{
  "best_metric": 0.6017911434173584,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8820286659316428,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004410143329658214,
      "grad_norm": 47.876609802246094,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 7.0707,
      "step": 1
    },
    {
      "epoch": 0.004410143329658214,
      "eval_loss": 1.7239896059036255,
      "eval_runtime": 120.2101,
      "eval_samples_per_second": 3.178,
      "eval_steps_per_second": 0.799,
      "step": 1
    },
    {
      "epoch": 0.008820286659316428,
      "grad_norm": 52.10292434692383,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 7.6856,
      "step": 2
    },
    {
      "epoch": 0.013230429988974642,
      "grad_norm": 37.03164291381836,
      "learning_rate": 3.3e-05,
      "loss": 7.0494,
      "step": 3
    },
    {
      "epoch": 0.017640573318632856,
      "grad_norm": 21.86464500427246,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 6.6184,
      "step": 4
    },
    {
      "epoch": 0.022050716648291068,
      "grad_norm": 28.910720825195312,
      "learning_rate": 5.5e-05,
      "loss": 6.3385,
      "step": 5
    },
    {
      "epoch": 0.026460859977949284,
      "grad_norm": 17.85317611694336,
      "learning_rate": 6.6e-05,
      "loss": 5.7431,
      "step": 6
    },
    {
      "epoch": 0.030871003307607496,
      "grad_norm": 8.013505935668945,
      "learning_rate": 7.7e-05,
      "loss": 5.4803,
      "step": 7
    },
    {
      "epoch": 0.03528114663726571,
      "grad_norm": 14.263498306274414,
      "learning_rate": 8.800000000000001e-05,
      "loss": 5.1798,
      "step": 8
    },
    {
      "epoch": 0.03969128996692393,
      "grad_norm": 12.464676856994629,
      "learning_rate": 9.900000000000001e-05,
      "loss": 4.7788,
      "step": 9
    },
    {
      "epoch": 0.044101433296582136,
      "grad_norm": 10.08563232421875,
      "learning_rate": 0.00011,
      "loss": 4.7925,
      "step": 10
    },
    {
      "epoch": 0.04851157662624035,
      "grad_norm": 7.417860507965088,
      "learning_rate": 0.0001099924817745858,
      "loss": 4.7195,
      "step": 11
    },
    {
      "epoch": 0.05292171995589857,
      "grad_norm": 8.151042938232422,
      "learning_rate": 0.00010996992915375093,
      "loss": 4.5419,
      "step": 12
    },
    {
      "epoch": 0.05733186328555678,
      "grad_norm": 5.868197917938232,
      "learning_rate": 0.00010993234830315676,
      "loss": 4.026,
      "step": 13
    },
    {
      "epoch": 0.06174200661521499,
      "grad_norm": 4.988924503326416,
      "learning_rate": 0.0001098797494970326,
      "loss": 3.8786,
      "step": 14
    },
    {
      "epoch": 0.06615214994487321,
      "grad_norm": 5.958721160888672,
      "learning_rate": 0.00010981214711536684,
      "loss": 3.9877,
      "step": 15
    },
    {
      "epoch": 0.07056229327453142,
      "grad_norm": 8.914958953857422,
      "learning_rate": 0.00010972955963997563,
      "loss": 3.5444,
      "step": 16
    },
    {
      "epoch": 0.07497243660418963,
      "grad_norm": 5.30311393737793,
      "learning_rate": 0.00010963200964945011,
      "loss": 3.5812,
      "step": 17
    },
    {
      "epoch": 0.07938257993384785,
      "grad_norm": 9.792562484741211,
      "learning_rate": 0.00010951952381298364,
      "loss": 3.6679,
      "step": 18
    },
    {
      "epoch": 0.08379272326350606,
      "grad_norm": 3.1872658729553223,
      "learning_rate": 0.00010939213288308077,
      "loss": 3.4618,
      "step": 19
    },
    {
      "epoch": 0.08820286659316427,
      "grad_norm": 4.939417362213135,
      "learning_rate": 0.00010924987168714973,
      "loss": 3.5312,
      "step": 20
    },
    {
      "epoch": 0.0926130099228225,
      "grad_norm": 4.3936543464660645,
      "learning_rate": 0.00010909277911798103,
      "loss": 3.3944,
      "step": 21
    },
    {
      "epoch": 0.0970231532524807,
      "grad_norm": 3.6638882160186768,
      "learning_rate": 0.00010892089812311451,
      "loss": 3.5882,
      "step": 22
    },
    {
      "epoch": 0.10143329658213891,
      "grad_norm": 3.5145962238311768,
      "learning_rate": 0.00010873427569309797,
      "loss": 3.4025,
      "step": 23
    },
    {
      "epoch": 0.10584343991179714,
      "grad_norm": 2.5855441093444824,
      "learning_rate": 0.00010853296284864032,
      "loss": 3.42,
      "step": 24
    },
    {
      "epoch": 0.11025358324145534,
      "grad_norm": 3.6106998920440674,
      "learning_rate": 0.00010831701462666318,
      "loss": 3.165,
      "step": 25
    },
    {
      "epoch": 0.11466372657111357,
      "grad_norm": 2.4775733947753906,
      "learning_rate": 0.00010808649006525419,
      "loss": 3.1228,
      "step": 26
    },
    {
      "epoch": 0.11907386990077178,
      "grad_norm": 3.008453607559204,
      "learning_rate": 0.00010784145218752665,
      "loss": 3.4477,
      "step": 27
    },
    {
      "epoch": 0.12348401323042998,
      "grad_norm": 2.6751112937927246,
      "learning_rate": 0.00010758196798438968,
      "loss": 3.1794,
      "step": 28
    },
    {
      "epoch": 0.1278941565600882,
      "grad_norm": 2.252016067504883,
      "learning_rate": 0.00010730810839623346,
      "loss": 3.2785,
      "step": 29
    },
    {
      "epoch": 0.13230429988974643,
      "grad_norm": 2.5002379417419434,
      "learning_rate": 0.0001070199482935349,
      "loss": 3.2805,
      "step": 30
    },
    {
      "epoch": 0.13671444321940462,
      "grad_norm": 2.217980146408081,
      "learning_rate": 0.00010671756645638888,
      "loss": 3.1141,
      "step": 31
    },
    {
      "epoch": 0.14112458654906285,
      "grad_norm": 2.0301949977874756,
      "learning_rate": 0.00010640104555297034,
      "loss": 3.2129,
      "step": 32
    },
    {
      "epoch": 0.14553472987872107,
      "grad_norm": 2.218367338180542,
      "learning_rate": 0.00010607047211693389,
      "loss": 3.4561,
      "step": 33
    },
    {
      "epoch": 0.14994487320837926,
      "grad_norm": 2.181100368499756,
      "learning_rate": 0.00010572593652375616,
      "loss": 3.0422,
      "step": 34
    },
    {
      "epoch": 0.1543550165380375,
      "grad_norm": 1.9729782342910767,
      "learning_rate": 0.00010536753296602816,
      "loss": 2.911,
      "step": 35
    },
    {
      "epoch": 0.1587651598676957,
      "grad_norm": 1.8118611574172974,
      "learning_rate": 0.00010499535942770394,
      "loss": 3.4308,
      "step": 36
    },
    {
      "epoch": 0.1631753031973539,
      "grad_norm": 1.7124440670013428,
      "learning_rate": 0.00010460951765731275,
      "loss": 3.3144,
      "step": 37
    },
    {
      "epoch": 0.16758544652701213,
      "grad_norm": 1.7353549003601074,
      "learning_rate": 0.000104210113140142,
      "loss": 2.8064,
      "step": 38
    },
    {
      "epoch": 0.17199558985667035,
      "grad_norm": 2.1845147609710693,
      "learning_rate": 0.00010379725506939865,
      "loss": 3.0968,
      "step": 39
    },
    {
      "epoch": 0.17640573318632854,
      "grad_norm": 3.1064226627349854,
      "learning_rate": 0.0001033710563163569,
      "loss": 1.8918,
      "step": 40
    },
    {
      "epoch": 0.18081587651598677,
      "grad_norm": 2.765388011932373,
      "learning_rate": 0.00010293163339950024,
      "loss": 1.9075,
      "step": 41
    },
    {
      "epoch": 0.185226019845645,
      "grad_norm": 5.030835151672363,
      "learning_rate": 0.00010247910645266658,
      "loss": 1.8417,
      "step": 42
    },
    {
      "epoch": 0.18963616317530319,
      "grad_norm": 3.176011323928833,
      "learning_rate": 0.00010201359919220464,
      "loss": 1.6077,
      "step": 43
    },
    {
      "epoch": 0.1940463065049614,
      "grad_norm": 2.074098587036133,
      "learning_rate": 0.00010153523888315144,
      "loss": 0.6585,
      "step": 44
    },
    {
      "epoch": 0.19845644983461963,
      "grad_norm": 2.3801369667053223,
      "learning_rate": 0.00010104415630443907,
      "loss": 0.8426,
      "step": 45
    },
    {
      "epoch": 0.20286659316427783,
      "grad_norm": 5.068062782287598,
      "learning_rate": 0.0001005404857131411,
      "loss": 0.4928,
      "step": 46
    },
    {
      "epoch": 0.20727673649393605,
      "grad_norm": 2.475529193878174,
      "learning_rate": 0.00010002436480776809,
      "loss": 0.5498,
      "step": 47
    },
    {
      "epoch": 0.21168687982359427,
      "grad_norm": 2.115375518798828,
      "learning_rate": 9.949593469062211e-05,
      "loss": 0.5205,
      "step": 48
    },
    {
      "epoch": 0.2160970231532525,
      "grad_norm": 2.468719005584717,
      "learning_rate": 9.895533982922087e-05,
      "loss": 0.4485,
      "step": 49
    },
    {
      "epoch": 0.2205071664829107,
      "grad_norm": 3.052607774734497,
      "learning_rate": 9.840272801680165e-05,
      "loss": 0.4655,
      "step": 50
    },
    {
      "epoch": 0.2205071664829107,
      "eval_loss": 0.7270308136940002,
      "eval_runtime": 112.569,
      "eval_samples_per_second": 3.393,
      "eval_steps_per_second": 0.853,
      "step": 50
    },
    {
      "epoch": 0.2249173098125689,
      "grad_norm": 6.1360578536987305,
      "learning_rate": 9.783825033191619e-05,
      "loss": 3.4272,
      "step": 51
    },
    {
      "epoch": 0.22932745314222713,
      "grad_norm": 3.558617353439331,
      "learning_rate": 9.726206109712725e-05,
      "loss": 3.4427,
      "step": 52
    },
    {
      "epoch": 0.23373759647188533,
      "grad_norm": 2.51824951171875,
      "learning_rate": 9.667431783681842e-05,
      "loss": 3.388,
      "step": 53
    },
    {
      "epoch": 0.23814773980154355,
      "grad_norm": 2.5455639362335205,
      "learning_rate": 9.607518123412847e-05,
      "loss": 3.4787,
      "step": 54
    },
    {
      "epoch": 0.24255788313120177,
      "grad_norm": 2.201443672180176,
      "learning_rate": 9.546481508702224e-05,
      "loss": 3.4015,
      "step": 55
    },
    {
      "epoch": 0.24696802646085997,
      "grad_norm": 1.9890426397323608,
      "learning_rate": 9.48433862635099e-05,
      "loss": 3.1958,
      "step": 56
    },
    {
      "epoch": 0.2513781697905182,
      "grad_norm": 1.727371096611023,
      "learning_rate": 9.421106465602684e-05,
      "loss": 2.8546,
      "step": 57
    },
    {
      "epoch": 0.2557883131201764,
      "grad_norm": 1.931094765663147,
      "learning_rate": 9.356802313498687e-05,
      "loss": 3.0559,
      "step": 58
    },
    {
      "epoch": 0.26019845644983464,
      "grad_norm": 2.186201333999634,
      "learning_rate": 9.291443750152112e-05,
      "loss": 3.0093,
      "step": 59
    },
    {
      "epoch": 0.26460859977949286,
      "grad_norm": 1.641863226890564,
      "learning_rate": 9.225048643941577e-05,
      "loss": 3.1301,
      "step": 60
    },
    {
      "epoch": 0.269018743109151,
      "grad_norm": 2.0277044773101807,
      "learning_rate": 9.157635146626164e-05,
      "loss": 2.9672,
      "step": 61
    },
    {
      "epoch": 0.27342888643880925,
      "grad_norm": 1.6549147367477417,
      "learning_rate": 9.089221688382928e-05,
      "loss": 3.0468,
      "step": 62
    },
    {
      "epoch": 0.27783902976846747,
      "grad_norm": 1.6773267984390259,
      "learning_rate": 9.019826972768242e-05,
      "loss": 3.0721,
      "step": 63
    },
    {
      "epoch": 0.2822491730981257,
      "grad_norm": 2.053152561187744,
      "learning_rate": 8.949469971604454e-05,
      "loss": 3.2846,
      "step": 64
    },
    {
      "epoch": 0.2866593164277839,
      "grad_norm": 1.6968820095062256,
      "learning_rate": 8.878169919793173e-05,
      "loss": 3.0149,
      "step": 65
    },
    {
      "epoch": 0.29106945975744214,
      "grad_norm": 2.266279697418213,
      "learning_rate": 8.805946310056638e-05,
      "loss": 3.167,
      "step": 66
    },
    {
      "epoch": 0.2954796030871003,
      "grad_norm": 1.6353049278259277,
      "learning_rate": 8.732818887608602e-05,
      "loss": 3.1197,
      "step": 67
    },
    {
      "epoch": 0.29988974641675853,
      "grad_norm": 1.604249119758606,
      "learning_rate": 8.65880764475619e-05,
      "loss": 2.6482,
      "step": 68
    },
    {
      "epoch": 0.30429988974641675,
      "grad_norm": 1.6863490343093872,
      "learning_rate": 8.583932815434201e-05,
      "loss": 3.0375,
      "step": 69
    },
    {
      "epoch": 0.308710033076075,
      "grad_norm": 1.7916977405548096,
      "learning_rate": 8.50821486967335e-05,
      "loss": 2.8989,
      "step": 70
    },
    {
      "epoch": 0.3131201764057332,
      "grad_norm": 1.5783381462097168,
      "learning_rate": 8.431674508003966e-05,
      "loss": 3.0388,
      "step": 71
    },
    {
      "epoch": 0.3175303197353914,
      "grad_norm": 1.6949927806854248,
      "learning_rate": 8.354332655796683e-05,
      "loss": 2.8297,
      "step": 72
    },
    {
      "epoch": 0.3219404630650496,
      "grad_norm": 1.6010867357254028,
      "learning_rate": 8.276210457541642e-05,
      "loss": 3.0765,
      "step": 73
    },
    {
      "epoch": 0.3263506063947078,
      "grad_norm": 1.8386789560317993,
      "learning_rate": 8.197329271067796e-05,
      "loss": 3.2591,
      "step": 74
    },
    {
      "epoch": 0.33076074972436603,
      "grad_norm": 1.5259605646133423,
      "learning_rate": 8.117710661703905e-05,
      "loss": 2.8493,
      "step": 75
    },
    {
      "epoch": 0.33517089305402425,
      "grad_norm": 1.7161970138549805,
      "learning_rate": 8.037376396382784e-05,
      "loss": 2.8058,
      "step": 76
    },
    {
      "epoch": 0.3395810363836825,
      "grad_norm": 1.714835524559021,
      "learning_rate": 7.956348437690437e-05,
      "loss": 3.2383,
      "step": 77
    },
    {
      "epoch": 0.3439911797133407,
      "grad_norm": 1.6482939720153809,
      "learning_rate": 7.87464893786171e-05,
      "loss": 3.2056,
      "step": 78
    },
    {
      "epoch": 0.3484013230429989,
      "grad_norm": 1.6448861360549927,
      "learning_rate": 7.792300232724097e-05,
      "loss": 2.8896,
      "step": 79
    },
    {
      "epoch": 0.3528114663726571,
      "grad_norm": 1.551682949066162,
      "learning_rate": 7.709324835591332e-05,
      "loss": 3.1397,
      "step": 80
    },
    {
      "epoch": 0.3572216097023153,
      "grad_norm": 1.4730720520019531,
      "learning_rate": 7.625745431108487e-05,
      "loss": 2.8795,
      "step": 81
    },
    {
      "epoch": 0.36163175303197354,
      "grad_norm": 1.538916826248169,
      "learning_rate": 7.541584869050213e-05,
      "loss": 3.0389,
      "step": 82
    },
    {
      "epoch": 0.36604189636163176,
      "grad_norm": 2.091306447982788,
      "learning_rate": 7.456866158073842e-05,
      "loss": 3.3668,
      "step": 83
    },
    {
      "epoch": 0.37045203969129,
      "grad_norm": 1.774909496307373,
      "learning_rate": 7.371612459429037e-05,
      "loss": 2.8627,
      "step": 84
    },
    {
      "epoch": 0.3748621830209482,
      "grad_norm": 1.583600401878357,
      "learning_rate": 7.28584708062576e-05,
      "loss": 2.9084,
      "step": 85
    },
    {
      "epoch": 0.37927232635060637,
      "grad_norm": 1.38632071018219,
      "learning_rate": 7.19959346906221e-05,
      "loss": 3.1065,
      "step": 86
    },
    {
      "epoch": 0.3836824696802646,
      "grad_norm": 1.57198965549469,
      "learning_rate": 7.112875205614558e-05,
      "loss": 3.3521,
      "step": 87
    },
    {
      "epoch": 0.3880926130099228,
      "grad_norm": 1.7089447975158691,
      "learning_rate": 7.025715998190145e-05,
      "loss": 2.5392,
      "step": 88
    },
    {
      "epoch": 0.39250275633958104,
      "grad_norm": 1.6826099157333374,
      "learning_rate": 6.938139675246009e-05,
      "loss": 1.5587,
      "step": 89
    },
    {
      "epoch": 0.39691289966923926,
      "grad_norm": 1.8765373229980469,
      "learning_rate": 6.850170179274395e-05,
      "loss": 1.6571,
      "step": 90
    },
    {
      "epoch": 0.4013230429988975,
      "grad_norm": 2.031247854232788,
      "learning_rate": 6.761831560257134e-05,
      "loss": 1.2091,
      "step": 91
    },
    {
      "epoch": 0.40573318632855565,
      "grad_norm": 1.892391562461853,
      "learning_rate": 6.673147969090608e-05,
      "loss": 0.8195,
      "step": 92
    },
    {
      "epoch": 0.4101433296582139,
      "grad_norm": 1.6355128288269043,
      "learning_rate": 6.584143650983141e-05,
      "loss": 0.4771,
      "step": 93
    },
    {
      "epoch": 0.4145534729878721,
      "grad_norm": 2.7239136695861816,
      "learning_rate": 6.494842938826605e-05,
      "loss": 0.4328,
      "step": 94
    },
    {
      "epoch": 0.4189636163175303,
      "grad_norm": 2.0957109928131104,
      "learning_rate": 6.405270246544037e-05,
      "loss": 0.3941,
      "step": 95
    },
    {
      "epoch": 0.42337375964718854,
      "grad_norm": 1.5724101066589355,
      "learning_rate": 6.31545006241513e-05,
      "loss": 0.388,
      "step": 96
    },
    {
      "epoch": 0.42778390297684676,
      "grad_norm": 2.5362913608551025,
      "learning_rate": 6.22540694238138e-05,
      "loss": 0.3823,
      "step": 97
    },
    {
      "epoch": 0.432194046306505,
      "grad_norm": 2.424511194229126,
      "learning_rate": 6.135165503332725e-05,
      "loss": 0.3324,
      "step": 98
    },
    {
      "epoch": 0.43660418963616315,
      "grad_norm": 1.4182337522506714,
      "learning_rate": 6.0447504163775465e-05,
      "loss": 0.2856,
      "step": 99
    },
    {
      "epoch": 0.4410143329658214,
      "grad_norm": 3.211148977279663,
      "learning_rate": 5.954186400097829e-05,
      "loss": 0.3445,
      "step": 100
    },
    {
      "epoch": 0.4410143329658214,
      "eval_loss": 0.6658266186714172,
      "eval_runtime": 107.4688,
      "eval_samples_per_second": 3.555,
      "eval_steps_per_second": 0.893,
      "step": 100
    },
    {
      "epoch": 0.4454244762954796,
      "grad_norm": 4.080659866333008,
      "learning_rate": 5.8634982137913465e-05,
      "loss": 3.2606,
      "step": 101
    },
    {
      "epoch": 0.4498346196251378,
      "grad_norm": 2.8597896099090576,
      "learning_rate": 5.772710650702723e-05,
      "loss": 3.2198,
      "step": 102
    },
    {
      "epoch": 0.45424476295479604,
      "grad_norm": 1.9857066869735718,
      "learning_rate": 5.681848531245195e-05,
      "loss": 3.1464,
      "step": 103
    },
    {
      "epoch": 0.45865490628445427,
      "grad_norm": 1.819825291633606,
      "learning_rate": 5.590936696214972e-05,
      "loss": 3.3014,
      "step": 104
    },
    {
      "epoch": 0.46306504961411243,
      "grad_norm": 1.7104649543762207,
      "learning_rate": 5.5e-05,
      "loss": 3.1921,
      "step": 105
    },
    {
      "epoch": 0.46747519294377066,
      "grad_norm": 1.5252690315246582,
      "learning_rate": 5.409063303785029e-05,
      "loss": 3.0135,
      "step": 106
    },
    {
      "epoch": 0.4718853362734289,
      "grad_norm": 1.5344245433807373,
      "learning_rate": 5.318151468754805e-05,
      "loss": 2.9589,
      "step": 107
    },
    {
      "epoch": 0.4762954796030871,
      "grad_norm": 1.3514833450317383,
      "learning_rate": 5.227289349297277e-05,
      "loss": 2.898,
      "step": 108
    },
    {
      "epoch": 0.4807056229327453,
      "grad_norm": 1.3298234939575195,
      "learning_rate": 5.136501786208654e-05,
      "loss": 2.9296,
      "step": 109
    },
    {
      "epoch": 0.48511576626240355,
      "grad_norm": 1.4843939542770386,
      "learning_rate": 5.045813599902173e-05,
      "loss": 3.2311,
      "step": 110
    },
    {
      "epoch": 0.4895259095920617,
      "grad_norm": 1.3323349952697754,
      "learning_rate": 4.955249583622455e-05,
      "loss": 2.8828,
      "step": 111
    },
    {
      "epoch": 0.49393605292171994,
      "grad_norm": 1.3595972061157227,
      "learning_rate": 4.8648344966672767e-05,
      "loss": 2.9486,
      "step": 112
    },
    {
      "epoch": 0.49834619625137816,
      "grad_norm": 1.3509643077850342,
      "learning_rate": 4.774593057618621e-05,
      "loss": 3.2162,
      "step": 113
    },
    {
      "epoch": 0.5027563395810364,
      "grad_norm": 1.3561177253723145,
      "learning_rate": 4.6845499375848686e-05,
      "loss": 2.854,
      "step": 114
    },
    {
      "epoch": 0.5071664829106945,
      "grad_norm": 1.4786139726638794,
      "learning_rate": 4.5947297534559625e-05,
      "loss": 3.1804,
      "step": 115
    },
    {
      "epoch": 0.5115766262403528,
      "grad_norm": 1.3076729774475098,
      "learning_rate": 4.5051570611733976e-05,
      "loss": 3.134,
      "step": 116
    },
    {
      "epoch": 0.515986769570011,
      "grad_norm": 1.4321703910827637,
      "learning_rate": 4.415856349016859e-05,
      "loss": 2.891,
      "step": 117
    },
    {
      "epoch": 0.5203969128996693,
      "grad_norm": 1.3954541683197021,
      "learning_rate": 4.326852030909393e-05,
      "loss": 2.9936,
      "step": 118
    },
    {
      "epoch": 0.5248070562293274,
      "grad_norm": 1.3153645992279053,
      "learning_rate": 4.238168439742867e-05,
      "loss": 3.1401,
      "step": 119
    },
    {
      "epoch": 0.5292171995589857,
      "grad_norm": 1.3978232145309448,
      "learning_rate": 4.149829820725605e-05,
      "loss": 3.2422,
      "step": 120
    },
    {
      "epoch": 0.5336273428886439,
      "grad_norm": 1.3654237985610962,
      "learning_rate": 4.0618603247539916e-05,
      "loss": 2.9804,
      "step": 121
    },
    {
      "epoch": 0.538037486218302,
      "grad_norm": 1.6010512113571167,
      "learning_rate": 3.9742840018098564e-05,
      "loss": 2.8879,
      "step": 122
    },
    {
      "epoch": 0.5424476295479603,
      "grad_norm": 1.2477831840515137,
      "learning_rate": 3.887124794385445e-05,
      "loss": 2.6711,
      "step": 123
    },
    {
      "epoch": 0.5468577728776185,
      "grad_norm": 1.3943934440612793,
      "learning_rate": 3.80040653093779e-05,
      "loss": 3.0408,
      "step": 124
    },
    {
      "epoch": 0.5512679162072768,
      "grad_norm": 1.632759690284729,
      "learning_rate": 3.714152919374241e-05,
      "loss": 3.076,
      "step": 125
    },
    {
      "epoch": 0.5556780595369349,
      "grad_norm": 1.3918051719665527,
      "learning_rate": 3.628387540570963e-05,
      "loss": 3.0342,
      "step": 126
    },
    {
      "epoch": 0.5600882028665931,
      "grad_norm": 1.2972222566604614,
      "learning_rate": 3.543133841926159e-05,
      "loss": 2.82,
      "step": 127
    },
    {
      "epoch": 0.5644983461962514,
      "grad_norm": 1.202620267868042,
      "learning_rate": 3.458415130949785e-05,
      "loss": 2.4912,
      "step": 128
    },
    {
      "epoch": 0.5689084895259096,
      "grad_norm": 1.3449923992156982,
      "learning_rate": 3.374254568891514e-05,
      "loss": 3.3369,
      "step": 129
    },
    {
      "epoch": 0.5733186328555678,
      "grad_norm": 1.3851882219314575,
      "learning_rate": 3.290675164408669e-05,
      "loss": 2.565,
      "step": 130
    },
    {
      "epoch": 0.577728776185226,
      "grad_norm": 1.2937684059143066,
      "learning_rate": 3.207699767275904e-05,
      "loss": 2.7112,
      "step": 131
    },
    {
      "epoch": 0.5821389195148843,
      "grad_norm": 1.4037163257598877,
      "learning_rate": 3.12535106213829e-05,
      "loss": 3.11,
      "step": 132
    },
    {
      "epoch": 0.5865490628445424,
      "grad_norm": 1.3117399215698242,
      "learning_rate": 3.0436515623095647e-05,
      "loss": 2.6965,
      "step": 133
    },
    {
      "epoch": 0.5909592061742006,
      "grad_norm": 1.3175640106201172,
      "learning_rate": 2.962623603617218e-05,
      "loss": 2.7956,
      "step": 134
    },
    {
      "epoch": 0.5953693495038589,
      "grad_norm": 1.3013272285461426,
      "learning_rate": 2.8822893382960955e-05,
      "loss": 2.6119,
      "step": 135
    },
    {
      "epoch": 0.5997794928335171,
      "grad_norm": 1.3417240381240845,
      "learning_rate": 2.802670728932207e-05,
      "loss": 2.7399,
      "step": 136
    },
    {
      "epoch": 0.6041896361631753,
      "grad_norm": 1.4003099203109741,
      "learning_rate": 2.723789542458361e-05,
      "loss": 3.2267,
      "step": 137
    },
    {
      "epoch": 0.6085997794928335,
      "grad_norm": 1.3760734796524048,
      "learning_rate": 2.6456673442033183e-05,
      "loss": 3.064,
      "step": 138
    },
    {
      "epoch": 0.6130099228224918,
      "grad_norm": 1.4111647605895996,
      "learning_rate": 2.5683254919960356e-05,
      "loss": 2.8482,
      "step": 139
    },
    {
      "epoch": 0.61742006615215,
      "grad_norm": 1.7103841304779053,
      "learning_rate": 2.4917851303266533e-05,
      "loss": 1.9783,
      "step": 140
    },
    {
      "epoch": 0.6218302094818081,
      "grad_norm": 1.6013977527618408,
      "learning_rate": 2.4160671845658007e-05,
      "loss": 1.8013,
      "step": 141
    },
    {
      "epoch": 0.6262403528114664,
      "grad_norm": 1.2906684875488281,
      "learning_rate": 2.3411923552438105e-05,
      "loss": 1.0097,
      "step": 142
    },
    {
      "epoch": 0.6306504961411246,
      "grad_norm": 1.3760795593261719,
      "learning_rate": 2.2671811123913983e-05,
      "loss": 1.2116,
      "step": 143
    },
    {
      "epoch": 0.6350606394707828,
      "grad_norm": 1.368346095085144,
      "learning_rate": 2.194053689943362e-05,
      "loss": 0.7298,
      "step": 144
    },
    {
      "epoch": 0.639470782800441,
      "grad_norm": 1.2973005771636963,
      "learning_rate": 2.121830080206827e-05,
      "loss": 0.3624,
      "step": 145
    },
    {
      "epoch": 0.6438809261300992,
      "grad_norm": 1.3431459665298462,
      "learning_rate": 2.0505300283955464e-05,
      "loss": 0.3439,
      "step": 146
    },
    {
      "epoch": 0.6482910694597575,
      "grad_norm": 2.1608824729919434,
      "learning_rate": 1.9801730272317585e-05,
      "loss": 0.3482,
      "step": 147
    },
    {
      "epoch": 0.6527012127894156,
      "grad_norm": 1.9314510822296143,
      "learning_rate": 1.910778311617072e-05,
      "loss": 0.3296,
      "step": 148
    },
    {
      "epoch": 0.6571113561190739,
      "grad_norm": 1.8207006454467773,
      "learning_rate": 1.8423648533738342e-05,
      "loss": 0.3223,
      "step": 149
    },
    {
      "epoch": 0.6615214994487321,
      "grad_norm": 2.0629324913024902,
      "learning_rate": 1.7749513560584252e-05,
      "loss": 0.2425,
      "step": 150
    },
    {
      "epoch": 0.6615214994487321,
      "eval_loss": 0.6403191685676575,
      "eval_runtime": 116.2907,
      "eval_samples_per_second": 3.285,
      "eval_steps_per_second": 0.826,
      "step": 150
    },
    {
      "epoch": 0.6659316427783903,
      "grad_norm": 3.8753366470336914,
      "learning_rate": 1.7085562498478883e-05,
      "loss": 3.1445,
      "step": 151
    },
    {
      "epoch": 0.6703417861080485,
      "grad_norm": 3.695657730102539,
      "learning_rate": 1.6431976865013128e-05,
      "loss": 3.1754,
      "step": 152
    },
    {
      "epoch": 0.6747519294377067,
      "grad_norm": 2.9441332817077637,
      "learning_rate": 1.5788935343973164e-05,
      "loss": 3.2525,
      "step": 153
    },
    {
      "epoch": 0.679162072767365,
      "grad_norm": 2.0445456504821777,
      "learning_rate": 1.5156613736490108e-05,
      "loss": 3.278,
      "step": 154
    },
    {
      "epoch": 0.6835722160970231,
      "grad_norm": 1.5744311809539795,
      "learning_rate": 1.4535184912977763e-05,
      "loss": 3.1633,
      "step": 155
    },
    {
      "epoch": 0.6879823594266814,
      "grad_norm": 1.7349194288253784,
      "learning_rate": 1.3924818765871553e-05,
      "loss": 2.5782,
      "step": 156
    },
    {
      "epoch": 0.6923925027563396,
      "grad_norm": 1.9107156991958618,
      "learning_rate": 1.3325682163181601e-05,
      "loss": 2.716,
      "step": 157
    },
    {
      "epoch": 0.6968026460859978,
      "grad_norm": 1.9235740900039673,
      "learning_rate": 1.2737938902872767e-05,
      "loss": 2.7254,
      "step": 158
    },
    {
      "epoch": 0.701212789415656,
      "grad_norm": 1.995104193687439,
      "learning_rate": 1.2161749668083823e-05,
      "loss": 3.1104,
      "step": 159
    },
    {
      "epoch": 0.7056229327453142,
      "grad_norm": 1.724377155303955,
      "learning_rate": 1.159727198319836e-05,
      "loss": 2.9403,
      "step": 160
    },
    {
      "epoch": 0.7100330760749725,
      "grad_norm": 1.5613574981689453,
      "learning_rate": 1.1044660170779142e-05,
      "loss": 3.037,
      "step": 161
    },
    {
      "epoch": 0.7144432194046306,
      "grad_norm": 1.3397530317306519,
      "learning_rate": 1.0504065309377897e-05,
      "loss": 3.0121,
      "step": 162
    },
    {
      "epoch": 0.7188533627342889,
      "grad_norm": 1.4853931665420532,
      "learning_rate": 9.97563519223192e-06,
      "loss": 2.9336,
      "step": 163
    },
    {
      "epoch": 0.7232635060639471,
      "grad_norm": 1.5100167989730835,
      "learning_rate": 9.459514286858898e-06,
      "loss": 2.8047,
      "step": 164
    },
    {
      "epoch": 0.7276736493936052,
      "grad_norm": 1.5139286518096924,
      "learning_rate": 8.95584369556093e-06,
      "loss": 3.251,
      "step": 165
    },
    {
      "epoch": 0.7320837927232635,
      "grad_norm": 1.4658764600753784,
      "learning_rate": 8.464761116848546e-06,
      "loss": 2.9285,
      "step": 166
    },
    {
      "epoch": 0.7364939360529217,
      "grad_norm": 1.382473111152649,
      "learning_rate": 7.986400807795349e-06,
      "loss": 2.9684,
      "step": 167
    },
    {
      "epoch": 0.74090407938258,
      "grad_norm": 1.3517184257507324,
      "learning_rate": 7.520893547333436e-06,
      "loss": 2.8531,
      "step": 168
    },
    {
      "epoch": 0.7453142227122381,
      "grad_norm": 1.2622511386871338,
      "learning_rate": 7.068366600499744e-06,
      "loss": 2.8949,
      "step": 169
    },
    {
      "epoch": 0.7497243660418964,
      "grad_norm": 1.2846345901489258,
      "learning_rate": 6.6289436836431076e-06,
      "loss": 2.7951,
      "step": 170
    },
    {
      "epoch": 0.7541345093715546,
      "grad_norm": 1.3046170473098755,
      "learning_rate": 6.20274493060135e-06,
      "loss": 3.0261,
      "step": 171
    },
    {
      "epoch": 0.7585446527012127,
      "grad_norm": 1.3198810815811157,
      "learning_rate": 5.789886859858009e-06,
      "loss": 3.0008,
      "step": 172
    },
    {
      "epoch": 0.762954796030871,
      "grad_norm": 1.3677592277526855,
      "learning_rate": 5.3904823426872605e-06,
      "loss": 2.9813,
      "step": 173
    },
    {
      "epoch": 0.7673649393605292,
      "grad_norm": 1.3680708408355713,
      "learning_rate": 5.004640572296062e-06,
      "loss": 2.9926,
      "step": 174
    },
    {
      "epoch": 0.7717750826901875,
      "grad_norm": 1.3535288572311401,
      "learning_rate": 4.632467033971838e-06,
      "loss": 3.0709,
      "step": 175
    },
    {
      "epoch": 0.7761852260198456,
      "grad_norm": 1.2958035469055176,
      "learning_rate": 4.274063476243839e-06,
      "loss": 3.0148,
      "step": 176
    },
    {
      "epoch": 0.7805953693495039,
      "grad_norm": 1.2792253494262695,
      "learning_rate": 3.929527883066117e-06,
      "loss": 2.8443,
      "step": 177
    },
    {
      "epoch": 0.7850055126791621,
      "grad_norm": 1.2686259746551514,
      "learning_rate": 3.5989544470296595e-06,
      "loss": 3.2438,
      "step": 178
    },
    {
      "epoch": 0.7894156560088202,
      "grad_norm": 1.3277795314788818,
      "learning_rate": 3.282433543611136e-06,
      "loss": 2.958,
      "step": 179
    },
    {
      "epoch": 0.7938257993384785,
      "grad_norm": 1.3202295303344727,
      "learning_rate": 2.980051706465095e-06,
      "loss": 2.8722,
      "step": 180
    },
    {
      "epoch": 0.7982359426681367,
      "grad_norm": 1.3094717264175415,
      "learning_rate": 2.691891603766556e-06,
      "loss": 2.6468,
      "step": 181
    },
    {
      "epoch": 0.802646085997795,
      "grad_norm": 1.2734664678573608,
      "learning_rate": 2.4180320156103298e-06,
      "loss": 3.0554,
      "step": 182
    },
    {
      "epoch": 0.8070562293274531,
      "grad_norm": 1.2695180177688599,
      "learning_rate": 2.158547812473352e-06,
      "loss": 3.4244,
      "step": 183
    },
    {
      "epoch": 0.8114663726571113,
      "grad_norm": 1.3620620965957642,
      "learning_rate": 1.9135099347458293e-06,
      "loss": 3.1298,
      "step": 184
    },
    {
      "epoch": 0.8158765159867696,
      "grad_norm": 1.1754707098007202,
      "learning_rate": 1.6829853733368294e-06,
      "loss": 2.9018,
      "step": 185
    },
    {
      "epoch": 0.8202866593164277,
      "grad_norm": 1.435257077217102,
      "learning_rate": 1.4670371513596842e-06,
      "loss": 2.7138,
      "step": 186
    },
    {
      "epoch": 0.824696802646086,
      "grad_norm": 1.3883874416351318,
      "learning_rate": 1.2657243069020402e-06,
      "loss": 3.158,
      "step": 187
    },
    {
      "epoch": 0.8291069459757442,
      "grad_norm": 1.3034346103668213,
      "learning_rate": 1.0791018768854855e-06,
      "loss": 3.162,
      "step": 188
    },
    {
      "epoch": 0.8335170893054025,
      "grad_norm": 1.2247211933135986,
      "learning_rate": 9.072208820189698e-07,
      "loss": 2.9649,
      "step": 189
    },
    {
      "epoch": 0.8379272326350606,
      "grad_norm": 1.425723671913147,
      "learning_rate": 7.501283128502722e-07,
      "loss": 1.9743,
      "step": 190
    },
    {
      "epoch": 0.8423373759647188,
      "grad_norm": 1.296178936958313,
      "learning_rate": 6.07867116919233e-07,
      "loss": 1.2535,
      "step": 191
    },
    {
      "epoch": 0.8467475192943771,
      "grad_norm": 1.4856878519058228,
      "learning_rate": 4.804761870163643e-07,
      "loss": 1.0276,
      "step": 192
    },
    {
      "epoch": 0.8511576626240352,
      "grad_norm": 1.459718942642212,
      "learning_rate": 3.6799035054990215e-07,
      "loss": 1.0288,
      "step": 193
    },
    {
      "epoch": 0.8555678059536935,
      "grad_norm": 1.4478521347045898,
      "learning_rate": 2.704403600243721e-07,
      "loss": 0.492,
      "step": 194
    },
    {
      "epoch": 0.8599779492833517,
      "grad_norm": 1.547826886177063,
      "learning_rate": 1.878528846331584e-07,
      "loss": 0.5052,
      "step": 195
    },
    {
      "epoch": 0.86438809261301,
      "grad_norm": 1.6399399042129517,
      "learning_rate": 1.202505029674006e-07,
      "loss": 0.3783,
      "step": 196
    },
    {
      "epoch": 0.8687982359426681,
      "grad_norm": 1.6422384977340698,
      "learning_rate": 6.765169684323947e-08,
      "loss": 0.3455,
      "step": 197
    },
    {
      "epoch": 0.8732083792723263,
      "grad_norm": 1.7368685007095337,
      "learning_rate": 3.007084624906731e-08,
      "loss": 0.3094,
      "step": 198
    },
    {
      "epoch": 0.8776185226019846,
      "grad_norm": 2.1973204612731934,
      "learning_rate": 7.518225414204771e-09,
      "loss": 0.3237,
      "step": 199
    },
    {
      "epoch": 0.8820286659316428,
      "grad_norm": 2.7537074089050293,
      "learning_rate": 0.0,
      "loss": 0.3645,
      "step": 200
    },
    {
      "epoch": 0.8820286659316428,
      "eval_loss": 0.6017911434173584,
      "eval_runtime": 127.4467,
      "eval_samples_per_second": 2.997,
      "eval_steps_per_second": 0.753,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.573740381675192e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}