": 1.3478581979320532e-05,
      "loss": 0.0081,
      "step": 4006
    },
    {
      "epoch": 1.4661544090742773,
      "grad_norm": 0.01861211098730564,
      "learning_rate": 1.3469350073855244e-05,
      "loss": 0.0004,
      "step": 4007
    },
    {
      "epoch": 1.4665203073545554,
      "grad_norm": 0.006720947567373514,
      "learning_rate": 1.3460118168389956e-05,
      "loss": 0.0002,
      "step": 4008
    },
    {
      "epoch": 1.4668862056348335,
      "grad_norm": 0.6064707636833191,
      "learning_rate": 1.345088626292467e-05,
      "loss": 0.0055,
      "step": 4009
    },
    {
      "epoch": 1.4672521039151116,
      "grad_norm": 0.005706157069653273,
      "learning_rate": 1.3441654357459379e-05,
      "loss": 0.0003,
      "step": 4010
    },
    {
      "epoch": 1.4676180021953897,
      "grad_norm": 0.06355342268943787,
      "learning_rate": 1.3432422451994093e-05,
      "loss": 0.0004,
      "step": 4011
    },
    {
      "epoch": 1.4679839004756678,
      "grad_norm": 0.20491044223308563,
      "learning_rate": 1.3423190546528805e-05,
      "loss": 0.0063,
      "step": 4012
    },
    {
      "epoch": 1.4683497987559457,
      "grad_norm": 0.005032045301049948,
      "learning_rate": 1.3413958641063516e-05,
      "loss": 0.0002,
      "step": 4013
    },
    {
      "epoch": 1.4687156970362238,
      "grad_norm": 0.005713785532861948,
      "learning_rate": 1.3404726735598228e-05,
      "loss": 0.0002,
      "step": 4014
    },
    {
      "epoch": 1.469081595316502,
      "grad_norm": 0.014670152217149734,
      "learning_rate": 1.339549483013294e-05,
      "loss": 0.0002,
      "step": 4015
    },
    {
      "epoch": 1.46944749359678,
      "grad_norm": 0.006246511358767748,
      "learning_rate": 1.3386262924667652e-05,
      "loss": 0.0002,
      "step": 4016
    },
    {
      "epoch": 1.4698133918770582,
      "grad_norm": 0.005324751604348421,
      "learning_rate": 1.3377031019202363e-05,
      "loss": 0.0002,
      "step": 4017
    },
    {
      "epoch": 1.4701792901573363,
      "grad_norm": 0.2713962495326996,
      "learning_rate": 1.3367799113737075e-05,
      "loss": 0.001,
      "step": 4018
    },
    {
      "epoch": 1.4705451884376144,
      "grad_norm": 0.00582299055531621,
      "learning_rate": 1.3358567208271789e-05,
      "loss": 0.0002,
      "step": 4019
    },
    {
      "epoch": 1.4709110867178925,
      "grad_norm": 0.06695003062486649,
      "learning_rate": 1.3349335302806501e-05,
      "loss": 0.0004,
      "step": 4020
    },
    {
      "epoch": 1.4712769849981706,
      "grad_norm": 0.04010280221700668,
      "learning_rate": 1.334010339734121e-05,
      "loss": 0.0005,
      "step": 4021
    },
    {
      "epoch": 1.4716428832784487,
      "grad_norm": 0.005584307014942169,
      "learning_rate": 1.3330871491875924e-05,
      "loss": 0.0002,
      "step": 4022
    },
    {
      "epoch": 1.4720087815587266,
      "grad_norm": 0.006222480442374945,
      "learning_rate": 1.3321639586410636e-05,
      "loss": 0.0002,
      "step": 4023
    },
    {
      "epoch": 1.4723746798390047,
      "grad_norm": 0.14586177468299866,
      "learning_rate": 1.3312407680945349e-05,
      "loss": 0.0005,
      "step": 4024
    },
    {
      "epoch": 1.4727405781192828,
      "grad_norm": 0.004807180259376764,
      "learning_rate": 1.330317577548006e-05,
      "loss": 0.0002,
      "step": 4025
    },
    {
      "epoch": 1.473106476399561,
      "grad_norm": 0.4414343535900116,
      "learning_rate": 1.3293943870014771e-05,
      "loss": 0.0104,
      "step": 4026
    },
    {
      "epoch": 1.473472374679839,
      "grad_norm": 0.004065286368131638,
      "learning_rate": 1.3284711964549484e-05,
      "loss": 0.0002,
      "step": 4027
    },
    {
      "epoch": 1.4738382729601172,
      "grad_norm": 0.004780706483870745,
      "learning_rate": 1.3275480059084194e-05,
      "loss": 0.0002,
      "step": 4028
    },
    {
      "epoch": 1.474204171240395,
      "grad_norm": 2.063446283340454,
      "learning_rate": 1.3266248153618906e-05,
      "loss": 0.5006,
      "step": 4029
    },
    {
      "epoch": 1.4745700695206732,
      "grad_norm": 0.008270715363323689,
      "learning_rate": 1.325701624815362e-05,
      "loss": 0.0002,
      "step": 4030
    },
    {
      "epoch": 1.4749359678009513,
      "grad_norm": 0.009865415282547474,
      "learning_rate": 1.3247784342688333e-05,
      "loss": 0.0003,
      "step": 4031
    },
    {
      "epoch": 1.4753018660812294,
      "grad_norm": 0.04416215419769287,
      "learning_rate": 1.3238552437223043e-05,
      "loss": 0.0006,
      "step": 4032
    },
    {
      "epoch": 1.4756677643615075,
      "grad_norm": 0.016438249498605728,
      "learning_rate": 1.3229320531757755e-05,
      "loss": 0.0005,
      "step": 4033
    },
    {
      "epoch": 1.4760336626417856,
      "grad_norm": 0.016711965203285217,
      "learning_rate": 1.3220088626292468e-05,
      "loss": 0.0005,
      "step": 4034
    },
    {
      "epoch": 1.4763995609220637,
      "grad_norm": 0.01224077120423317,
      "learning_rate": 1.321085672082718e-05,
      "loss": 0.0004,
      "step": 4035
    },
    {
      "epoch": 1.4767654592023418,
      "grad_norm": 0.013150466606020927,
      "learning_rate": 1.320162481536189e-05,
      "loss": 0.0004,
      "step": 4036
    },
    {
      "epoch": 1.47713135748262,
      "grad_norm": 0.011002597399055958,
      "learning_rate": 1.3192392909896603e-05,
      "loss": 0.0004,
      "step": 4037
    },
    {
      "epoch": 1.477497255762898,
      "grad_norm": 0.01436053216457367,
      "learning_rate": 1.3183161004431317e-05,
      "loss": 0.0005,
      "step": 4038
    },
    {
      "epoch": 1.477863154043176,
      "grad_norm": 0.01951855607330799,
      "learning_rate": 1.3173929098966029e-05,
      "loss": 0.0006,
      "step": 4039
    },
    {
      "epoch": 1.478229052323454,
      "grad_norm": 0.02391449734568596,
      "learning_rate": 1.3164697193500738e-05,
      "loss": 0.0006,
      "step": 4040
    },
    {
      "epoch": 1.4785949506037321,
      "grad_norm": 0.423060804605484,
      "learning_rate": 1.3155465288035452e-05,
      "loss": 0.0056,
      "step": 4041
    },
    {
      "epoch": 1.4789608488840102,
      "grad_norm": 0.024264562875032425,
      "learning_rate": 1.3146233382570164e-05,
      "loss": 0.0007,
      "step": 4042
    },
    {
      "epoch": 1.4793267471642884,
      "grad_norm": 16.752573013305664,
      "learning_rate": 1.3137001477104874e-05,
      "loss": 0.0948,
      "step": 4043
    },
    {
      "epoch": 1.4796926454445665,
      "grad_norm": 0.4409504234790802,
      "learning_rate": 1.3127769571639587e-05,
      "loss": 0.0017,
      "step": 4044
    },
    {
      "epoch": 1.4800585437248444,
      "grad_norm": 0.01628195121884346,
      "learning_rate": 1.3118537666174299e-05,
      "loss": 0.0006,
      "step": 4045
    },
    {
      "epoch": 1.4804244420051225,
      "grad_norm": 0.014977598562836647,
      "learning_rate": 1.3109305760709011e-05,
      "loss": 0.0005,
      "step": 4046
    },
    {
      "epoch": 1.4807903402854006,
      "grad_norm": 0.01095468271523714,
      "learning_rate": 1.3100073855243722e-05,
      "loss": 0.0004,
      "step": 4047
    },
    {
      "epoch": 1.4811562385656787,
      "grad_norm": 0.016695957630872726,
      "learning_rate": 1.3090841949778434e-05,
      "loss": 0.0004,
      "step": 4048
    },
    {
      "epoch": 1.4815221368459568,
      "grad_norm": 0.22029897570610046,
      "learning_rate": 1.3081610044313148e-05,
      "loss": 0.0007,
      "step": 4049
    },
    {
      "epoch": 1.481888035126235,
      "grad_norm": 0.00882063340395689,
      "learning_rate": 1.307237813884786e-05,
      "loss": 0.0003,
      "step": 4050
    },
    {
      "epoch": 1.482253933406513,
      "grad_norm": 0.01163645088672638,
      "learning_rate": 1.306314623338257e-05,
      "loss": 0.0004,
      "step": 4051
    },
    {
      "epoch": 1.4826198316867911,
      "grad_norm": 0.009244863875210285,
      "learning_rate": 1.3053914327917283e-05,
      "loss": 0.0004,
      "step": 4052
    },
    {
      "epoch": 1.4829857299670692,
      "grad_norm": 0.009741991758346558,
      "learning_rate": 1.3044682422451995e-05,
      "loss": 0.0004,
      "step": 4053
    },
    {
      "epoch": 1.4833516282473473,
      "grad_norm": 0.006087000481784344,
      "learning_rate": 1.3035450516986707e-05,
      "loss": 0.0003,
      "step": 4054
    },
    {
      "epoch": 1.4837175265276255,
      "grad_norm": 0.011317790485918522,
      "learning_rate": 1.3026218611521418e-05,
      "loss": 0.0005,
      "step": 4055
    },
    {
      "epoch": 1.4840834248079033,
      "grad_norm": 0.006071112118661404,
      "learning_rate": 1.301698670605613e-05,
      "loss": 0.0003,
      "step": 4056
    },
    {
      "epoch": 1.4844493230881814,
      "grad_norm": 0.008585026487708092,
      "learning_rate": 1.3007754800590844e-05,
      "loss": 0.0003,
      "step": 4057
    },
    {
      "epoch": 1.4848152213684596,
      "grad_norm": 0.005917617119848728,
      "learning_rate": 1.2998522895125553e-05,
      "loss": 0.0002,
      "step": 4058
    },
    {
      "epoch": 1.4851811196487377,
      "grad_norm": 0.007087185047566891,
      "learning_rate": 1.2989290989660265e-05,
      "loss": 0.0003,
      "step": 4059
    },
    {
      "epoch": 1.4855470179290158,
      "grad_norm": 0.006649922579526901,
      "learning_rate": 1.298005908419498e-05,
      "loss": 0.0003,
      "step": 4060
    },
    {
      "epoch": 1.4859129162092939,
      "grad_norm": 0.006873312406241894,
      "learning_rate": 1.2970827178729691e-05,
      "loss": 0.0003,
      "step": 4061
    },
    {
      "epoch": 1.4862788144895718,
      "grad_norm": 0.007801490370184183,
      "learning_rate": 1.2961595273264402e-05,
      "loss": 0.0003,
      "step": 4062
    },
    {
      "epoch": 1.4866447127698499,
      "grad_norm": 0.3873565196990967,
      "learning_rate": 1.2952363367799114e-05,
      "loss": 0.0044,
      "step": 4063
    },
    {
      "epoch": 1.487010611050128,
      "grad_norm": 0.03383038938045502,
      "learning_rate": 1.2943131462333826e-05,
      "loss": 0.0004,
      "step": 4064
    },
    {
      "epoch": 1.487376509330406,
      "grad_norm": 0.04976731166243553,
      "learning_rate": 1.2933899556868539e-05,
      "loss": 0.0004,
      "step": 4065
    },
    {
      "epoch": 1.4877424076106842,
      "grad_norm": 0.028861351311206818,
      "learning_rate": 1.292466765140325e-05,
      "loss": 0.0003,
      "step": 4066
    },
    {
      "epoch": 1.4881083058909623,
      "grad_norm": 0.004153182730078697,
      "learning_rate": 1.2915435745937962e-05,
      "loss": 0.0002,
      "step": 4067
    },
    {
      "epoch": 1.4884742041712404,
      "grad_norm": 0.008651182055473328,
      "learning_rate": 1.2906203840472675e-05,
      "loss": 0.0003,
      "step": 4068
    },
    {
      "epoch": 1.4888401024515185,
      "grad_norm": 0.008907508105039597,
      "learning_rate": 1.2896971935007384e-05,
      "loss": 0.0004,
      "step": 4069
    },
    {
      "epoch": 1.4892060007317967,
      "grad_norm": 0.007471663411706686,
      "learning_rate": 1.2887740029542098e-05,
      "loss": 0.0003,
      "step": 4070
    },
    {
      "epoch": 1.4895718990120748,
      "grad_norm": 0.00720871752128005,
      "learning_rate": 1.287850812407681e-05,
      "loss": 0.0003,
      "step": 4071
    },
    {
      "epoch": 1.4899377972923527,
      "grad_norm": 0.004582891706377268,
      "learning_rate": 1.2869276218611523e-05,
      "loss": 0.0002,
      "step": 4072
    },
    {
      "epoch": 1.4903036955726308,
      "grad_norm": 19.472122192382812,
      "learning_rate": 1.2860044313146233e-05,
      "loss": 0.4884,
      "step": 4073
    },
    {
      "epoch": 1.4906695938529089,
      "grad_norm": 0.0052816844545304775,
      "learning_rate": 1.2850812407680946e-05,
      "loss": 0.0002,
      "step": 4074
    },
    {
      "epoch": 1.491035492133187,
      "grad_norm": 0.014528868719935417,
      "learning_rate": 1.2841580502215658e-05,
      "loss": 0.0003,
      "step": 4075
    },
    {
      "epoch": 1.491401390413465,
      "grad_norm": 0.005661723669618368,
      "learning_rate": 1.2832348596750372e-05,
      "loss": 0.0003,
      "step": 4076
    },
    {
      "epoch": 1.4917672886937432,
      "grad_norm": 0.004942134488373995,
      "learning_rate": 1.282311669128508e-05,
      "loss": 0.0002,
      "step": 4077
    },
    {
      "epoch": 1.492133186974021,
      "grad_norm": 0.008330590091645718,
      "learning_rate": 1.2813884785819793e-05,
      "loss": 0.0003,
      "step": 4078
    },
    {
      "epoch": 1.4924990852542992,
      "grad_norm": 0.004515483044087887,
      "learning_rate": 1.2804652880354507e-05,
      "loss": 0.0002,
      "step": 4079
    },
    {
      "epoch": 1.4928649835345773,
      "grad_norm": 0.5530557632446289,
      "learning_rate": 1.2795420974889219e-05,
      "loss": 0.0078,
      "step": 4080
    },
    {
      "epoch": 1.4932308818148554,
      "grad_norm": 0.5011402368545532,
      "learning_rate": 1.278618906942393e-05,
      "loss": 0.0034,
      "step": 4081
    },
    {
      "epoch": 1.4935967800951335,
      "grad_norm": 0.00947494339197874,
      "learning_rate": 1.2776957163958642e-05,
      "loss": 0.0002,
      "step": 4082
    },
    {
      "epoch": 1.4939626783754116,
      "grad_norm": 0.0051788389682769775,
      "learning_rate": 1.2767725258493354e-05,
      "loss": 0.0002,
      "step": 4083
    },
    {
      "epoch": 1.4943285766556897,
      "grad_norm": 1.1262362003326416,
      "learning_rate": 1.2758493353028065e-05,
      "loss": 0.0132,
      "step": 4084
    },
    {
      "epoch": 1.4946944749359679,
      "grad_norm": 0.008075696416199207,
      "learning_rate": 1.2749261447562777e-05,
      "loss": 0.0003,
      "step": 4085
    },
    {
      "epoch": 1.495060373216246,
      "grad_norm": 0.00341073889285326,
      "learning_rate": 1.2740029542097489e-05,
      "loss": 0.0002,
      "step": 4086
    },
    {
      "epoch": 1.495426271496524,
      "grad_norm": 0.006022040266543627,
      "learning_rate": 1.2730797636632203e-05,
      "loss": 0.0002,
      "step": 4087
    },
    {
      "epoch": 1.4957921697768022,
      "grad_norm": 0.007270899135619402,
      "learning_rate": 1.2721565731166912e-05,
      "loss": 0.0003,
      "step": 4088
    },
    {
      "epoch": 1.49615806805708,
      "grad_norm": 0.0057664127089083195,
      "learning_rate": 1.2712333825701626e-05,
      "loss": 0.0003,
      "step": 4089
    },
    {
      "epoch": 1.4965239663373582,
      "grad_norm": 0.006337427068501711,
      "learning_rate": 1.2703101920236338e-05,
      "loss": 0.0003,
      "step": 4090
    },
    {
      "epoch": 1.4968898646176363,
      "grad_norm": 0.008483794517815113,
      "learning_rate": 1.269387001477105e-05,
      "loss": 0.0003,
      "step": 4091
    },
    {
      "epoch": 1.4972557628979144,
      "grad_norm": 3.8785924911499023,
      "learning_rate": 1.268463810930576e-05,
      "loss": 0.5694,
      "step": 4092
    },
    {
      "epoch": 1.4976216611781925,
      "grad_norm": 0.40543216466903687,
      "learning_rate": 1.2675406203840473e-05,
      "loss": 0.0023,
      "step": 4093
    },
    {
      "epoch": 1.4979875594584706,
      "grad_norm": 0.005519910715520382,
      "learning_rate": 1.2666174298375185e-05,
      "loss": 0.0003,
      "step": 4094
    },
    {
      "epoch": 1.4983534577387485,
      "grad_norm": 0.010195664130151272,
      "learning_rate": 1.26569423929099e-05,
      "loss": 0.0003,
      "step": 4095
    },
    {
      "epoch": 1.4987193560190266,
      "grad_norm": 0.006370835471898317,
      "learning_rate": 1.2647710487444608e-05,
      "loss": 0.0003,
      "step": 4096
    },
    {
      "epoch": 1.4990852542993047,
      "grad_norm": 0.007535924669355154,
      "learning_rate": 1.263847858197932e-05,
      "loss": 0.0002,
      "step": 4097
    },
    {
      "epoch": 1.4994511525795828,
      "grad_norm": 0.3105514645576477,
      "learning_rate": 1.2629246676514034e-05,
      "loss": 0.0043,
      "step": 4098
    },
    {
      "epoch": 1.499817050859861,
      "grad_norm": 0.00550704775378108,
      "learning_rate": 1.2620014771048743e-05,
      "loss": 0.0002,
      "step": 4099
    },
    {
      "epoch": 1.500182949140139,
      "grad_norm": 0.19786562025547028,
      "learning_rate": 1.2610782865583457e-05,
      "loss": 0.0011,
      "step": 4100
    },
    {
      "epoch": 1.5005488474204172,
      "grad_norm": 8.518290519714355,
      "learning_rate": 1.260155096011817e-05,
      "loss": 0.1691,
      "step": 4101
    },
    {
      "epoch": 1.5009147457006953,
      "grad_norm": 0.006310344208031893,
      "learning_rate": 1.2592319054652882e-05,
      "loss": 0.0003,
      "step": 4102
    },
    {
      "epoch": 1.5012806439809734,
      "grad_norm": 0.005832385737448931,
      "learning_rate": 1.2583087149187592e-05,
      "loss": 0.0003,
      "step": 4103
    },
    {
      "epoch": 1.5016465422612515,
      "grad_norm": 0.006942113861441612,
      "learning_rate": 1.2573855243722304e-05,
      "loss": 0.0003,
      "step": 4104
    },
    {
      "epoch": 1.5020124405415296,
      "grad_norm": 0.005675166845321655,
      "learning_rate": 1.2564623338257017e-05,
      "loss": 0.0003,
      "step": 4105
    },
    {
      "epoch": 1.5023783388218075,
      "grad_norm": 0.0056724101305007935,
      "learning_rate": 1.255539143279173e-05,
      "loss": 0.0003,
      "step": 4106
    },
    {
      "epoch": 1.5027442371020856,
      "grad_norm": 0.005400795955210924,
      "learning_rate": 1.254615952732644e-05,
      "loss": 0.0002,
      "step": 4107
    },
    {
      "epoch": 1.5031101353823637,
      "grad_norm": 6.566115856170654,
      "learning_rate": 1.2536927621861153e-05,
      "loss": 0.1784,
      "step": 4108
    },
    {
      "epoch": 1.5034760336626418,
      "grad_norm": 0.006318472325801849,
      "learning_rate": 1.2527695716395866e-05,
      "loss": 0.0003,
      "step": 4109
    },
    {
      "epoch": 1.5038419319429197,
      "grad_norm": 0.009025814011693,
      "learning_rate": 1.2518463810930578e-05,
      "loss": 0.0003,
      "step": 4110
    },
    {
      "epoch": 1.5042078302231978,
      "grad_norm": 0.00852295383810997,
      "learning_rate": 1.2509231905465288e-05,
      "loss": 0.0002,
      "step": 4111
    },
    {
      "epoch": 1.504573728503476,
      "grad_norm": 0.006994515657424927,
      "learning_rate": 1.25e-05,
      "loss": 0.0003,
      "step": 4112
    },
    {
      "epoch": 1.504939626783754,
      "grad_norm": 0.0072339302860200405,
      "learning_rate": 1.2490768094534711e-05,
      "loss": 0.0003,
      "step": 4113
    },
    {
      "epoch": 1.5053055250640321,
      "grad_norm": 2.060432195663452,
      "learning_rate": 1.2481536189069425e-05,
      "loss": 0.5013,
      "step": 4114
    },
    {
      "epoch": 1.5056714233443103,
      "grad_norm": 0.007652924861758947,
      "learning_rate": 1.2472304283604136e-05,
      "loss": 0.0003,
      "step": 4115
    },
    {
      "epoch": 1.5060373216245884,
      "grad_norm": 0.00558836292475462,
      "learning_rate": 1.2463072378138848e-05,
      "loss": 0.0003,
      "step": 4116
    },
    {
      "epoch": 1.5064032199048665,
      "grad_norm": 0.01022849790751934,
      "learning_rate": 1.245384047267356e-05,
      "loss": 0.0005,
      "step": 4117
    },
    {
      "epoch": 1.5067691181851446,
      "grad_norm": 0.017481600865721703,
      "learning_rate": 1.2444608567208272e-05,
      "loss": 0.0004,
      "step": 4118
    },
    {
      "epoch": 1.5071350164654227,
      "grad_norm": 0.031028522178530693,
      "learning_rate": 1.2435376661742985e-05,
      "loss": 0.0008,
      "step": 4119
    },
    {
      "epoch": 1.5075009147457008,
      "grad_norm": 0.009204420261085033,
      "learning_rate": 1.2426144756277697e-05,
      "loss": 0.0004,
      "step": 4120
    },
    {
      "epoch": 1.507866813025979,
      "grad_norm": 0.008147639222443104,
      "learning_rate": 1.2416912850812407e-05,
      "loss": 0.0004,
      "step": 4121
    },
    {
      "epoch": 1.508232711306257,
      "grad_norm": 1.9780819416046143,
      "learning_rate": 1.2407680945347121e-05,
      "loss": 0.0091,
      "step": 4122
    },
    {
      "epoch": 1.508598609586535,
      "grad_norm": 0.011534777469933033,
      "learning_rate": 1.2398449039881832e-05,
      "loss": 0.0005,
      "step": 4123
    },
    {
      "epoch": 1.508964507866813,
      "grad_norm": 0.009805556386709213,
      "learning_rate": 1.2389217134416544e-05,
      "loss": 0.0004,
      "step": 4124
    },
    {
      "epoch": 1.5093304061470911,
      "grad_norm": 0.01196227129548788,
      "learning_rate": 1.2379985228951256e-05,
      "loss": 0.0005,
      "step": 4125
    },
    {
      "epoch": 1.5096963044273692,
      "grad_norm": 1.949657678604126,
      "learning_rate": 1.2370753323485969e-05,
      "loss": 0.4733,
      "step": 4126
    },
    {
      "epoch": 1.5100622027076471,
      "grad_norm": 0.016394270583987236,
      "learning_rate": 1.2361521418020681e-05,
      "loss": 0.0006,
      "step": 4127
    },
    {
      "epoch": 1.5104281009879252,
      "grad_norm": 0.2951868772506714,
      "learning_rate": 1.2352289512555391e-05,
      "loss": 0.0011,
      "step": 4128
    },
    {
      "epoch": 1.5107939992682033,
      "grad_norm": 0.291135311126709,
      "learning_rate": 1.2343057607090104e-05,
      "loss": 0.0022,
      "step": 4129
    },
    {
      "epoch": 1.5111598975484815,
      "grad_norm": 0.018954655155539513,
      "learning_rate": 1.2333825701624816e-05,
      "loss": 0.0008,
      "step": 4130
    },
    {
      "epoch": 1.5115257958287596,
      "grad_norm": 2.2837772369384766,
      "learning_rate": 1.2324593796159528e-05,
      "loss": 0.5023,
      "step": 4131
    },
    {
      "epoch": 1.5118916941090377,
      "grad_norm": 0.017834356054663658,
      "learning_rate": 1.2315361890694239e-05,
      "loss": 0.0007,
      "step": 4132
    },
    {
      "epoch": 1.5122575923893158,
      "grad_norm": 9.231467247009277,
      "learning_rate": 1.2306129985228953e-05,
      "loss": 0.0939,
      "step": 4133
    },
    {
      "epoch": 1.512623490669594,
      "grad_norm": 0.02884552627801895,
      "learning_rate": 1.2296898079763663e-05,
      "loss": 0.0012,
      "step": 4134
    },
    {
      "epoch": 1.512989388949872,
      "grad_norm": 0.025607185438275337,
      "learning_rate": 1.2287666174298375e-05,
      "loss": 0.0011,
      "step": 4135
    },
    {
      "epoch": 1.5133552872301501,
      "grad_norm": 0.018650773912668228,
      "learning_rate": 1.2278434268833088e-05,
      "loss": 0.0009,
      "step": 4136
    },
    {
      "epoch": 1.5137211855104282,
      "grad_norm": 0.032731566578149796,
      "learning_rate": 1.22692023633678e-05,
      "loss": 0.0014,
      "step": 4137
    },
    {
      "epoch": 1.5140870837907063,
      "grad_norm": 0.02608663961291313,
      "learning_rate": 1.2259970457902512e-05,
      "loss": 0.0013,
      "step": 4138
    },
    {
      "epoch": 1.5144529820709842,
      "grad_norm": 0.03403254598379135,
      "learning_rate": 1.2250738552437224e-05,
      "loss": 0.0015,
      "step": 4139
    },
    {
      "epoch": 1.5148188803512623,
      "grad_norm": 0.053966324776411057,
      "learning_rate": 1.2241506646971935e-05,
      "loss": 0.0021,
      "step": 4140
    },
    {
      "epoch": 1.5151847786315404,
      "grad_norm": 0.02973151206970215,
      "learning_rate": 1.2232274741506647e-05,
      "loss": 0.0014,
      "step": 4141
    },
    {
      "epoch": 1.5155506769118186,
      "grad_norm": 0.029878560453653336,
      "learning_rate": 1.222304283604136e-05,
      "loss": 0.0015,
      "step": 4142
    },
    {
      "epoch": 1.5159165751920964,
      "grad_norm": 0.06901603192090988,
      "learning_rate": 1.2213810930576072e-05,
      "loss": 0.0014,
      "step": 4143
    },
    {
      "epoch": 1.5162824734723745,
      "grad_norm": 0.45763126015663147,
      "learning_rate": 1.2204579025110784e-05,
      "loss": 0.0065,
      "step": 4144
    },
    {
      "epoch": 1.5166483717526527,
      "grad_norm": 0.025101467967033386,
      "learning_rate": 1.2195347119645494e-05,
      "loss": 0.0012,
      "step": 4145
    },
    {
      "epoch": 1.5170142700329308,
      "grad_norm": 0.02907131239771843,
      "learning_rate": 1.2186115214180208e-05,
      "loss": 0.0015,
      "step": 4146
    },
    {
      "epoch": 1.5173801683132089,
      "grad_norm": 0.06582856923341751,
      "learning_rate": 1.2176883308714919e-05,
      "loss": 0.0019,
      "step": 4147
    },
    {
      "epoch": 1.517746066593487,
      "grad_norm": 0.04410015791654587,
      "learning_rate": 1.2167651403249631e-05,
      "loss": 0.0013,
      "step": 4148
    },
    {
      "epoch": 1.518111964873765,
      "grad_norm": 0.024688785895705223,
      "learning_rate": 1.2158419497784343e-05,
      "loss": 0.0012,
      "step": 4149
    },
    {
      "epoch": 1.5184778631540432,
      "grad_norm": 0.026860520243644714,
      "learning_rate": 1.2149187592319056e-05,
      "loss": 0.0014,
      "step": 4150
    },
    {
      "epoch": 1.5188437614343213,
      "grad_norm": 0.9214026927947998,
      "learning_rate": 1.2139955686853766e-05,
      "loss": 0.0036,
      "step": 4151
    },
    {
      "epoch": 1.5192096597145994,
      "grad_norm": 0.024360457435250282,
      "learning_rate": 1.213072378138848e-05,
      "loss": 0.0012,
      "step": 4152
    },
    {
      "epoch": 1.5195755579948775,
      "grad_norm": 0.04207649827003479,
      "learning_rate": 1.212149187592319e-05,
      "loss": 0.0016,
      "step": 4153
    },
    {
      "epoch": 1.5199414562751556,
      "grad_norm": 0.024578111246228218,
      "learning_rate": 1.2112259970457903e-05,
      "loss": 0.0011,
      "step": 4154
    },
    {
      "epoch": 1.5203073545554338,
      "grad_norm": 1.186159610748291,
      "learning_rate": 1.2103028064992615e-05,
      "loss": 0.0089,
      "step": 4155
    },
    {
      "epoch": 1.5206732528357116,
      "grad_norm": 0.025355331599712372,
      "learning_rate": 1.2093796159527326e-05,
      "loss": 0.0012,
      "step": 4156
    },
    {
      "epoch": 1.5210391511159898,
      "grad_norm": 0.027537334710359573,
      "learning_rate": 1.208456425406204e-05,
      "loss": 0.0012,
      "step": 4157
    },
    {
      "epoch": 1.5214050493962679,
      "grad_norm": 0.01840866729617119,
      "learning_rate": 1.207533234859675e-05,
      "loss": 0.0009,
      "step": 4158
    },
    {
      "epoch": 1.521770947676546,
      "grad_norm": 0.034658245742321014,
      "learning_rate": 1.2066100443131462e-05,
      "loss": 0.0014,
      "step": 4159
    },
    {
      "epoch": 1.5221368459568239,
      "grad_norm": 0.017184440046548843,
      "learning_rate": 1.2056868537666175e-05,
      "loss": 0.0008,
      "step": 4160
    },
    {
      "epoch": 1.522502744237102,
      "grad_norm": 24.508968353271484,
      "learning_rate": 1.2047636632200887e-05,
      "loss": 0.0861,
      "step": 4161
    },
    {
      "epoch": 1.52286864251738,
      "grad_norm": 0.019967524334788322,
      "learning_rate": 1.20384047267356e-05,
      "loss": 0.001,
      "step": 4162
    },
    {
      "epoch": 1.5232345407976582,
      "grad_norm": 0.016429394483566284,
      "learning_rate": 1.2029172821270311e-05,
      "loss": 0.0008,
      "step": 4163
    },
    {
      "epoch": 1.5236004390779363,
      "grad_norm": 0.030657611787319183,
      "learning_rate": 1.2019940915805022e-05,
      "loss": 0.0011,
      "step": 4164
    },
    {
      "epoch": 1.5239663373582144,
      "grad_norm": 0.018099453300237656,
      "learning_rate": 1.2010709010339736e-05,
      "loss": 0.0009,
      "step": 4165
    },
    {
      "epoch": 1.5243322356384925,
      "grad_norm": 0.018163349479436874,
      "learning_rate": 1.2001477104874446e-05,
      "loss": 0.0009,
      "step": 4166
    },
    {
      "epoch": 1.5246981339187706,
      "grad_norm": 23.59604835510254,
      "learning_rate": 1.1992245199409159e-05,
      "loss": 0.3197,
      "step": 4167
    },
    {
      "epoch": 1.5250640321990487,
      "grad_norm": 0.018667615950107574,
      "learning_rate": 1.1983013293943871e-05,
      "loss": 0.0008,
      "step": 4168
    },
    {
      "epoch": 1.5254299304793268,
      "grad_norm": 0.3163060247898102,
      "learning_rate": 1.1973781388478583e-05,
      "loss": 0.0035,
      "step": 4169
    },
    {
      "epoch": 1.525795828759605,
      "grad_norm": 0.021852584555745125,
      "learning_rate": 1.1964549483013294e-05,
      "loss": 0.0009,
      "step": 4170
    },
    {
      "epoch": 1.526161727039883,
      "grad_norm": 0.01592862978577614,
      "learning_rate": 1.1955317577548006e-05,
      "loss": 0.0008,
      "step": 4171
    },
    {
      "epoch": 1.526527625320161,
      "grad_norm": 0.015683064237236977,
      "learning_rate": 1.1946085672082718e-05,
      "loss": 0.0007,
      "step": 4172
    },
    {
      "epoch": 1.526893523600439,
      "grad_norm": 0.01684763841331005,
      "learning_rate": 1.193685376661743e-05,
      "loss": 0.0007,
      "step": 4173
    },
    {
      "epoch": 1.5272594218807172,
      "grad_norm": 0.02224970981478691,
      "learning_rate": 1.1927621861152143e-05,
      "loss": 0.0011,
      "step": 4174
    },
    {
      "epoch": 1.5276253201609953,
      "grad_norm": 0.015960045158863068,
      "learning_rate": 1.1918389955686853e-05,
      "loss": 0.0008,
      "step": 4175
    },
    {
      "epoch": 1.5279912184412732,
      "grad_norm": 0.014054426923394203,
      "learning_rate": 1.1909158050221567e-05,
      "loss": 0.0006,
      "step": 4176
    },
    {
      "epoch": 1.5283571167215513,
      "grad_norm": 0.012393604032695293,
      "learning_rate": 1.1899926144756278e-05,
      "loss": 0.0006,
      "step": 4177
    },
    {
      "epoch": 1.5287230150018294,
      "grad_norm": 0.011261655017733574,
      "learning_rate": 1.189069423929099e-05,
      "loss": 0.0005,
      "step": 4178
    },
    {
      "epoch": 1.5290889132821075,
      "grad_norm": 0.011908812448382378,
      "learning_rate": 1.1881462333825702e-05,
      "loss": 0.0006,
      "step": 4179
    },
    {
      "epoch": 1.5294548115623856,
      "grad_norm": 0.835659384727478,
      "learning_rate": 1.1872230428360414e-05,
      "loss": 0.0026,
      "step": 4180
    },
    {
      "epoch": 1.5298207098426637,
      "grad_norm": 0.015431888401508331,
      "learning_rate": 1.1862998522895127e-05,
      "loss": 0.0007,
      "step": 4181
    },
    {
      "epoch": 1.5301866081229418,
      "grad_norm": 0.01238157693296671,
      "learning_rate": 1.1853766617429839e-05,
      "loss": 0.0006,
      "step": 4182
    },
    {
      "epoch": 1.53055250640322,
      "grad_norm": 0.011478048749268055,
      "learning_rate": 1.184453471196455e-05,
      "loss": 0.0006,
      "step": 4183
    },
    {
      "epoch": 1.530918404683498,
      "grad_norm": 0.012292876839637756,
      "learning_rate": 1.1835302806499263e-05,
      "loss": 0.0006,
      "step": 4184
    },
    {
      "epoch": 1.5312843029637762,
      "grad_norm": 0.012991370633244514,
      "learning_rate": 1.1826070901033974e-05,
      "loss": 0.0006,
      "step": 4185
    },
    {
      "epoch": 1.5316502012440543,
      "grad_norm": 0.3256487548351288,
      "learning_rate": 1.1816838995568685e-05,
      "loss": 0.0029,
      "step": 4186
    },
    {
      "epoch": 1.5320160995243324,
      "grad_norm": 0.020694145932793617,
      "learning_rate": 1.1807607090103399e-05,
      "loss": 0.0008,
      "step": 4187
    },
    {
      "epoch": 1.5323819978046103,
      "grad_norm": 0.00999025534838438,
      "learning_rate": 1.1798375184638109e-05,
      "loss": 0.0005,
      "step": 4188
    },
    {
      "epoch": 1.5327478960848884,
      "grad_norm": 0.011175127699971199,
      "learning_rate": 1.1789143279172821e-05,
      "loss": 0.0005,
      "step": 4189
    },
    {
      "epoch": 1.5331137943651665,
      "grad_norm": 0.011808721348643303,
      "learning_rate": 1.1779911373707534e-05,
      "loss": 0.0006,
      "step": 4190
    },
    {
      "epoch": 1.5334796926454446,
      "grad_norm": 0.01025282870978117,
      "learning_rate": 1.1770679468242246e-05,
      "loss": 0.0005,
      "step": 4191
    },
    {
      "epoch": 1.5338455909257227,
      "grad_norm": 0.011670083738863468,
      "learning_rate": 1.1761447562776958e-05,
      "loss": 0.0005,
      "step": 4192
    },
    {
      "epoch": 1.5342114892060006,
      "grad_norm": 0.017352033406496048,
      "learning_rate": 1.175221565731167e-05,
      "loss": 0.0007,
      "step": 4193
    },
    {
      "epoch": 1.5345773874862787,
      "grad_norm": 0.011579904705286026,
      "learning_rate": 1.174298375184638e-05,
      "loss": 0.0005,
      "step": 4194
    },
    {
      "epoch": 1.5349432857665568,
      "grad_norm": 0.008238845504820347,
      "learning_rate": 1.1733751846381095e-05,
      "loss": 0.0004,
      "step": 4195
    },
    {
      "epoch": 1.535309184046835,
      "grad_norm": 12.339556694030762,
      "learning_rate": 1.1724519940915805e-05,
      "loss": 0.3212,
      "step": 4196
    },
    {
      "epoch": 1.535675082327113,
      "grad_norm": 0.011021404527127743,
      "learning_rate": 1.1715288035450518e-05,
      "loss": 0.0005,
      "step": 4197
    },
    {
      "epoch": 1.5360409806073911,
      "grad_norm": 0.011344093829393387,
      "learning_rate": 1.170605612998523e-05,
      "loss": 0.0005,
      "step": 4198
    },
    {
      "epoch": 1.5364068788876692,
      "grad_norm": 0.009537976235151291,
      "learning_rate": 1.169682422451994e-05,
      "loss": 0.0004,
      "step": 4199
    },
    {
      "epoch": 1.5367727771679474,
      "grad_norm": 3.255119562149048,
      "learning_rate": 1.1687592319054654e-05,
      "loss": 0.0065,
      "step": 4200
    },
    {
      "epoch": 1.5371386754482255,
      "grad_norm": 0.00803777389228344,
      "learning_rate": 1.1678360413589365e-05,
      "loss": 0.0004,
      "step": 4201
    },
    {
      "epoch": 1.5375045737285036,
      "grad_norm": 0.012356054969131947,
      "learning_rate": 1.1669128508124077e-05,
      "loss": 0.0006,
      "step": 4202
    },
    {
      "epoch": 1.5378704720087817,
      "grad_norm": 0.032926175743341446,
      "learning_rate": 1.165989660265879e-05,
      "loss": 0.0007,
      "step": 4203
    },
    {
      "epoch": 1.5382363702890598,
      "grad_norm": 0.01003547664731741,
      "learning_rate": 1.1650664697193502e-05,
      "loss": 0.0005,
      "step": 4204
    },
    {
      "epoch": 1.5386022685693377,
      "grad_norm": 0.008241957984864712,
      "learning_rate": 1.1641432791728212e-05,
      "loss": 0.0004,
      "step": 4205
    },
    {
      "epoch": 1.5389681668496158,
      "grad_norm": 0.00957463402301073,
      "learning_rate": 1.1632200886262926e-05,
      "loss": 0.0004,
      "step": 4206
    },
    {
      "epoch": 1.539334065129894,
      "grad_norm": 5.783638000488281,
      "learning_rate": 1.1622968980797637e-05,
      "loss": 0.173,
      "step": 4207
    },
    {
      "epoch": 1.539699963410172,
      "grad_norm": 0.013450651429593563,
      "learning_rate": 1.1613737075332349e-05,
      "loss": 0.0006,
      "step": 4208
    },
    {
      "epoch": 1.54006586169045,
      "grad_norm": 0.01488453708589077,
      "learning_rate": 1.1604505169867061e-05,
      "loss": 0.0005,
      "step": 4209
    },
    {
      "epoch": 1.540431759970728,
      "grad_norm": 0.01084329653531313,
      "learning_rate": 1.1595273264401773e-05,
      "loss": 0.0005,
      "step": 4210
    },
    {
      "epoch": 1.5407976582510061,
      "grad_norm": 0.0075876605696976185,
      "learning_rate": 1.1586041358936486e-05,
      "loss": 0.0003,
      "step": 4211
    },
    {
      "epoch": 1.5411635565312842,
      "grad_norm": 0.007607039995491505,
      "learning_rate": 1.1576809453471198e-05,
      "loss": 0.0003,
      "step": 4212
    },
    {
      "epoch": 1.5415294548115623,
      "grad_norm": 0.008688348345458508,
      "learning_rate": 1.1567577548005908e-05,
      "loss": 0.0004,
      "step": 4213
    },
    {
      "epoch": 1.5418953530918404,
      "grad_norm": 0.019376404583454132,
      "learning_rate": 1.155834564254062e-05,
      "loss": 0.0007,
      "step": 4214
    },
    {
      "epoch": 1.5422612513721186,
      "grad_norm": 0.23779122531414032,
      "learning_rate": 1.1549113737075333e-05,
      "loss": 0.0053,
      "step": 4215
    },
    {
      "epoch": 1.5426271496523967,
      "grad_norm": 0.14023597538471222,
      "learning_rate": 1.1539881831610045e-05,
      "loss": 0.001,
      "step": 4216
    },
    {
      "epoch": 1.5429930479326748,
      "grad_norm": 0.02228502184152603,
      "learning_rate": 1.1530649926144757e-05,
      "loss": 0.0006,
      "step": 4217
    },
    {
      "epoch": 1.5433589462129529,
      "grad_norm": 0.12156853079795837,
      "learning_rate": 1.1521418020679468e-05,
      "loss": 0.0011,
      "step": 4218
    },
    {
      "epoch": 1.543724844493231,
      "grad_norm": 0.008330394513905048,
      "learning_rate": 1.1512186115214182e-05,
      "loss": 0.0004,
      "step": 4219
    },
    {
      "epoch": 1.544090742773509,
      "grad_norm": 0.008115258999168873,
      "learning_rate": 1.1502954209748892e-05,
      "loss": 0.0004,
      "step": 4220
    },
    {
      "epoch": 1.544456641053787,
      "grad_norm": 0.00919254869222641,
      "learning_rate": 1.1493722304283605e-05,
      "loss": 0.0004,
      "step": 4221
    },
    {
      "epoch": 1.544822539334065,
      "grad_norm": 0.011681226082146168,
      "learning_rate": 1.1484490398818317e-05,
      "loss": 0.0005,
      "step": 4222
    },
    {
      "epoch": 1.5451884376143432,
      "grad_norm": 15.470967292785645,
      "learning_rate": 1.1475258493353029e-05,
      "loss": 0.2893,
      "step": 4223
    },
    {
      "epoch": 1.5455543358946213,
      "grad_norm": 0.010101084597408772,
      "learning_rate": 1.146602658788774e-05,
      "loss": 0.0005,
      "step": 4224
    },
    {
      "epoch": 1.5459202341748994,
      "grad_norm": 0.05883502587676048,
      "learning_rate": 1.1456794682422454e-05,
      "loss": 0.0008,
      "step": 4225
    },
    {
      "epoch": 1.5462861324551773,
      "grad_norm": 0.013879473321139812,
      "learning_rate": 1.1447562776957164e-05,
      "loss": 0.0006,
      "step": 4226
    },
    {
      "epoch": 1.5466520307354554,
      "grad_norm": 0.15318000316619873,
      "learning_rate": 1.1438330871491876e-05,
      "loss": 0.0012,
      "step": 4227
    },
    {
      "epoch": 1.5470179290157335,
      "grad_norm": 0.013172488659620285,
      "learning_rate": 1.1429098966026589e-05,
      "loss": 0.0006,
      "step": 4228
    },
    {
      "epoch": 1.5473838272960116,
      "grad_norm": 0.008628927171230316,
      "learning_rate": 1.14198670605613e-05,
      "loss": 0.0004,
      "step": 4229
    },
    {
      "epoch": 1.5477497255762898,
      "grad_norm": 0.007501145359128714,
      "learning_rate": 1.1410635155096013e-05,
      "loss": 0.0004,
      "step": 4230
    },
    {
      "epoch": 1.5481156238565679,
      "grad_norm": 0.020321523770689964,
      "learning_rate": 1.1401403249630724e-05,
      "loss": 0.0005,
      "step": 4231
    },
    {
      "epoch": 1.548481522136846,
      "grad_norm": 0.008191731758415699,
      "learning_rate": 1.1392171344165436e-05,
      "loss": 0.0003,
      "step": 4232
    },
    {
      "epoch": 1.548847420417124,
      "grad_norm": 0.011581468395888805,
      "learning_rate": 1.1382939438700148e-05,
      "loss": 0.0005,
      "step": 4233
    },
    {
      "epoch": 1.5492133186974022,
      "grad_norm": 0.1193598136305809,
      "learning_rate": 1.137370753323486e-05,
      "loss": 0.0013,
      "step": 4234
    },
    {
      "epoch": 1.5495792169776803,
      "grad_norm": 0.007257279474288225,
      "learning_rate": 1.1364475627769573e-05,
      "loss": 0.0003,
      "step": 4235
    },
    {
      "epoch": 1.5499451152579584,
      "grad_norm": 0.011282816529273987,
      "learning_rate": 1.1355243722304285e-05,
      "loss": 0.0005,
      "step": 4236
    },
    {
      "epoch": 1.5503110135382365,
      "grad_norm": 0.009161350317299366,
      "learning_rate": 1.1346011816838995e-05,
      "loss": 0.0004,
      "step": 4237
    },
    {
      "epoch": 1.5506769118185144,
      "grad_norm": 4.51605749130249,
      "learning_rate": 1.133677991137371e-05,
      "loss": 0.1391,
      "step": 4238
    },
    {
      "epoch": 1.5510428100987925,
      "grad_norm": 0.0100539680570364,
      "learning_rate": 1.132754800590842e-05,
      "loss": 0.0004,
      "step": 4239
    },
    {
      "epoch": 1.5514087083790706,
      "grad_norm": 0.010599909350275993,
      "learning_rate": 1.1318316100443132e-05,
      "loss": 0.0005,
      "step": 4240
    },
    {
      "epoch": 1.5517746066593487,
      "grad_norm": 0.11632754653692245,
      "learning_rate": 1.1309084194977844e-05,
      "loss": 0.001,
      "step": 4241
    },
    {
      "epoch": 1.5521405049396266,
      "grad_norm": 0.01466225367039442,
      "learning_rate": 1.1299852289512555e-05,
      "loss": 0.0006,
      "step": 4242
    },
    {
      "epoch": 1.5525064032199047,
      "grad_norm": 0.01072614174336195,
      "learning_rate": 1.1290620384047267e-05,
      "loss": 0.0004,
      "step": 4243
    },
    {
      "epoch": 1.5528723015001829,
      "grad_norm": 0.007335635833442211,
      "learning_rate": 1.128138847858198e-05,
      "loss": 0.0003,
      "step": 4244
    },
    {
      "epoch": 1.553238199780461,
      "grad_norm": 0.28219231963157654,
      "learning_rate": 1.1272156573116692e-05,
      "loss": 0.0015,
      "step": 4245
    },
    {
      "epoch": 1.553604098060739,
      "grad_norm": 0.01077958568930626,
      "learning_rate": 1.1262924667651404e-05,
      "loss": 0.0005,
      "step": 4246
    },
    {
      "epoch": 1.5539699963410172,
      "grad_norm": 0.417696475982666,
      "learning_rate": 1.1253692762186116e-05,
      "loss": 0.0024,
      "step": 4247
    },
    {
      "epoch": 1.5543358946212953,
      "grad_norm": 0.010790879838168621,
      "learning_rate": 1.1244460856720827e-05,
      "loss": 0.0004,
      "step": 4248
    },
    {
      "epoch": 1.5547017929015734,
      "grad_norm": 0.006550358142703772,
      "learning_rate": 1.123522895125554e-05,
      "loss": 0.0003,
      "step": 4249
    },
    {
      "epoch": 1.5550676911818515,
      "grad_norm": 0.0076394532807171345,
      "learning_rate": 1.1225997045790251e-05,
      "loss": 0.0004,
      "step": 4250
    },
    {
      "epoch": 1.5554335894621296,
      "grad_norm": 0.006504204124212265,
      "learning_rate": 1.1216765140324963e-05,
      "loss": 0.0003,
      "step": 4251
    },
    {
      "epoch": 1.5557994877424077,
      "grad_norm": 0.007811976596713066,
      "learning_rate": 1.1207533234859676e-05,
      "loss": 0.0003,
      "step": 4252
    },
    {
      "epoch": 1.5561653860226858,
      "grad_norm": 0.007002044003456831,
      "learning_rate": 1.1198301329394388e-05,
      "loss": 0.0003,
      "step": 4253
    },
    {
      "epoch": 1.5565312843029637,
      "grad_norm": 0.010772104375064373,
      "learning_rate": 1.11890694239291e-05,
      "loss": 0.0004,
      "step": 4254
    },
    {
      "epoch": 1.5568971825832418,
      "grad_norm": 0.006836721207946539,
      "learning_rate": 1.1179837518463812e-05,
      "loss": 0.0003,
      "step": 4255
    },
    {
      "epoch": 1.55726308086352,
      "grad_norm": 0.007320896256715059,
      "learning_rate": 1.1170605612998523e-05,
      "loss": 0.0003,
      "step": 4256
    },
    {
      "epoch": 1.557628979143798,
      "grad_norm": 0.007968125864863396,
      "learning_rate": 1.1161373707533235e-05,
      "loss": 0.0004,
      "step": 4257
    },
    {
      "epoch": 1.557994877424076,
      "grad_norm": 0.2609386444091797,
      "learning_rate": 1.1152141802067947e-05,
      "loss": 0.0061,
      "step": 4258
    },
    {
      "epoch": 1.558360775704354,
      "grad_norm": 0.007521696388721466,
      "learning_rate": 1.1142909896602658e-05,
      "loss": 0.0003,
      "step": 4259
    },
    {
      "epoch": 1.5587266739846322,
      "grad_norm": 0.006479410920292139,
      "learning_rate": 1.1133677991137372e-05,
      "loss": 0.0003,
      "step": 4260
    },
    {
      "epoch": 1.5590925722649103,
      "grad_norm": 2.8100638389587402,
      "learning_rate": 1.1124446085672082e-05,
      "loss": 0.0162,
      "step": 4261
    },
    {
      "epoch": 1.5594584705451884,
      "grad_norm": 0.009766236878931522,
      "learning_rate": 1.1115214180206795e-05,
      "loss": 0.0004,
      "step": 4262
    },
    {
      "epoch": 1.5598243688254665,
      "grad_norm": 0.008163949474692345,
      "learning_rate": 1.1105982274741507e-05,
      "loss": 0.0004,
      "step": 4263
    },
    {
      "epoch": 1.5601902671057446,
      "grad_norm": 0.00864297803491354,
      "learning_rate": 1.109675036927622e-05,
      "loss": 0.0004,
      "step": 4264
    },
    {
      "epoch": 1.5605561653860227,
      "grad_norm": 0.008528376929461956,
      "learning_rate": 1.1087518463810931e-05,
      "loss": 0.0004,
      "step": 4265
    },
    {
      "epoch": 1.5609220636663008,
      "grad_norm": 0.00961605366319418,
      "learning_rate": 1.1078286558345644e-05,
      "loss": 0.0004,
      "step": 4266
    },
    {
      "epoch": 1.561287961946579,
      "grad_norm": 0.0073863849975168705,
      "learning_rate": 1.1069054652880354e-05,
      "loss": 0.0003,
      "step": 4267
    },
    {
      "epoch": 1.561653860226857,
      "grad_norm": 0.006379057187587023,
      "learning_rate": 1.1059822747415068e-05,
      "loss": 0.0003,
      "step": 4268
    },
    {
      "epoch": 1.5620197585071351,
      "grad_norm": 0.007590513676404953,
      "learning_rate": 1.1050590841949779e-05,
      "loss": 0.0003,
      "step": 4269
    },
    {
      "epoch": 1.5623856567874133,
      "grad_norm": 0.009180577471852303,
      "learning_rate": 1.1041358936484491e-05,
      "loss": 0.0004,
      "step": 4270
    },
    {
      "epoch": 1.5627515550676911,
      "grad_norm": 0.19803494215011597,
      "learning_rate": 1.1032127031019203e-05,
      "loss": 0.0008,
      "step": 4271
    },
    {
      "epoch": 1.5631174533479693,
      "grad_norm": 0.007785739377140999,
      "learning_rate": 1.1022895125553914e-05,
      "loss": 0.0003,
      "step": 4272
    },
    {
      "epoch": 1.5634833516282474,
      "grad_norm": 0.006774968467652798,
      "learning_rate": 1.1013663220088628e-05,
      "loss": 0.0003,
      "step": 4273
    },
    {
      "epoch": 1.5638492499085255,
      "grad_norm": 0.0050655254162848,
      "learning_rate": 1.1004431314623338e-05,
      "loss": 0.0002,
      "step": 4274
    },
    {
      "epoch": 1.5642151481888034,
      "grad_norm": 0.0069184754975140095,
      "learning_rate": 1.099519940915805e-05,
      "loss": 0.0003,
      "step": 4275
    },
    {
      "epoch": 1.5645810464690815,
      "grad_norm": 118.92799377441406,
      "learning_rate": 1.0985967503692763e-05,
      "loss": 0.0944,
      "step": 4276
    },
    {
      "epoch": 1.5649469447493596,
      "grad_norm": 0.12640751898288727,
      "learning_rate": 1.0976735598227475e-05,
      "loss": 0.0011,
      "step": 4277
    },
    {
      "epoch": 1.5653128430296377,
      "grad_norm": 0.00711703859269619,
      "learning_rate": 1.0967503692762186e-05,
      "loss": 0.0003,
      "step": 4278
    },
    {
      "epoch": 1.5656787413099158,
      "grad_norm": 0.007869500666856766,
      "learning_rate": 1.09582717872969e-05,
      "loss": 0.0003,
      "step": 4279
    },
    {
      "epoch": 1.566044639590194,
      "grad_norm": 0.006346103269606829,
      "learning_rate": 1.094903988183161e-05,
      "loss": 0.0003,
      "step": 4280
    },
    {
      "epoch": 1.566410537870472,
      "grad_norm": 7.133592128753662,
      "learning_rate": 1.0939807976366322e-05,
      "loss": 0.1173,
      "step": 4281
    },
    {
      "epoch": 1.5667764361507501,
      "grad_norm": 0.007481199223548174,
      "learning_rate": 1.0930576070901034e-05,
      "loss": 0.0003,
      "step": 4282
    },
    {
      "epoch": 1.5671423344310282,
      "grad_norm": 0.008243652991950512,
      "learning_rate": 1.0921344165435747e-05,
      "loss": 0.0004,
      "step": 4283
    },
    {
      "epoch": 1.5675082327113063,
      "grad_norm": 0.005795022938400507,
      "learning_rate": 1.0912112259970459e-05,
      "loss": 0.0003,
      "step": 4284
    },
    {
      "epoch": 1.5678741309915845,
      "grad_norm": 0.0060423873364925385,
      "learning_rate": 1.090288035450517e-05,
      "loss": 0.0003,
      "step": 4285
    },
    {
      "epoch": 1.5682400292718626,
      "grad_norm": 0.03285020962357521,
      "learning_rate": 1.0893648449039882e-05,
      "loss": 0.0005,
      "step": 4286
    },
    {
      "epoch": 1.5686059275521405,
      "grad_norm": 0.005292508751153946,
      "learning_rate": 1.0884416543574594e-05,
      "loss": 0.0002,
      "step": 4287
    },
    {
      "epoch": 1.5689718258324186,
      "grad_norm": 0.006903129164129496,
      "learning_rate": 1.0875184638109306e-05,
      "loss": 0.0003,
      "step": 4288
    },
    {
      "epoch": 1.5693377241126967,
      "grad_norm": 0.008401859551668167,
      "learning_rate": 1.0865952732644019e-05,
      "loss": 0.0003,
      "step": 4289
    },
    {
      "epoch": 1.5697036223929748,
      "grad_norm": 0.008255522698163986,
      "learning_rate": 1.085672082717873e-05,
      "loss": 0.0003,
      "step": 4290
    },
    {
      "epoch": 1.5700695206732527,
      "grad_norm": 0.005895090755075216,
      "learning_rate": 1.0847488921713441e-05,
      "loss": 0.0003,
      "step": 4291
    },
    {
      "epoch": 1.5704354189535308,
      "grad_norm": 0.006948414258658886,
      "learning_rate": 1.0838257016248155e-05,
      "loss": 0.0003,
      "step": 4292
    },
    {
      "epoch": 1.570801317233809,
      "grad_norm": 0.007088435348123312,
      "learning_rate": 1.0829025110782866e-05,
      "loss": 0.0003,
      "step": 4293
    },
    {
      "epoch": 1.571167215514087,
      "grad_norm": 0.004950673319399357,
      "learning_rate": 1.0819793205317578e-05,
      "loss": 0.0002,
      "step": 4294
    },
    {
      "epoch": 1.571533113794365,
      "grad_norm": 0.009465060196816921,
      "learning_rate": 1.081056129985229e-05,
      "loss": 0.0004,
      "step": 4295
    },
    {
      "epoch": 1.5718990120746432,
      "grad_norm": 0.014333104714751244,
      "learning_rate": 1.0801329394387003e-05,
      "loss": 0.0005,
      "step": 4296
    },
    {
      "epoch": 1.5722649103549213,
      "grad_norm": 16.387531280517578,
      "learning_rate": 1.0792097488921713e-05,
      "loss": 0.1928,
      "step": 4297
    },
    {
      "epoch": 1.5726308086351994,
      "grad_norm": 0.01927109621465206,
      "learning_rate": 1.0782865583456427e-05,
      "loss": 0.0005,
      "step": 4298
    },
    {
      "epoch": 1.5729967069154775,
      "grad_norm": 0.007577107287943363,
      "learning_rate": 1.0773633677991138e-05,
      "loss": 0.0003,
      "step": 4299
    },
    {
      "epoch": 1.5733626051957557,
      "grad_norm": 3.111433506011963,
      "learning_rate": 1.076440177252585e-05,
      "loss": 0.1239,
      "step": 4300
    },
    {
      "epoch": 1.5737285034760338,
      "grad_norm": 0.006041236687451601,
      "learning_rate": 1.0755169867060562e-05,
      "loss": 0.0003,
      "step": 4301
    },
    {
      "epoch": 1.5740944017563119,
      "grad_norm": 0.005879019852727652,
      "learning_rate": 1.0745937961595273e-05,
      "loss": 0.0003,
      "step": 4302
    },
    {
      "epoch": 1.57446030003659,
      "grad_norm": 0.03885069489479065,
      "learning_rate": 1.0736706056129987e-05,
      "loss": 0.0005,
      "step": 4303
    },
    {
      "epoch": 1.5748261983168679,
      "grad_norm": 0.009503970853984356,
      "learning_rate": 1.0727474150664697e-05,
      "loss": 0.0004,
      "step": 4304
    },
    {
      "epoch": 1.575192096597146,
      "grad_norm": 0.1656082272529602,
      "learning_rate": 1.071824224519941e-05,
      "loss": 0.0014,
      "step": 4305
    },
    {
      "epoch": 1.575557994877424,
      "grad_norm": 0.006995531264692545,
      "learning_rate": 1.0709010339734122e-05,
      "loss": 0.0003,
      "step": 4306
    },
    {
      "epoch": 1.5759238931577022,
      "grad_norm": 0.006300643086433411,
      "learning_rate": 1.0699778434268834e-05,
      "loss": 0.0003,
      "step": 4307
    },
    {
      "epoch": 1.57628979143798,
      "grad_norm": 0.07675051689147949,
      "learning_rate": 1.0690546528803546e-05,
      "loss": 0.0006,
      "step": 4308
    },
    {
      "epoch": 1.5766556897182582,
      "grad_norm": 0.003760649124160409,
      "learning_rate": 1.0681314623338258e-05,
      "loss": 0.0002,
      "step": 4309
    },
    {
      "epoch": 1.5770215879985363,
      "grad_norm": 0.006570950150489807,
      "learning_rate": 1.0672082717872969e-05,
      "loss": 0.0003,
      "step": 4310
    },
    {
      "epoch": 1.5773874862788144,
      "grad_norm": 17.094308853149414,
      "learning_rate": 1.0662850812407683e-05,
      "loss": 0.097,
      "step": 4311
    },
    {
      "epoch": 1.5777533845590925,
      "grad_norm": 0.006013748701661825,
      "learning_rate": 1.0653618906942393e-05,
      "loss": 0.0003,
      "step": 4312
    },
    {
      "epoch": 1.5781192828393706,
      "grad_norm": 0.009581812657415867,
      "learning_rate": 1.0644387001477106e-05,
      "loss": 0.0004,
      "step": 4313
    },
    {
      "epoch": 1.5784851811196488,
      "grad_norm": 0.005693680606782436,
      "learning_rate": 1.0635155096011818e-05,
      "loss": 0.0003,
      "step": 4314
    },
    {
      "epoch": 1.5788510793999269,
      "grad_norm": 0.6317214369773865,
      "learning_rate": 1.0625923190546528e-05,
      "loss": 0.0023,
      "step": 4315
    },
    {
      "epoch": 1.579216977680205,
      "grad_norm": 0.004997295793145895,
      "learning_rate": 1.061669128508124e-05,
      "loss": 0.0002,
      "step": 4316
    },
    {
      "epoch": 1.579582875960483,
      "grad_norm": 0.005739511922001839,
      "learning_rate": 1.0607459379615953e-05,
      "loss": 0.0003,
      "step": 4317
    },
    {
      "epoch": 1.5799487742407612,
      "grad_norm": 0.006593408994376659,
      "learning_rate": 1.0598227474150665e-05,
      "loss": 0.0003,
      "step": 4318
    },
    {
      "epoch": 1.5803146725210393,
      "grad_norm": 0.005064493510872126,
      "learning_rate": 1.0588995568685377e-05,
      "loss": 0.0002,
      "step": 4319
    },
    {
      "epoch": 1.5806805708013172,
      "grad_norm": 0.006353434175252914,
      "learning_rate": 1.057976366322009e-05,
      "loss": 0.0003,
      "step": 4320
    },
    {
      "epoch": 1.5810464690815953,
      "grad_norm": 0.06663119792938232,
      "learning_rate": 1.05705317577548e-05,
      "loss": 0.0006,
      "step": 4321
    },
    {
      "epoch": 1.5814123673618734,
      "grad_norm": 0.007123278453946114,
      "learning_rate": 1.0561299852289514e-05,
      "loss": 0.0003,
      "step": 4322
    },
    {
      "epoch": 1.5817782656421515,
      "grad_norm": 0.004908145871013403,
      "learning_rate": 1.0552067946824225e-05,
      "loss": 0.0002,
      "step": 4323
    },
    {
      "epoch": 1.5821441639224294,
      "grad_norm": 0.1926468163728714,
      "learning_rate": 1.0542836041358937e-05,
      "loss": 0.0009,
      "step": 4324
    },
    {
      "epoch": 1.5825100622027075,
      "grad_norm": 0.005240020342171192,
      "learning_rate": 1.0533604135893649e-05,
      "loss": 0.0002,
      "step": 4325
    },
    {
      "epoch": 1.5828759604829856,
      "grad_norm": 0.8287988305091858,
      "learning_rate": 1.0524372230428361e-05,
      "loss": 0.0043,
      "step": 4326
    },
    {
      "epoch": 1.5832418587632637,
      "grad_norm": 0.005162778310477734,
      "learning_rate": 1.0515140324963074e-05,
      "loss": 0.0002,
      "step": 4327
    },
    {
      "epoch": 1.5836077570435418,
      "grad_norm": 0.10692349821329117,
      "learning_rate": 1.0505908419497786e-05,
      "loss": 0.0006,
      "step": 4328
    },
    {
      "epoch": 1.58397365532382,
      "grad_norm": 0.02229653112590313,
      "learning_rate": 1.0496676514032496e-05,
      "loss": 0.0004,
      "step": 4329
    },
    {
      "epoch": 1.584339553604098,
      "grad_norm": 0.005361636634916067,
      "learning_rate": 1.0487444608567209e-05,
      "loss": 0.0002,
      "step": 4330
    },
    {
      "epoch": 1.5847054518843762,
      "grad_norm": 0.004315141588449478,
      "learning_rate": 1.0478212703101921e-05,
      "loss": 0.0002,
      "step": 4331
    },
    {
      "epoch": 1.5850713501646543,
      "grad_norm": 0.005441086832433939,
      "learning_rate": 1.0468980797636631e-05,
      "loss": 0.0002,
      "step": 4332
    },
    {
      "epoch": 1.5854372484449324,
      "grad_norm": 0.0049467491917312145,
      "learning_rate": 1.0459748892171345e-05,
      "loss": 0.0002,
      "step": 4333
    },
    {
      "epoch": 1.5858031467252105,
      "grad_norm": 0.006922232452780008,
      "learning_rate": 1.0450516986706056e-05,
      "loss": 0.0003,
      "step": 4334
    },
    {
      "epoch": 1.5861690450054886,
      "grad_norm": 0.006518320180475712,
      "learning_rate": 1.0441285081240768e-05,
      "loss": 0.0003,
      "step": 4335
    },
    {
      "epoch": 1.5865349432857667,
      "grad_norm": 0.13601723313331604,
      "learning_rate": 1.043205317577548e-05,
      "loss": 0.0008,
      "step": 4336
    },
    {
      "epoch": 1.5869008415660446,
      "grad_norm": 0.0161801278591156,
      "learning_rate": 1.0422821270310193e-05,
      "loss": 0.0004,
      "step": 4337
    },
    {
      "epoch": 1.5872667398463227,
      "grad_norm": 0.5610435009002686,
      "learning_rate": 1.0413589364844905e-05,
      "loss": 0.0031,
      "step": 4338
    },
    {
      "epoch": 1.5876326381266008,
      "grad_norm": 13.773670196533203,
      "learning_rate": 1.0404357459379617e-05,
      "loss": 0.0179,
      "step": 4339
    },
    {
      "epoch": 1.587998536406879,
      "grad_norm": 0.004510920960456133,
      "learning_rate": 1.0395125553914328e-05,
      "loss": 0.0002,
      "step": 4340
    },
    {
      "epoch": 1.5883644346871568,
      "grad_norm": 0.0069051641039550304,
      "learning_rate": 1.0385893648449042e-05,
      "loss": 0.0003,
      "step": 4341
    },
    {
      "epoch": 1.588730332967435,
      "grad_norm": 0.005801075603812933,
      "learning_rate": 1.0376661742983752e-05,
      "loss": 0.0002,
      "step": 4342
    },
    {
      "epoch": 1.589096231247713,
      "grad_norm": 0.006554202176630497,
      "learning_rate": 1.0367429837518464e-05,
      "loss": 0.0002,
      "step": 4343
    },
    {
      "epoch": 1.5894621295279912,
      "grad_norm": 0.004349564202129841,
      "learning_rate": 1.0358197932053177e-05,
      "loss": 0.0002,
      "step": 4344
    },
    {
      "epoch": 1.5898280278082693,
      "grad_norm": 0.029759228229522705,
      "learning_rate": 1.0348966026587887e-05,
      "loss": 0.0003,
      "step": 4345
    },
    {
      "epoch": 1.5901939260885474,
      "grad_norm": 0.005914822220802307,
      "learning_rate": 1.0339734121122601e-05,
      "loss": 0.0002,
      "step": 4346
    },
    {
      "epoch": 1.5905598243688255,
      "grad_norm": 0.00445357384160161,
      "learning_rate": 1.0330502215657312e-05,
      "loss": 0.0002,
      "step": 4347
    },
    {
      "epoch": 1.5909257226491036,
      "grad_norm": 0.0076176621951162815,
      "learning_rate": 1.0321270310192024e-05,
      "loss": 0.0003,
      "step": 4348
    },
    {
      "epoch": 1.5912916209293817,
      "grad_norm": 0.004314923193305731,
      "learning_rate": 1.0312038404726736e-05,
      "loss": 0.0002,
      "step": 4349
    },
    {
      "epoch": 1.5916575192096598,
      "grad_norm": 0.0054754167795181274,
      "learning_rate": 1.0302806499261448e-05,
      "loss": 0.0002,
      "step": 4350
    },
    {
      "epoch": 1.592023417489938,
      "grad_norm": 21.14862632751465,
      "learning_rate": 1.0293574593796159e-05,
      "loss": 0.2307,
      "step": 4351
    },
    {
      "epoch": 1.592389315770216,
      "grad_norm": 0.005628217943012714,
      "learning_rate": 1.0284342688330873e-05,
      "loss": 0.0003,
      "step": 4352
    },
    {
      "epoch": 1.592755214050494,
      "grad_norm": 0.005074667278677225,
      "learning_rate": 1.0275110782865583e-05,
      "loss": 0.0002,
      "step": 4353
    },
    {
      "epoch": 1.593121112330772,
      "grad_norm": 0.0037077544257044792,
      "learning_rate": 1.0265878877400296e-05,
      "loss": 0.0002,
      "step": 4354
    },
    {
      "epoch": 1.5934870106110501,
      "grad_norm": 0.006281067617237568,
      "learning_rate": 1.0256646971935008e-05,
      "loss": 0.0003,
      "step": 4355
    },
    {
      "epoch": 1.5938529088913282,
      "grad_norm": 0.0036905331071466208,
      "learning_rate": 1.024741506646972e-05,
      "loss": 0.0002,
      "step": 4356
    },
    {
      "epoch": 1.5942188071716061,
      "grad_norm": 0.005260481499135494,
      "learning_rate": 1.0238183161004432e-05,
      "loss": 0.0002,
      "step": 4357
    },
    {
      "epoch": 1.5945847054518842,
      "grad_norm": 0.1713351458311081,
      "learning_rate": 1.0228951255539143e-05,
      "loss": 0.0006,
      "step": 4358
    },
    {
      "epoch": 1.5949506037321624,
      "grad_norm": 0.0069582536816596985,
      "learning_rate": 1.0219719350073855e-05,
      "loss": 0.0003,
      "step": 4359
    },
    {
      "epoch": 1.5953165020124405,
      "grad_norm": 0.006184925325214863,
      "learning_rate": 1.0210487444608567e-05,
      "loss": 0.0003,
      "step": 4360
    },
    {
      "epoch": 1.5956824002927186,
      "grad_norm": 0.0038786064833402634,
      "learning_rate": 1.020125553914328e-05,
      "loss": 0.0002,
      "step": 4361
    },
    {
      "epoch": 1.5960482985729967,
      "grad_norm": 0.006582365836948156,
      "learning_rate": 1.0192023633677992e-05,
      "loss": 0.0002,
      "step": 4362
    },
    {
      "epoch": 1.5964141968532748,
      "grad_norm": 0.005275289993733168,
      "learning_rate": 1.0182791728212704e-05,
      "loss": 0.0002,
      "step": 4363
    },
    {
      "epoch": 1.596780095133553,
      "grad_norm": 0.005290469154715538,
      "learning_rate": 1.0173559822747415e-05,
      "loss": 0.0002,
      "step": 4364
    },
    {
      "epoch": 1.597145993413831,
      "grad_norm": 0.003930186852812767,
      "learning_rate": 1.0164327917282129e-05,
      "loss": 0.0002,
      "step": 4365
    },
    {
      "epoch": 1.5975118916941091,
      "grad_norm": 5.43695592880249,
      "learning_rate": 1.015509601181684e-05,
      "loss": 0.4188,
      "step": 4366
    },
    {
      "epoch": 1.5978777899743872,
      "grad_norm": 0.005406483542174101,
      "learning_rate": 1.0145864106351551e-05,
      "loss": 0.0002,
      "step": 4367
    },
    {
      "epoch": 1.5982436882546653,
      "grad_norm": 0.004952817689627409,
      "learning_rate": 1.0136632200886264e-05,
      "loss": 0.0002,
      "step": 4368
    },
    {
      "epoch": 1.5986095865349434,
      "grad_norm": 0.004541268106549978,
      "learning_rate": 1.0127400295420976e-05,
      "loss": 0.0002,
      "step": 4369
    },
    {
      "epoch": 1.5989754848152213,
      "grad_norm": 0.0054872650653123856,
      "learning_rate": 1.0118168389955686e-05,
      "loss": 0.0002,
      "step": 4370
    },
    {
      "epoch": 1.5993413830954994,
      "grad_norm": 0.08721670508384705,
      "learning_rate": 1.01089364844904e-05,
      "loss": 0.0008,
      "step": 4371
    },
    {
      "epoch": 1.5997072813757776,
      "grad_norm": 0.0044119893573224545,
      "learning_rate": 1.0099704579025111e-05,
      "loss": 0.0002,
      "step": 4372
    },
    {
      "epoch": 1.6000731796560557,
      "grad_norm": 0.22228577733039856,
      "learning_rate": 1.0090472673559823e-05,
      "loss": 0.0011,
      "step": 4373
    },
    {
      "epoch": 1.6004390779363336,
      "grad_norm": 0.1991717368364334,
      "learning_rate": 1.0081240768094535e-05,
      "loss": 0.001,
      "step": 4374
    },
    {
      "epoch": 1.6008049762166117,
      "grad_norm": 0.005463206209242344,
      "learning_rate": 1.0072008862629246e-05,
      "loss": 0.0002,
      "step": 4375
    },
    {
      "epoch": 1.6011708744968898,
      "grad_norm": 0.004322153516113758,
      "learning_rate": 1.006277695716396e-05,
      "loss": 0.0002,
      "step": 4376
    },
    {
      "epoch": 1.6015367727771679,
      "grad_norm": 0.0059922244399785995,
      "learning_rate": 1.005354505169867e-05,
      "loss": 0.0002,
      "step": 4377
    },
    {
      "epoch": 1.601902671057446,
      "grad_norm": 0.004264464136213064,
      "learning_rate": 1.0044313146233383e-05,
      "loss": 0.0002,
      "step": 4378
    },
    {
      "epoch": 1.602268569337724,
      "grad_norm": 0.005099993199110031,
      "learning_rate": 1.0035081240768095e-05,
      "loss": 0.0002,
      "step": 4379
    },
    {
      "epoch": 1.6026344676180022,
      "grad_norm": 0.005181861110031605,
      "learning_rate": 1.0025849335302807e-05,
      "loss": 0.0002,
      "step": 4380
    },
    {
      "epoch": 1.6030003658982803,
      "grad_norm": 0.11081348359584808,
      "learning_rate": 1.001661742983752e-05,
      "loss": 0.0008,
      "step": 4381
    },
    {
      "epoch": 1.6033662641785584,
      "grad_norm": 0.006734611000865698,
      "learning_rate": 1.0007385524372232e-05,
      "loss": 0.0003,
      "step": 4382
    },
    {
      "epoch": 1.6037321624588365,
      "grad_norm": 0.004833712242543697,
      "learning_rate": 9.998153618906942e-06,
      "loss": 0.0002,
      "step": 4383
    },
    {
      "epoch": 1.6040980607391147,
      "grad_norm": 0.005628752056509256,
      "learning_rate": 9.988921713441656e-06,
      "loss": 0.0003,
      "step": 4384
    },
    {
      "epoch": 1.6044639590193928,
      "grad_norm": 4.065608501434326,
      "learning_rate": 9.979689807976367e-06,
      "loss": 0.9815,
      "step": 4385
    },
    {
      "epoch": 1.6048298572996706,
      "grad_norm": 0.005502123851329088,
      "learning_rate": 9.970457902511077e-06,
      "loss": 0.0002,
      "step": 4386
    },
    {
      "epoch": 1.6051957555799488,
      "grad_norm": 0.005191100295633078,
      "learning_rate": 9.961225997045791e-06,
      "loss": 0.0002,
      "step": 4387
    },
    {
      "epoch": 1.6055616538602269,
      "grad_norm": 0.005521942395716906,
      "learning_rate": 9.951994091580502e-06,
      "loss": 0.0003,
      "step": 4388
    },
    {
      "epoch": 1.605927552140505,
      "grad_norm": 0.006163351703435183,
      "learning_rate": 9.942762186115214e-06,
      "loss": 0.0003,
      "step": 4389
    },
    {
      "epoch": 1.6062934504207829,
      "grad_norm": 0.029227787628769875,
      "learning_rate": 9.933530280649926e-06,
      "loss": 0.0006,
      "step": 4390
    },
    {
      "epoch": 1.606659348701061,
      "grad_norm": 0.009125648066401482,
      "learning_rate": 9.924298375184639e-06,
      "loss": 0.0004,
      "step": 4391
    },
    {
      "epoch": 1.607025246981339,
      "grad_norm": 0.007114452309906483,
      "learning_rate": 9.91506646971935e-06,
      "loss": 0.0003,
      "step": 4392
    },
    {
      "epoch": 1.6073911452616172,
      "grad_norm": 0.007204586174339056,
      "learning_rate": 9.905834564254063e-06,
      "loss": 0.0003,
      "step": 4393
    },
    {
      "epoch": 1.6077570435418953,
      "grad_norm": 0.04918231442570686,
      "learning_rate": 9.896602658788774e-06,
      "loss": 0.0005,
      "step": 4394
    },
    {
      "epoch": 1.6081229418221734,
      "grad_norm": 0.009902381338179111,
      "learning_rate": 9.887370753323487e-06,
      "loss": 0.0004,
      "step": 4395
    },
    {
      "epoch": 1.6084888401024515,
      "grad_norm": 0.006972847506403923,
      "learning_rate": 9.878138847858198e-06,
      "loss": 0.0003,
      "step": 4396
    },
    {
      "epoch": 1.6088547383827296,
      "grad_norm": 0.007943124510347843,
      "learning_rate": 9.86890694239291e-06,
      "loss": 0.0004,
      "step": 4397
    },
    {
      "epoch": 1.6092206366630077,
      "grad_norm": 0.005947232246398926,
      "learning_rate": 9.859675036927623e-06,
      "loss": 0.0003,
      "step": 4398
    },
    {
      "epoch": 1.6095865349432859,
      "grad_norm": 0.011068726889789104,
      "learning_rate": 9.850443131462335e-06,
      "loss": 0.0004,
      "step": 4399
    },
    {
      "epoch": 1.609952433223564,
      "grad_norm": 0.010359075851738453,
      "learning_rate": 9.841211225997047e-06,
      "loss": 0.0004,
      "step": 4400
    },
    {
      "epoch": 1.610318331503842,
      "grad_norm": 0.039299335330724716,
      "learning_rate": 9.831979320531758e-06,
      "loss": 0.0006,
      "step": 4401
    },
    {
      "epoch": 1.6106842297841202,
      "grad_norm": 0.013788397423923016,
      "learning_rate": 9.82274741506647e-06,
      "loss": 0.0004,
      "step": 4402
    },
    {
      "epoch": 1.611050128064398,
      "grad_norm": 0.00730047607794404,
      "learning_rate": 9.813515509601182e-06,
      "loss": 0.0003,
      "step": 4403
    },
    {
      "epoch": 1.6114160263446762,
      "grad_norm": 0.008971883915364742,
      "learning_rate": 9.804283604135894e-06,
      "loss": 0.0004,
      "step": 4404
    },
    {
      "epoch": 1.6117819246249543,
      "grad_norm": 47.716888427734375,
      "learning_rate": 9.795051698670605e-06,
      "loss": 0.043,
      "step": 4405
    },
    {
      "epoch": 1.6121478229052324,
      "grad_norm": 0.007507238071411848,
      "learning_rate": 9.785819793205319e-06,
      "loss": 0.0003,
      "step": 4406
    },
    {
      "epoch": 1.6125137211855103,
      "grad_norm": 0.021081972867250443,
      "learning_rate": 9.77658788774003e-06,
      "loss": 0.0007,
      "step": 4407
    },
    {
      "epoch": 1.6128796194657884,
      "grad_norm": 0.00879986584186554,
      "learning_rate": 9.767355982274742e-06,
      "loss": 0.0004,
      "step": 4408
    },
    {
      "epoch": 1.6132455177460665,
      "grad_norm": 0.009278547018766403,
      "learning_rate": 9.758124076809454e-06,
      "loss": 0.0003,
      "step": 4409
    },
    {
      "epoch": 1.6136114160263446,
      "grad_norm": 0.009231370873749256,
      "learning_rate": 9.748892171344166e-06,
      "loss": 0.0004,
      "step": 4410
    },
    {
      "epoch": 1.6139773143066227,
      "grad_norm": 0.009425104595720768,
      "learning_rate": 9.739660265878878e-06,
      "loss": 0.0004,
      "step": 4411
    },
    {
      "epoch": 1.6143432125869008,
      "grad_norm": 0.011997558176517487,
      "learning_rate": 9.73042836041359e-06,
      "loss": 0.0004,
      "step": 4412
    },
    {
      "epoch": 1.614709110867179,
      "grad_norm": 0.006419930141419172,
      "learning_rate": 9.721196454948301e-06,
      "loss": 0.0003,
      "step": 4413
    },
    {
      "epoch": 1.615075009147457,
      "grad_norm": 2.0033493041992188,
      "learning_rate": 9.711964549483015e-06,
      "loss": 0.4946,
      "step": 4414
    },
    {
      "epoch": 1.6154409074277352,
      "grad_norm": 0.02485267072916031,
      "learning_rate": 9.702732644017726e-06,
      "loss": 0.0004,
      "step": 4415
    },
    {
      "epoch": 1.6158068057080133,
      "grad_norm": 0.0066540720872581005,
      "learning_rate": 9.693500738552438e-06,
      "loss": 0.0003,
      "step": 4416
    },
    {
      "epoch": 1.6161727039882914,
      "grad_norm": 0.031524013727903366,
      "learning_rate": 9.68426883308715e-06,
      "loss": 0.0008,
      "step": 4417
    },
    {
      "epoch": 1.6165386022685695,
      "grad_norm": 2.86027455329895,
      "learning_rate": 9.67503692762186e-06,
      "loss": 0.0059,
      "step": 4418
    },
    {
      "epoch": 1.6169045005488474,
      "grad_norm": 0.02141481079161167,
      "learning_rate": 9.665805022156575e-06,
      "loss": 0.0007,
      "step": 4419
    },
    {
      "epoch": 1.6172703988291255,
      "grad_norm": 0.010534746572375298,
      "learning_rate": 9.656573116691285e-06,
      "loss": 0.0005,
      "step": 4420
    },
    {
      "epoch": 1.6176362971094036,
      "grad_norm": 19.241533279418945,
      "learning_rate": 9.647341211225997e-06,
      "loss": 0.1517,
      "step": 4421
    },
    {
      "epoch": 1.6180021953896817,
      "grad_norm": 0.00927891954779625,
      "learning_rate": 9.63810930576071e-06,
      "loss": 0.0004,
      "step": 4422
    },
    {
      "epoch": 1.6183680936699596,
      "grad_norm": 0.010488525964319706,
      "learning_rate": 9.628877400295422e-06,
      "loss": 0.0005,
      "step": 4423
    },
    {
      "epoch": 1.6187339919502377,
      "grad_norm": 6.567628383636475,
      "learning_rate": 9.619645494830132e-06,
      "loss": 0.1095,
      "step": 4424
    },
    {
      "epoch": 1.6190998902305158,
      "grad_norm": 0.012389236129820347,
      "learning_rate": 9.610413589364846e-06,
      "loss": 0.0005,
      "step": 4425
    },
    {
      "epoch": 1.619465788510794,
      "grad_norm": 0.007879192009568214,
      "learning_rate": 9.601181683899557e-06,
      "loss": 0.0004,
      "step": 4426
    },
    {
      "epoch": 1.619831686791072,
      "grad_norm": 0.008951444178819656,
      "learning_rate": 9.591949778434269e-06,
      "loss": 0.0004,
      "step": 4427
    },
    {
      "epoch": 1.6201975850713501,
      "grad_norm": 0.009362036362290382,
      "learning_rate": 9.582717872968981e-06,
      "loss": 0.0004,
      "step": 4428
    },
    {
      "epoch": 1.6205634833516283,
      "grad_norm": 6.251626014709473,
      "learning_rate": 9.573485967503692e-06,
      "loss": 0.1478,
      "step": 4429
    },
    {
      "epoch": 1.6209293816319064,
      "grad_norm": 0.007350089494138956,
      "learning_rate": 9.564254062038406e-06,
      "loss": 0.0003,
      "step": 4430
    },
    {
      "epoch": 1.6212952799121845,
      "grad_norm": 0.01068863831460476,
      "learning_rate": 9.555022156573116e-06,
      "loss": 0.0005,
      "step": 4431
    },
    {
      "epoch": 1.6216611781924626,
      "grad_norm": 15.16765308380127,
      "learning_rate": 9.545790251107829e-06,
      "loss": 0.3485,
      "step": 4432
    },
    {
      "epoch": 1.6220270764727407,
      "grad_norm": 0.010714532807469368,
      "learning_rate": 9.536558345642541e-06,
      "loss": 0.0005,
      "step": 4433
    },
    {
      "epoch": 1.6223929747530188,
      "grad_norm": 11.605268478393555,
      "learning_rate": 9.527326440177253e-06,
      "loss": 0.0174,
      "step": 4434
    },
    {
      "epoch": 1.6227588730332967,
      "grad_norm": 0.39440107345581055,
      "learning_rate": 9.518094534711965e-06,
      "loss": 0.0023,
      "step": 4435
    },
    {
      "epoch": 1.6231247713135748,
      "grad_norm": 0.00827914196997881,
      "learning_rate": 9.508862629246678e-06,
      "loss": 0.0004,
      "step": 4436
    },
    {
      "epoch": 1.623490669593853,
      "grad_norm": 0.011481781490147114,
      "learning_rate": 9.499630723781388e-06,
      "loss": 0.0005,
      "step": 4437
    },
    {
      "epoch": 1.623856567874131,
      "grad_norm": 0.7931006550788879,
      "learning_rate": 9.490398818316102e-06,
      "loss": 0.0034,
      "step": 4438
    },
    {
      "epoch": 1.6242224661544091,
      "grad_norm": 0.018145792186260223,
      "learning_rate": 9.481166912850813e-06,
      "loss": 0.0004,
      "step": 4439
    },
    {
      "epoch": 1.624588364434687,
      "grad_norm": 0.02056349255144596,
      "learning_rate": 9.471935007385525e-06,
      "loss": 0.0005,
      "step": 4440
    },
    {
      "epoch": 1.6249542627149651,
      "grad_norm": 0.2634795010089874,
      "learning_rate": 9.462703101920237e-06,
      "loss": 0.0011,
      "step": 4441
    },
    {
      "epoch": 1.6253201609952432,
      "grad_norm": 0.1115393117070198,
      "learning_rate": 9.45347119645495e-06,
      "loss": 0.0009,
      "step": 4442
    },
    {
      "epoch": 1.6256860592755213,
      "grad_norm": 0.010160387493669987,
      "learning_rate": 9.44423929098966e-06,
      "loss": 0.0004,
      "step": 4443
    },
    {
      "epoch": 1.6260519575557995,
      "grad_norm": 0.009279830381274223,
      "learning_rate": 9.435007385524372e-06,
      "loss": 0.0004,
      "step": 4444
    },
    {
      "epoch": 1.6264178558360776,
      "grad_norm": 0.048141613602638245,
      "learning_rate": 9.425775480059084e-06,
      "loss": 0.0007,
      "step": 4445
    },
    {
      "epoch": 1.6267837541163557,
      "grad_norm": 0.008512926287949085,
      "learning_rate": 9.416543574593797e-06,
      "loss": 0.0004,
      "step": 4446
    },
    {
      "epoch": 1.6271496523966338,
      "grad_norm": 0.8791294693946838,
      "learning_rate": 9.407311669128509e-06,
      "loss": 0.0029,
      "step": 4447
    },
    {
      "epoch": 1.627515550676912,
      "grad_norm": 0.11479802429676056,
      "learning_rate": 9.39807976366322e-06,
      "loss": 0.0009,
      "step": 4448
    },
    {
      "epoch": 1.62788144895719,
      "grad_norm": 0.033305905759334564,
      "learning_rate": 9.388847858197933e-06,
      "loss": 0.0005,
      "step": 4449
    },
    {
      "epoch": 1.6282473472374681,
      "grad_norm": 4.2923994064331055,
      "learning_rate": 9.379615952732644e-06,
      "loss": 0.0313,
      "step": 4450
    },
    {
      "epoch": 1.6286132455177462,
      "grad_norm": 0.4132571518421173,
      "learning_rate": 9.370384047267356e-06,
      "loss": 0.0025,
      "step": 4451
    },
    {
      "epoch": 1.628979143798024,
      "grad_norm": 10.868939399719238,
      "learning_rate": 9.361152141802068e-06,
      "loss": 0.0225,
      "step": 4452
    },
    {
      "epoch": 1.6293450420783022,
      "grad_norm": 0.07956430315971375,
      "learning_rate": 9.35192023633678e-06,
      "loss": 0.0005,
      "step": 4453
    },
    {
      "epoch": 1.6297109403585803,
      "grad_norm": 0.006961274892091751,
      "learning_rate": 9.342688330871493e-06,
      "loss": 0.0003,
      "step": 4454
    },
    {
      "epoch": 1.6300768386388584,
      "grad_norm": 0.007975859567523003,
      "learning_rate": 9.333456425406205e-06,
      "loss": 0.0003,
      "step": 4455
    },
    {
      "epoch": 1.6304427369191363,
      "grad_norm": 0.009905374608933926,
      "learning_rate": 9.324224519940916e-06,
      "loss": 0.0003,
      "step": 4456
    },
    {
      "epoch": 1.6308086351994144,
      "grad_norm": 0.010641681030392647,
      "learning_rate": 9.31499261447563e-06,
      "loss": 0.0004,
      "step": 4457
    },
    {
      "epoch": 1.6311745334796925,
      "grad_norm": 0.00572163425385952,
      "learning_rate": 9.30576070901034e-06,
      "loss": 0.0003,
      "step": 4458
    },
    {
      "epoch": 1.6315404317599707,
      "grad_norm": 0.1717863380908966,
      "learning_rate": 9.29652880354505e-06,
      "loss": 0.0008,
      "step": 4459
    },
    {
      "epoch": 1.6319063300402488,
      "grad_norm": 0.009449026547372341,
      "learning_rate": 9.287296898079765e-06,
      "loss": 0.0004,
      "step": 4460
    },
    {
      "epoch": 1.6322722283205269,
      "grad_norm": 5.709840297698975,
      "learning_rate": 9.278064992614475e-06,
      "loss": 0.3643,
      "step": 4461
    },
    {
      "epoch": 1.632638126600805,
      "grad_norm": 0.009869184345006943,
      "learning_rate": 9.268833087149187e-06,
      "loss": 0.0004,
      "step": 4462
    },
    {
      "epoch": 1.633004024881083,
      "grad_norm": 0.1946403980255127,
      "learning_rate": 9.2596011816839e-06,
      "loss": 0.0009,
      "step": 4463
    },
    {
      "epoch": 1.6333699231613612,
      "grad_norm": 0.9688766002655029,
      "learning_rate": 9.250369276218612e-06,
      "loss": 0.0064,
      "step": 4464
    },
    {
      "epoch": 1.6337358214416393,
      "grad_norm": 0.6821357607841492,
      "learning_rate": 9.241137370753324e-06,
      "loss": 0.002,
      "step": 4465
    },
    {
      "epoch": 1.6341017197219174,
      "grad_norm": 0.009056284092366695,
      "learning_rate": 9.231905465288036e-06,
      "loss": 0.0004,
      "step": 4466
    },
    {
      "epoch": 1.6344676180021955,
      "grad_norm": 0.009355796501040459,
      "learning_rate": 9.222673559822747e-06,
      "loss": 0.0004,
      "step": 4467
    },
    {
      "epoch": 1.6348335162824734,
      "grad_norm": 0.00930742546916008,
      "learning_rate": 9.213441654357461e-06,
      "loss": 0.0004,
      "step": 4468
    },
    {
      "epoch": 1.6351994145627515,
      "grad_norm": 0.010128830559551716,
      "learning_rate": 9.204209748892171e-06,
      "loss": 0.0004,
      "step": 4469
    },
    {
      "epoch": 1.6355653128430296,
      "grad_norm": 0.007309223059564829,
      "learning_rate": 9.194977843426884e-06,
      "loss": 0.0003,
      "step": 4470
    },
    {
      "epoch": 1.6359312111233077,
      "grad_norm": 15.231865882873535,
      "learning_rate": 9.185745937961596e-06,
      "loss": 0.1531,
      "step": 4471
    },
    {
      "epoch": 1.6362971094035856,
      "grad_norm": 0.007490131538361311,
      "learning_rate": 9.176514032496308e-06,
      "loss": 0.0003,
      "step": 4472
    },
    {
      "epoch": 1.6366630076838637,
      "grad_norm": 0.00740177184343338,
      "learning_rate": 9.16728212703102e-06,
      "loss": 0.0003,
      "step": 4473
    },
    {
      "epoch": 1.6370289059641419,
      "grad_norm": 0.006844229530543089,
      "learning_rate": 9.158050221565731e-06,
      "loss": 0.0003,
      "step": 4474
    },
    {
      "epoch": 1.63739480424442,
      "grad_norm": 0.007207219488918781,
      "learning_rate": 9.148818316100443e-06,
      "loss": 0.0003,
      "step": 4475
    },
    {
      "epoch": 1.637760702524698,
      "grad_norm": 0.07500728219747543,
      "learning_rate": 9.139586410635155e-06,
      "loss": 0.0009,
      "step": 4476
    },
    {
      "epoch": 1.6381266008049762,
      "grad_norm": 0.0582161620259285,
      "learning_rate": 9.130354505169868e-06,
      "loss": 0.0009,
      "step": 4477
    },
    {
      "epoch": 1.6384924990852543,
      "grad_norm": 0.01431242935359478,
      "learning_rate": 9.121122599704578e-06,
      "loss": 0.0006,
      "step": 4478
    },
    {
      "epoch": 1.6388583973655324,
      "grad_norm": 0.011119847185909748,
      "learning_rate": 9.111890694239292e-06,
      "loss": 0.0005,
      "step": 4479
    },
    {
      "epoch": 1.6392242956458105,
      "grad_norm": 0.008373083546757698,
      "learning_rate": 9.102658788774003e-06,
      "loss": 0.0004,
      "step": 4480
    },
    {
      "epoch": 1.6395901939260886,
      "grad_norm": 0.006957348436117172,
      "learning_rate": 9.093426883308715e-06,
      "loss": 0.0003,
      "step": 4481
    },
    {
      "epoch": 1.6399560922063667,
      "grad_norm": 0.007344801444560289,
      "learning_rate": 9.084194977843427e-06,
      "loss": 0.0003,
      "step": 4482
    },
    {
      "epoch": 1.6403219904866448,
      "grad_norm": 0.17848123610019684,
      "learning_rate": 9.07496307237814e-06,
      "loss": 0.0015,
      "step": 4483
    },
    {
      "epoch": 1.640687888766923,
      "grad_norm": 0.02898194082081318,
      "learning_rate": 9.065731166912852e-06,
      "loss": 0.0008,
      "step": 4484
    },
    {
      "epoch": 1.6410537870472008,
      "grad_norm": 0.12856213748455048,
      "learning_rate": 9.056499261447564e-06,
      "loss": 0.0009,
      "step": 4485
    },
    {
      "epoch": 1.641419685327479,
      "grad_norm": 0.01558457501232624,
      "learning_rate": 9.047267355982274e-06,
      "loss": 0.0006,
      "step": 4486
    },
    {
      "epoch": 1.641785583607757,
      "grad_norm": 0.00863824412226677,
      "learning_rate": 9.038035450516987e-06,
      "loss": 0.0004,
      "step": 4487
    },
    {
      "epoch": 1.6421514818880352,
      "grad_norm": 0.006767268292605877,
      "learning_rate": 9.028803545051699e-06,
      "loss": 0.0003,
      "step": 4488
    },
    {
      "epoch": 1.642517380168313,
      "grad_norm": 14.403444290161133,
      "learning_rate": 9.019571639586411e-06,
      "loss": 0.4535,
      "step": 4489
    },
    {
      "epoch": 1.6428832784485912,
      "grad_norm": 0.044978946447372437,
      "learning_rate": 9.010339734121123e-06,
      "loss": 0.0007,
      "step": 4490
    },
    {
      "epoch": 1.6432491767288693,
      "grad_norm": 0.014297238551080227,
      "learning_rate": 9.001107828655834e-06,
      "loss": 0.0005,
      "step": 4491
    },
    {
      "epoch": 1.6436150750091474,
      "grad_norm": 0.01208081841468811,
      "learning_rate": 8.991875923190548e-06,
      "loss": 0.0005,
      "step": 4492
    },
    {
      "epoch": 1.6439809732894255,
      "grad_norm": 0.00855469610542059,
      "learning_rate": 8.982644017725258e-06,
      "loss": 0.0004,
      "step": 4493
    },
    {
      "epoch": 1.6443468715697036,
      "grad_norm": 0.00860854797065258,
      "learning_rate": 8.97341211225997e-06,
      "loss": 0.0004,
      "step": 4494
    },
    {
      "epoch": 1.6447127698499817,
      "grad_norm": 0.006547768134623766,
      "learning_rate": 8.964180206794683e-06,
      "loss": 0.0003,
      "step": 4495
    },
    {
      "epoch": 1.6450786681302598,
      "grad_norm": 14.945968627929688,
      "learning_rate": 8.954948301329395e-06,
      "loss": 0.0942,
      "step": 4496
    },
    {
      "epoch": 1.645444566410538,
      "grad_norm": 0.0065134549513459206,
      "learning_rate": 8.945716395864106e-06,
      "loss": 0.0003,
      "step": 4497
    },
    {
      "epoch": 1.645810464690816,
      "grad_norm": 0.005646414123475552,
      "learning_rate": 8.93648449039882e-06,
      "loss": 0.0003,
      "step": 4498
    },
    {
      "epoch": 1.6461763629710942,
      "grad_norm": 0.011770959943532944,
      "learning_rate": 8.92725258493353e-06,
      "loss": 0.0004,
      "step": 4499
    },
    {
      "epoch": 1.6465422612513723,
      "grad_norm": 0.010787442326545715,
      "learning_rate": 8.918020679468243e-06,
      "loss": 0.0005,
      "step": 4500
    },
    {
      "epoch": 1.6469081595316502,
      "grad_norm": 0.009309494867920876,
      "learning_rate": 8.908788774002955e-06,
      "loss": 0.0004,
      "step": 4501
    },
    {
      "epoch": 1.6472740578119283,
      "grad_norm": 0.08641643077135086,
      "learning_rate": 8.899556868537665e-06,
      "loss": 0.0006,
      "step": 4502
    },
    {
      "epoch": 1.6476399560922064,
      "grad_norm": 0.009784706868231297,
      "learning_rate": 8.89032496307238e-06,
      "loss": 0.0004,
      "step": 4503
    },
    {
      "epoch": 1.6480058543724845,
      "grad_norm": 0.011152708902955055,
      "learning_rate": 8.88109305760709e-06,
      "loss": 0.0004,
      "step": 4504
    },
    {
      "epoch": 1.6483717526527624,
      "grad_norm": 5.073084831237793,
      "learning_rate": 8.871861152141802e-06,
      "loss": 0.4394,
      "step": 4505
    },
    {
      "epoch": 1.6487376509330405,
      "grad_norm": 7.133708953857422,
      "learning_rate": 8.862629246676514e-06,
      "loss": 0.1394,
      "step": 4506
    },
    {
      "epoch": 1.6491035492133186,
      "grad_norm": 0.007607525680214167,
      "learning_rate": 8.853397341211227e-06,
      "loss": 0.0004,
      "step": 4507
    },
    {
      "epoch": 1.6494694474935967,
      "grad_norm": 0.02253725938498974,
      "learning_rate": 8.844165435745939e-06,
      "loss": 0.0005,
      "step": 4508
    },
    {
      "epoch": 1.6498353457738748,
      "grad_norm": 0.012776762247085571,
      "learning_rate": 8.834933530280651e-06,
      "loss": 0.0005,
      "step": 4509
    },
    {
      "epoch": 1.650201244054153,
      "grad_norm": 0.010790606960654259,
      "learning_rate": 8.825701624815362e-06,
      "loss": 0.0005,
      "step": 4510
    },
    {
      "epoch": 1.650567142334431,
      "grad_norm": 0.007219522725790739,
      "learning_rate": 8.816469719350075e-06,
      "loss": 0.0003,
      "step": 4511
    },
    {
      "epoch": 1.6509330406147091,
      "grad_norm": 23.304773330688477,
      "learning_rate": 8.807237813884786e-06,
      "loss": 0.2707,
      "step": 4512
    },
    {
      "epoch": 1.6512989388949872,
      "grad_norm": 0.008355841971933842,
      "learning_rate": 8.798005908419498e-06,
      "loss": 0.0004,
      "step": 4513
    },
    {
      "epoch": 1.6516648371752654,
      "grad_norm": 0.008584462106227875,
      "learning_rate": 8.78877400295421e-06,
      "loss": 0.0004,
      "step": 4514
    },
    {
      "epoch": 1.6520307354555435,
      "grad_norm": 0.0077319215051829815,
      "learning_rate": 8.779542097488923e-06,
      "loss": 0.0003,
      "step": 4515
    },
    {
      "epoch": 1.6523966337358216,
      "grad_norm": 0.8011586666107178,
      "learning_rate": 8.770310192023633e-06,
      "loss": 0.0018,
      "step": 4516
    },
    {
      "epoch": 1.6527625320160997,
      "grad_norm": 0.007111769635230303,
      "learning_rate": 8.761078286558346e-06,
      "loss": 0.0003,
      "step": 4517
    },
    {
      "epoch": 1.6531284302963776,
      "grad_norm": 0.007021291647106409,
      "learning_rate": 8.751846381093058e-06,
      "loss": 0.0003,
      "step": 4518
    },
    {
      "epoch": 1.6534943285766557,
      "grad_norm": 0.22383703291416168,
      "learning_rate": 8.74261447562777e-06,
      "loss": 0.0008,
      "step": 4519
    },
    {
      "epoch": 1.6538602268569338,
      "grad_norm": 0.009058067575097084,
      "learning_rate": 8.733382570162482e-06,
      "loss": 0.0004,
      "step": 4520
    },
    {
      "epoch": 1.654226125137212,
      "grad_norm": 0.013349374756217003,
      "learning_rate": 8.724150664697193e-06,
      "loss": 0.0005,
      "step": 4521
    },
    {
      "epoch": 1.6545920234174898,
      "grad_norm": 0.008930698037147522,
      "learning_rate": 8.714918759231907e-06,
      "loss": 0.0004,
      "step": 4522
    },
    {
      "epoch": 1.654957921697768,
      "grad_norm": 0.00897971261292696,
      "learning_rate": 8.705686853766617e-06,
      "loss": 0.0004,
      "step": 4523
    },
    {
      "epoch": 1.655323819978046,
      "grad_norm": 0.009361157193779945,
      "learning_rate": 8.69645494830133e-06,
      "loss": 0.0004,
      "step": 4524
    },
    {
      "epoch": 1.6556897182583241,
      "grad_norm": 21.1439151763916,
      "learning_rate": 8.687223042836042e-06,
      "loss": 0.061,
      "step": 4525
    },
    {
      "epoch": 1.6560556165386022,
      "grad_norm": 0.0090032322332263,
      "learning_rate": 8.677991137370754e-06,
      "loss": 0.0004,
      "step": 4526
    },
    {
      "epoch": 1.6564215148188803,
      "grad_norm": 0.013535914942622185,
      "learning_rate": 8.668759231905466e-06,
      "loss": 0.0006,
      "step": 4527
    },
    {
      "epoch": 1.6567874130991584,
      "grad_norm": 0.008313198573887348,
      "learning_rate": 8.659527326440179e-06,
      "loss": 0.0003,
      "step": 4528
    },
    {
      "epoch": 1.6571533113794366,
      "grad_norm": 0.010990486480295658,
      "learning_rate": 8.650295420974889e-06,
      "loss": 0.0005,
      "step": 4529
    },
    {
      "epoch": 1.6575192096597147,
      "grad_norm": 0.007656985428184271,
      "learning_rate": 8.641063515509601e-06,
      "loss": 0.0004,
      "step": 4530
    },
    {
      "epoch": 1.6578851079399928,
      "grad_norm": 0.008708354085683823,
      "learning_rate": 8.631831610044314e-06,
      "loss": 0.0004,
      "step": 4531
    },
    {
      "epoch": 1.6582510062202709,
      "grad_norm": 0.008620901964604855,
      "learning_rate": 8.622599704579024e-06,
      "loss": 0.0004,
      "step": 4532
    },
    {
      "epoch": 1.658616904500549,
      "grad_norm": 0.011251299642026424,
      "learning_rate": 8.613367799113738e-06,
      "loss": 0.0005,
      "step": 4533
    },
    {
      "epoch": 1.6589828027808269,
      "grad_norm": 30.240386962890625,
      "learning_rate": 8.604135893648449e-06,
      "loss": 0.0288,
      "step": 4534
    },
    {
      "epoch": 1.659348701061105,
      "grad_norm": 0.025175360962748528,
      "learning_rate": 8.594903988183161e-06,
      "loss": 0.0007,
      "step": 4535
    },
    {
      "epoch": 1.659714599341383,
      "grad_norm": 0.00971155520528555,
      "learning_rate": 8.585672082717873e-06,
      "loss": 0.0004,
      "step": 4536
    },
    {
      "epoch": 1.6600804976216612,
      "grad_norm": 0.012505855411291122,
      "learning_rate": 8.576440177252585e-06,
      "loss": 0.0005,
      "step": 4537
    },
    {
      "epoch": 1.660446395901939,
      "grad_norm": 0.01703392155468464,
      "learning_rate": 8.567208271787298e-06,
      "loss": 0.0005,
      "step": 4538
    },
    {
      "epoch": 1.6608122941822172,
      "grad_norm": 0.018347354605793953,
      "learning_rate": 8.55797636632201e-06,
      "loss": 0.0005,
      "step": 4539
    },
    {
      "epoch": 1.6611781924624953,
      "grad_norm": 0.008090237155556679,
      "learning_rate": 8.54874446085672e-06,
      "loss": 0.0004,
      "step": 4540
    },
    {
      "epoch": 1.6615440907427734,
      "grad_norm": 0.007688513025641441,
      "learning_rate": 8.539512555391434e-06,
      "loss": 0.0004,
      "step": 4541
    },
    {
      "epoch": 1.6619099890230515,
      "grad_norm": 10.92933177947998,
      "learning_rate": 8.530280649926145e-06,
      "loss": 0.2086,
      "step": 4542
    },
    {
      "epoch": 1.6622758873033296,
      "grad_norm": 11.59370231628418,
      "learning_rate": 8.521048744460857e-06,
      "loss": 0.0221,
      "step": 4543
    },
    {
      "epoch": 1.6626417855836078,
      "grad_norm": 0.014081118628382683,
      "learning_rate": 8.51181683899557e-06,
      "loss": 0.0005,
      "step": 4544
    },
    {
      "epoch": 1.6630076838638859,
      "grad_norm": 0.008165184408426285,
      "learning_rate": 8.50258493353028e-06,
      "loss": 0.0004,
      "step": 4545
    },
    {
      "epoch": 1.663373582144164,
      "grad_norm": 0.010892054997384548,
      "learning_rate": 8.493353028064994e-06,
      "loss": 0.0004,
      "step": 4546
    },
    {
      "epoch": 1.663739480424442,
      "grad_norm": 0.032807815819978714,
      "learning_rate": 8.484121122599704e-06,
      "loss": 0.0008,
      "step": 4547
    },
    {
      "epoch": 1.6641053787047202,
      "grad_norm": 0.008408739231526852,
      "learning_rate": 8.474889217134417e-06,
      "loss": 0.0004,
      "step": 4548
    },
    {
      "epoch": 1.6644712769849983,
      "grad_norm": 0.02096468396484852,
      "learning_rate": 8.465657311669129e-06,
      "loss": 0.0005,
      "step": 4549
    },
    {
      "epoch": 1.6648371752652764,
      "grad_norm": 0.01338039617985487,
      "learning_rate": 8.456425406203841e-06,
      "loss": 0.0005,
      "step": 4550
    },
    {
      "epoch": 1.6652030735455543,
      "grad_norm": 0.34912508726119995,
      "learning_rate": 8.447193500738552e-06,
      "loss": 0.0093,
      "step": 4551
    },
    {
      "epoch": 1.6655689718258324,
      "grad_norm": 0.00792862381786108,
      "learning_rate": 8.437961595273266e-06,
      "loss": 0.0003,
      "step": 4552
    },
    {
      "epoch": 1.6659348701061105,
      "grad_norm": 0.007235822267830372,
      "learning_rate": 8.428729689807976e-06,
      "loss": 0.0003,
      "step": 4553
    },
    {
      "epoch": 1.6663007683863886,
      "grad_norm": 0.009063982404768467,
      "learning_rate": 8.419497784342688e-06,
      "loss": 0.0003,
      "step": 4554
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 10.895363807678223,
      "learning_rate": 8.4102658788774e-06,
      "loss": 0.0911,
      "step": 4555
    },
    {
      "epoch": 1.6670325649469446,
      "grad_norm": 0.006677835248410702,
      "learning_rate": 8.401033973412113e-06,
      "loss": 0.0003,
      "step": 4556
    },
    {
      "epoch": 1.6673984632272227,
      "grad_norm": 0.9212242364883423,
      "learning_rate": 8.391802067946825e-06,
      "loss": 0.0038,
      "step": 4557
    },
    {
      "epoch": 1.6677643615075008,
      "grad_norm": 3.5028417110443115,
      "learning_rate": 8.382570162481537e-06,
      "loss": 0.414,
      "step": 4558
    },
    {
      "epoch": 1.668130259787779,
      "grad_norm": 11.1292085647583,
      "learning_rate": 8.373338257016248e-06,
      "loss": 0.1839,
      "step": 4559
    },
    {
      "epoch": 1.668496158068057,
      "grad_norm": 1.6970285177230835,
      "learning_rate": 8.36410635155096e-06,
      "loss": 0.0149,
      "step": 4560
    },
    {
      "epoch": 1.6688620563483352,
      "grad_norm": 4.725955486297607,
      "learning_rate": 8.354874446085672e-06,
      "loss": 0.1169,
      "step": 4561
    },
    {
      "epoch": 1.6692279546286133,
      "grad_norm": 0.031149225309491158,
      "learning_rate": 8.345642540620385e-06,
      "loss": 0.0006,
      "step": 4562
    },
    {
      "epoch": 1.6695938529088914,
      "grad_norm": 0.0552603043615818,
      "learning_rate": 8.336410635155097e-06,
      "loss": 0.0009,
      "step": 4563
    },
    {
      "epoch": 1.6699597511891695,
      "grad_norm": 0.011931651271879673,
      "learning_rate": 8.327178729689807e-06,
      "loss": 0.0004,
      "step": 4564
    },
    {
      "epoch": 1.6703256494694476,
      "grad_norm": 0.12963737547397614,
      "learning_rate": 8.317946824224521e-06,
      "loss": 0.0026,
      "step": 4565
    },
    {
      "epoch": 1.6706915477497257,
      "grad_norm": 0.008385048247873783,
      "learning_rate": 8.308714918759232e-06,
      "loss": 0.0003,
      "step": 4566
    },
    {
      "epoch": 1.6710574460300036,
      "grad_norm": 0.015075314790010452,
      "learning_rate": 8.299483013293944e-06,
      "loss": 0.0005,
      "step": 4567
    },
    {
      "epoch": 1.6714233443102817,
      "grad_norm": 0.010090948082506657,
      "learning_rate": 8.290251107828656e-06,
      "loss": 0.0004,
      "step": 4568
    },
    {
      "epoch": 1.6717892425905598,
      "grad_norm": 1.0835062265396118,
      "learning_rate": 8.281019202363369e-06,
      "loss": 0.0051,
      "step": 4569
    },
    {
      "epoch": 1.672155140870838,
      "grad_norm": 0.015821881592273712,
      "learning_rate": 8.27178729689808e-06,
      "loss": 0.0006,
      "step": 4570
    },
    {
      "epoch": 1.6725210391511158,
      "grad_norm": 0.011317677795886993,
      "learning_rate": 8.262555391432793e-06,
      "loss": 0.0005,
      "step": 4571
    },
    {
      "epoch": 1.672886937431394,
      "grad_norm": 0.02474306896328926,
      "learning_rate": 8.253323485967504e-06,
      "loss": 0.0008,
      "step": 4572
    },
    {
      "epoch": 1.673252835711672,
      "grad_norm": 0.08794669061899185,
      "learning_rate": 8.244091580502216e-06,
      "loss": 0.0015,
      "step": 4573
    },
    {
      "epoch": 1.6736187339919502,
      "grad_norm": 0.6491462588310242,
      "learning_rate": 8.234859675036928e-06,
      "loss": 0.0137,
      "step": 4574
    },
    {
      "epoch": 1.6739846322722283,
      "grad_norm": 0.02334779128432274,
      "learning_rate": 8.225627769571639e-06,
      "loss": 0.0008,
      "step": 4575
    },
    {
      "epoch": 1.6743505305525064,
      "grad_norm": 0.00793495774269104,
      "learning_rate": 8.216395864106353e-06,
      "loss": 0.0003,
      "step": 4576
    },
    {
      "epoch": 1.6747164288327845,
      "grad_norm": 0.010593737475574017,
      "learning_rate": 8.207163958641063e-06,
      "loss": 0.0005,
      "step": 4577
    },
    {
      "epoch": 1.6750823271130626,
      "grad_norm": 0.010367199778556824,
      "learning_rate": 8.197932053175775e-06,
      "loss": 0.0005,
      "step": 4578
    },
    {
      "epoch": 1.6754482253933407,
      "grad_norm": 0.007179709617048502,
      "learning_rate": 8.188700147710488e-06,
      "loss": 0.0003,
      "step": 4579
    },
    {
      "epoch": 1.6758141236736188,
      "grad_norm": 5.291762828826904,
      "learning_rate": 8.1794682422452e-06,
      "loss": 0.106,
      "step": 4580
    },
    {
      "epoch": 1.676180021953897,
      "grad_norm": 8.631414413452148,
      "learning_rate": 8.170236336779912e-06,
      "loss": 0.1642,
      "step": 4581
    },
    {
      "epoch": 1.676545920234175,
      "grad_norm": 0.04212210699915886,
      "learning_rate": 8.161004431314624e-06,
      "loss": 0.0012,
      "step": 4582
    },
    {
      "epoch": 1.6769118185144531,
      "grad_norm": 0.12582211196422577,
      "learning_rate": 8.151772525849335e-06,
      "loss": 0.0016,
      "step": 4583
    },
    {
      "epoch": 1.677277716794731,
      "grad_norm": 0.027096783742308617,
      "learning_rate": 8.142540620384049e-06,
      "loss": 0.0008,
      "step": 4584
    },
    {
      "epoch": 1.6776436150750091,
      "grad_norm": 0.016842084005475044,
      "learning_rate": 8.13330871491876e-06,
      "loss": 0.0006,
      "step": 4585
    },
    {
      "epoch": 1.6780095133552873,
      "grad_norm": 0.0068128169514238834,
      "learning_rate": 8.124076809453472e-06,
      "loss": 0.0003,
      "step": 4586
    },
    {
      "epoch": 1.6783754116355654,
      "grad_norm": 0.01378614827990532,
      "learning_rate": 8.114844903988184e-06,
      "loss": 0.0005,
      "step": 4587
    },
    {
      "epoch": 1.6787413099158432,
      "grad_norm": 0.06070421636104584,
      "learning_rate": 8.105612998522894e-06,
      "loss": 0.0011,
      "step": 4588
    },
    {
      "epoch": 1.6791072081961214,
      "grad_norm": 9.009408950805664,
      "learning_rate": 8.096381093057607e-06,
      "loss": 0.1069,
      "step": 4589
    },
    {
      "epoch": 1.6794731064763995,
      "grad_norm": 0.0222628116607666,
      "learning_rate": 8.087149187592319e-06,
      "loss": 0.0007,
      "step": 4590
    },
    {
      "epoch": 1.6798390047566776,
      "grad_norm": 3.092712640762329,
      "learning_rate": 8.077917282127031e-06,
      "loss": 0.4991,
      "step": 4591
    },
    {
      "epoch": 1.6802049030369557,
      "grad_norm": 0.04631847143173218,
      "learning_rate": 8.068685376661743e-06,
      "loss": 0.001,
      "step": 4592
    },
    {
      "epoch": 1.6805708013172338,
      "grad_norm": 1.5640000104904175,
      "learning_rate": 8.059453471196456e-06,
      "loss": 0.0092,
      "step": 4593
    },
    {
      "epoch": 1.680936699597512,
      "grad_norm": 0.005580914672464132,
      "learning_rate": 8.050221565731166e-06,
      "loss": 0.0002,
      "step": 4594
    },
    {
      "epoch": 1.68130259787779,
      "grad_norm": 0.007082208059728146,
      "learning_rate": 8.04098966026588e-06,
      "loss": 0.0003,
      "step": 4595
    },
    {
      "epoch": 1.6816684961580681,
      "grad_norm": 0.0067643229849636555,
      "learning_rate": 8.03175775480059e-06,
      "loss": 0.0003,
      "step": 4596
    },
    {
      "epoch": 1.6820343944383462,
      "grad_norm": 0.061567772179841995,
      "learning_rate": 8.022525849335303e-06,
      "loss": 0.0009,
      "step": 4597
    },
    {
      "epoch": 1.6824002927186243,
      "grad_norm": 0.008541042916476727,
      "learning_rate": 8.013293943870015e-06,
      "loss": 0.0004,
      "step": 4598
    },
    {
      "epoch": 1.6827661909989025,
      "grad_norm": 0.031039448454976082,
      "learning_rate": 8.004062038404727e-06,
      "loss": 0.0011,
      "step": 4599
    },
    {
      "epoch": 1.6831320892791803,
      "grad_norm": 0.008477128110826015,
      "learning_rate": 7.99483013293944e-06,
      "loss": 0.0004,
      "step": 4600
    },
    {
      "epoch": 1.6834979875594585,
      "grad_norm": 0.046541497111320496,
      "learning_rate": 7.985598227474152e-06,
      "loss": 0.001,
      "step": 4601
    },
    {
      "epoch": 1.6838638858397366,
      "grad_norm": 0.008879323489964008,
      "learning_rate": 7.976366322008863e-06,
      "loss": 0.0004,
      "step": 4602
    },
    {
      "epoch": 1.6842297841200147,
      "grad_norm": 47.098148345947266,
      "learning_rate": 7.967134416543575e-06,
      "loss": 0.1355,
      "step": 4603
    },
    {
      "epoch": 1.6845956824002926,
      "grad_norm": 1.2735346555709839,
      "learning_rate": 7.957902511078287e-06,
      "loss": 0.0038,
      "step": 4604
    },
    {
      "epoch": 1.6849615806805707,
      "grad_norm": 1.5033329725265503,
      "learning_rate": 7.948670605612998e-06,
      "loss": 0.0043,
      "step": 4605
    },
    {
      "epoch": 1.6853274789608488,
      "grad_norm": 0.007697482127696276,
      "learning_rate": 7.939438700147711e-06,
      "loss": 0.0003,
      "step": 4606
    },
    {
      "epoch": 1.685693377241127,
      "grad_norm": 0.008911695331335068,
      "learning_rate": 7.930206794682422e-06,
      "loss": 0.0004,
      "step": 4607
    },
    {
      "epoch": 1.686059275521405,
      "grad_norm": 0.9810544848442078,
      "learning_rate": 7.920974889217134e-06,
      "loss": 0.0038,
      "step": 4608
    },
    {
      "epoch": 1.686425173801683,
      "grad_norm": 0.009392629377543926,
      "learning_rate": 7.911742983751847e-06,
      "loss": 0.0004,
      "step": 4609
    },
    {
      "epoch": 1.6867910720819612,
      "grad_norm": 0.006914577446877956,
      "learning_rate": 7.902511078286559e-06,
      "loss": 0.0003,
      "step": 4610
    },
    {
      "epoch": 1.6871569703622393,
      "grad_norm": 0.008699582889676094,
      "learning_rate": 7.893279172821271e-06,
      "loss": 0.0004,
      "step": 4611
    },
    {
      "epoch": 1.6875228686425174,
      "grad_norm": 0.006871065590530634,
      "learning_rate": 7.884047267355983e-06,
      "loss": 0.0003,
      "step": 4612
    },
    {
      "epoch": 1.6878887669227955,
      "grad_norm": 0.011279397644102573,
      "learning_rate": 7.874815361890694e-06,
      "loss": 0.0004,
      "step": 4613
    },
    {
      "epoch": 1.6882546652030737,
      "grad_norm": 0.007884033024311066,
      "learning_rate": 7.865583456425408e-06,
      "loss": 0.0004,
      "step": 4614
    },
    {
      "epoch": 1.6886205634833518,
      "grad_norm": 0.011324825696647167,
      "learning_rate": 7.856351550960118e-06,
      "loss": 0.0005,
      "step": 4615
    },
    {
      "epoch": 1.6889864617636299,
      "grad_norm": 0.01308431476354599,
      "learning_rate": 7.84711964549483e-06,
      "loss": 0.0005,
      "step": 4616
    },
    {
      "epoch": 1.6893523600439078,
      "grad_norm": 0.050224483013153076,
      "learning_rate": 7.837887740029543e-06,
      "loss": 0.0009,
      "step": 4617
    },
    {
      "epoch": 1.6897182583241859,
      "grad_norm": 0.0313914529979229,
      "learning_rate": 7.828655834564253e-06,
      "loss": 0.0005,
      "step": 4618
    },
    {
      "epoch": 1.690084156604464,
      "grad_norm": 0.022516680881381035,
      "learning_rate": 7.819423929098967e-06,
      "loss": 0.0006,
      "step": 4619
    },
    {
      "epoch": 1.690450054884742,
      "grad_norm": 0.012958125211298466,
      "learning_rate": 7.810192023633678e-06,
      "loss": 0.0004,
      "step": 4620
    },
    {
      "epoch": 1.69081595316502,
      "grad_norm": 0.02422594279050827,
      "learning_rate": 7.80096011816839e-06,
      "loss": 0.0006,
      "step": 4621
    },
    {
      "epoch": 1.691181851445298,
      "grad_norm": 6.849592208862305,
      "learning_rate": 7.791728212703102e-06,
      "loss": 0.1486,
      "step": 4622
    },
    {
      "epoch": 1.6915477497255762,
      "grad_norm": 0.005865386221557856,
      "learning_rate": 7.782496307237815e-06,
      "loss": 0.0003,
      "step": 4623
    },
    {
      "epoch": 1.6919136480058543,
      "grad_norm": 23.148677825927734,
      "learning_rate": 7.773264401772525e-06,
      "loss": 0.5749,
      "step": 4624
    },
    {
      "epoch": 1.6922795462861324,
      "grad_norm": 0.01482174452394247,
      "learning_rate": 7.764032496307239e-06,
      "loss": 0.0005,
      "step": 4625
    },
    {
      "epoch": 1.6926454445664105,
      "grad_norm": 0.010615844279527664,
      "learning_rate": 7.75480059084195e-06,
      "loss": 0.0005,
      "step": 4626
    },
    {
      "epoch": 1.6930113428466886,
      "grad_norm": 0.010749369859695435,
      "learning_rate": 7.745568685376662e-06,
      "loss": 0.0004,
      "step": 4627
    },
    {
      "epoch": 1.6933772411269667,
      "grad_norm": 0.014085769653320312,
      "learning_rate": 7.736336779911374e-06,
      "loss": 0.0006,
      "step": 4628
    },
    {
      "epoch": 1.6937431394072449,
      "grad_norm": 0.009447756223380566,
      "learning_rate": 7.727104874446086e-06,
      "loss": 0.0004,
      "step": 4629
    },
    {
      "epoch": 1.694109037687523,
      "grad_norm": 0.008274707943201065,
      "learning_rate": 7.717872968980799e-06,
      "loss": 0.0004,
      "step": 4630
    },
    {
      "epoch": 1.694474935967801,
      "grad_norm": 0.0212831012904644,
      "learning_rate": 7.708641063515509e-06,
      "loss": 0.0006,
      "step": 4631
    },
    {
      "epoch": 1.6948408342480792,
      "grad_norm": 0.1920301914215088,
      "learning_rate": 7.699409158050221e-06,
      "loss": 0.0014,
      "step": 4632
    },
    {
      "epoch": 1.695206732528357,
      "grad_norm": 0.01569179818034172,
      "learning_rate": 7.690177252584934e-06,
      "loss": 0.0004,
      "step": 4633
    },
    {
      "epoch": 1.6955726308086352,
      "grad_norm": 0.009947187267243862,
      "learning_rate": 7.680945347119646e-06,
      "loss": 0.0004,
      "step": 4634
    },
    {
      "epoch": 1.6959385290889133,
      "grad_norm": 0.006528674624860287,
      "learning_rate": 7.671713441654358e-06,
      "loss": 0.0003,
      "step": 4635
    },
    {
      "epoch": 1.6963044273691914,
      "grad_norm": 0.007171184290200472,
      "learning_rate": 7.66248153618907e-06,
      "loss": 0.0003,
      "step": 4636
    },
    {
      "epoch": 1.6966703256494693,
      "grad_norm": 0.05082575976848602,
      "learning_rate": 7.653249630723781e-06,
      "loss": 0.0007,
      "step": 4637
    },
    {
      "epoch": 1.6970362239297474,
      "grad_norm": 0.007685566321015358,
      "learning_rate": 7.644017725258495e-06,
      "loss": 0.0003,
      "step": 4638
    },
    {
      "epoch": 1.6974021222100255,
      "grad_norm": 0.010215269401669502,
      "learning_rate": 7.634785819793205e-06,
      "loss": 0.0004,
      "step": 4639
    },
    {
      "epoch": 1.6977680204903036,
      "grad_norm": 0.008297431282699108,
      "learning_rate": 7.6255539143279184e-06,
      "loss": 0.0004,
      "step": 4640
    },
    {
      "epoch": 1.6981339187705817,
      "grad_norm": 0.006869670934975147,
      "learning_rate": 7.61632200886263e-06,
      "loss": 0.0003,
      "step": 4641
    },
    {
      "epoch": 1.6984998170508598,
      "grad_norm": 0.005899280309677124,
      "learning_rate": 7.607090103397342e-06,
      "loss": 0.0003,
      "step": 4642
    },
    {
      "epoch": 1.698865715331138,
      "grad_norm": 0.38755473494529724,
      "learning_rate": 7.5978581979320535e-06,
      "loss": 0.0023,
      "step": 4643
    },
    {
      "epoch": 1.699231613611416,
      "grad_norm": 0.010320802219212055,
      "learning_rate": 7.5886262924667666e-06,
      "loss": 0.0004,
      "step": 4644
    },
    {
      "epoch": 1.6995975118916942,
      "grad_norm": 5.389504432678223,
      "learning_rate": 7.579394387001477e-06,
      "loss": 0.1027,
      "step": 4645
    },
    {
      "epoch": 1.6999634101719723,
      "grad_norm": 0.006604920607060194,
      "learning_rate": 7.5701624815361885e-06,
      "loss": 0.0003,
      "step": 4646
    },
    {
      "epoch": 1.7003293084522504,
      "grad_norm": 10.986390113830566,
      "learning_rate": 7.560930576070902e-06,
      "loss": 0.255,
      "step": 4647
    },
    {
      "epoch": 1.7006952067325285,
      "grad_norm": 0.0073693906888365746,
      "learning_rate": 7.551698670605613e-06,
      "loss": 0.0003,
      "step": 4648
    },
    {
      "epoch": 1.7010611050128064,
      "grad_norm": 9.82042121887207,
      "learning_rate": 7.542466765140325e-06,
      "loss": 0.1379,
      "step": 4649
    },
    {
      "epoch": 1.7014270032930845,
      "grad_norm": 0.006397013086825609,
      "learning_rate": 7.533234859675037e-06,
      "loss": 0.0003,
      "step": 4650
    },
    {
      "epoch": 1.7017929015733626,
      "grad_norm": 0.007113938219845295,
      "learning_rate": 7.52400295420975e-06,
      "loss": 0.0003,
      "step": 4651
    },
    {
      "epoch": 1.7021587998536407,
      "grad_norm": 0.008801727555692196,
      "learning_rate": 7.514771048744461e-06,
      "loss": 0.0004,
      "step": 4652
    },
    {
      "epoch": 1.7025246981339188,
      "grad_norm": 0.0076668234542012215,
      "learning_rate": 7.505539143279173e-06,
      "loss": 0.0003,
      "step": 4653
    },
    {
      "epoch": 1.7028905964141967,
      "grad_norm": 15.856414794921875,
      "learning_rate": 7.496307237813885e-06,
      "loss": 0.4447,
      "step": 4654
    },
    {
      "epoch": 1.7032564946944748,
      "grad_norm": 0.007742827292531729,
      "learning_rate": 7.487075332348598e-06,
      "loss": 0.0003,
      "step": 4655
    },
    {
      "epoch": 1.703622392974753,
      "grad_norm": 0.010594294406473637,
      "learning_rate": 7.477843426883309e-06,
      "loss": 0.0004,
      "step": 4656
    },
    {
      "epoch": 1.703988291255031,
      "grad_norm": 11.157734870910645,
      "learning_rate": 7.4686115214180215e-06,
      "loss": 0.0103,
      "step": 4657
    },
    {
      "epoch": 1.7043541895353091,
      "grad_norm": 48.06556701660156,
      "learning_rate": 7.459379615952733e-06,
      "loss": 0.117,
      "step": 4658
    },
    {
      "epoch": 1.7047200878155873,
      "grad_norm": 0.005951519589871168,
      "learning_rate": 7.450147710487446e-06,
      "loss": 0.0003,
      "step": 4659
    },
    {
      "epoch": 1.7050859860958654,
      "grad_norm": 0.14568471908569336,
      "learning_rate": 7.440915805022157e-06,
      "loss": 0.0009,
      "step": 4660
    },
    {
      "epoch": 1.7054518843761435,
      "grad_norm": 0.006761434953659773,
      "learning_rate": 7.431683899556868e-06,
      "loss": 0.0003,
      "step": 4661
    },
    {
      "epoch": 1.7058177826564216,
      "grad_norm": 0.6902306079864502,
      "learning_rate": 7.422451994091581e-06,
      "loss": 0.0041,
      "step": 4662
    },
    {
      "epoch": 1.7061836809366997,
      "grad_norm": 0.008671006187796593,
      "learning_rate": 7.413220088626292e-06,
      "loss": 0.0004,
      "step": 4663
    },
    {
      "epoch": 1.7065495792169778,
      "grad_norm": 0.05123284086585045,
      "learning_rate": 7.403988183161005e-06,
      "loss": 0.0008,
      "step": 4664
    },
    {
      "epoch": 1.706915477497256,
      "grad_norm": 0.7514568567276001,
      "learning_rate": 7.394756277695716e-06,
      "loss": 0.0121,
      "step": 4665
    },
    {
      "epoch": 1.7072813757775338,
      "grad_norm": 0.008397466503083706,
      "learning_rate": 7.385524372230429e-06,
      "loss": 0.0003,
      "step": 4666
    },
    {
      "epoch": 1.707647274057812,
      "grad_norm": 0.008747774176299572,
      "learning_rate": 7.3762924667651405e-06,
      "loss": 0.0003,
      "step": 4667
    },
    {
      "epoch": 1.70801317233809,
      "grad_norm": 0.008028978481888771,
      "learning_rate": 7.367060561299853e-06,
      "loss": 0.0003,
      "step": 4668
    },
    {
      "epoch": 1.7083790706183681,
      "grad_norm": 0.005971522536128759,
      "learning_rate": 7.357828655834564e-06,
      "loss": 0.0003,
      "step": 4669
    },
    {
      "epoch": 1.708744968898646,
      "grad_norm": 0.009175727143883705,
      "learning_rate": 7.348596750369277e-06,
      "loss": 0.0004,
      "step": 4670
    },
    {
      "epoch": 1.7091108671789241,
      "grad_norm": 0.2543427646160126,
      "learning_rate": 7.339364844903989e-06,
      "loss": 0.0021,
      "step": 4671
    },
    {
      "epoch": 1.7094767654592022,
      "grad_norm": 0.016028646379709244,
      "learning_rate": 7.330132939438701e-06,
      "loss": 0.0004,
      "step": 4672
    },
    {
      "epoch": 1.7098426637394804,
      "grad_norm": 8.769341468811035,
      "learning_rate": 7.320901033973412e-06,
      "loss": 0.084,
      "step": 4673
    },
    {
      "epoch": 1.7102085620197585,
      "grad_norm": 0.027341917157173157,
      "learning_rate": 7.311669128508124e-06,
      "loss": 0.0005,
      "step": 4674
    },
    {
      "epoch": 1.7105744603000366,
      "grad_norm": 5.228568077087402,
      "learning_rate": 7.302437223042837e-06,
      "loss": 0.38,
      "step": 4675
    },
    {
      "epoch": 1.7109403585803147,
      "grad_norm": 0.010488537140190601,
      "learning_rate": 7.293205317577548e-06,
      "loss": 0.0004,
      "step": 4676
    },
    {
      "epoch": 1.7113062568605928,
      "grad_norm": 3.7975244522094727,
      "learning_rate": 7.28397341211226e-06,
      "loss": 0.0172,
      "step": 4677
    },
    {
      "epoch": 1.711672155140871,
      "grad_norm": 0.007664793636649847,
      "learning_rate": 7.274741506646972e-06,
      "loss": 0.0003,
      "step": 4678
    },
    {
      "epoch": 1.712038053421149,
      "grad_norm": 0.14994093775749207,
      "learning_rate": 7.265509601181685e-06,
      "loss": 0.0011,
      "step": 4679
    },
    {
      "epoch": 1.7124039517014271,
      "grad_norm": 14.376771926879883,
      "learning_rate": 7.2562776957163954e-06,
      "loss": 0.2558,
      "step": 4680
    },
    {
      "epoch": 1.7127698499817052,
      "grad_norm": 0.2513490915298462,
      "learning_rate": 7.2470457902511085e-06,
      "loss": 0.0009,
      "step": 4681
    },
    {
      "epoch": 1.7131357482619831,
      "grad_norm": 0.008480214513838291,
      "learning_rate": 7.23781388478582e-06,
      "loss": 0.0004,
      "step": 4682
    },
    {
      "epoch": 1.7135016465422612,
      "grad_norm": 0.005163677968084812,
      "learning_rate": 7.228581979320532e-06,
      "loss": 0.0002,
      "step": 4683
    },
    {
      "epoch": 1.7138675448225393,
      "grad_norm": 0.5056856274604797,
      "learning_rate": 7.2193500738552436e-06,
      "loss": 0.0029,
      "step": 4684
    },
    {
      "epoch": 1.7142334431028174,
      "grad_norm": 4.244954586029053,
      "learning_rate": 7.210118168389957e-06,
      "loss": 0.0122,
      "step": 4685
    },
    {
      "epoch": 1.7145993413830956,
      "grad_norm": 0.006114634685218334,
      "learning_rate": 7.200886262924668e-06,
      "loss": 0.0003,
      "step": 4686
    },
    {
      "epoch": 1.7149652396633734,
      "grad_norm": 0.011173063889145851,
      "learning_rate": 7.19165435745938e-06,
      "loss": 0.0003,
      "step": 4687
    },
    {
      "epoch": 1.7153311379436516,
      "grad_norm": 0.016961142420768738,
      "learning_rate": 7.182422451994092e-06,
      "loss": 0.0006,
      "step": 4688
    },
    {
      "epoch": 1.7156970362239297,
      "grad_norm": 0.027059851214289665,
      "learning_rate": 7.173190546528803e-06,
      "loss": 0.0007,
      "step": 4689
    },
    {
      "epoch": 1.7160629345042078,
      "grad_norm": 0.006611773744225502,
      "learning_rate": 7.163958641063516e-06,
      "loss": 0.0003,
      "step": 4690
    },
    {
      "epoch": 1.7164288327844859,
      "grad_norm": 0.6792314648628235,
      "learning_rate": 7.154726735598228e-06,
      "loss": 0.0081,
      "step": 4691
    },
    {
      "epoch": 1.716794731064764,
      "grad_norm": 0.005907207261770964,
      "learning_rate": 7.14549483013294e-06,
      "loss": 0.0003,
      "step": 4692
    },
    {
      "epoch": 1.717160629345042,
      "grad_norm": 0.006221989169716835,
      "learning_rate": 7.136262924667651e-06,
      "loss": 0.0003,
      "step": 4693
    },
    {
      "epoch": 1.7175265276253202,
      "grad_norm": 0.011968581937253475,
      "learning_rate": 7.127031019202364e-06,
      "loss": 0.0005,
      "step": 4694
    },
    {
      "epoch": 1.7178924259055983,
      "grad_norm": 0.16431429982185364,
      "learning_rate": 7.117799113737076e-06,
      "loss": 0.0008,
      "step": 4695
    },
    {
      "epoch": 1.7182583241858764,
      "grad_norm": 0.019603531807661057,
      "learning_rate": 7.108567208271788e-06,
      "loss": 0.0005,
      "step": 4696
    },
    {
      "epoch": 1.7186242224661545,
      "grad_norm": 0.007898983545601368,
      "learning_rate": 7.099335302806499e-06,
      "loss": 0.0003,
      "step": 4697
    },
    {
      "epoch": 1.7189901207464326,
      "grad_norm": 0.008078859187662601,
      "learning_rate": 7.0901033973412124e-06,
      "loss": 0.0003,
      "step": 4698
    },
    {
      "epoch": 1.7193560190267105,
      "grad_norm": 2.3134567737579346,
      "learning_rate": 7.080871491875923e-06,
      "loss": 0.009,
      "step": 4699
    },
    {
      "epoch": 1.7197219173069886,
      "grad_norm": 0.01679077558219433,
      "learning_rate": 7.071639586410636e-06,
      "loss": 0.0005,
      "step": 4700
    },
    {
      "epoch": 1.7200878155872668,
      "grad_norm": 0.01670665293931961,
      "learning_rate": 7.0624076809453475e-06,
      "loss": 0.0005,
      "step": 4701
    },
    {
      "epoch": 1.7204537138675449,
      "grad_norm": 0.006987068802118301,
      "learning_rate": 7.05317577548006e-06,
      "loss": 0.0003,
      "step": 4702
    },
    {
      "epoch": 1.7208196121478228,
      "grad_norm": 0.009951764717698097,
      "learning_rate": 7.043943870014771e-06,
      "loss": 0.0003,
      "step": 4703
    },
    {
      "epoch": 1.7211855104281009,
      "grad_norm": 0.007833958603441715,
      "learning_rate": 7.0347119645494825e-06,
      "loss": 0.0003,
      "step": 4704
    },
    {
      "epoch": 1.721551408708379,
      "grad_norm": 0.01449516136199236,
      "learning_rate": 7.025480059084196e-06,
      "loss": 0.0004,
      "step": 4705
    },
    {
      "epoch": 1.721917306988657,
      "grad_norm": 0.006029164418578148,
      "learning_rate": 7.016248153618907e-06,
      "loss": 0.0003,
      "step": 4706
    },
    {
      "epoch": 1.7222832052689352,
      "grad_norm": 0.006420355290174484,
      "learning_rate": 7.007016248153619e-06,
      "loss": 0.0003,
      "step": 4707
    },
    {
      "epoch": 1.7226491035492133,
      "grad_norm": 18.260730743408203,
      "learning_rate": 6.997784342688331e-06,
      "loss": 0.1735,
      "step": 4708
    },
    {
      "epoch": 1.7230150018294914,
      "grad_norm": 0.009233391843736172,
      "learning_rate": 6.988552437223044e-06,
      "loss": 0.0004,
      "step": 4709
    },
    {
      "epoch": 1.7233809001097695,
      "grad_norm": 0.015854958444833755,
      "learning_rate": 6.979320531757755e-06,
      "loss": 0.0004,
      "step": 4710
    },
    {
      "epoch": 1.7237467983900476,
      "grad_norm": 9.600006103515625,
      "learning_rate": 6.970088626292467e-06,
      "loss": 0.0282,
      "step": 4711
    },
    {
      "epoch": 1.7241126966703257,
      "grad_norm": 0.015269165858626366,
      "learning_rate": 6.960856720827179e-06,
      "loss": 0.0005,
      "step": 4712
    },
    {
      "epoch": 1.7244785949506038,
      "grad_norm": 0.023591242730617523,
      "learning_rate": 6.951624815361892e-06,
      "loss": 0.0006,
      "step": 4713
    },
    {
      "epoch": 1.724844493230882,
      "grad_norm": 0.3472841680049896,
      "learning_rate": 6.942392909896603e-06,
      "loss": 0.0025,
      "step": 4714
    },
    {
      "epoch": 1.7252103915111598,
      "grad_norm": 0.004404689185321331,
      "learning_rate": 6.9331610044313155e-06,
      "loss": 0.0002,
      "step": 4715
    },
    {
      "epoch": 1.725576289791438,
      "grad_norm": 0.025684839114546776,
      "learning_rate": 6.923929098966027e-06,
      "loss": 0.0005,
      "step": 4716
    },
    {
      "epoch": 1.725942188071716,
      "grad_norm": 15.968782424926758,
      "learning_rate": 6.914697193500738e-06,
      "loss": 0.351,
      "step": 4717
    },
    {
      "epoch": 1.7263080863519942,
      "grad_norm": 0.008770691230893135,
      "learning_rate": 6.9054652880354505e-06,
      "loss": 0.0003,
      "step": 4718
    },
    {
      "epoch": 1.726673984632272,
      "grad_norm": 0.00570042384788394,
      "learning_rate": 6.896233382570162e-06,
      "loss": 0.0003,
      "step": 4719
    },
    {
      "epoch": 1.7270398829125502,
      "grad_norm": 0.007861558347940445,
      "learning_rate": 6.887001477104875e-06,
      "loss": 0.0003,
      "step": 4720
    },
    {
      "epoch": 1.7274057811928283,
      "grad_norm": 0.0139191634953022,
      "learning_rate": 6.877769571639586e-06,
      "loss": 0.0004,
      "step": 4721
    },
    {
      "epoch": 1.7277716794731064,
      "grad_norm": 0.02224995754659176,
      "learning_rate": 6.868537666174299e-06,
      "loss": 0.0005,
      "step": 4722
    },
    {
      "epoch": 1.7281375777533845,
      "grad_norm": 14.749820709228516,
      "learning_rate": 6.85930576070901e-06,
      "loss": 0.477,
      "step": 4723
    },
    {
      "epoch": 1.7285034760336626,
      "grad_norm": 0.06390965729951859,
      "learning_rate": 6.850073855243723e-06,
      "loss": 0.0006,
      "step": 4724
    },
    {
      "epoch": 1.7288693743139407,
      "grad_norm": 0.006414422765374184,
      "learning_rate": 6.8408419497784345e-06,
      "loss": 0.0003,
      "step": 4725
    },
    {
      "epoch": 1.7292352725942188,
      "grad_norm": 2.0967893600463867,
      "learning_rate": 6.831610044313147e-06,
      "loss": 0.5227,
      "step": 4726
    },
    {
      "epoch": 1.729601170874497,
      "grad_norm": 0.007710200268775225,
      "learning_rate": 6.822378138847858e-06,
      "loss": 0.0003,
      "step": 4727
    },
    {
      "epoch": 1.729967069154775,
      "grad_norm": 0.008473210036754608,
      "learning_rate": 6.813146233382571e-06,
      "loss": 0.0004,
      "step": 4728
    },
    {
      "epoch": 1.7303329674350532,
      "grad_norm": 4.665215969085693,
      "learning_rate": 6.803914327917283e-06,
      "loss": 0.1023,
      "step": 4729
    },
    {
      "epoch": 1.7306988657153313,
      "grad_norm": 0.01598418317735195,
      "learning_rate": 6.794682422451995e-06,
      "loss": 0.0004,
      "step": 4730
    },
    {
      "epoch": 1.7310647639956094,
      "grad_norm": 0.030173409730196,
      "learning_rate": 6.785450516986706e-06,
      "loss": 0.0008,
      "step": 4731
    },
    {
      "epoch": 1.7314306622758873,
      "grad_norm": 0.13859571516513824,
      "learning_rate": 6.776218611521418e-06,
      "loss": 0.0011,
      "step": 4732
    },
    {
      "epoch": 1.7317965605561654,
      "grad_norm": 0.007722522132098675,
      "learning_rate": 6.766986706056131e-06,
      "loss": 0.0003,
      "step": 4733
    },
    {
      "epoch": 1.7321624588364435,
      "grad_norm": 0.010388514958322048,
      "learning_rate": 6.757754800590841e-06,
      "loss": 0.0004,
      "step": 4734
    },
    {
      "epoch": 1.7325283571167216,
      "grad_norm": 1.0364561080932617,
      "learning_rate": 6.748522895125554e-06,
      "loss": 0.0048,
      "step": 4735
    },
    {
      "epoch": 1.7328942553969995,
      "grad_norm": 0.008154502138495445,
      "learning_rate": 6.739290989660266e-06,
      "loss": 0.0003,
      "step": 4736
    },
    {
      "epoch": 1.7332601536772776,
      "grad_norm": 0.006269177887588739,
      "learning_rate": 6.730059084194978e-06,
      "loss": 0.0003,
      "step": 4737
    },
    {
      "epoch": 1.7336260519575557,
      "grad_norm": 0.012951074168086052,
      "learning_rate": 6.7208271787296895e-06,
      "loss": 0.0005,
      "step": 4738
    },
    {
      "epoch": 1.7339919502378338,
      "grad_norm": 0.010530957952141762,
      "learning_rate": 6.7115952732644025e-06,
      "loss": 0.0004,
      "step": 4739
    },
    {
      "epoch": 1.734357848518112,
      "grad_norm": 0.01361516211181879,
      "learning_rate": 6.702363367799114e-06,
      "loss": 0.0005,
      "step": 4740
    },
    {
      "epoch": 1.73472374679839,
      "grad_norm": 0.006779481656849384,
      "learning_rate": 6.693131462333826e-06,
      "loss": 0.0003,
      "step": 4741
    },
    {
      "epoch": 1.7350896450786681,
      "grad_norm": 0.11548011004924774,
      "learning_rate": 6.683899556868538e-06,
      "loss": 0.0008,
      "step": 4742
    },
    {
      "epoch": 1.7354555433589463,
      "grad_norm": 24.390642166137695,
      "learning_rate": 6.674667651403251e-06,
      "loss": 0.1519,
      "step": 4743
    },
    {
      "epoch": 1.7358214416392244,
      "grad_norm": 12.77892017364502,
      "learning_rate": 6.665435745937962e-06,
      "loss": 0.1046,
      "step": 4744
    },
    {
      "epoch": 1.7361873399195025,
      "grad_norm": 2.069943428039551,
      "learning_rate": 6.656203840472674e-06,
      "loss": 0.4971,
      "step": 4745
    },
    {
      "epoch": 1.7365532381997806,
      "grad_norm": 12.073755264282227,
      "learning_rate": 6.646971935007386e-06,
      "loss": 0.1669,
      "step": 4746
    },
    {
      "epoch": 1.7369191364800587,
      "grad_norm": 0.028730526566505432,
      "learning_rate": 6.637740029542097e-06,
      "loss": 0.0006,
      "step": 4747
    },
    {
      "epoch": 1.7372850347603366,
      "grad_norm": 0.6954273581504822,
      "learning_rate": 6.62850812407681e-06,
      "loss": 0.0056,
      "step": 4748
    },
    {
      "epoch": 1.7376509330406147,
      "grad_norm": 9.859169960021973,
      "learning_rate": 6.619276218611522e-06,
      "loss": 0.2809,
      "step": 4749
    },
    {
      "epoch": 1.7380168313208928,
      "grad_norm": 0.012257029302418232,
      "learning_rate": 6.610044313146234e-06,
      "loss": 0.0004,
      "step": 4750
    },
    {
      "epoch": 1.738382729601171,
      "grad_norm": 0.01762661524116993,
      "learning_rate": 6.600812407680945e-06,
      "loss": 0.0007,
      "step": 4751
    },
    {
      "epoch": 1.7387486278814488,
      "grad_norm": 0.012276790104806423,
      "learning_rate": 6.591580502215658e-06,
      "loss": 0.0005,
      "step": 4752
    },
    {
      "epoch": 1.739114526161727,
      "grad_norm": 0.014293325133621693,
      "learning_rate": 6.582348596750369e-06,
      "loss": 0.0006,
      "step": 4753
    },
    {
      "epoch": 1.739480424442005,
      "grad_norm": 0.011826969683170319,
      "learning_rate": 6.573116691285082e-06,
      "loss": 0.0005,
      "step": 4754
    },
    {
      "epoch": 1.7398463227222831,
      "grad_norm": 0.043212782591581345,
      "learning_rate": 6.563884785819793e-06,
      "loss": 0.0009,
      "step": 4755
    },
    {
      "epoch": 1.7402122210025612,
      "grad_norm": 6.8053998947143555,
      "learning_rate": 6.554652880354506e-06,
      "loss": 0.2571,
      "step": 4756
    },
    {
      "epoch": 1.7405781192828393,
      "grad_norm": 0.6561671495437622,
      "learning_rate": 6.545420974889217e-06,
      "loss": 0.0044,
      "step": 4757
    },
    {
      "epoch": 1.7409440175631175,
      "grad_norm": 0.012691590934991837,
      "learning_rate": 6.53618906942393e-06,
      "loss": 0.0005,
      "step": 4758
    },
    {
      "epoch": 1.7413099158433956,
      "grad_norm": 0.02600737474858761,
      "learning_rate": 6.5269571639586415e-06,
      "loss": 0.001,
      "step": 4759
    },
    {
      "epoch": 1.7416758141236737,
      "grad_norm": 7.4412031173706055,
      "learning_rate": 6.517725258493354e-06,
      "loss": 0.0752,
      "step": 4760
    },
    {
      "epoch": 1.7420417124039518,
      "grad_norm": 0.017459142953157425,
      "learning_rate": 6.508493353028065e-06,
      "loss": 0.0007,
      "step": 4761
    },
    {
      "epoch": 1.74240761068423,
      "grad_norm": 0.1243622824549675,
      "learning_rate": 6.4992614475627765e-06,
      "loss": 0.0012,
      "step": 4762
    },
    {
      "epoch": 1.742773508964508,
      "grad_norm": 0.27783194184303284,
      "learning_rate": 6.49002954209749e-06,
      "loss": 0.0019,
      "step": 4763
    },
    {
      "epoch": 1.743139407244786,
      "grad_norm": 1.9676599502563477,
      "learning_rate": 6.480797636632201e-06,
      "loss": 0.4375,
      "step": 4764
    },
    {
      "epoch": 1.743505305525064,
      "grad_norm": 0.01626843586564064,
      "learning_rate": 6.471565731166913e-06,
      "loss": 0.0007,
      "step": 4765
    },
    {
      "epoch": 1.743871203805342,
      "grad_norm": 0.48939114809036255,
      "learning_rate": 6.462333825701625e-06,
      "loss": 0.0029,
      "step": 4766
    },
    {
      "epoch": 1.7442371020856202,
      "grad_norm": 0.11537767946720123,
      "learning_rate": 6.453101920236338e-06,
      "loss": 0.0022,
      "step": 4767
    },
    {
      "epoch": 1.7446030003658983,
      "grad_norm": 0.024120191112160683,
      "learning_rate": 6.443870014771049e-06,
      "loss": 0.0009,
      "step": 4768
    },
    {
      "epoch": 1.7449688986461762,
      "grad_norm": 0.4134479761123657,
      "learning_rate": 6.434638109305761e-06,
      "loss": 0.0039,
      "step": 4769
    },
    {
      "epoch": 1.7453347969264543,
      "grad_norm": 0.03034045733511448,
      "learning_rate": 6.425406203840473e-06,
      "loss": 0.0011,
      "step": 4770
    },
    {
      "epoch": 1.7457006952067324,
      "grad_norm": 0.020920658484101295,
      "learning_rate": 6.416174298375186e-06,
      "loss": 0.0009,
      "step": 4771
    },
    {
      "epoch": 1.7460665934870105,
      "grad_norm": 0.027574889361858368,
      "learning_rate": 6.406942392909896e-06,
      "loss": 0.0011,
      "step": 4772
    },
    {
      "epoch": 1.7464324917672887,
      "grad_norm": 0.028416795656085014,
      "learning_rate": 6.3977104874446095e-06,
      "loss": 0.0012,
      "step": 4773
    },
    {
      "epoch": 1.7467983900475668,
      "grad_norm": 0.04137808457016945,
      "learning_rate": 6.388478581979321e-06,
      "loss": 0.0014,
      "step": 4774
    },
    {
      "epoch": 1.7471642883278449,
      "grad_norm": 0.042137037962675095,
      "learning_rate": 6.379246676514032e-06,
      "loss": 0.0017,
      "step": 4775
    },
    {
      "epoch": 1.747530186608123,
      "grad_norm": 0.06183304637670517,
      "learning_rate": 6.3700147710487445e-06,
      "loss": 0.0014,
      "step": 4776
    },
    {
      "epoch": 1.747896084888401,
      "grad_norm": 0.1677701622247696,
      "learning_rate": 6.360782865583456e-06,
      "loss": 0.0022,
      "step": 4777
    },
    {
      "epoch": 1.7482619831686792,
      "grad_norm": 0.02942570298910141,
      "learning_rate": 6.351550960118169e-06,
      "loss": 0.0011,
      "step": 4778
    },
    {
      "epoch": 1.7486278814489573,
      "grad_norm": 0.04701344668865204,
      "learning_rate": 6.34231905465288e-06,
      "loss": 0.0014,
      "step": 4779
    },
    {
      "epoch": 1.7489937797292354,
      "grad_norm": 5.5656819343566895,
      "learning_rate": 6.333087149187593e-06,
      "loss": 0.0274,
      "step": 4780
    },
    {
      "epoch": 1.7493596780095133,
      "grad_norm": 0.04037758708000183,
      "learning_rate": 6.323855243722304e-06,
      "loss": 0.0018,
      "step": 4781
    },
    {
      "epoch": 1.7497255762897914,
      "grad_norm": 0.020080994814634323,
      "learning_rate": 6.314623338257017e-06,
      "loss": 0.0009,
      "step": 4782
    },
    {
      "epoch": 1.7500914745700695,
      "grad_norm": 0.02857017144560814,
      "learning_rate": 6.3053914327917285e-06,
      "loss": 0.0012,
      "step": 4783
    },
    {
      "epoch": 1.7504573728503476,
      "grad_norm": 0.023747719824314117,
      "learning_rate": 6.296159527326441e-06,
      "loss": 0.0009,
      "step": 4784
    },
    {
      "epoch": 1.7508232711306255,
      "grad_norm": 0.023162133991718292,
      "learning_rate": 6.286927621861152e-06,
      "loss": 0.0009,
      "step": 4785
    },
    {
      "epoch": 1.7511891694109036,
      "grad_norm": 0.03551102057099342,
      "learning_rate": 6.277695716395865e-06,
      "loss": 0.0014,
      "step": 4786
    },
    {
      "epoch": 1.7515550676911817,
      "grad_norm": 0.019589504227042198,
      "learning_rate": 6.268463810930577e-06,
      "loss": 0.0008,
      "step": 4787
    },
    {
      "epoch": 1.7519209659714599,
      "grad_norm": 0.3660137951374054,
      "learning_rate": 6.259231905465289e-06,
      "loss": 0.005,
      "step": 4788
    },
    {
      "epoch": 1.752286864251738,
      "grad_norm": 0.024247797206044197,
      "learning_rate": 6.25e-06,
      "loss": 0.001,
      "step": 4789
    },
    {
      "epoch": 1.752652762532016,
      "grad_norm": 0.8783857822418213,
      "learning_rate": 6.2407680945347125e-06,
      "loss": 0.01,
      "step": 4790
    },
    {
      "epoch": 1.7530186608122942,
      "grad_norm": 0.03814731910824776,
      "learning_rate": 6.231536189069424e-06,
      "loss": 0.0016,
      "step": 4791
    },
    {
      "epoch": 1.7533845590925723,
      "grad_norm": 0.03536950424313545,
      "learning_rate": 6.222304283604136e-06,
      "loss": 0.001,
      "step": 4792
    },
    {
      "epoch": 1.7537504573728504,
      "grad_norm": 0.02069792151451111,
      "learning_rate": 6.2130723781388484e-06,
      "loss": 0.0009,
      "step": 4793
    },
    {
      "epoch": 1.7541163556531285,
      "grad_norm": 0.024890253320336342,
      "learning_rate": 6.203840472673561e-06,
      "loss": 0.001,
      "step": 4794
    },
    {
      "epoch": 1.7544822539334066,
      "grad_norm": 0.018287187442183495,
      "learning_rate": 6.194608567208272e-06,
      "loss": 0.0007,
      "step": 4795
    },
    {
      "epoch": 1.7548481522136847,
      "grad_norm": 0.04734965041279793,
      "learning_rate": 6.185376661742984e-06,
      "loss": 0.0018,
      "step": 4796
    },
    {
      "epoch": 1.7552140504939628,
      "grad_norm": 0.02629242092370987,
      "learning_rate": 6.176144756277696e-06,
      "loss": 0.0009,
      "step": 4797
    },
    {
      "epoch": 1.7555799487742407,
      "grad_norm": 0.01703261211514473,
      "learning_rate": 6.166912850812408e-06,
      "loss": 0.0007,
      "step": 4798
    },
    {
      "epoch": 1.7559458470545188,
      "grad_norm": 0.0178449135273695,
      "learning_rate": 6.157680945347119e-06,
      "loss": 0.0007,
      "step": 4799
    },
    {
      "epoch": 1.756311745334797,
      "grad_norm": 0.07781192660331726,
      "learning_rate": 6.148449039881832e-06,
      "loss": 0.0017,
      "step": 4800
    },
    {
      "epoch": 1.756677643615075,
      "grad_norm": 0.01660052314400673,
      "learning_rate": 6.139217134416544e-06,
      "loss": 0.0007,
      "step": 4801
    },
    {
      "epoch": 1.757043541895353,
      "grad_norm": 0.012701459228992462,
      "learning_rate": 6.129985228951256e-06,
      "loss": 0.0006,
      "step": 4802
    },
    {
      "epoch": 1.757409440175631,
      "grad_norm": 0.015046636573970318,
      "learning_rate": 6.1207533234859675e-06,
      "loss": 0.0006,
      "step": 4803
    },
    {
      "epoch": 1.7577753384559092,
      "grad_norm": 0.013078782707452774,
      "learning_rate": 6.11152141802068e-06,
      "loss": 0.0006,
      "step": 4804
    },
    {
      "epoch": 1.7581412367361873,
      "grad_norm": 0.009254069067537785,
      "learning_rate": 6.102289512555392e-06,
      "loss": 0.0004,
      "step": 4805
    },
    {
      "epoch": 1.7585071350164654,
      "grad_norm": 0.008881776593625546,
      "learning_rate": 6.093057607090104e-06,
      "loss": 0.0004,
      "step": 4806
    },
    {
      "epoch": 1.7588730332967435,
      "grad_norm": 0.01391610223799944,
      "learning_rate": 6.083825701624816e-06,
      "loss": 0.0006,
      "step": 4807
    },
    {
      "epoch": 1.7592389315770216,
      "grad_norm": 0.5184784531593323,
      "learning_rate": 6.074593796159528e-06,
      "loss": 0.003,
      "step": 4808
    },
    {
      "epoch": 1.7596048298572997,
      "grad_norm": 0.016794724389910698,
      "learning_rate": 6.06536189069424e-06,
      "loss": 0.0007,
      "step": 4809
    },
    {
      "epoch": 1.7599707281375778,
      "grad_norm": 0.01809258759021759,
      "learning_rate": 6.0561299852289515e-06,
      "loss": 0.0007,
      "step": 4810
    },
    {
      "epoch": 1.760336626417856,
      "grad_norm": 0.013469984754920006,
      "learning_rate": 6.046898079763663e-06,
      "loss": 0.0006,
      "step": 4811
    },
    {
      "epoch": 1.760702524698134,
      "grad_norm": 0.018194260075688362,
      "learning_rate": 6.037666174298375e-06,
      "loss": 0.0007,
      "step": 4812
    },
    {
      "epoch": 1.7610684229784122,
      "grad_norm": 0.01608111895620823,
      "learning_rate": 6.028434268833087e-06,
      "loss": 0.0006,
      "step": 4813
    },
    {
      "epoch": 1.76143432125869,
      "grad_norm": 0.017594754695892334,
      "learning_rate": 6.0192023633678e-06,
      "loss": 0.0007,
      "step": 4814
    },
    {
      "epoch": 1.7618002195389681,
      "grad_norm": 0.01097502838820219,
      "learning_rate": 6.009970457902511e-06,
      "loss": 0.0004,
      "step": 4815
    },
    {
      "epoch": 1.7621661178192463,
      "grad_norm": 0.013898756355047226,
      "learning_rate": 6.000738552437223e-06,
      "loss": 0.0005,
      "step": 4816
    },
    {
      "epoch": 1.7625320160995244,
      "grad_norm": 0.024400416761636734,
      "learning_rate": 5.9915066469719355e-06,
      "loss": 0.0007,
      "step": 4817
    },
    {
      "epoch": 1.7628979143798023,
      "grad_norm": 0.049069635570049286,
      "learning_rate": 5.982274741506647e-06,
      "loss": 0.0007,
      "step": 4818
    },
    {
      "epoch": 1.7632638126600804,
      "grad_norm": 0.014142762869596481,
      "learning_rate": 5.973042836041359e-06,
      "loss": 0.0006,
      "step": 4819
    },
    {
      "epoch": 1.7636297109403585,
      "grad_norm": 0.012584250420331955,
      "learning_rate": 5.963810930576071e-06,
      "loss": 0.0005,
      "step": 4820
    },
    {
      "epoch": 1.7639956092206366,
      "grad_norm": 0.01697264239192009,
      "learning_rate": 5.954579025110784e-06,
      "loss": 0.0006,
      "step": 4821
    },
    {
      "epoch": 1.7643615075009147,
      "grad_norm": 0.016599752008914948,
      "learning_rate": 5.945347119645495e-06,
      "loss": 0.0007,
      "step": 4822
    },
    {
      "epoch": 1.7647274057811928,
      "grad_norm": 0.07528334856033325,
      "learning_rate": 5.936115214180207e-06,
      "loss": 0.0011,
      "step": 4823
    },
    {
      "epoch": 1.765093304061471,
      "grad_norm": 0.018646573647856712,
      "learning_rate": 5.9268833087149195e-06,
      "loss": 0.0008,
      "step": 4824
    },
    {
      "epoch": 1.765459202341749,
      "grad_norm": 0.2575817406177521,
      "learning_rate": 5.917651403249632e-06,
      "loss": 0.0018,
      "step": 4825
    },
    {
      "epoch": 1.7658251006220271,
      "grad_norm": 0.022359369322657585,
      "learning_rate": 5.908419497784342e-06,
      "loss": 0.0009,
      "step": 4826
    },
    {
      "epoch": 1.7661909989023052,
      "grad_norm": 0.009948653168976307,
      "learning_rate": 5.8991875923190545e-06,
      "loss": 0.0004,
      "step": 4827
    },
    {
      "epoch": 1.7665568971825834,
      "grad_norm": 0.022069571539759636,
      "learning_rate": 5.889955686853767e-06,
      "loss": 0.0007,
      "step": 4828
    },
    {
      "epoch": 1.7669227954628615,
      "grad_norm": 0.24851514399051666,
      "learning_rate": 5.880723781388479e-06,
      "loss": 0.0025,
      "step": 4829
    },
    {
      "epoch": 1.7672886937431396,
      "grad_norm": 0.01603434421122074,
      "learning_rate": 5.87149187592319e-06,
      "loss": 0.0007,
      "step": 4830
    },
    {
      "epoch": 1.7676545920234175,
      "grad_norm": 0.010069811716675758,
      "learning_rate": 5.862259970457903e-06,
      "loss": 0.0004,
      "step": 4831
    },
    {
      "epoch": 1.7680204903036956,
      "grad_norm": 0.0132294911891222,
      "learning_rate": 5.853028064992615e-06,
      "loss": 0.0005,
      "step": 4832
    },
    {
      "epoch": 1.7683863885839737,
      "grad_norm": 0.007609691936522722,
      "learning_rate": 5.843796159527327e-06,
      "loss": 0.0003,
      "step": 4833
    },
    {
      "epoch": 1.7687522868642518,
      "grad_norm": 0.013900397345423698,
      "learning_rate": 5.8345642540620385e-06,
      "loss": 0.0005,
      "step": 4834
    },
    {
      "epoch": 1.7691181851445297,
      "grad_norm": 0.013567395508289337,
      "learning_rate": 5.825332348596751e-06,
      "loss": 0.0005,
      "step": 4835
    },
    {
      "epoch": 1.7694840834248078,
      "grad_norm": 0.04742130637168884,
      "learning_rate": 5.816100443131463e-06,
      "loss": 0.0006,
      "step": 4836
    },
    {
      "epoch": 1.769849981705086,
      "grad_norm": 0.01821953058242798,
      "learning_rate": 5.806868537666174e-06,
      "loss": 0.0007,
      "step": 4837
    },
    {
      "epoch": 1.770215879985364,
      "grad_norm": 0.009889429435133934,
      "learning_rate": 5.797636632200887e-06,
      "loss": 0.0004,
      "step": 4838
    },
    {
      "epoch": 1.7705817782656421,
      "grad_norm": 0.012130777351558208,
      "learning_rate": 5.788404726735599e-06,
      "loss": 0.0005,
      "step": 4839
    },
    {
      "epoch": 1.7709476765459202,
      "grad_norm": 0.2021130621433258,
      "learning_rate": 5.77917282127031e-06,
      "loss": 0.0012,
      "step": 4840
    },
    {
      "epoch": 1.7713135748261983,
      "grad_norm": 0.012512750923633575,
      "learning_rate": 5.7699409158050225e-06,
      "loss": 0.0005,
      "step": 4841
    },
    {
      "epoch": 1.7716794731064764,
      "grad_norm": 0.01460105087608099,
      "learning_rate": 5.760709010339734e-06,
      "loss": 0.0005,
      "step": 4842
    },
    {
      "epoch": 1.7720453713867546,
      "grad_norm": 0.010385612025856972,
      "learning_rate": 5.751477104874446e-06,
      "loss": 0.0004,
      "step": 4843
    },
    {
      "epoch": 1.7724112696670327,
      "grad_norm": 0.01248529925942421,
      "learning_rate": 5.7422451994091584e-06,
      "loss": 0.0005,
      "step": 4844
    },
    {
      "epoch": 1.7727771679473108,
      "grad_norm": 0.01604003645479679,
      "learning_rate": 5.73301329394387e-06,
      "loss": 0.0006,
      "step": 4845
    },
    {
      "epoch": 1.7731430662275889,
      "grad_norm": 0.01304524578154087,
      "learning_rate": 5.723781388478582e-06,
      "loss": 0.0005,
      "step": 4846
    },
    {
      "epoch": 1.7735089645078668,
      "grad_norm": 0.030267605558037758,
      "learning_rate": 5.714549483013294e-06,
      "loss": 0.0008,
      "step": 4847
    },
    {
      "epoch": 1.7738748627881449,
      "grad_norm": 0.014906746335327625,
      "learning_rate": 5.7053175775480065e-06,
      "loss": 0.0006,
      "step": 4848
    },
    {
      "epoch": 1.774240761068423,
      "grad_norm": 0.1036069244146347,
      "learning_rate": 5.696085672082718e-06,
      "loss": 0.0012,
      "step": 4849
    },
    {
      "epoch": 1.774606659348701,
      "grad_norm": 0.012994305230677128,
      "learning_rate": 5.68685376661743e-06,
      "loss": 0.0005,
      "step": 4850
    },
    {
      "epoch": 1.774972557628979,
      "grad_norm": 0.006454347632825375,
      "learning_rate": 5.6776218611521424e-06,
      "loss": 0.0003,
      "step": 4851
    },
    {
      "epoch": 1.775338455909257,
      "grad_norm": 1.391870379447937,
      "learning_rate": 5.668389955686855e-06,
      "loss": 0.0057,
      "step": 4852
    },
    {
      "epoch": 1.7757043541895352,
      "grad_norm": 0.011097656562924385,
      "learning_rate": 5.659158050221566e-06,
      "loss": 0.0004,
      "step": 4853
    },
    {
      "epoch": 1.7760702524698133,
      "grad_norm": 0.5588937401771545,
      "learning_rate": 5.6499261447562775e-06,
      "loss": 0.0091,
      "step": 4854
    },
    {
      "epoch": 1.7764361507500914,
      "grad_norm": 0.008543158881366253,
      "learning_rate": 5.64069423929099e-06,
      "loss": 0.0004,
      "step": 4855
    },
    {
      "epoch": 1.7768020490303695,
      "grad_norm": 0.036191124469041824,
      "learning_rate": 5.631462333825702e-06,
      "loss": 0.0009,
      "step": 4856
    },
    {
      "epoch": 1.7771679473106476,
      "grad_norm": 0.01418816763907671,
      "learning_rate": 5.622230428360413e-06,
      "loss": 0.0005,
      "step": 4857
    },
    {
      "epoch": 1.7775338455909258,
      "grad_norm": 0.008733364753425121,
      "learning_rate": 5.612998522895126e-06,
      "loss": 0.0003,
      "step": 4858
    },
    {
      "epoch": 1.7778997438712039,
      "grad_norm": 0.011488579213619232,
      "learning_rate": 5.603766617429838e-06,
      "loss": 0.0005,
      "step": 4859
    },
    {
      "epoch": 1.778265642151482,
      "grad_norm": 0.010236766189336777,
      "learning_rate": 5.59453471196455e-06,
      "loss": 0.0004,
      "step": 4860
    },
    {
      "epoch": 1.77863154043176,
      "grad_norm": 0.012938511557877064,
      "learning_rate": 5.5853028064992615e-06,
      "loss": 0.0005,
      "step": 4861
    },
    {
      "epoch": 1.7789974387120382,
      "grad_norm": 0.012436051853001118,
      "learning_rate": 5.576070901033974e-06,
      "loss": 0.0005,
      "step": 4862
    },
    {
      "epoch": 1.7793633369923163,
      "grad_norm": 10.745953559875488,
      "learning_rate": 5.566838995568686e-06,
      "loss": 0.1254,
      "step": 4863
    },
    {
      "epoch": 1.7797292352725942,
      "grad_norm": 0.008603501133620739,
      "learning_rate": 5.557607090103397e-06,
      "loss": 0.0003,
      "step": 4864
    },
    {
      "epoch": 1.7800951335528723,
      "grad_norm": 0.01137317530810833,
      "learning_rate": 5.54837518463811e-06,
      "loss": 0.0005,
      "step": 4865
    },
    {
      "epoch": 1.7804610318331504,
      "grad_norm": 0.010331947356462479,
      "learning_rate": 5.539143279172822e-06,
      "loss": 0.0004,
      "step": 4866
    },
    {
      "epoch": 1.7808269301134285,
      "grad_norm": 0.013525458052754402,
      "learning_rate": 5.529911373707534e-06,
      "loss": 0.0004,
      "step": 4867
    },
    {
      "epoch": 1.7811928283937064,
      "grad_norm": 0.017687194049358368,
      "learning_rate": 5.5206794682422455e-06,
      "loss": 0.0005,
      "step": 4868
    },
    {
      "epoch": 1.7815587266739845,
      "grad_norm": 0.14228355884552002,
      "learning_rate": 5.511447562776957e-06,
      "loss": 0.0015,
      "step": 4869
    },
    {
      "epoch": 1.7819246249542626,
      "grad_norm": 0.015248517505824566,
      "learning_rate": 5.502215657311669e-06,
      "loss": 0.0005,
      "step": 4870
    },
    {
      "epoch": 1.7822905232345407,
      "grad_norm": 0.013238871470093727,
      "learning_rate": 5.492983751846381e-06,
      "loss": 0.0004,
      "step": 4871
    },
    {
      "epoch": 1.7826564215148188,
      "grad_norm": 0.013202013447880745,
      "learning_rate": 5.483751846381093e-06,
      "loss": 0.0005,
      "step": 4872
    },
    {
      "epoch": 1.783022319795097,
      "grad_norm": 0.013062858022749424,
      "learning_rate": 5.474519940915805e-06,
      "loss": 0.0004,
      "step": 4873
    },
    {
      "epoch": 1.783388218075375,
      "grad_norm": 0.008363430388271809,
      "learning_rate": 5.465288035450517e-06,
      "loss": 0.0004,
      "step": 4874
    },
    {
      "epoch": 1.7837541163556532,
      "grad_norm": 1.9572614431381226,
      "learning_rate": 5.4560561299852295e-06,
      "loss": 0.4475,
      "step": 4875
    },
    {
      "epoch": 1.7841200146359313,
      "grad_norm": 0.009263087995350361,
      "learning_rate": 5.446824224519941e-06,
      "loss": 0.0004,
      "step": 4876
    },
    {
      "epoch": 1.7844859129162094,
      "grad_norm": 0.00846378318965435,
      "learning_rate": 5.437592319054653e-06,
      "loss": 0.0004,
      "step": 4877
    },
    {
      "epoch": 1.7848518111964875,
      "grad_norm": 0.00985461100935936,
      "learning_rate": 5.428360413589365e-06,
      "loss": 0.0004,
      "step": 4878
    },
    {
      "epoch": 1.7852177094767656,
      "grad_norm": 0.014840880408883095,
      "learning_rate": 5.419128508124078e-06,
      "loss": 0.0006,
      "step": 4879
    },
    {
      "epoch": 1.7855836077570435,
      "grad_norm": 0.009368476457893848,
      "learning_rate": 5.409896602658789e-06,
      "loss": 0.0004,
      "step": 4880
    },
    {
      "epoch": 1.7859495060373216,
      "grad_norm": 14.354782104492188,
      "learning_rate": 5.400664697193501e-06,
      "loss": 0.3605,
      "step": 4881
    },
    {
      "epoch": 1.7863154043175997,
      "grad_norm": 2.239686965942383,
      "learning_rate": 5.3914327917282135e-06,
      "loss": 0.419,
      "step": 4882
    },
    {
      "epoch": 1.7866813025978778,
      "grad_norm": 0.008027661591768265,
      "learning_rate": 5.382200886262925e-06,
      "loss": 0.0004,
      "step": 4883
    },
    {
      "epoch": 1.7870472008781557,
      "grad_norm": 0.018590319901704788,
      "learning_rate": 5.372968980797636e-06,
      "loss": 0.0007,
      "step": 4884
    },
    {
      "epoch": 1.7874130991584338,
      "grad_norm": 0.00981917418539524,
      "learning_rate": 5.3637370753323485e-06,
      "loss": 0.0004,
      "step": 4885
    },
    {
      "epoch": 1.787778997438712,
      "grad_norm": 0.015987545251846313,
      "learning_rate": 5.354505169867061e-06,
      "loss": 0.0006,
      "step": 4886
    },
    {
      "epoch": 1.78814489571899,
      "grad_norm": 0.01028453093022108,
      "learning_rate": 5.345273264401773e-06,
      "loss": 0.0004,
      "step": 4887
    },
    {
      "epoch": 1.7885107939992682,
      "grad_norm": 0.021465029567480087,
      "learning_rate": 5.336041358936484e-06,
      "loss": 0.0008,
      "step": 4888
    },
    {
      "epoch": 1.7888766922795463,
      "grad_norm": 0.015362144447863102,
      "learning_rate": 5.326809453471197e-06,
      "loss": 0.0006,
      "step": 4889
    },
    {
      "epoch": 1.7892425905598244,
      "grad_norm": 0.01837424747645855,
      "learning_rate": 5.317577548005909e-06,
      "loss": 0.0008,
      "step": 4890
    },
    {
      "epoch": 1.7896084888401025,
      "grad_norm": 0.030078314244747162,
      "learning_rate": 5.30834564254062e-06,
      "loss": 0.0009,
      "step": 4891
    },
    {
      "epoch": 1.7899743871203806,
      "grad_norm": 0.022654345259070396,
      "learning_rate": 5.2991137370753325e-06,
      "loss": 0.0009,
      "step": 4892
    },
    {
      "epoch": 1.7903402854006587,
      "grad_norm": 0.013421183452010155,
      "learning_rate": 5.289881831610045e-06,
      "loss": 0.0006,
      "step": 4893
    },
    {
      "epoch": 1.7907061836809368,
      "grad_norm": 0.018194321542978287,
      "learning_rate": 5.280649926144757e-06,
      "loss": 0.0007,
      "step": 4894
    },
    {
      "epoch": 1.791072081961215,
      "grad_norm": 0.016654353588819504,
      "learning_rate": 5.2714180206794684e-06,
      "loss": 0.0006,
      "step": 4895
    },
    {
      "epoch": 1.7914379802414928,
      "grad_norm": 0.015356470830738544,
      "learning_rate": 5.262186115214181e-06,
      "loss": 0.0006,
      "step": 4896
    },
    {
      "epoch": 1.791803878521771,
      "grad_norm": 0.021786728873848915,
      "learning_rate": 5.252954209748893e-06,
      "loss": 0.0008,
      "step": 4897
    },
    {
      "epoch": 1.792169776802049,
      "grad_norm": 0.04768887907266617,
      "learning_rate": 5.243722304283604e-06,
      "loss": 0.0013,
      "step": 4898
    },
    {
      "epoch": 1.7925356750823271,
      "grad_norm": 0.01827170141041279,
      "learning_rate": 5.234490398818316e-06,
      "loss": 0.0007,
      "step": 4899
    },
    {
      "epoch": 1.7929015733626052,
      "grad_norm": 12.378630638122559,
      "learning_rate": 5.225258493353028e-06,
      "loss": 0.1757,
      "step": 4900
    },
    {
      "epoch": 1.7932674716428831,
      "grad_norm": 0.008133374154567719,
      "learning_rate": 5.21602658788774e-06,
      "loss": 0.0003,
      "step": 4901
    },
    {
      "epoch": 1.7936333699231612,
      "grad_norm": 0.022028792649507523,
      "learning_rate": 5.2067946824224524e-06,
      "loss": 0.0008,
      "step": 4902
    },
    {
      "epoch": 1.7939992682034394,
      "grad_norm": 0.011177590116858482,
      "learning_rate": 5.197562776957164e-06,
      "loss": 0.0005,
      "step": 4903
    },
    {
      "epoch": 1.7943651664837175,
      "grad_norm": 0.02515394054353237,
      "learning_rate": 5.188330871491876e-06,
      "loss": 0.001,
      "step": 4904
    },
    {
      "epoch": 1.7947310647639956,
      "grad_norm": 0.5188920497894287,
      "learning_rate": 5.179098966026588e-06,
      "loss": 0.0031,
      "step": 4905
    },
    {
      "epoch": 1.7950969630442737,
      "grad_norm": 0.02330586314201355,
      "learning_rate": 5.1698670605613006e-06,
      "loss": 0.0009,
      "step": 4906
    },
    {
      "epoch": 1.7954628613245518,
      "grad_norm": 4.957740783691406,
      "learning_rate": 5.160635155096012e-06,
      "loss": 0.0117,
      "step": 4907
    },
    {
      "epoch": 1.79582875960483,
      "grad_norm": 0.01668916642665863,
      "learning_rate": 5.151403249630724e-06,
      "loss": 0.0006,
      "step": 4908
    },
    {
      "epoch": 1.796194657885108,
      "grad_norm": 0.025648757815361023,
      "learning_rate": 5.1421713441654364e-06,
      "loss": 0.001,
      "step": 4909
    },
    {
      "epoch": 1.7965605561653861,
      "grad_norm": 0.0173764880746603,
      "learning_rate": 5.132939438700148e-06,
      "loss": 0.0007,
      "step": 4910
    },
    {
      "epoch": 1.7969264544456642,
      "grad_norm": 0.8617385625839233,
      "learning_rate": 5.12370753323486e-06,
      "loss": 0.004,
      "step": 4911
    },
    {
      "epoch": 1.7972923527259423,
      "grad_norm": 0.04201683774590492,
      "learning_rate": 5.1144756277695715e-06,
      "loss": 0.0007,
      "step": 4912
    },
    {
      "epoch": 1.7976582510062202,
      "grad_norm": 0.01674499735236168,
      "learning_rate": 5.105243722304284e-06,
      "loss": 0.0007,
      "step": 4913
    },
    {
      "epoch": 1.7980241492864983,
      "grad_norm": 0.02477574348449707,
      "learning_rate": 5.096011816838996e-06,
      "loss": 0.0008,
      "step": 4914
    },
    {
      "epoch": 1.7983900475667765,
      "grad_norm": 0.03906891122460365,
      "learning_rate": 5.086779911373707e-06,
      "loss": 0.0008,
      "step": 4915
    },
    {
      "epoch": 1.7987559458470546,
      "grad_norm": 0.01542702503502369,
      "learning_rate": 5.07754800590842e-06,
      "loss": 0.0006,
      "step": 4916
    },
    {
      "epoch": 1.7991218441273324,
      "grad_norm": 0.012354147620499134,
      "learning_rate": 5.068316100443132e-06,
      "loss": 0.0005,
      "step": 4917
    },
    {
      "epoch": 1.7994877424076106,
      "grad_norm": 0.7439084053039551,
      "learning_rate": 5.059084194977843e-06,
      "loss": 0.0032,
      "step": 4918
    },
    {
      "epoch": 1.7998536406878887,
      "grad_norm": 0.01663884148001671,
      "learning_rate": 5.0498522895125555e-06,
      "loss": 0.0006,
      "step": 4919
    },
    {
      "epoch": 1.8002195389681668,
      "grad_norm": 0.013184069655835629,
      "learning_rate": 5.040620384047268e-06,
      "loss": 0.0005,
      "step": 4920
    },
    {
      "epoch": 1.8005854372484449,
      "grad_norm": 7.665675163269043,
      "learning_rate": 5.03138847858198e-06,
      "loss": 0.1972,
      "step": 4921
    },
    {
      "epoch": 1.800951335528723,
      "grad_norm": 5.657801628112793,
      "learning_rate": 5.022156573116691e-06,
      "loss": 0.4009,
      "step": 4922
    },
    {
      "epoch": 1.801317233809001,
      "grad_norm": 0.012794293463230133,
      "learning_rate": 5.012924667651404e-06,
      "loss": 0.0005,
      "step": 4923
    },
    {
      "epoch": 1.8016831320892792,
      "grad_norm": 0.021114248782396317,
      "learning_rate": 5.003692762186116e-06,
      "loss": 0.0008,
      "step": 4924
    },
    {
      "epoch": 1.8020490303695573,
      "grad_norm": 0.01568901166319847,
      "learning_rate": 4.994460856720828e-06,
      "loss": 0.0006,
      "step": 4925
    },
    {
      "epoch": 1.8024149286498354,
      "grad_norm": 0.026749324053525925,
      "learning_rate": 4.985228951255539e-06,
      "loss": 0.0011,
      "step": 4926
    },
    {
      "epoch": 1.8027808269301135,
      "grad_norm": 0.024909749627113342,
      "learning_rate": 4.975997045790251e-06,
      "loss": 0.0009,
      "step": 4927
    },
    {
      "epoch": 1.8031467252103917,
      "grad_norm": 0.018157774582505226,
      "learning_rate": 4.966765140324963e-06,
      "loss": 0.0008,
      "step": 4928
    },
    {
      "epoch": 1.8035126234906695,
      "grad_norm": 0.3597557246685028,
      "learning_rate": 4.957533234859675e-06,
      "loss": 0.005,
      "step": 4929
    },
    {
      "epoch": 1.8038785217709477,
      "grad_norm": 0.018401585519313812,
      "learning_rate": 4.948301329394387e-06,
      "loss": 0.0007,
      "step": 4930
    },
    {
      "epoch": 1.8042444200512258,
      "grad_norm": 0.01871640980243683,
      "learning_rate": 4.939069423929099e-06,
      "loss": 0.0007,
      "step": 4931
    },
    {
      "epoch": 1.8046103183315039,
      "grad_norm": 0.012966001406311989,
      "learning_rate": 4.929837518463811e-06,
      "loss": 0.0005,
      "step": 4932
    },
    {
      "epoch": 1.8049762166117818,
      "grad_norm": 0.04625115543603897,
      "learning_rate": 4.9206056129985235e-06,
      "loss": 0.001,
      "step": 4933
    },
    {
      "epoch": 1.8053421148920599,
      "grad_norm": 0.021082838997244835,
      "learning_rate": 4.911373707533235e-06,
      "loss": 0.0007,
      "step": 4934
    },
    {
      "epoch": 1.805708013172338,
      "grad_norm": 0.013976296409964561,
      "learning_rate": 4.902141802067947e-06,
      "loss": 0.0005,
      "step": 4935
    },
    {
      "epoch": 1.806073911452616,
      "grad_norm": 0.023857982829213142,
      "learning_rate": 4.892909896602659e-06,
      "loss": 0.0008,
      "step": 4936
    },
    {
      "epoch": 1.8064398097328942,
      "grad_norm": 0.013137158937752247,
      "learning_rate": 4.883677991137371e-06,
      "loss": 0.0005,
      "step": 4937
    },
    {
      "epoch": 1.8068057080131723,
      "grad_norm": 0.04421281814575195,
      "learning_rate": 4.874446085672083e-06,
      "loss": 0.0009,
      "step": 4938
    },
    {
      "epoch": 1.8071716062934504,
      "grad_norm": 0.01756793074309826,
      "learning_rate": 4.865214180206795e-06,
      "loss": 0.0007,
      "step": 4939
    },
    {
      "epoch": 1.8075375045737285,
      "grad_norm": 0.08212804794311523,
      "learning_rate": 4.8559822747415075e-06,
      "loss": 0.001,
      "step": 4940
    },
    {
      "epoch": 1.8079034028540066,
      "grad_norm": 19.071083068847656,
      "learning_rate": 4.846750369276219e-06,
      "loss": 0.2194,
      "step": 4941
    },
    {
      "epoch": 1.8082693011342847,
      "grad_norm": 0.012774254195392132,
      "learning_rate": 4.83751846381093e-06,
      "loss": 0.0005,
      "step": 4942
    },
    {
      "epoch": 1.8086351994145629,
      "grad_norm": 0.0089265126734972,
      "learning_rate": 4.8282865583456425e-06,
      "loss": 0.0004,
      "step": 4943
    },
    {
      "epoch": 1.809001097694841,
      "grad_norm": 24.24550437927246,
      "learning_rate": 4.819054652880355e-06,
      "loss": 0.1616,
      "step": 4944
    },
    {
      "epoch": 1.809366995975119,
      "grad_norm": 34.21855926513672,
      "learning_rate": 4.809822747415066e-06,
      "loss": 0.1067,
      "step": 4945
    },
    {
      "epoch": 1.809732894255397,
      "grad_norm": 0.018026933073997498,
      "learning_rate": 4.800590841949778e-06,
      "loss": 0.0007,
      "step": 4946
    },
    {
      "epoch": 1.810098792535675,
      "grad_norm": 0.5506243705749512,
      "learning_rate": 4.791358936484491e-06,
      "loss": 0.0036,
      "step": 4947
    },
    {
      "epoch": 1.8104646908159532,
      "grad_norm": 0.01574215665459633,
      "learning_rate": 4.782127031019203e-06,
      "loss": 0.0006,
      "step": 4948
    },
    {
      "epoch": 1.8108305890962313,
      "grad_norm": 14.369535446166992,
      "learning_rate": 4.772895125553914e-06,
      "loss": 0.033,
      "step": 4949
    },
    {
      "epoch": 1.8111964873765092,
      "grad_norm": 0.01334299985319376,
      "learning_rate": 4.7636632200886265e-06,
      "loss": 0.0005,
      "step": 4950
    },
    {
      "epoch": 1.8115623856567873,
      "grad_norm": 0.02556615322828293,
      "learning_rate": 4.754431314623339e-06,
      "loss": 0.0007,
      "step": 4951
    },
    {
      "epoch": 1.8119282839370654,
      "grad_norm": 0.014703619293868542,
      "learning_rate": 4.745199409158051e-06,
      "loss": 0.0006,
      "step": 4952
    },
    {
      "epoch": 1.8122941822173435,
      "grad_norm": 0.017920127138495445,
      "learning_rate": 4.7359675036927624e-06,
      "loss": 0.0007,
      "step": 4953
    },
    {
      "epoch": 1.8126600804976216,
      "grad_norm": 0.015789056196808815,
      "learning_rate": 4.726735598227475e-06,
      "loss": 0.0004,
      "step": 4954
    },
    {
      "epoch": 1.8130259787778997,
      "grad_norm": 54.5548095703125,
      "learning_rate": 4.717503692762186e-06,
      "loss": 0.1132,
      "step": 4955
    },
    {
      "epoch": 1.8133918770581778,
      "grad_norm": 0.01211879588663578,
      "learning_rate": 4.708271787296898e-06,
      "loss": 0.0005,
      "step": 4956
    },
    {
      "epoch": 1.813757775338456,
      "grad_norm": 0.014441709034144878,
      "learning_rate": 4.69903988183161e-06,
      "loss": 0.0005,
      "step": 4957
    },
    {
      "epoch": 1.814123673618734,
      "grad_norm": 0.013029094785451889,
      "learning_rate": 4.689807976366322e-06,
      "loss": 0.0005,
      "step": 4958
    },
    {
      "epoch": 1.8144895718990122,
      "grad_norm": 0.019359661266207695,
      "learning_rate": 4.680576070901034e-06,
      "loss": 0.0007,
      "step": 4959
    },
    {
      "epoch": 1.8148554701792903,
      "grad_norm": 0.014423462562263012,
      "learning_rate": 4.6713441654357464e-06,
      "loss": 0.0006,
      "step": 4960
    },
    {
      "epoch": 1.8152213684595684,
      "grad_norm": 0.010835761204361916,
      "learning_rate": 4.662112259970458e-06,
      "loss": 0.0004,
      "step": 4961
    },
    {
      "epoch": 1.8155872667398463,
      "grad_norm": 0.009685625322163105,
      "learning_rate": 4.65288035450517e-06,
      "loss": 0.0004,
      "step": 4962
    },
    {
      "epoch": 1.8159531650201244,
      "grad_norm": 0.01050441712141037,
      "learning_rate": 4.643648449039882e-06,
      "loss": 0.0004,
      "step": 4963
    },
    {
      "epoch": 1.8163190633004025,
      "grad_norm": 0.011874141171574593,
      "learning_rate": 4.634416543574594e-06,
      "loss": 0.0005,
      "step": 4964
    },
    {
      "epoch": 1.8166849615806806,
      "grad_norm": 0.04071520268917084,
      "learning_rate": 4.625184638109306e-06,
      "loss": 0.0009,
      "step": 4965
    },
    {
      "epoch": 1.8170508598609585,
      "grad_norm": 0.009245752356946468,
      "learning_rate": 4.615952732644018e-06,
      "loss": 0.0004,
      "step": 4966
    },
    {
      "epoch": 1.8174167581412366,
      "grad_norm": 0.010844600386917591,
      "learning_rate": 4.6067208271787304e-06,
      "loss": 0.0004,
      "step": 4967
    },
    {
      "epoch": 1.8177826564215147,
      "grad_norm": 0.019993355497717857,
      "learning_rate": 4.597488921713442e-06,
      "loss": 0.0008,
      "step": 4968
    },
    {
      "epoch": 1.8181485547017928,
      "grad_norm": 0.013079262338578701,
      "learning_rate": 4.588257016248154e-06,
      "loss": 0.0005,
      "step": 4969
    },
    {
      "epoch": 1.818514452982071,
      "grad_norm": 0.007195820100605488,
      "learning_rate": 4.5790251107828655e-06,
      "loss": 0.0003,
      "step": 4970
    },
    {
      "epoch": 1.818880351262349,
      "grad_norm": 0.011214781552553177,
      "learning_rate": 4.569793205317578e-06,
      "loss": 0.0005,
      "step": 4971
    },
    {
      "epoch": 1.8192462495426271,
      "grad_norm": 0.011769532226026058,
      "learning_rate": 4.560561299852289e-06,
      "loss": 0.0005,
      "step": 4972
    },
    {
      "epoch": 1.8196121478229053,
      "grad_norm": 0.016082413494586945,
      "learning_rate": 4.551329394387001e-06,
      "loss": 0.0006,
      "step": 4973
    },
    {
      "epoch": 1.8199780461031834,
      "grad_norm": 0.011728244833648205,
      "learning_rate": 4.542097488921714e-06,
      "loss": 0.0005,
      "step": 4974
    },
    {
      "epoch": 1.8203439443834615,
      "grad_norm": 0.0060065798461437225,
      "learning_rate": 4.532865583456426e-06,
      "loss": 0.0003,
      "step": 4975
    },
    {
      "epoch": 1.8207098426637396,
      "grad_norm": 0.01385668944567442,
      "learning_rate": 4.523633677991137e-06,
      "loss": 0.0006,
      "step": 4976
    },
    {
      "epoch": 1.8210757409440177,
      "grad_norm": 9.329787254333496,
      "learning_rate": 4.5144017725258495e-06,
      "loss": 0.2355,
      "step": 4977
    },
    {
      "epoch": 1.8214416392242958,
      "grad_norm": 0.009828499518334866,
      "learning_rate": 4.505169867060562e-06,
      "loss": 0.0004,
      "step": 4978
    },
    {
      "epoch": 1.8218075375045737,
      "grad_norm": 0.06524066627025604,
      "learning_rate": 4.495937961595274e-06,
      "loss": 0.0008,
      "step": 4979
    },
    {
      "epoch": 1.8221734357848518,
      "grad_norm": 0.012082725763320923,
      "learning_rate": 4.486706056129985e-06,
      "loss": 0.0005,
      "step": 4980
    },
    {
      "epoch": 1.82253933406513,
      "grad_norm": 0.008907588198781013,
      "learning_rate": 4.477474150664698e-06,
      "loss": 0.0003,
      "step": 4981
    },
    {
      "epoch": 1.822905232345408,
      "grad_norm": 0.006568908225744963,
      "learning_rate": 4.46824224519941e-06,
      "loss": 0.0003,
      "step": 4982
    },
    {
      "epoch": 1.823271130625686,
      "grad_norm": 0.013481094501912594,
      "learning_rate": 4.459010339734121e-06,
      "loss": 0.0005,
      "step": 4983
    },
    {
      "epoch": 1.823637028905964,
      "grad_norm": 0.0527813658118248,
      "learning_rate": 4.449778434268833e-06,
      "loss": 0.0006,
      "step": 4984
    },
    {
      "epoch": 1.8240029271862421,
      "grad_norm": 0.03299397975206375,
      "learning_rate": 4.440546528803545e-06,
      "loss": 0.0008,
      "step": 4985
    },
    {
      "epoch": 1.8243688254665202,
      "grad_norm": 0.8790319561958313,
      "learning_rate": 4.431314623338257e-06,
      "loss": 0.0033,
      "step": 4986
    },
    {
      "epoch": 1.8247347237467983,
      "grad_norm": 0.010665668174624443,
      "learning_rate": 4.422082717872969e-06,
      "loss": 0.0004,
      "step": 4987
    },
    {
      "epoch": 1.8251006220270765,
      "grad_norm": 0.01669255457818508,
      "learning_rate": 4.412850812407681e-06,
      "loss": 0.0006,
      "step": 4988
    },
    {
      "epoch": 1.8254665203073546,
      "grad_norm": 0.0186922587454319,
      "learning_rate": 4.403618906942393e-06,
      "loss": 0.0006,
      "step": 4989
    },
    {
      "epoch": 1.8258324185876327,
      "grad_norm": 0.011855486780405045,
      "learning_rate": 4.394387001477105e-06,
      "loss": 0.0004,
      "step": 4990
    },
    {
      "epoch": 1.8261983168679108,
      "grad_norm": 0.004236925393342972,
      "learning_rate": 4.385155096011817e-06,
      "loss": 0.0002,
      "step": 4991
    },
    {
      "epoch": 1.826564215148189,
      "grad_norm": 0.012294919230043888,
      "learning_rate": 4.375923190546529e-06,
      "loss": 0.0005,
      "step": 4992
    },
    {
      "epoch": 1.826930113428467,
      "grad_norm": 0.008618909865617752,
      "learning_rate": 4.366691285081241e-06,
      "loss": 0.0003,
      "step": 4993
    },
    {
      "epoch": 1.8272960117087451,
      "grad_norm": 0.03694150969386101,
      "learning_rate": 4.357459379615953e-06,
      "loss": 0.0009,
      "step": 4994
    },
    {
      "epoch": 1.827661909989023,
      "grad_norm": 0.006904667243361473,
      "learning_rate": 4.348227474150665e-06,
      "loss": 0.0003,
      "step": 4995
    },
    {
      "epoch": 1.8280278082693011,
      "grad_norm": 0.012119163759052753,
      "learning_rate": 4.338995568685377e-06,
      "loss": 0.0005,
      "step": 4996
    },
    {
      "epoch": 1.8283937065495792,
      "grad_norm": 0.009820038452744484,
      "learning_rate": 4.329763663220089e-06,
      "loss": 0.0004,
      "step": 4997
    },
    {
      "epoch": 1.8287596048298573,
      "grad_norm": 0.01330367848277092,
      "learning_rate": 4.320531757754801e-06,
      "loss": 0.0005,
      "step": 4998
    },
    {
      "epoch": 1.8291255031101352,
      "grad_norm": 0.058927036821842194,
      "learning_rate": 4.311299852289512e-06,
      "loss": 0.0011,
      "step": 4999
    },
    {
      "epoch": 1.8294914013904133,
      "grad_norm": 0.5028088092803955,
      "learning_rate": 4.302067946824224e-06,
      "loss": 0.0028,
      "step": 5000
    },
    {
      "epoch": 1.8298572996706914,
      "grad_norm": 0.011324873194098473,
      "learning_rate": 4.2928360413589365e-06,
      "loss": 0.0004,
      "step": 5001
    },
    {
      "epoch": 1.8302231979509695,
      "grad_norm": 0.01082000881433487,
      "learning_rate": 4.283604135893649e-06,
      "loss": 0.0005,
      "step": 5002
    },
    {
      "epoch": 1.8305890962312477,
      "grad_norm": 0.006334720645099878,
      "learning_rate": 4.27437223042836e-06,
      "loss": 0.0003,
      "step": 5003
    },
    {
      "epoch": 1.8309549945115258,
      "grad_norm": 0.016308190301060677,
      "learning_rate": 4.2651403249630724e-06,
      "loss": 0.0006,
      "step": 5004
    },
    {
      "epoch": 1.8313208927918039,
      "grad_norm": 0.012001242488622665,
      "learning_rate": 4.255908419497785e-06,
      "loss": 0.0004,
      "step": 5005
    },
    {
      "epoch": 1.831686791072082,
      "grad_norm": 0.03685678914189339,
      "learning_rate": 4.246676514032497e-06,
      "loss": 0.0006,
      "step": 5006
    },
    {
      "epoch": 1.83205268935236,
      "grad_norm": 0.22523503005504608,
      "learning_rate": 4.237444608567208e-06,
      "loss": 0.0023,
      "step": 5007
    },
    {
      "epoch": 1.8324185876326382,
      "grad_norm": 0.009907945059239864,
      "learning_rate": 4.2282127031019206e-06,
      "loss": 0.0004,
      "step": 5008
    },
    {
      "epoch": 1.8327844859129163,
      "grad_norm": 0.21735338866710663,
      "learning_rate": 4.218980797636633e-06,
      "loss": 0.0019,
      "step": 5009
    },
    {
      "epoch": 1.8331503841931944,
      "grad_norm": 0.08237295597791672,
      "learning_rate": 4.209748892171344e-06,
      "loss": 0.0012,
      "step": 5010
    },
    {
      "epoch": 1.8335162824734725,
      "grad_norm": 0.0189270731061697,
      "learning_rate": 4.2005169867060564e-06,
      "loss": 0.0007,
      "step": 5011
    },
    {
      "epoch": 1.8338821807537504,
      "grad_norm": 0.00776682049036026,
      "learning_rate": 4.191285081240769e-06,
      "loss": 0.0003,
      "step": 5012
    },
    {
      "epoch": 1.8342480790340285,
      "grad_norm": 0.18900567293167114,
      "learning_rate": 4.18205317577548e-06,
      "loss": 0.0016,
      "step": 5013
    },
    {
      "epoch": 1.8346139773143066,
      "grad_norm": 0.012558123096823692,
      "learning_rate": 4.172821270310192e-06,
      "loss": 0.0005,
      "step": 5014
    },
    {
      "epoch": 1.8349798755945848,
      "grad_norm": 1.2016758918762207,
      "learning_rate": 4.163589364844904e-06,
      "loss": 0.0024,
      "step": 5015
    },
    {
      "epoch": 1.8353457738748626,
      "grad_norm": 0.012260349467396736,
      "learning_rate": 4.154357459379616e-06,
      "loss": 0.0005,
      "step": 5016
    },
    {
      "epoch": 1.8357116721551408,
      "grad_norm": 0.011501764878630638,
      "learning_rate": 4.145125553914328e-06,
      "loss": 0.0004,
      "step": 5017
    },
    {
      "epoch": 1.8360775704354189,
      "grad_norm": 0.010524746961891651,
      "learning_rate": 4.13589364844904e-06,
      "loss": 0.0004,
      "step": 5018
    },
    {
      "epoch": 1.836443468715697,
      "grad_norm": 0.9243431687355042,
      "learning_rate": 4.126661742983752e-06,
      "loss": 0.006,
      "step": 5019
    },
    {
      "epoch": 1.836809366995975,
      "grad_norm": 0.5888090133666992,
      "learning_rate": 4.117429837518464e-06,
      "loss": 0.0096,
      "step": 5020
    },
    {
      "epoch": 1.8371752652762532,
      "grad_norm": 0.06762746721506119,
      "learning_rate": 4.108197932053176e-06,
      "loss": 0.0009,
      "step": 5021
    },
    {
      "epoch": 1.8375411635565313,
      "grad_norm": 0.011188309639692307,
      "learning_rate": 4.098966026587888e-06,
      "loss": 0.0004,
      "step": 5022
    },
    {
      "epoch": 1.8379070618368094,
      "grad_norm": 0.008358005434274673,
      "learning_rate": 4.0897341211226e-06,
      "loss": 0.0003,
      "step": 5023
    },
    {
      "epoch": 1.8382729601170875,
      "grad_norm": 0.017588231712579727,
      "learning_rate": 4.080502215657312e-06,
      "loss": 0.0006,
      "step": 5024
    },
    {
      "epoch": 1.8386388583973656,
      "grad_norm": 0.008168831467628479,
      "learning_rate": 4.0712703101920245e-06,
      "loss": 0.0003,
      "step": 5025
    },
    {
      "epoch": 1.8390047566776437,
      "grad_norm": 0.010645205155014992,
      "learning_rate": 4.062038404726736e-06,
      "loss": 0.0004,
      "step": 5026
    },
    {
      "epoch": 1.8393706549579218,
      "grad_norm": 5.335343360900879,
      "learning_rate": 4.052806499261447e-06,
      "loss": 0.5154,
      "step": 5027
    },
    {
      "epoch": 1.8397365532381997,
      "grad_norm": 0.007521784398704767,
      "learning_rate": 4.0435745937961595e-06,
      "loss": 0.0003,
      "step": 5028
    },
    {
      "epoch": 1.8401024515184778,
      "grad_norm": 0.030782731249928474,
      "learning_rate": 4.034342688330872e-06,
      "loss": 0.0005,
      "step": 5029
    },
    {
      "epoch": 1.840468349798756,
      "grad_norm": 0.013444789685308933,
      "learning_rate": 4.025110782865583e-06,
      "loss": 0.0005,
      "step": 5030
    },
    {
      "epoch": 1.840834248079034,
      "grad_norm": 0.009412032552063465,
      "learning_rate": 4.015878877400295e-06,
      "loss": 0.0004,
      "step": 5031
    },
    {
      "epoch": 1.841200146359312,
      "grad_norm": 0.38536807894706726,
      "learning_rate": 4.006646971935008e-06,
      "loss": 0.0022,
      "step": 5032
    },
    {
      "epoch": 1.84156604463959,
      "grad_norm": 0.009716100990772247,
      "learning_rate": 3.99741506646972e-06,
      "loss": 0.0004,
      "step": 5033
    },
    {
      "epoch": 1.8419319429198682,
      "grad_norm": 22.82943344116211,
      "learning_rate": 3.988183161004431e-06,
      "loss": 0.0305,
      "step": 5034
    },
    {
      "epoch": 1.8422978412001463,
      "grad_norm": 0.010846910998225212,
      "learning_rate": 3.9789512555391435e-06,
      "loss": 0.0004,
      "step": 5035
    },
    {
      "epoch": 1.8426637394804244,
      "grad_norm": 0.02457268536090851,
      "learning_rate": 3.969719350073856e-06,
      "loss": 0.0006,
      "step": 5036
    },
    {
      "epoch": 1.8430296377607025,
      "grad_norm": 0.012077580206096172,
      "learning_rate": 3.960487444608567e-06,
      "loss": 0.0004,
      "step": 5037
    },
    {
      "epoch": 1.8433955360409806,
      "grad_norm": 0.007139825262129307,
      "learning_rate": 3.951255539143279e-06,
      "loss": 0.0003,
      "step": 5038
    },
    {
      "epoch": 1.8437614343212587,
      "grad_norm": 3.826831102371216,
      "learning_rate": 3.942023633677992e-06,
      "loss": 0.0136,
      "step": 5039
    },
    {
      "epoch": 1.8441273326015368,
      "grad_norm": 0.007196071557700634,
      "learning_rate": 3.932791728212704e-06,
      "loss": 0.0003,
      "step": 5040
    },
    {
      "epoch": 1.844493230881815,
      "grad_norm": 0.008487931452691555,
      "learning_rate": 3.923559822747415e-06,
      "loss": 0.0003,
      "step": 5041
    },
    {
      "epoch": 1.844859129162093,
      "grad_norm": 0.016270693391561508,
      "learning_rate": 3.914327917282127e-06,
      "loss": 0.0005,
      "step": 5042
    },
    {
      "epoch": 1.8452250274423712,
      "grad_norm": 0.041083116084337234,
      "learning_rate": 3.905096011816839e-06,
      "loss": 0.0007,
      "step": 5043
    },
    {
      "epoch": 1.8455909257226493,
      "grad_norm": 0.0051384721882641315,
      "learning_rate": 3.895864106351551e-06,
      "loss": 0.0002,
      "step": 5044
    },
    {
      "epoch": 1.8459568240029272,
      "grad_norm": 0.011425524950027466,
      "learning_rate": 3.8866322008862625e-06,
      "loss": 0.0004,
      "step": 5045
    },
    {
      "epoch": 1.8463227222832053,
      "grad_norm": 0.17098508775234222,
      "learning_rate": 3.877400295420975e-06,
      "loss": 0.0009,
      "step": 5046
    },
    {
      "epoch": 1.8466886205634834,
      "grad_norm": 0.028710080310702324,
      "learning_rate": 3.868168389955687e-06,
      "loss": 0.0006,
      "step": 5047
    },
    {
      "epoch": 1.8470545188437615,
      "grad_norm": 0.011860388331115246,
      "learning_rate": 3.858936484490399e-06,
      "loss": 0.0004,
      "step": 5048
    },
    {
      "epoch": 1.8474204171240394,
      "grad_norm": 0.01071089319884777,
      "learning_rate": 3.849704579025111e-06,
      "loss": 0.0004,
      "step": 5049
    },
    {
      "epoch": 1.8477863154043175,
      "grad_norm": 0.011242726817727089,
      "learning_rate": 3.840472673559823e-06,
      "loss": 0.0004,
      "step": 5050
    },
    {
      "epoch": 1.8481522136845956,
      "grad_norm": 0.006919392850250006,
      "learning_rate": 3.831240768094535e-06,
      "loss": 0.0003,
      "step": 5051
    },
    {
      "epoch": 1.8485181119648737,
      "grad_norm": 3.6260664463043213,
      "learning_rate": 3.822008862629247e-06,
      "loss": 0.4961,
      "step": 5052
    },
    {
      "epoch": 1.8488840102451518,
      "grad_norm": 0.02517409808933735,
      "learning_rate": 3.8127769571639592e-06,
      "loss": 0.0005,
      "step": 5053
    },
    {
      "epoch": 1.84924990852543,
      "grad_norm": 0.021045800298452377,
      "learning_rate": 3.803545051698671e-06,
      "loss": 0.0005,
      "step": 5054
    },
    {
      "epoch": 1.849615806805708,
      "grad_norm": 0.009917892515659332,
      "learning_rate": 3.7943131462333833e-06,
      "loss": 0.0004,
      "step": 5055
    },
    {
      "epoch": 1.8499817050859861,
      "grad_norm": 0.016097920015454292,
      "learning_rate": 3.7850812407680942e-06,
      "loss": 0.0005,
      "step": 5056
    },
    {
      "epoch": 1.8503476033662642,
      "grad_norm": 0.03117789700627327,
      "learning_rate": 3.7758493353028065e-06,
      "loss": 0.0008,
      "step": 5057
    },
    {
      "epoch": 1.8507135016465424,
      "grad_norm": 0.008891085162758827,
      "learning_rate": 3.7666174298375183e-06,
      "loss": 0.0004,
      "step": 5058
    },
    {
      "epoch": 1.8510793999268205,
      "grad_norm": 0.021641777828335762,
      "learning_rate": 3.7573855243722306e-06,
      "loss": 0.0006,
      "step": 5059
    },
    {
      "epoch": 1.8514452982070986,
      "grad_norm": 0.007842743769288063,
      "learning_rate": 3.7481536189069424e-06,
      "loss": 0.0003,
      "step": 5060
    },
    {
      "epoch": 1.8518111964873765,
      "grad_norm": 0.008815458044409752,
      "learning_rate": 3.7389217134416546e-06,
      "loss": 0.0003,
      "step": 5061
    },
    {
      "epoch": 1.8521770947676546,
      "grad_norm": 0.012225762940943241,
      "learning_rate": 3.7296898079763664e-06,
      "loss": 0.0004,
      "step": 5062
    },
    {
      "epoch": 1.8525429930479327,
      "grad_norm": 0.016953784972429276,
      "learning_rate": 3.7204579025110787e-06,
      "loss": 0.0005,
      "step": 5063
    },
    {
      "epoch": 1.8529088913282108,
      "grad_norm": 0.02103891409933567,
      "learning_rate": 3.7112259970457905e-06,
      "loss": 0.0006,
      "step": 5064
    },
    {
      "epoch": 1.8532747896084887,
      "grad_norm": 0.011366695165634155,
      "learning_rate": 3.7019940915805023e-06,
      "loss": 0.0005,
      "step": 5065
    },
    {
      "epoch": 1.8536406878887668,
      "grad_norm": 0.009914198890328407,
      "learning_rate": 3.6927621861152146e-06,
      "loss": 0.0004,
      "step": 5066
    },
    {
      "epoch": 1.854006586169045,
      "grad_norm": 0.014020469971001148,
      "learning_rate": 3.6835302806499264e-06,
      "loss": 0.0006,
      "step": 5067
    },
    {
      "epoch": 1.854372484449323,
      "grad_norm": 0.014142590574920177,
      "learning_rate": 3.6742983751846386e-06,
      "loss": 0.0005,
      "step": 5068
    },
    {
      "epoch": 1.8547383827296011,
      "grad_norm": 0.03985031321644783,
      "learning_rate": 3.6650664697193504e-06,
      "loss": 0.0009,
      "step": 5069
    },
    {
      "epoch": 1.8551042810098792,
      "grad_norm": 0.08879678696393967,
      "learning_rate": 3.655834564254062e-06,
      "loss": 0.0013,
      "step": 5070
    },
    {
      "epoch": 1.8554701792901573,
      "grad_norm": 0.019029667600989342,
      "learning_rate": 3.646602658788774e-06,
      "loss": 0.0006,
      "step": 5071
    },
    {
      "epoch": 1.8558360775704354,
      "grad_norm": 0.008925882168114185,
      "learning_rate": 3.637370753323486e-06,
      "loss": 0.0004,
      "step": 5072
    },
    {
      "epoch": 1.8562019758507136,
      "grad_norm": 0.007864211685955524,
      "learning_rate": 3.6281388478581977e-06,
      "loss": 0.0003,
      "step": 5073
    },
    {
      "epoch": 1.8565678741309917,
      "grad_norm": 0.008260690607130527,
      "learning_rate": 3.61890694239291e-06,
      "loss": 0.0004,
      "step": 5074
    },
    {
      "epoch": 1.8569337724112698,
      "grad_norm": 0.016473721712827682,
      "learning_rate": 3.6096750369276218e-06,
      "loss": 0.0005,
      "step": 5075
    },
    {
      "epoch": 1.8572996706915479,
      "grad_norm": 0.01793615147471428,
      "learning_rate": 3.600443131462334e-06,
      "loss": 0.0007,
      "step": 5076
    },
    {
      "epoch": 1.857665568971826,
      "grad_norm": 0.006978594698011875,
      "learning_rate": 3.591211225997046e-06,
      "loss": 0.0003,
      "step": 5077
    },
    {
      "epoch": 1.8580314672521039,
      "grad_norm": 0.012403883039951324,
      "learning_rate": 3.581979320531758e-06,
      "loss": 0.0005,
      "step": 5078
    },
    {
      "epoch": 1.858397365532382,
      "grad_norm": 0.009026795625686646,
      "learning_rate": 3.57274741506647e-06,
      "loss": 0.0004,
      "step": 5079
    },
    {
      "epoch": 1.85876326381266,
      "grad_norm": 0.009616249240934849,
      "learning_rate": 3.563515509601182e-06,
      "loss": 0.0004,
      "step": 5080
    },
    {
      "epoch": 1.8591291620929382,
      "grad_norm": 0.2538212537765503,
      "learning_rate": 3.554283604135894e-06,
      "loss": 0.0025,
      "step": 5081
    },
    {
      "epoch": 1.859495060373216,
      "grad_norm": 0.007050897926092148,
      "learning_rate": 3.5450516986706062e-06,
      "loss": 0.0003,
      "step": 5082
    },
    {
      "epoch": 1.8598609586534942,
      "grad_norm": 0.01750732772052288,
      "learning_rate": 3.535819793205318e-06,
      "loss": 0.0006,
      "step": 5083
    },
    {
      "epoch": 1.8602268569337723,
      "grad_norm": 0.014450296759605408,
      "learning_rate": 3.52658788774003e-06,
      "loss": 0.0006,
      "step": 5084
    },
    {
      "epoch": 1.8605927552140504,
      "grad_norm": 0.00748571427538991,
      "learning_rate": 3.5173559822747413e-06,
      "loss": 0.0003,
      "step": 5085
    },
    {
      "epoch": 1.8609586534943285,
      "grad_norm": 0.008782071061432362,
      "learning_rate": 3.5081240768094535e-06,
      "loss": 0.0004,
      "step": 5086
    },
    {
      "epoch": 1.8613245517746067,
      "grad_norm": 0.01705392450094223,
      "learning_rate": 3.4988921713441653e-06,
      "loss": 0.0006,
      "step": 5087
    },
    {
      "epoch": 1.8616904500548848,
      "grad_norm": 0.011830395087599754,
      "learning_rate": 3.4896602658788776e-06,
      "loss": 0.0005,
      "step": 5088
    },
    {
      "epoch": 1.8620563483351629,
      "grad_norm": 0.014655773527920246,
      "learning_rate": 3.4804283604135894e-06,
      "loss": 0.0006,
      "step": 5089
    },
    {
      "epoch": 1.862422246615441,
      "grad_norm": 0.014868979342281818,
      "learning_rate": 3.4711964549483016e-06,
      "loss": 0.0005,
      "step": 5090
    },
    {
      "epoch": 1.862788144895719,
      "grad_norm": 0.01027233898639679,
      "learning_rate": 3.4619645494830134e-06,
      "loss": 0.0004,
      "step": 5091
    },
    {
      "epoch": 1.8631540431759972,
      "grad_norm": 0.00595919881016016,
      "learning_rate": 3.4527326440177253e-06,
      "loss": 0.0003,
      "step": 5092
    },
    {
      "epoch": 1.8635199414562753,
      "grad_norm": 0.00589431868866086,
      "learning_rate": 3.4435007385524375e-06,
      "loss": 0.0003,
      "step": 5093
    },
    {
      "epoch": 1.8638858397365532,
      "grad_norm": 19.11582374572754,
      "learning_rate": 3.4342688330871493e-06,
      "loss": 0.0679,
      "step": 5094
    },
    {
      "epoch": 1.8642517380168313,
      "grad_norm": 3.5694501399993896,
      "learning_rate": 3.4250369276218616e-06,
      "loss": 0.3172,
      "step": 5095
    },
    {
      "epoch": 1.8646176362971094,
      "grad_norm": 0.014734541065990925,
      "learning_rate": 3.4158050221565734e-06,
      "loss": 0.0004,
      "step": 5096
    },
    {
      "epoch": 1.8649835345773875,
      "grad_norm": 0.012401354499161243,
      "learning_rate": 3.4065731166912856e-06,
      "loss": 0.0005,
      "step": 5097
    },
    {
      "epoch": 1.8653494328576654,
      "grad_norm": 0.0088847316801548,
      "learning_rate": 3.3973412112259974e-06,
      "loss": 0.0004,
      "step": 5098
    },
    {
      "epoch": 1.8657153311379435,
      "grad_norm": 1.5265170335769653,
      "learning_rate": 3.388109305760709e-06,
      "loss": 0.007,
      "step": 5099
    },
    {
      "epoch": 1.8660812294182216,
      "grad_norm": 0.025453675538301468,
      "learning_rate": 3.3788774002954207e-06,
      "loss": 0.0006,
      "step": 5100
    },
    {
      "epoch": 1.8664471276984997,
      "grad_norm": 0.014173824340105057,
      "learning_rate": 3.369645494830133e-06,
      "loss": 0.0005,
      "step": 5101
    },
    {
      "epoch": 1.8668130259787779,
      "grad_norm": 0.007999676279723644,
      "learning_rate": 3.3604135893648447e-06,
      "loss": 0.0004,
      "step": 5102
    },
    {
      "epoch": 1.867178924259056,
      "grad_norm": 0.006381881423294544,
      "learning_rate": 3.351181683899557e-06,
      "loss": 0.0003,
      "step": 5103
    },
    {
      "epoch": 1.867544822539334,
      "grad_norm": 0.008163963444530964,
      "learning_rate": 3.341949778434269e-06,
      "loss": 0.0004,
      "step": 5104
    },
    {
      "epoch": 1.8679107208196122,
      "grad_norm": 0.007611056789755821,
      "learning_rate": 3.332717872968981e-06,
      "loss": 0.0003,
      "step": 5105
    },
    {
      "epoch": 1.8682766190998903,
      "grad_norm": 0.00958904717117548,
      "learning_rate": 3.323485967503693e-06,
      "loss": 0.0004,
      "step": 5106
    },
    {
      "epoch": 1.8686425173801684,
      "grad_norm": 0.012523400597274303,
      "learning_rate": 3.314254062038405e-06,
      "loss": 0.0004,
      "step": 5107
    },
    {
      "epoch": 1.8690084156604465,
      "grad_norm": 0.043625764548778534,
      "learning_rate": 3.305022156573117e-06,
      "loss": 0.0007,
      "step": 5108
    },
    {
      "epoch": 1.8693743139407246,
      "grad_norm": 0.015626773238182068,
      "learning_rate": 3.295790251107829e-06,
      "loss": 0.0005,
      "step": 5109
    },
    {
      "epoch": 1.8697402122210025,
      "grad_norm": 0.013709991239011288,
      "learning_rate": 3.286558345642541e-06,
      "loss": 0.0005,
      "step": 5110
    },
    {
      "epoch": 1.8701061105012806,
      "grad_norm": 0.008316886611282825,
      "learning_rate": 3.277326440177253e-06,
      "loss": 0.0003,
      "step": 5111
    },
    {
      "epoch": 1.8704720087815587,
      "grad_norm": 0.007241636980324984,
      "learning_rate": 3.268094534711965e-06,
      "loss": 0.0003,
      "step": 5112
    },
    {
      "epoch": 1.8708379070618368,
      "grad_norm": 0.03919963911175728,
      "learning_rate": 3.258862629246677e-06,
      "loss": 0.0009,
      "step": 5113
    },
    {
      "epoch": 1.871203805342115,
      "grad_norm": 0.021791113540530205,
      "learning_rate": 3.2496307237813883e-06,
      "loss": 0.0008,
      "step": 5114
    },
    {
      "epoch": 1.8715697036223928,
      "grad_norm": 0.01419104915112257,
      "learning_rate": 3.2403988183161005e-06,
      "loss": 0.0005,
      "step": 5115
    },
    {
      "epoch": 1.871935601902671,
      "grad_norm": 25.76705551147461,
      "learning_rate": 3.2311669128508123e-06,
      "loss": 0.1856,
      "step": 5116
    },
    {
      "epoch": 1.872301500182949,
      "grad_norm": 0.00753635773435235,
      "learning_rate": 3.2219350073855246e-06,
      "loss": 0.0003,
      "step": 5117
    },
    {
      "epoch": 1.8726673984632272,
      "grad_norm": 0.010492231696844101,
      "learning_rate": 3.2127031019202364e-06,
      "loss": 0.0004,
      "step": 5118
    },
    {
      "epoch": 1.8730332967435053,
      "grad_norm": 0.017284484580159187,
      "learning_rate": 3.203471196454948e-06,
      "loss": 0.0006,
      "step": 5119
    },
    {
      "epoch": 1.8733991950237834,
      "grad_norm": 0.010709918104112148,
      "learning_rate": 3.1942392909896604e-06,
      "loss": 0.0004,
      "step": 5120
    },
    {
      "epoch": 1.8737650933040615,
      "grad_norm": 0.01750991865992546,
      "learning_rate": 3.1850073855243723e-06,
      "loss": 0.0006,
      "step": 5121
    },
    {
      "epoch": 1.8741309915843396,
      "grad_norm": 10.181550025939941,
      "learning_rate": 3.1757754800590845e-06,
      "loss": 0.193,
      "step": 5122
    },
    {
      "epoch": 1.8744968898646177,
      "grad_norm": 0.007698504254221916,
      "learning_rate": 3.1665435745937963e-06,
      "loss": 0.0003,
      "step": 5123
    },
    {
      "epoch": 1.8748627881448958,
      "grad_norm": 0.008313181810081005,
      "learning_rate": 3.1573116691285086e-06,
      "loss": 0.0003,
      "step": 5124
    },
    {
      "epoch": 1.875228686425174,
      "grad_norm": 0.006061617750674486,
      "learning_rate": 3.1480797636632204e-06,
      "loss": 0.0003,
      "step": 5125
    },
    {
      "epoch": 1.875594584705452,
      "grad_norm": 0.38911858201026917,
      "learning_rate": 3.1388478581979326e-06,
      "loss": 0.0029,
      "step": 5126
    },
    {
      "epoch": 1.87596048298573,
      "grad_norm": 0.01761011965572834,
      "learning_rate": 3.1296159527326445e-06,
      "loss": 0.0005,
      "step": 5127
    },
    {
      "epoch": 1.876326381266008,
      "grad_norm": 0.009253369644284248,
      "learning_rate": 3.1203840472673563e-06,
      "loss": 0.0004,
      "step": 5128
    },
    {
      "epoch": 1.8766922795462861,
      "grad_norm": 0.7711682319641113,
      "learning_rate": 3.111152141802068e-06,
      "loss": 0.0085,
      "step": 5129
    },
    {
      "epoch": 1.8770581778265643,
      "grad_norm": 0.005971061065793037,
      "learning_rate": 3.1019202363367803e-06,
      "loss": 0.0003,
      "step": 5130
    },
    {
      "epoch": 1.8774240761068421,
      "grad_norm": 16.967679977416992,
      "learning_rate": 3.092688330871492e-06,
      "loss": 0.7057,
      "step": 5131
    },
    {
      "epoch": 1.8777899743871203,
      "grad_norm": 0.009698725305497646,
      "learning_rate": 3.083456425406204e-06,
      "loss": 0.0004,
      "step": 5132
    },
    {
      "epoch": 1.8781558726673984,
      "grad_norm": 0.022692225873470306,
      "learning_rate": 3.074224519940916e-06,
      "loss": 0.0005,
      "step": 5133
    },
    {
      "epoch": 1.8785217709476765,
      "grad_norm": 0.01165812648832798,
      "learning_rate": 3.064992614475628e-06,
      "loss": 0.0003,
      "step": 5134
    },
    {
      "epoch": 1.8788876692279546,
      "grad_norm": 0.010103319771587849,
      "learning_rate": 3.05576070901034e-06,
      "loss": 0.0004,
      "step": 5135
    },
    {
      "epoch": 1.8792535675082327,
      "grad_norm": 0.007668691221624613,
      "learning_rate": 3.046528803545052e-06,
      "loss": 0.0003,
      "step": 5136
    },
    {
      "epoch": 1.8796194657885108,
      "grad_norm": 0.01179431565105915,
      "learning_rate": 3.037296898079764e-06,
      "loss": 0.0004,
      "step": 5137
    },
    {
      "epoch": 1.879985364068789,
      "grad_norm": 21.914453506469727,
      "learning_rate": 3.0280649926144757e-06,
      "loss": 0.549,
      "step": 5138
    },
    {
      "epoch": 1.880351262349067,
      "grad_norm": 0.006778615992516279,
      "learning_rate": 3.0188330871491876e-06,
      "loss": 0.0003,
      "step": 5139
    },
    {
      "epoch": 1.8807171606293451,
      "grad_norm": 0.01605774462223053,
      "learning_rate": 3.0096011816839e-06,
      "loss": 0.0005,
      "step": 5140
    },
    {
      "epoch": 1.8810830589096232,
      "grad_norm": 0.03182213380932808,
      "learning_rate": 3.0003692762186116e-06,
      "loss": 0.0008,
      "step": 5141
    },
    {
      "epoch": 1.8814489571899013,
      "grad_norm": 0.012729140929877758,
      "learning_rate": 2.9911373707533234e-06,
      "loss": 0.0004,
      "step": 5142
    },
    {
      "epoch": 1.8818148554701792,
      "grad_norm": 0.430316299200058,
      "learning_rate": 2.9819054652880357e-06,
      "loss": 0.0017,
      "step": 5143
    },
    {
      "epoch": 1.8821807537504573,
      "grad_norm": 0.00798310711979866,
      "learning_rate": 2.9726735598227475e-06,
      "loss": 0.0003,
      "step": 5144
    },
    {
      "epoch": 1.8825466520307355,
      "grad_norm": 2.9827160835266113,
      "learning_rate": 2.9634416543574597e-06,
      "loss": 0.4279,
      "step": 5145
    },
    {
      "epoch": 1.8829125503110136,
      "grad_norm": 0.05113670229911804,
      "learning_rate": 2.954209748892171e-06,
      "loss": 0.0011,
      "step": 5146
    },
    {
      "epoch": 1.8832784485912917,
      "grad_norm": 0.011413239873945713,
      "learning_rate": 2.9449778434268834e-06,
      "loss": 0.0005,
      "step": 5147
    },
    {
      "epoch": 1.8836443468715696,
      "grad_norm": 0.006928952410817146,
      "learning_rate": 2.935745937961595e-06,
      "loss": 0.0003,
      "step": 5148
    },
    {
      "epoch": 1.8840102451518477,
      "grad_norm": 0.011766173876821995,
      "learning_rate": 2.9265140324963074e-06,
      "loss": 0.0003,
      "step": 5149
    },
    {
      "epoch": 1.8843761434321258,
      "grad_norm": 8.16541576385498,
      "learning_rate": 2.9172821270310193e-06,
      "loss": 0.01,
      "step": 5150
    },
    {
      "epoch": 1.884742041712404,
      "grad_norm": 0.29142576456069946,
      "learning_rate": 2.9080502215657315e-06,
      "loss": 0.0014,
      "step": 5151
    },
    {
      "epoch": 1.885107939992682,
      "grad_norm": 0.00971097033470869,
      "learning_rate": 2.8988183161004433e-06,
      "loss": 0.0004,
      "step": 5152
    },
    {
      "epoch": 1.8854738382729601,
      "grad_norm": 0.009982203133404255,
      "learning_rate": 2.889586410635155e-06,
      "loss": 0.0004,
      "step": 5153
    },
    {
      "epoch": 1.8858397365532382,
      "grad_norm": 0.011407872661948204,
      "learning_rate": 2.880354505169867e-06,
      "loss": 0.0004,
      "step": 5154
    },
    {
      "epoch": 1.8862056348335163,
      "grad_norm": 0.011885032057762146,
      "learning_rate": 2.8711225997045792e-06,
      "loss": 0.0004,
      "step": 5155
    },
    {
      "epoch": 1.8865715331137944,
      "grad_norm": 3.0598981380462646,
      "learning_rate": 2.861890694239291e-06,
      "loss": 0.4893,
      "step": 5156
    },
    {
      "epoch": 1.8869374313940726,
      "grad_norm": 0.008814419619739056,
      "learning_rate": 2.8526587887740033e-06,
      "loss": 0.0004,
      "step": 5157
    },
    {
      "epoch": 1.8873033296743507,
      "grad_norm": 0.011101115494966507,
      "learning_rate": 2.843426883308715e-06,
      "loss": 0.0004,
      "step": 5158
    },
    {
      "epoch": 1.8876692279546288,
      "grad_norm": 0.1710323691368103,
      "learning_rate": 2.8341949778434273e-06,
      "loss": 0.0014,
      "step": 5159
    },
    {
      "epoch": 1.8880351262349067,
      "grad_norm": 0.006814558524638414,
      "learning_rate": 2.8249630723781387e-06,
      "loss": 0.0003,
      "step": 5160
    },
    {
      "epoch": 1.8884010245151848,
      "grad_norm": 0.020459983497858047,
      "learning_rate": 2.815731166912851e-06,
      "loss": 0.0006,
      "step": 5161
    },
    {
      "epoch": 1.8887669227954629,
      "grad_norm": 0.012995966710150242,
      "learning_rate": 2.806499261447563e-06,
      "loss": 0.0005,
      "step": 5162
    },
    {
      "epoch": 1.889132821075741,
      "grad_norm": 0.013399191200733185,
      "learning_rate": 2.797267355982275e-06,
      "loss": 0.0005,
      "step": 5163
    },
    {
      "epoch": 1.8894987193560189,
      "grad_norm": 0.014374487102031708,
      "learning_rate": 2.788035450516987e-06,
      "loss": 0.0005,
      "step": 5164
    },
    {
      "epoch": 1.889864617636297,
      "grad_norm": 0.007866152562201023,
      "learning_rate": 2.7788035450516987e-06,
      "loss": 0.0004,
      "step": 5165
    },
    {
      "epoch": 1.890230515916575,
      "grad_norm": 0.038694098591804504,
      "learning_rate": 2.769571639586411e-06,
      "loss": 0.0009,
      "step": 5166
    },
    {
      "epoch": 1.8905964141968532,
      "grad_norm": 0.009094960987567902,
      "learning_rate": 2.7603397341211227e-06,
      "loss": 0.0003,
      "step": 5167
    },
    {
      "epoch": 1.8909623124771313,
      "grad_norm": 0.008256380446255207,
      "learning_rate": 2.7511078286558346e-06,
      "loss": 0.0004,
      "step": 5168
    },
    {
      "epoch": 1.8913282107574094,
      "grad_norm": 0.00996987521648407,
      "learning_rate": 2.7418759231905464e-06,
      "loss": 0.0004,
      "step": 5169
    },
    {
      "epoch": 1.8916941090376875,
      "grad_norm": 0.09840153157711029,
      "learning_rate": 2.7326440177252586e-06,
      "loss": 0.0009,
      "step": 5170
    },
    {
      "epoch": 1.8920600073179656,
      "grad_norm": 19.115055084228516,
      "learning_rate": 2.7234121122599704e-06,
      "loss": 0.0414,
      "step": 5171
    },
    {
      "epoch": 1.8924259055982438,
      "grad_norm": 0.04656176641583443,
      "learning_rate": 2.7141802067946827e-06,
      "loss": 0.0011,
      "step": 5172
    },
    {
      "epoch": 1.8927918038785219,
      "grad_norm": 0.01058531180024147,
      "learning_rate": 2.7049483013293945e-06,
      "loss": 0.0004,
      "step": 5173
    },
    {
      "epoch": 1.8931577021588,
      "grad_norm": 0.02276856079697609,
      "learning_rate": 2.6957163958641067e-06,
      "loss": 0.0009,
      "step": 5174
    },
    {
      "epoch": 1.893523600439078,
      "grad_norm": 0.0114536602050066,
      "learning_rate": 2.686484490398818e-06,
      "loss": 0.0004,
      "step": 5175
    },
    {
      "epoch": 1.893889498719356,
      "grad_norm": 0.010819872841238976,
      "learning_rate": 2.6772525849335304e-06,
      "loss": 0.0004,
      "step": 5176
    },
    {
      "epoch": 1.894255396999634,
      "grad_norm": 0.010767301544547081,
      "learning_rate": 2.668020679468242e-06,
      "loss": 0.0004,
      "step": 5177
    },
    {
      "epoch": 1.8946212952799122,
      "grad_norm": 0.062285471707582474,
      "learning_rate": 2.6587887740029545e-06,
      "loss": 0.0013,
      "step": 5178
    },
    {
      "epoch": 1.8949871935601903,
      "grad_norm": 0.011190147139132023,
      "learning_rate": 2.6495568685376663e-06,
      "loss": 0.0005,
      "step": 5179
    },
    {
      "epoch": 1.8953530918404682,
      "grad_norm": 12.81799602508545,
      "learning_rate": 2.6403249630723785e-06,
      "loss": 0.1357,
      "step": 5180
    },
    {
      "epoch": 1.8957189901207463,
      "grad_norm": 0.012076293118298054,
      "learning_rate": 2.6310930576070903e-06,
      "loss": 0.0005,
      "step": 5181
    },
    {
      "epoch": 1.8960848884010244,
      "grad_norm": 0.010767386294901371,
      "learning_rate": 2.621861152141802e-06,
      "loss": 0.0005,
      "step": 5182
    },
    {
      "epoch": 1.8964507866813025,
      "grad_norm": 0.010695559903979301,
      "learning_rate": 2.612629246676514e-06,
      "loss": 0.0004,
      "step": 5183
    },
    {
      "epoch": 1.8968166849615806,
      "grad_norm": 0.015602247789502144,
      "learning_rate": 2.6033973412112262e-06,
      "loss": 0.0005,
      "step": 5184
    },
    {
      "epoch": 1.8971825832418587,
      "grad_norm": 0.03818860650062561,
      "learning_rate": 2.594165435745938e-06,
      "loss": 0.0011,
      "step": 5185
    },
    {
      "epoch": 1.8975484815221368,
      "grad_norm": 0.014721919782459736,
      "learning_rate": 2.5849335302806503e-06,
      "loss": 0.0005,
      "step": 5186
    },
    {
      "epoch": 1.897914379802415,
      "grad_norm": 0.016690032556653023,
      "learning_rate": 2.575701624815362e-06,
      "loss": 0.0005,
      "step": 5187
    },
    {
      "epoch": 1.898280278082693,
      "grad_norm": 0.004312015604227781,
      "learning_rate": 2.566469719350074e-06,
      "loss": 0.0002,
      "step": 5188
    },
    {
      "epoch": 1.8986461763629712,
      "grad_norm": 0.012290812097489834,
      "learning_rate": 2.5572378138847857e-06,
      "loss": 0.0004,
      "step": 5189
    },
    {
      "epoch": 1.8990120746432493,
      "grad_norm": 0.014366446994245052,
      "learning_rate": 2.548005908419498e-06,
      "loss": 0.0005,
      "step": 5190
    },
    {
      "epoch": 1.8993779729235274,
      "grad_norm": 0.03142852708697319,
      "learning_rate": 2.53877400295421e-06,
      "loss": 0.0009,
      "step": 5191
    },
    {
      "epoch": 1.8997438712038055,
      "grad_norm": 0.01668711192905903,
      "learning_rate": 2.5295420974889216e-06,
      "loss": 0.0006,
      "step": 5192
    },
    {
      "epoch": 1.9001097694840834,
      "grad_norm": 0.2878037393093109,
      "learning_rate": 2.520310192023634e-06,
      "loss": 0.0034,
      "step": 5193
    },
    {
      "epoch": 1.9004756677643615,
      "grad_norm": 0.013914012350142002,
      "learning_rate": 2.5110782865583457e-06,
      "loss": 0.0006,
      "step": 5194
    },
    {
      "epoch": 1.9008415660446396,
      "grad_norm": 0.009939026087522507,
      "learning_rate": 2.501846381093058e-06,
      "loss": 0.0004,
      "step": 5195
    },
    {
      "epoch": 1.9012074643249177,
      "grad_norm": 0.010153340175747871,
      "learning_rate": 2.4926144756277693e-06,
      "loss": 0.0004,
      "step": 5196
    },
    {
      "epoch": 1.9015733626051956,
      "grad_norm": 0.14568956196308136,
      "learning_rate": 2.4833825701624816e-06,
      "loss": 0.0016,
      "step": 5197
    },
    {
      "epoch": 1.9019392608854737,
      "grad_norm": 0.021244032308459282,
      "learning_rate": 2.4741506646971934e-06,
      "loss": 0.0008,
      "step": 5198
    },
    {
      "epoch": 1.9023051591657518,
      "grad_norm": 0.009836114943027496,
      "learning_rate": 2.4649187592319056e-06,
      "loss": 0.0004,
      "step": 5199
    },
    {
      "epoch": 1.90267105744603,
      "grad_norm": 0.021608291193842888,
      "learning_rate": 2.4556868537666174e-06,
      "loss": 0.0006,
      "step": 5200
    },
    {
      "epoch": 1.903036955726308,
      "grad_norm": 0.02599077671766281,
      "learning_rate": 2.4464549483013297e-06,
      "loss": 0.0007,
      "step": 5201
    },
    {
      "epoch": 1.9034028540065862,
      "grad_norm": 0.012964565306901932,
      "learning_rate": 2.4372230428360415e-06,
      "loss": 0.0006,
      "step": 5202
    },
    {
      "epoch": 1.9037687522868643,
      "grad_norm": 0.010857940651476383,
      "learning_rate": 2.4279911373707538e-06,
      "loss": 0.0004,
      "step": 5203
    },
    {
      "epoch": 1.9041346505671424,
      "grad_norm": 2.221482992172241,
      "learning_rate": 2.418759231905465e-06,
      "loss": 0.3984,
      "step": 5204
    },
    {
      "epoch": 1.9045005488474205,
      "grad_norm": 0.021662293002009392,
      "learning_rate": 2.4095273264401774e-06,
      "loss": 0.0007,
      "step": 5205
    },
    {
      "epoch": 1.9048664471276986,
      "grad_norm": 0.03282339498400688,
      "learning_rate": 2.400295420974889e-06,
      "loss": 0.0006,
      "step": 5206
    },
    {
      "epoch": 1.9052323454079767,
      "grad_norm": 0.008751489222049713,
      "learning_rate": 2.3910635155096015e-06,
      "loss": 0.0003,
      "step": 5207
    },
    {
      "epoch": 1.9055982436882548,
      "grad_norm": 0.03691980242729187,
      "learning_rate": 2.3818316100443133e-06,
      "loss": 0.0009,
      "step": 5208
    },
    {
      "epoch": 1.9059641419685327,
      "grad_norm": 0.009221796877682209,
      "learning_rate": 2.3725997045790255e-06,
      "loss": 0.0004,
      "step": 5209
    },
    {
      "epoch": 1.9063300402488108,
      "grad_norm": 0.008774777874350548,
      "learning_rate": 2.3633677991137373e-06,
      "loss": 0.0003,
      "step": 5210
    },
    {
      "epoch": 1.906695938529089,
      "grad_norm": 0.03732734173536301,
      "learning_rate": 2.354135893648449e-06,
      "loss": 0.0008,
      "step": 5211
    },
    {
      "epoch": 1.907061836809367,
      "grad_norm": 0.006250884849578142,
      "learning_rate": 2.344903988183161e-06,
      "loss": 0.0003,
      "step": 5212
    },
    {
      "epoch": 1.907427735089645,
      "grad_norm": 0.029560359194874763,
      "learning_rate": 2.3356720827178732e-06,
      "loss": 0.0007,
      "step": 5213
    },
    {
      "epoch": 1.907793633369923,
      "grad_norm": 0.012678188271820545,
      "learning_rate": 2.326440177252585e-06,
      "loss": 0.0005,
      "step": 5214
    },
    {
      "epoch": 1.9081595316502011,
      "grad_norm": 0.018977506086230278,
      "learning_rate": 2.317208271787297e-06,
      "loss": 0.0005,
      "step": 5215
    },
    {
      "epoch": 1.9085254299304792,
      "grad_norm": 0.0062491376884281635,
      "learning_rate": 2.307976366322009e-06,
      "loss": 0.0003,
      "step": 5216
    },
    {
      "epoch": 1.9088913282107574,
      "grad_norm": 0.3525281846523285,
      "learning_rate": 2.298744460856721e-06,
      "loss": 0.0016,
      "step": 5217
    },
    {
      "epoch": 1.9092572264910355,
      "grad_norm": 0.012839959003031254,
      "learning_rate": 2.2895125553914327e-06,
      "loss": 0.0005,
      "step": 5218
    },
    {
      "epoch": 1.9096231247713136,
      "grad_norm": 0.029803775250911713,
      "learning_rate": 2.2802806499261446e-06,
      "loss": 0.0008,
      "step": 5219
    },
    {
      "epoch": 1.9099890230515917,
      "grad_norm": 0.021568510681390762,
      "learning_rate": 2.271048744460857e-06,
      "loss": 0.0007,
      "step": 5220
    },
    {
      "epoch": 1.9103549213318698,
      "grad_norm": 0.01666373200714588,
      "learning_rate": 2.2618168389955686e-06,
      "loss": 0.0007,
      "step": 5221
    },
    {
      "epoch": 1.910720819612148,
      "grad_norm": 0.018550831824541092,
      "learning_rate": 2.252584933530281e-06,
      "loss": 0.0007,
      "step": 5222
    },
    {
      "epoch": 1.911086717892426,
      "grad_norm": 0.009356333874166012,
      "learning_rate": 2.2433530280649927e-06,
      "loss": 0.0004,
      "step": 5223
    },
    {
      "epoch": 1.9114526161727041,
      "grad_norm": 0.009737898595631123,
      "learning_rate": 2.234121122599705e-06,
      "loss": 0.0004,
      "step": 5224
    },
    {
      "epoch": 1.9118185144529822,
      "grad_norm": 0.01143872831016779,
      "learning_rate": 2.2248892171344163e-06,
      "loss": 0.0004,
      "step": 5225
    },
    {
      "epoch": 1.9121844127332601,
      "grad_norm": 0.8297363519668579,
      "learning_rate": 2.2156573116691286e-06,
      "loss": 0.0093,
      "step": 5226
    },
    {
      "epoch": 1.9125503110135382,
      "grad_norm": 0.014524830505251884,
      "learning_rate": 2.2064254062038404e-06,
      "loss": 0.0006,
      "step": 5227
    },
    {
      "epoch": 1.9129162092938163,
      "grad_norm": 0.012210020795464516,
      "learning_rate": 2.1971935007385526e-06,
      "loss": 0.0005,
      "step": 5228
    },
    {
      "epoch": 1.9132821075740944,
      "grad_norm": 0.01334438007324934,
      "learning_rate": 2.1879615952732645e-06,
      "loss": 0.0005,
      "step": 5229
    },
    {
      "epoch": 1.9136480058543723,
      "grad_norm": 0.015489533543586731,
      "learning_rate": 2.1787296898079767e-06,
      "loss": 0.0006,
      "step": 5230
    },
    {
      "epoch": 1.9140139041346504,
      "grad_norm": 0.020008428022265434,
      "learning_rate": 2.1694977843426885e-06,
      "loss": 0.0007,
      "step": 5231
    },
    {
      "epoch": 1.9143798024149286,
      "grad_norm": 1.0323432683944702,
      "learning_rate": 2.1602658788774003e-06,
      "loss": 0.0063,
      "step": 5232
    },
    {
      "epoch": 1.9147457006952067,
      "grad_norm": 0.010932007804512978,
      "learning_rate": 2.151033973412112e-06,
      "loss": 0.0004,
      "step": 5233
    },
    {
      "epoch": 1.9151115989754848,
      "grad_norm": 0.009459799155592918,
      "learning_rate": 2.1418020679468244e-06,
      "loss": 0.0004,
      "step": 5234
    },
    {
      "epoch": 1.9154774972557629,
      "grad_norm": 3.9323856830596924,
      "learning_rate": 2.1325701624815362e-06,
      "loss": 0.0147,
      "step": 5235
    },
    {
      "epoch": 1.915843395536041,
      "grad_norm": 0.011774441227316856,
      "learning_rate": 2.1233382570162485e-06,
      "loss": 0.0004,
      "step": 5236
    },
    {
      "epoch": 1.916209293816319,
      "grad_norm": 0.012152764946222305,
      "learning_rate": 2.1141063515509603e-06,
      "loss": 0.0005,
      "step": 5237
    },
    {
      "epoch": 1.9165751920965972,
      "grad_norm": 6.904515743255615,
      "learning_rate": 2.104874446085672e-06,
      "loss": 0.0721,
      "step": 5238
    },
    {
      "epoch": 1.9169410903768753,
      "grad_norm": 0.00911310501396656,
      "learning_rate": 2.0956425406203843e-06,
      "loss": 0.0004,
      "step": 5239
    },
    {
      "epoch": 1.9173069886571534,
      "grad_norm": 0.00960900541394949,
      "learning_rate": 2.086410635155096e-06,
      "loss": 0.0004,
      "step": 5240
    },
    {
      "epoch": 1.9176728869374315,
      "grad_norm": 0.011617099866271019,
      "learning_rate": 2.077178729689808e-06,
      "loss": 0.0005,
      "step": 5241
    },
    {
      "epoch": 1.9180387852177094,
      "grad_norm": 0.0100978659465909,
      "learning_rate": 2.06794682422452e-06,
      "loss": 0.0004,
      "step": 5242
    },
    {
      "epoch": 1.9184046834979875,
      "grad_norm": 3.3173937797546387,
      "learning_rate": 2.058714918759232e-06,
      "loss": 0.3495,
      "step": 5243
    },
    {
      "epoch": 1.9187705817782656,
      "grad_norm": 0.014637225307524204,
      "learning_rate": 2.049483013293944e-06,
      "loss": 0.0006,
      "step": 5244
    },
    {
      "epoch": 1.9191364800585438,
      "grad_norm": 0.010818821378052235,
      "learning_rate": 2.040251107828656e-06,
      "loss": 0.0005,
      "step": 5245
    },
    {
      "epoch": 1.9195023783388216,
      "grad_norm": 0.013562880456447601,
      "learning_rate": 2.031019202363368e-06,
      "loss": 0.0006,
      "step": 5246
    },
    {
      "epoch": 1.9198682766190998,
      "grad_norm": 0.009408848360180855,
      "learning_rate": 2.0217872968980797e-06,
      "loss": 0.0004,
      "step": 5247
    },
    {
      "epoch": 1.9202341748993779,
      "grad_norm": 0.042087700217962265,
      "learning_rate": 2.0125553914327916e-06,
      "loss": 0.0009,
      "step": 5248
    },
    {
      "epoch": 1.920600073179656,
      "grad_norm": 6.6471266746521,
      "learning_rate": 2.003323485967504e-06,
      "loss": 0.2474,
      "step": 5249
    },
    {
      "epoch": 1.920965971459934,
      "grad_norm": 0.009686136618256569,
      "learning_rate": 1.9940915805022156e-06,
      "loss": 0.0004,
      "step": 5250
    },
    {
      "epoch": 1.9213318697402122,
      "grad_norm": 0.006213278044015169,
      "learning_rate": 1.984859675036928e-06,
      "loss": 0.0003,
      "step": 5251
    },
    {
      "epoch": 1.9216977680204903,
      "grad_norm": 0.02403700351715088,
      "learning_rate": 1.9756277695716397e-06,
      "loss": 0.0006,
      "step": 5252
    },
    {
      "epoch": 1.9220636663007684,
      "grad_norm": 0.00897147785872221,
      "learning_rate": 1.966395864106352e-06,
      "loss": 0.0004,
      "step": 5253
    },
    {
      "epoch": 1.9224295645810465,
      "grad_norm": 0.012461263686418533,
      "learning_rate": 1.9571639586410633e-06,
      "loss": 0.0005,
      "step": 5254
    },
    {
      "epoch": 1.9227954628613246,
      "grad_norm": 5.701416492462158,
      "learning_rate": 1.9479320531757756e-06,
      "loss": 0.3464,
      "step": 5255
    },
    {
      "epoch": 1.9231613611416027,
      "grad_norm": 0.014536569826304913,
      "learning_rate": 1.9387001477104874e-06,
      "loss": 0.0005,
      "step": 5256
    },
    {
      "epoch": 1.9235272594218809,
      "grad_norm": 0.010455353185534477,
      "learning_rate": 1.9294682422451996e-06,
      "loss": 0.0004,
      "step": 5257
    },
    {
      "epoch": 1.923893157702159,
      "grad_norm": 0.019249431788921356,
      "learning_rate": 1.9202363367799115e-06,
      "loss": 0.0007,
      "step": 5258
    },
    {
      "epoch": 1.9242590559824369,
      "grad_norm": 0.01203659363090992,
      "learning_rate": 1.9110044313146237e-06,
      "loss": 0.0005,
      "step": 5259
    },
    {
      "epoch": 1.924624954262715,
      "grad_norm": 5.6898298263549805,
      "learning_rate": 1.9017725258493355e-06,
      "loss": 0.4445,
      "step": 5260
    },
    {
      "epoch": 1.924990852542993,
      "grad_norm": 26.4736270904541,
      "learning_rate": 1.8925406203840471e-06,
      "loss": 0.2876,
      "step": 5261
    },
    {
      "epoch": 1.9253567508232712,
      "grad_norm": 0.012161375023424625,
      "learning_rate": 1.8833087149187592e-06,
      "loss": 0.0005,
      "step": 5262
    },
    {
      "epoch": 1.925722649103549,
      "grad_norm": 20.018978118896484,
      "learning_rate": 1.8740768094534712e-06,
      "loss": 0.0672,
      "step": 5263
    },
    {
      "epoch": 1.9260885473838272,
      "grad_norm": 0.021083995699882507,
      "learning_rate": 1.8648449039881832e-06,
      "loss": 0.0007,
      "step": 5264
    },
    {
      "epoch": 1.9264544456641053,
      "grad_norm": 0.010028752498328686,
      "learning_rate": 1.8556129985228952e-06,
      "loss": 0.0004,
      "step": 5265
    },
    {
      "epoch": 1.9268203439443834,
      "grad_norm": 0.018823346123099327,
      "learning_rate": 1.8463810930576073e-06,
      "loss": 0.0006,
      "step": 5266
    },
    {
      "epoch": 1.9271862422246615,
      "grad_norm": 0.009161002933979034,
      "learning_rate": 1.8371491875923193e-06,
      "loss": 0.0004,
      "step": 5267
    },
    {
      "epoch": 1.9275521405049396,
      "grad_norm": 0.013507588766515255,
      "learning_rate": 1.827917282127031e-06,
      "loss": 0.0006,
      "step": 5268
    },
    {
      "epoch": 1.9279180387852177,
      "grad_norm": 0.010838381946086884,
      "learning_rate": 1.818685376661743e-06,
      "loss": 0.0004,
      "step": 5269
    },
    {
      "epoch": 1.9282839370654958,
      "grad_norm": 0.016163690015673637,
      "learning_rate": 1.809453471196455e-06,
      "loss": 0.0005,
      "step": 5270
    },
    {
      "epoch": 1.928649835345774,
      "grad_norm": 0.028679531067609787,
      "learning_rate": 1.800221565731167e-06,
      "loss": 0.001,
      "step": 5271
    },
    {
      "epoch": 1.929015733626052,
      "grad_norm": 0.008840259164571762,
      "learning_rate": 1.790989660265879e-06,
      "loss": 0.0004,
      "step": 5272
    },
    {
      "epoch": 1.9293816319063302,
      "grad_norm": 14.7344388961792,
      "learning_rate": 1.781757754800591e-06,
      "loss": 0.2841,
      "step": 5273
    },
    {
      "epoch": 1.9297475301866083,
      "grad_norm": 0.012019124813377857,
      "learning_rate": 1.7725258493353031e-06,
      "loss": 0.0005,
      "step": 5274
    },
    {
      "epoch": 1.9301134284668862,
      "grad_norm": 0.010730467736721039,
      "learning_rate": 1.763293943870015e-06,
      "loss": 0.0004,
      "step": 5275
    },
    {
      "epoch": 1.9304793267471643,
      "grad_norm": 2.4340174198150635,
      "learning_rate": 1.7540620384047267e-06,
      "loss": 0.015,
      "step": 5276
    },
    {
      "epoch": 1.9308452250274424,
      "grad_norm": 0.010322725400328636,
      "learning_rate": 1.7448301329394388e-06,
      "loss": 0.0004,
      "step": 5277
    },
    {
      "epoch": 1.9312111233077205,
      "grad_norm": 0.012112687341868877,
      "learning_rate": 1.7355982274741508e-06,
      "loss": 0.0005,
      "step": 5278
    },
    {
      "epoch": 1.9315770215879984,
      "grad_norm": 0.012053713202476501,
      "learning_rate": 1.7263663220088626e-06,
      "loss": 0.0005,
      "step": 5279
    },
    {
      "epoch": 1.9319429198682765,
      "grad_norm": 0.0142282759770751,
      "learning_rate": 1.7171344165435747e-06,
      "loss": 0.0006,
      "step": 5280
    },
    {
      "epoch": 1.9323088181485546,
      "grad_norm": 0.018396850675344467,
      "learning_rate": 1.7079025110782867e-06,
      "loss": 0.0007,
      "step": 5281
    },
    {
      "epoch": 1.9326747164288327,
      "grad_norm": 0.030754562467336655,
      "learning_rate": 1.6986706056129987e-06,
      "loss": 0.0008,
      "step": 5282
    },
    {
      "epoch": 1.9330406147091108,
      "grad_norm": 0.013925961218774319,
      "learning_rate": 1.6894387001477103e-06,
      "loss": 0.0006,
      "step": 5283
    },
    {
      "epoch": 1.933406512989389,
      "grad_norm": 0.24978764355182648,
      "learning_rate": 1.6802067946824224e-06,
      "loss": 0.0022,
      "step": 5284
    },
    {
      "epoch": 1.933772411269667,
      "grad_norm": 9.14721393585205,
      "learning_rate": 1.6709748892171344e-06,
      "loss": 0.2919,
      "step": 5285
    },
    {
      "epoch": 1.9341383095499451,
      "grad_norm": 1.9623044729232788,
      "learning_rate": 1.6617429837518464e-06,
      "loss": 0.461,
      "step": 5286
    },
    {
      "epoch": 1.9345042078302233,
      "grad_norm": 6.941864013671875,
      "learning_rate": 1.6525110782865585e-06,
      "loss": 0.2935,
      "step": 5287
    },
    {
      "epoch": 1.9348701061105014,
      "grad_norm": 0.010338682681322098,
      "learning_rate": 1.6432791728212705e-06,
      "loss": 0.0004,
      "step": 5288
    },
    {
      "epoch": 1.9352360043907795,
      "grad_norm": 0.011714549735188484,
      "learning_rate": 1.6340472673559825e-06,
      "loss": 0.0005,
      "step": 5289
    },
    {
      "epoch": 1.9356019026710576,
      "grad_norm": 0.020740963518619537,
      "learning_rate": 1.6248153618906941e-06,
      "loss": 0.0008,
      "step": 5290
    },
    {
      "epoch": 1.9359678009513357,
      "grad_norm": 0.02140139602124691,
      "learning_rate": 1.6155834564254062e-06,
      "loss": 0.0008,
      "step": 5291
    },
    {
      "epoch": 1.9363336992316136,
      "grad_norm": 0.015339767560362816,
      "learning_rate": 1.6063515509601182e-06,
      "loss": 0.0006,
      "step": 5292
    },
    {
      "epoch": 1.9366995975118917,
      "grad_norm": 0.012354783713817596,
      "learning_rate": 1.5971196454948302e-06,
      "loss": 0.0005,
      "step": 5293
    },
    {
      "epoch": 1.9370654957921698,
      "grad_norm": 0.018447499722242355,
      "learning_rate": 1.5878877400295423e-06,
      "loss": 0.0006,
      "step": 5294
    },
    {
      "epoch": 1.937431394072448,
      "grad_norm": 0.020076684653759003,
      "learning_rate": 1.5786558345642543e-06,
      "loss": 0.0008,
      "step": 5295
    },
    {
      "epoch": 1.9377972923527258,
      "grad_norm": 0.012952711433172226,
      "learning_rate": 1.5694239290989663e-06,
      "loss": 0.0005,
      "step": 5296
    },
    {
      "epoch": 1.938163190633004,
      "grad_norm": 0.04428311809897423,
      "learning_rate": 1.5601920236336781e-06,
      "loss": 0.0009,
      "step": 5297
    },
    {
      "epoch": 1.938529088913282,
      "grad_norm": 0.015189536847174168,
      "learning_rate": 1.5509601181683902e-06,
      "loss": 0.0006,
      "step": 5298
    },
    {
      "epoch": 1.9388949871935601,
      "grad_norm": 0.023720690980553627,
      "learning_rate": 1.541728212703102e-06,
      "loss": 0.0009,
      "step": 5299
    },
    {
      "epoch": 1.9392608854738382,
      "grad_norm": 0.015895472839474678,
      "learning_rate": 1.532496307237814e-06,
      "loss": 0.0007,
      "step": 5300
    },
    {
      "epoch": 1.9396267837541163,
      "grad_norm": 0.01231888122856617,
      "learning_rate": 1.523264401772526e-06,
      "loss": 0.0005,
      "step": 5301
    },
    {
      "epoch": 1.9399926820343945,
      "grad_norm": 0.01591840386390686,
      "learning_rate": 1.5140324963072379e-06,
      "loss": 0.0006,
      "step": 5302
    },
    {
      "epoch": 1.9403585803146726,
      "grad_norm": 0.006812781561166048,
      "learning_rate": 1.50480059084195e-06,
      "loss": 0.0003,
      "step": 5303
    },
    {
      "epoch": 1.9407244785949507,
      "grad_norm": 0.01141938753426075,
      "learning_rate": 1.4955686853766617e-06,
      "loss": 0.0005,
      "step": 5304
    },
    {
      "epoch": 1.9410903768752288,
      "grad_norm": 0.042907316237688065,
      "learning_rate": 1.4863367799113738e-06,
      "loss": 0.0013,
      "step": 5305
    },
    {
      "epoch": 1.941456275155507,
      "grad_norm": 10.730822563171387,
      "learning_rate": 1.4771048744460856e-06,
      "loss": 0.1933,
      "step": 5306
    },
    {
      "epoch": 1.941822173435785,
      "grad_norm": 0.3243115246295929,
      "learning_rate": 1.4678729689807976e-06,
      "loss": 0.0019,
      "step": 5307
    },
    {
      "epoch": 1.942188071716063,
      "grad_norm": 0.011510038748383522,
      "learning_rate": 1.4586410635155096e-06,
      "loss": 0.0005,
      "step": 5308
    },
    {
      "epoch": 1.942553969996341,
      "grad_norm": 1.6246668100357056,
      "learning_rate": 1.4494091580502217e-06,
      "loss": 0.0194,
      "step": 5309
    },
    {
      "epoch": 1.942919868276619,
      "grad_norm": 0.026976587250828743,
      "learning_rate": 1.4401772525849335e-06,
      "loss": 0.0009,
      "step": 5310
    },
    {
      "epoch": 1.9432857665568972,
      "grad_norm": 0.019449923187494278,
      "learning_rate": 1.4309453471196455e-06,
      "loss": 0.0007,
      "step": 5311
    },
    {
      "epoch": 1.943651664837175,
      "grad_norm": 0.053342778235673904,
      "learning_rate": 1.4217134416543575e-06,
      "loss": 0.0013,
      "step": 5312
    },
    {
      "epoch": 1.9440175631174532,
      "grad_norm": 0.01956375315785408,
      "learning_rate": 1.4124815361890694e-06,
      "loss": 0.0008,
      "step": 5313
    },
    {
      "epoch": 1.9443834613977313,
      "grad_norm": 0.02061992511153221,
      "learning_rate": 1.4032496307237814e-06,
      "loss": 0.0008,
      "step": 5314
    },
    {
      "epoch": 1.9447493596780094,
      "grad_norm": 0.019361337646842003,
      "learning_rate": 1.3940177252584934e-06,
      "loss": 0.0006,
      "step": 5315
    },
    {
      "epoch": 1.9451152579582875,
      "grad_norm": 0.34227901697158813,
      "learning_rate": 1.3847858197932055e-06,
      "loss": 0.0028,
      "step": 5316
    },
    {
      "epoch": 1.9454811562385657,
      "grad_norm": 0.019348187372088432,
      "learning_rate": 1.3755539143279173e-06,
      "loss": 0.0007,
      "step": 5317
    },
    {
      "epoch": 1.9458470545188438,
      "grad_norm": 0.05452587828040123,
      "learning_rate": 1.3663220088626293e-06,
      "loss": 0.0012,
      "step": 5318
    },
    {
      "epoch": 1.9462129527991219,
      "grad_norm": 0.012075516395270824,
      "learning_rate": 1.3570901033973413e-06,
      "loss": 0.0005,
      "step": 5319
    },
    {
      "epoch": 1.9465788510794,
      "grad_norm": 0.6838570833206177,
      "learning_rate": 1.3478581979320534e-06,
      "loss": 0.0044,
      "step": 5320
    },
    {
      "epoch": 1.946944749359678,
      "grad_norm": 52.198020935058594,
      "learning_rate": 1.3386262924667652e-06,
      "loss": 0.0902,
      "step": 5321
    },
    {
      "epoch": 1.9473106476399562,
      "grad_norm": 0.019869336858391762,
      "learning_rate": 1.3293943870014772e-06,
      "loss": 0.0008,
      "step": 5322
    },
    {
      "epoch": 1.9476765459202343,
      "grad_norm": 0.032592155039310455,
      "learning_rate": 1.3201624815361893e-06,
      "loss": 0.0009,
      "step": 5323
    },
    {
      "epoch": 1.9480424442005122,
      "grad_norm": 0.01352542731910944,
      "learning_rate": 1.310930576070901e-06,
      "loss": 0.0005,
      "step": 5324
    },
    {
      "epoch": 1.9484083424807903,
      "grad_norm": 0.010239598341286182,
      "learning_rate": 1.3016986706056131e-06,
      "loss": 0.0004,
      "step": 5325
    },
    {
      "epoch": 1.9487742407610684,
      "grad_norm": 0.016973337158560753,
      "learning_rate": 1.2924667651403251e-06,
      "loss": 0.0006,
      "step": 5326
    },
    {
      "epoch": 1.9491401390413465,
      "grad_norm": 0.017838990315794945,
      "learning_rate": 1.283234859675037e-06,
      "loss": 0.0006,
      "step": 5327
    },
    {
      "epoch": 1.9495060373216246,
      "grad_norm": 0.024048147723078728,
      "learning_rate": 1.274002954209749e-06,
      "loss": 0.0009,
      "step": 5328
    },
    {
      "epoch": 1.9498719356019025,
      "grad_norm": 0.01078273169696331,
      "learning_rate": 1.2647710487444608e-06,
      "loss": 0.0004,
      "step": 5329
    },
    {
      "epoch": 1.9502378338821806,
      "grad_norm": 6.257836818695068,
      "learning_rate": 1.2555391432791728e-06,
      "loss": 0.22,
      "step": 5330
    },
    {
      "epoch": 1.9506037321624587,
      "grad_norm": 0.014711443334817886,
      "learning_rate": 1.2463072378138847e-06,
      "loss": 0.0006,
      "step": 5331
    },
    {
      "epoch": 1.9509696304427369,
      "grad_norm": 0.022848550230264664,
      "learning_rate": 1.2370753323485967e-06,
      "loss": 0.0008,
      "step": 5332
    },
    {
      "epoch": 1.951335528723015,
      "grad_norm": 0.020977934822440147,
      "learning_rate": 1.2278434268833087e-06,
      "loss": 0.0007,
      "step": 5333
    },
    {
      "epoch": 1.951701427003293,
      "grad_norm": 0.016205161809921265,
      "learning_rate": 1.2186115214180208e-06,
      "loss": 0.0007,
      "step": 5334
    },
    {
      "epoch": 1.9520673252835712,
      "grad_norm": 2.155777931213379,
      "learning_rate": 1.2093796159527326e-06,
      "loss": 0.4152,
      "step": 5335
    },
    {
      "epoch": 1.9524332235638493,
      "grad_norm": 0.017148340120911598,
      "learning_rate": 1.2001477104874446e-06,
      "loss": 0.0006,
      "step": 5336
    },
    {
      "epoch": 1.9527991218441274,
      "grad_norm": 0.2698966860771179,
      "learning_rate": 1.1909158050221566e-06,
      "loss": 0.002,
      "step": 5337
    },
    {
      "epoch": 1.9531650201244055,
      "grad_norm": 0.015424508601427078,
      "learning_rate": 1.1816838995568687e-06,
      "loss": 0.0006,
      "step": 5338
    },
    {
      "epoch": 1.9535309184046836,
      "grad_norm": 0.02253361977636814,
      "learning_rate": 1.1724519940915805e-06,
      "loss": 0.0008,
      "step": 5339
    },
    {
      "epoch": 1.9538968166849617,
      "grad_norm": 0.01476624608039856,
      "learning_rate": 1.1632200886262925e-06,
      "loss": 0.0006,
      "step": 5340
    },
    {
      "epoch": 1.9542627149652396,
      "grad_norm": 0.04255818575620651,
      "learning_rate": 1.1539881831610046e-06,
      "loss": 0.001,
      "step": 5341
    },
    {
      "epoch": 1.9546286132455177,
      "grad_norm": 0.017261240631341934,
      "learning_rate": 1.1447562776957164e-06,
      "loss": 0.0006,
      "step": 5342
    },
    {
      "epoch": 1.9549945115257958,
      "grad_norm": 0.015055711381137371,
      "learning_rate": 1.1355243722304284e-06,
      "loss": 0.0006,
      "step": 5343
    },
    {
      "epoch": 1.955360409806074,
      "grad_norm": 0.015713682398200035,
      "learning_rate": 1.1262924667651404e-06,
      "loss": 0.0006,
      "step": 5344
    },
    {
      "epoch": 1.9557263080863518,
      "grad_norm": 0.017316820099949837,
      "learning_rate": 1.1170605612998525e-06,
      "loss": 0.0006,
      "step": 5345
    },
    {
      "epoch": 1.95609220636663,
      "grad_norm": 2.9518470764160156,
      "learning_rate": 1.1078286558345643e-06,
      "loss": 0.5695,
      "step": 5346
    },
    {
      "epoch": 1.956458104646908,
      "grad_norm": 0.016444582492113113,
      "learning_rate": 1.0985967503692763e-06,
      "loss": 0.0006,
      "step": 5347
    },
    {
      "epoch": 1.9568240029271862,
      "grad_norm": 0.028525125235319138,
      "learning_rate": 1.0893648449039883e-06,
      "loss": 0.0009,
      "step": 5348
    },
    {
      "epoch": 1.9571899012074643,
      "grad_norm": 0.015421335585415363,
      "learning_rate": 1.0801329394387002e-06,
      "loss": 0.0006,
      "step": 5349
    },
    {
      "epoch": 1.9575557994877424,
      "grad_norm": 0.029087848961353302,
      "learning_rate": 1.0709010339734122e-06,
      "loss": 0.0005,
      "step": 5350
    },
    {
      "epoch": 1.9579216977680205,
      "grad_norm": 7.231049537658691,
      "learning_rate": 1.0616691285081242e-06,
      "loss": 0.3113,
      "step": 5351
    },
    {
      "epoch": 1.9582875960482986,
      "grad_norm": 0.3398507535457611,
      "learning_rate": 1.052437223042836e-06,
      "loss": 0.005,
      "step": 5352
    },
    {
      "epoch": 1.9586534943285767,
      "grad_norm": 0.015057605691254139,
      "learning_rate": 1.043205317577548e-06,
      "loss": 0.0006,
      "step": 5353
    },
    {
      "epoch": 1.9590193926088548,
      "grad_norm": 7.583098411560059,
      "learning_rate": 1.03397341211226e-06,
      "loss": 0.7029,
      "step": 5354
    },
    {
      "epoch": 1.959385290889133,
      "grad_norm": 0.03017222322523594,
      "learning_rate": 1.024741506646972e-06,
      "loss": 0.0008,
      "step": 5355
    },
    {
      "epoch": 1.959751189169411,
      "grad_norm": 0.009579743258655071,
      "learning_rate": 1.015509601181684e-06,
      "loss": 0.0004,
      "step": 5356
    },
    {
      "epoch": 1.960117087449689,
      "grad_norm": 0.16290214657783508,
      "learning_rate": 1.0062776957163958e-06,
      "loss": 0.0019,
      "step": 5357
    },
    {
      "epoch": 1.960482985729967,
      "grad_norm": 13.12922477722168,
      "learning_rate": 9.970457902511078e-07,
      "loss": 0.187,
      "step": 5358
    },
    {
      "epoch": 1.9608488840102452,
      "grad_norm": 0.01674077846109867,
      "learning_rate": 9.878138847858198e-07,
      "loss": 0.0007,
      "step": 5359
    },
    {
      "epoch": 1.9612147822905233,
      "grad_norm": 0.01544404961168766,
      "learning_rate": 9.785819793205317e-07,
      "loss": 0.0007,
      "step": 5360
    },
    {
      "epoch": 1.9615806805708014,
      "grad_norm": 0.04299624636769295,
      "learning_rate": 9.693500738552437e-07,
      "loss": 0.0009,
      "step": 5361
    },
    {
      "epoch": 1.9619465788510793,
      "grad_norm": 0.01341270562261343,
      "learning_rate": 9.601181683899557e-07,
      "loss": 0.0006,
      "step": 5362
    },
    {
      "epoch": 1.9623124771313574,
      "grad_norm": 0.01866983249783516,
      "learning_rate": 9.508862629246678e-07,
      "loss": 0.0006,
      "step": 5363
    },
    {
      "epoch": 1.9626783754116355,
      "grad_norm": 0.020396703854203224,
      "learning_rate": 9.416543574593796e-07,
      "loss": 0.0009,
      "step": 5364
    },
    {
      "epoch": 1.9630442736919136,
      "grad_norm": 2.311105728149414,
      "learning_rate": 9.324224519940916e-07,
      "loss": 0.4368,
      "step": 5365
    },
    {
      "epoch": 1.9634101719721917,
      "grad_norm": 0.03230583667755127,
      "learning_rate": 9.231905465288036e-07,
      "loss": 0.0009,
      "step": 5366
    },
    {
      "epoch": 1.9637760702524698,
      "grad_norm": 0.011731013655662537,
      "learning_rate": 9.139586410635155e-07,
      "loss": 0.0005,
      "step": 5367
    },
    {
      "epoch": 1.964141968532748,
      "grad_norm": 0.017579808831214905,
      "learning_rate": 9.047267355982275e-07,
      "loss": 0.0007,
      "step": 5368
    },
    {
      "epoch": 1.964507866813026,
      "grad_norm": 0.29352834820747375,
      "learning_rate": 8.954948301329395e-07,
      "loss": 0.0021,
      "step": 5369
    },
    {
      "epoch": 1.9648737650933041,
      "grad_norm": 0.03730339556932449,
      "learning_rate": 8.862629246676516e-07,
      "loss": 0.0009,
      "step": 5370
    },
    {
      "epoch": 1.9652396633735822,
      "grad_norm": 0.01425280049443245,
      "learning_rate": 8.770310192023634e-07,
      "loss": 0.0006,
      "step": 5371
    },
    {
      "epoch": 1.9656055616538604,
      "grad_norm": 0.023387299850583076,
      "learning_rate": 8.677991137370754e-07,
      "loss": 0.0008,
      "step": 5372
    },
    {
      "epoch": 1.9659714599341385,
      "grad_norm": 0.019751235842704773,
      "learning_rate": 8.585672082717873e-07,
      "loss": 0.0008,
      "step": 5373
    },
    {
      "epoch": 1.9663373582144164,
      "grad_norm": 0.01657267101109028,
      "learning_rate": 8.493353028064994e-07,
      "loss": 0.0007,
      "step": 5374
    },
    {
      "epoch": 1.9667032564946945,
      "grad_norm": 0.02813330851495266,
      "learning_rate": 8.401033973412112e-07,
      "loss": 0.0009,
      "step": 5375
    },
    {
      "epoch": 1.9670691547749726,
      "grad_norm": 0.016556769609451294,
      "learning_rate": 8.308714918759232e-07,
      "loss": 0.0007,
      "step": 5376
    },
    {
      "epoch": 1.9674350530552507,
      "grad_norm": 0.048347488045692444,
      "learning_rate": 8.216395864106352e-07,
      "loss": 0.0009,
      "step": 5377
    },
    {
      "epoch": 1.9678009513355286,
      "grad_norm": 0.013235826045274734,
      "learning_rate": 8.124076809453471e-07,
      "loss": 0.0006,
      "step": 5378
    },
    {
      "epoch": 1.9681668496158067,
      "grad_norm": 0.025654161348938942,
      "learning_rate": 8.031757754800591e-07,
      "loss": 0.001,
      "step": 5379
    },
    {
      "epoch": 1.9685327478960848,
      "grad_norm": 0.013327337801456451,
      "learning_rate": 7.939438700147711e-07,
      "loss": 0.0005,
      "step": 5380
    },
    {
      "epoch": 1.968898646176363,
      "grad_norm": 0.046013038605451584,
      "learning_rate": 7.847119645494832e-07,
      "loss": 0.0014,
      "step": 5381
    },
    {
      "epoch": 1.969264544456641,
      "grad_norm": 4.8843817710876465,
      "learning_rate": 7.754800590841951e-07,
      "loss": 0.3483,
      "step": 5382
    },
    {
      "epoch": 1.9696304427369191,
      "grad_norm": 0.011638229712843895,
      "learning_rate": 7.66248153618907e-07,
      "loss": 0.0005,
      "step": 5383
    },
    {
      "epoch": 1.9699963410171972,
      "grad_norm": 0.028632434085011482,
      "learning_rate": 7.570162481536189e-07,
      "loss": 0.0011,
      "step": 5384
    },
    {
      "epoch": 1.9703622392974753,
      "grad_norm": 0.05857148766517639,
      "learning_rate": 7.477843426883309e-07,
      "loss": 0.0012,
      "step": 5385
    },
    {
      "epoch": 1.9707281375777534,
      "grad_norm": 0.3064340353012085,
      "learning_rate": 7.385524372230428e-07,
      "loss": 0.0015,
      "step": 5386
    },
    {
      "epoch": 1.9710940358580316,
      "grad_norm": 0.017917724326252937,
      "learning_rate": 7.293205317577548e-07,
      "loss": 0.0006,
      "step": 5387
    },
    {
      "epoch": 1.9714599341383097,
      "grad_norm": 0.022650981321930885,
      "learning_rate": 7.200886262924667e-07,
      "loss": 0.0009,
      "step": 5388
    },
    {
      "epoch": 1.9718258324185878,
      "grad_norm": 0.025652781128883362,
      "learning_rate": 7.108567208271788e-07,
      "loss": 0.0009,
      "step": 5389
    },
    {
      "epoch": 1.9721917306988657,
      "grad_norm": 0.028575466945767403,
      "learning_rate": 7.016248153618907e-07,
      "loss": 0.0011,
      "step": 5390
    },
    {
      "epoch": 1.9725576289791438,
      "grad_norm": 0.013371365144848824,
      "learning_rate": 6.923929098966027e-07,
      "loss": 0.0006,
      "step": 5391
    },
    {
      "epoch": 1.9729235272594219,
      "grad_norm": 0.020344020798802376,
      "learning_rate": 6.831610044313147e-07,
      "loss": 0.0007,
      "step": 5392
    },
    {
      "epoch": 1.9732894255397,
      "grad_norm": 0.016781890764832497,
      "learning_rate": 6.739290989660267e-07,
      "loss": 0.0007,
      "step": 5393
    },
    {
      "epoch": 1.9736553238199779,
      "grad_norm": 0.018778637051582336,
      "learning_rate": 6.646971935007386e-07,
      "loss": 0.0007,
      "step": 5394
    },
    {
      "epoch": 1.974021222100256,
      "grad_norm": 0.3651171326637268,
      "learning_rate": 6.554652880354505e-07,
      "loss": 0.0028,
      "step": 5395
    },
    {
      "epoch": 1.974387120380534,
      "grad_norm": 0.03786512091755867,
      "learning_rate": 6.462333825701626e-07,
      "loss": 0.0007,
      "step": 5396
    },
    {
      "epoch": 1.9747530186608122,
      "grad_norm": 37.64456558227539,
      "learning_rate": 6.370014771048745e-07,
      "loss": 0.0731,
      "step": 5397
    },
    {
      "epoch": 1.9751189169410903,
      "grad_norm": 0.011650120839476585,
      "learning_rate": 6.277695716395864e-07,
      "loss": 0.0005,
      "step": 5398
    },
    {
      "epoch": 1.9754848152213684,
      "grad_norm": 0.013130156323313713,
      "learning_rate": 6.185376661742983e-07,
      "loss": 0.0005,
      "step": 5399
    },
    {
      "epoch": 1.9758507135016465,
      "grad_norm": 0.22776231169700623,
      "learning_rate": 6.093057607090104e-07,
      "loss": 0.0017,
      "step": 5400
    },
    {
      "epoch": 1.9762166117819246,
      "grad_norm": 0.01044655591249466,
      "learning_rate": 6.000738552437223e-07,
      "loss": 0.0005,
      "step": 5401
    },
    {
      "epoch": 1.9765825100622028,
      "grad_norm": 0.1790652573108673,
      "learning_rate": 5.908419497784343e-07,
      "loss": 0.0021,
      "step": 5402
    },
    {
      "epoch": 1.9769484083424809,
      "grad_norm": 0.023690462112426758,
      "learning_rate": 5.816100443131463e-07,
      "loss": 0.0007,
      "step": 5403
    },
    {
      "epoch": 1.977314306622759,
      "grad_norm": 0.01882549561560154,
      "learning_rate": 5.723781388478582e-07,
      "loss": 0.0006,
      "step": 5404
    },
    {
      "epoch": 1.977680204903037,
      "grad_norm": 0.020074056461453438,
      "learning_rate": 5.631462333825702e-07,
      "loss": 0.0007,
      "step": 5405
    },
    {
      "epoch": 1.9780461031833152,
      "grad_norm": 0.01011939998716116,
      "learning_rate": 5.539143279172821e-07,
      "loss": 0.0004,
      "step": 5406
    },
    {
      "epoch": 1.978412001463593,
      "grad_norm": 0.012768508866429329,
      "learning_rate": 5.446824224519942e-07,
      "loss": 0.0006,
      "step": 5407
    },
    {
      "epoch": 1.9787778997438712,
      "grad_norm": 0.014665121212601662,
      "learning_rate": 5.354505169867061e-07,
      "loss": 0.0005,
      "step": 5408
    },
    {
      "epoch": 1.9791437980241493,
      "grad_norm": 0.0161882434040308,
      "learning_rate": 5.26218611521418e-07,
      "loss": 0.0007,
      "step": 5409
    },
    {
      "epoch": 1.9795096963044274,
      "grad_norm": 0.039443179965019226,
      "learning_rate": 5.1698670605613e-07,
      "loss": 0.0014,
      "step": 5410
    },
    {
      "epoch": 1.9798755945847053,
      "grad_norm": 0.10345933586359024,
      "learning_rate": 5.07754800590842e-07,
      "loss": 0.0011,
      "step": 5411
    },
    {
      "epoch": 1.9802414928649834,
      "grad_norm": 0.011724976822733879,
      "learning_rate": 4.985228951255539e-07,
      "loss": 0.0005,
      "step": 5412
    },
    {
      "epoch": 1.9806073911452615,
      "grad_norm": 0.025410091504454613,
      "learning_rate": 4.892909896602658e-07,
      "loss": 0.001,
      "step": 5413
    },
    {
      "epoch": 1.9809732894255396,
      "grad_norm": 5.833539009094238,
      "learning_rate": 4.800590841949779e-07,
      "loss": 0.2012,
      "step": 5414
    },
    {
      "epoch": 1.9813391877058177,
      "grad_norm": 0.4036056697368622,
      "learning_rate": 4.708271787296898e-07,
      "loss": 0.0035,
      "step": 5415
    },
    {
      "epoch": 1.9817050859860958,
      "grad_norm": 0.015076389536261559,
      "learning_rate": 4.615952732644018e-07,
      "loss": 0.0006,
      "step": 5416
    },
    {
      "epoch": 1.982070984266374,
      "grad_norm": 0.5591112971305847,
      "learning_rate": 4.5236336779911375e-07,
      "loss": 0.0089,
      "step": 5417
    },
    {
      "epoch": 1.982436882546652,
      "grad_norm": 6.391634941101074,
      "learning_rate": 4.431314623338258e-07,
      "loss": 0.108,
      "step": 5418
    },
    {
      "epoch": 1.9828027808269302,
      "grad_norm": 0.010698670521378517,
      "learning_rate": 4.338995568685377e-07,
      "loss": 0.0005,
      "step": 5419
    },
    {
      "epoch": 1.9831686791072083,
      "grad_norm": 0.029627928510308266,
      "learning_rate": 4.246676514032497e-07,
      "loss": 0.001,
      "step": 5420
    },
    {
      "epoch": 1.9835345773874864,
      "grad_norm": 9.318329811096191,
      "learning_rate": 4.154357459379616e-07,
      "loss": 0.1601,
      "step": 5421
    },
    {
      "epoch": 1.9839004756677645,
      "grad_norm": 0.02555830590426922,
      "learning_rate": 4.0620384047267353e-07,
      "loss": 0.001,
      "step": 5422
    },
    {
      "epoch": 1.9842663739480424,
      "grad_norm": 0.02717740647494793,
      "learning_rate": 3.9697193500738556e-07,
      "loss": 0.001,
      "step": 5423
    },
    {
      "epoch": 1.9846322722283205,
      "grad_norm": 0.011603794991970062,
      "learning_rate": 3.8774002954209754e-07,
      "loss": 0.0005,
      "step": 5424
    },
    {
      "epoch": 1.9849981705085986,
      "grad_norm": 0.04246613383293152,
      "learning_rate": 3.7850812407680947e-07,
      "loss": 0.0011,
      "step": 5425
    },
    {
      "epoch": 1.9853640687888767,
      "grad_norm": 0.030128775164484978,
      "learning_rate": 3.692762186115214e-07,
      "loss": 0.0011,
      "step": 5426
    },
    {
      "epoch": 1.9857299670691546,
      "grad_norm": 0.27091172337532043,
      "learning_rate": 3.6004431314623337e-07,
      "loss": 0.0037,
      "step": 5427
    },
    {
      "epoch": 1.9860958653494327,
      "grad_norm": 5.759576320648193,
      "learning_rate": 3.5081240768094535e-07,
      "loss": 0.309,
      "step": 5428
    },
    {
      "epoch": 1.9864617636297108,
      "grad_norm": 0.010403675027191639,
      "learning_rate": 3.4158050221565733e-07,
      "loss": 0.0005,
      "step": 5429
    },
    {
      "epoch": 1.986827661909989,
      "grad_norm": 1.358656883239746,
      "learning_rate": 3.323485967503693e-07,
      "loss": 0.0077,
      "step": 5430
    },
    {
      "epoch": 1.987193560190267,
      "grad_norm": 0.016457565128803253,
      "learning_rate": 3.231166912850813e-07,
      "loss": 0.0006,
      "step": 5431
    },
    {
      "epoch": 1.9875594584705452,
      "grad_norm": 0.014839529059827328,
      "learning_rate": 3.138847858197932e-07,
      "loss": 0.0006,
      "step": 5432
    },
    {
      "epoch": 1.9879253567508233,
      "grad_norm": 0.007468000985682011,
      "learning_rate": 3.046528803545052e-07,
      "loss": 0.0003,
      "step": 5433
    },
    {
      "epoch": 1.9882912550311014,
      "grad_norm": 0.021238919347524643,
      "learning_rate": 2.9542097488921717e-07,
      "loss": 0.0008,
      "step": 5434
    },
    {
      "epoch": 1.9886571533113795,
      "grad_norm": 0.009958191774785519,
      "learning_rate": 2.861890694239291e-07,
      "loss": 0.0004,
      "step": 5435
    },
    {
      "epoch": 1.9890230515916576,
      "grad_norm": 13.307206153869629,
      "learning_rate": 2.7695716395864107e-07,
      "loss": 0.1989,
      "step": 5436
    },
    {
      "epoch": 1.9893889498719357,
      "grad_norm": 6.694438934326172,
      "learning_rate": 2.6772525849335305e-07,
      "loss": 0.3231,
      "step": 5437
    },
    {
      "epoch": 1.9897548481522138,
      "grad_norm": 0.17296062409877777,
      "learning_rate": 2.58493353028065e-07,
      "loss": 0.0022,
      "step": 5438
    },
    {
      "epoch": 1.990120746432492,
      "grad_norm": 0.04368946701288223,
      "learning_rate": 2.4926144756277695e-07,
      "loss": 0.0013,
      "step": 5439
    },
    {
      "epoch": 1.9904866447127698,
      "grad_norm": 0.22266912460327148,
      "learning_rate": 2.4002954209748893e-07,
      "loss": 0.0025,
      "step": 5440
    },
    {
      "epoch": 1.990852542993048,
      "grad_norm": 1.2233010530471802,
      "learning_rate": 2.307976366322009e-07,
      "loss": 0.0079,
      "step": 5441
    },
    {
      "epoch": 1.991218441273326,
      "grad_norm": 0.016002260148525238,
      "learning_rate": 2.215657311669129e-07,
      "loss": 0.0007,
      "step": 5442
    },
    {
      "epoch": 1.9915843395536041,
      "grad_norm": 0.02424187771975994,
      "learning_rate": 2.1233382570162484e-07,
      "loss": 0.001,
      "step": 5443
    },
    {
      "epoch": 1.991950237833882,
      "grad_norm": 0.01426553912460804,
      "learning_rate": 2.0310192023633677e-07,
      "loss": 0.0006,
      "step": 5444
    },
    {
      "epoch": 1.9923161361141601,
      "grad_norm": 0.012174546718597412,
      "learning_rate": 1.9387001477104877e-07,
      "loss": 0.0005,
      "step": 5445
    },
    {
      "epoch": 1.9926820343944383,
      "grad_norm": 7.063535690307617,
      "learning_rate": 1.846381093057607e-07,
      "loss": 0.222,
      "step": 5446
    },
    {
      "epoch": 1.9930479326747164,
      "grad_norm": 0.043873853981494904,
      "learning_rate": 1.7540620384047267e-07,
      "loss": 0.0013,
      "step": 5447
    },
    {
      "epoch": 1.9934138309549945,
      "grad_norm": 0.01740357279777527,
      "learning_rate": 1.6617429837518465e-07,
      "loss": 0.0007,
      "step": 5448
    },
    {
      "epoch": 1.9937797292352726,
      "grad_norm": 0.007199384272098541,
      "learning_rate": 1.569423929098966e-07,
      "loss": 0.0003,
      "step": 5449
    },
    {
      "epoch": 1.9941456275155507,
      "grad_norm": 0.014597057364881039,
      "learning_rate": 1.4771048744460858e-07,
      "loss": 0.0005,
      "step": 5450
    },
    {
      "epoch": 1.9945115257958288,
      "grad_norm": 0.01421300321817398,
      "learning_rate": 1.3847858197932054e-07,
      "loss": 0.0006,
      "step": 5451
    },
    {
      "epoch": 1.994877424076107,
      "grad_norm": 0.018708467483520508,
      "learning_rate": 1.292466765140325e-07,
      "loss": 0.0007,
      "step": 5452
    },
    {
      "epoch": 1.995243322356385,
      "grad_norm": 0.04663774371147156,
      "learning_rate": 1.2001477104874447e-07,
      "loss": 0.0011,
      "step": 5453
    },
    {
      "epoch": 1.9956092206366631,
      "grad_norm": 0.14848817884922028,
      "learning_rate": 1.1078286558345644e-07,
      "loss": 0.0026,
      "step": 5454
    },
    {
      "epoch": 1.9959751189169412,
      "grad_norm": 0.014926262199878693,
      "learning_rate": 1.0155096011816838e-07,
      "loss": 0.0006,
      "step": 5455
    },
    {
      "epoch": 1.9963410171972191,
      "grad_norm": 0.01872282661497593,
      "learning_rate": 9.231905465288035e-08,
      "loss": 0.0007,
      "step": 5456
    },
    {
      "epoch": 1.9967069154774972,
      "grad_norm": 0.5250523686408997,
      "learning_rate": 8.308714918759233e-08,
      "loss": 0.0034,
      "step": 5457
    },
    {
      "epoch": 1.9970728137577753,
      "grad_norm": 0.10431642830371857,
      "learning_rate": 7.385524372230429e-08,
      "loss": 0.0014,
      "step": 5458
    },
    {
      "epoch": 1.9974387120380535,
      "grad_norm": 5.642697811126709,
      "learning_rate": 6.462333825701624e-08,
      "loss": 0.0182,
      "step": 5459
    },
    {
      "epoch": 1.9978046103183313,
      "grad_norm": 0.4915127754211426,
      "learning_rate": 5.539143279172822e-08,
      "loss": 0.003,
      "step": 5460
    },
    {
      "epoch": 1.9981705085986095,
      "grad_norm": 0.008746049366891384,
      "learning_rate": 4.6159527326440174e-08,
      "loss": 0.0003,
      "step": 5461
    },
    {
      "epoch": 1.9985364068788876,
      "grad_norm": 5.926745891571045,
      "learning_rate": 3.6927621861152146e-08,
      "loss": 0.0092,
      "step": 5462
    },
    {
      "epoch": 1.9989023051591657,
      "grad_norm": 0.029802171513438225,
      "learning_rate": 2.769571639586411e-08,
      "loss": 0.0011,
      "step": 5463
    },
    {
      "epoch": 1.9992682034394438,
      "grad_norm": 0.011407977901399136,
      "learning_rate": 1.8463810930576073e-08,
      "loss": 0.0005,
      "step": 5464
    },
    {
      "epoch": 1.999634101719722,
      "grad_norm": 2.3521955013275146,
      "learning_rate": 9.231905465288036e-09,
      "loss": 0.0195,
      "step": 5465
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.015526242554187775,
      "learning_rate": 0.0,
      "loss": 0.0006,
      "step": 5466
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9901193906957596,
      "eval_loss": 0.04109187051653862,
      "eval_runtime": 47.0176,
      "eval_samples_per_second": 103.323,
      "eval_steps_per_second": 6.466,
      "step": 5466
    }
  ],
  "logging_steps": 1,
  "max_steps": 5466,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1571333603460992e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}