{
  "best_metric": 0.33371686935424805,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.32232070910556004,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016116035455278,
      "grad_norm": 0.3171309530735016,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 1.0204,
      "step": 1
    },
    {
      "epoch": 0.0016116035455278,
      "eval_loss": 1.5084112882614136,
      "eval_runtime": 135.8281,
      "eval_samples_per_second": 61.534,
      "eval_steps_per_second": 1.929,
      "step": 1
    },
    {
      "epoch": 0.0032232070910556,
      "grad_norm": 0.4342203438282013,
      "learning_rate": 2.0200000000000003e-05,
      "loss": 1.1485,
      "step": 2
    },
    {
      "epoch": 0.004834810636583401,
      "grad_norm": 0.5146673321723938,
      "learning_rate": 3.0299999999999998e-05,
      "loss": 1.1971,
      "step": 3
    },
    {
      "epoch": 0.0064464141821112,
      "grad_norm": 0.7256948947906494,
      "learning_rate": 4.0400000000000006e-05,
      "loss": 1.4328,
      "step": 4
    },
    {
      "epoch": 0.008058017727639,
      "grad_norm": 1.1304517984390259,
      "learning_rate": 5.05e-05,
      "loss": 1.596,
      "step": 5
    },
    {
      "epoch": 0.009669621273166801,
      "grad_norm": 1.6577770709991455,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 1.845,
      "step": 6
    },
    {
      "epoch": 0.011281224818694601,
      "grad_norm": 0.36109769344329834,
      "learning_rate": 7.07e-05,
      "loss": 0.9896,
      "step": 7
    },
    {
      "epoch": 0.0128928283642224,
      "grad_norm": 0.34879496693611145,
      "learning_rate": 8.080000000000001e-05,
      "loss": 0.9208,
      "step": 8
    },
    {
      "epoch": 0.014504431909750202,
      "grad_norm": 0.4985645115375519,
      "learning_rate": 9.09e-05,
      "loss": 0.895,
      "step": 9
    },
    {
      "epoch": 0.016116035455278,
      "grad_norm": 0.45808470249176025,
      "learning_rate": 0.000101,
      "loss": 0.8154,
      "step": 10
    },
    {
      "epoch": 0.017727639000805803,
      "grad_norm": 0.3932338058948517,
      "learning_rate": 0.00010046842105263158,
      "loss": 0.8742,
      "step": 11
    },
    {
      "epoch": 0.019339242546333603,
      "grad_norm": 0.45358890295028687,
      "learning_rate": 9.993684210526315e-05,
      "loss": 0.9433,
      "step": 12
    },
    {
      "epoch": 0.020950846091861403,
      "grad_norm": 0.29831695556640625,
      "learning_rate": 9.940526315789473e-05,
      "loss": 0.8187,
      "step": 13
    },
    {
      "epoch": 0.022562449637389202,
      "grad_norm": 0.22101442515850067,
      "learning_rate": 9.887368421052632e-05,
      "loss": 0.8009,
      "step": 14
    },
    {
      "epoch": 0.024174053182917002,
      "grad_norm": 0.27133405208587646,
      "learning_rate": 9.83421052631579e-05,
      "loss": 0.8114,
      "step": 15
    },
    {
      "epoch": 0.0257856567284448,
      "grad_norm": 0.23551404476165771,
      "learning_rate": 9.781052631578948e-05,
      "loss": 0.7651,
      "step": 16
    },
    {
      "epoch": 0.0273972602739726,
      "grad_norm": 0.3073457181453705,
      "learning_rate": 9.727894736842106e-05,
      "loss": 0.8419,
      "step": 17
    },
    {
      "epoch": 0.029008863819500404,
      "grad_norm": 0.3708623945713043,
      "learning_rate": 9.674736842105263e-05,
      "loss": 0.8201,
      "step": 18
    },
    {
      "epoch": 0.030620467365028204,
      "grad_norm": 0.22654138505458832,
      "learning_rate": 9.621578947368421e-05,
      "loss": 0.7523,
      "step": 19
    },
    {
      "epoch": 0.032232070910556,
      "grad_norm": 0.15454810857772827,
      "learning_rate": 9.568421052631578e-05,
      "loss": 0.7827,
      "step": 20
    },
    {
      "epoch": 0.03384367445608381,
      "grad_norm": 0.17125999927520752,
      "learning_rate": 9.515263157894737e-05,
      "loss": 0.7706,
      "step": 21
    },
    {
      "epoch": 0.035455278001611606,
      "grad_norm": 0.19242525100708008,
      "learning_rate": 9.462105263157895e-05,
      "loss": 0.7639,
      "step": 22
    },
    {
      "epoch": 0.037066881547139406,
      "grad_norm": 0.24774906039237976,
      "learning_rate": 9.408947368421054e-05,
      "loss": 0.7952,
      "step": 23
    },
    {
      "epoch": 0.038678485092667206,
      "grad_norm": 0.25956159830093384,
      "learning_rate": 9.355789473684211e-05,
      "loss": 0.7721,
      "step": 24
    },
    {
      "epoch": 0.040290088638195005,
      "grad_norm": 0.4470839202404022,
      "learning_rate": 9.302631578947369e-05,
      "loss": 0.7735,
      "step": 25
    },
    {
      "epoch": 0.041901692183722805,
      "grad_norm": 0.12824320793151855,
      "learning_rate": 9.249473684210526e-05,
      "loss": 0.7376,
      "step": 26
    },
    {
      "epoch": 0.043513295729250605,
      "grad_norm": 0.17759135365486145,
      "learning_rate": 9.196315789473685e-05,
      "loss": 0.7544,
      "step": 27
    },
    {
      "epoch": 0.045124899274778404,
      "grad_norm": 0.2012881338596344,
      "learning_rate": 9.143157894736843e-05,
      "loss": 0.7582,
      "step": 28
    },
    {
      "epoch": 0.046736502820306204,
      "grad_norm": 0.22395461797714233,
      "learning_rate": 9.09e-05,
      "loss": 0.764,
      "step": 29
    },
    {
      "epoch": 0.048348106365834004,
      "grad_norm": 0.3060719072818756,
      "learning_rate": 9.036842105263158e-05,
      "loss": 0.7868,
      "step": 30
    },
    {
      "epoch": 0.0499597099113618,
      "grad_norm": 0.327498197555542,
      "learning_rate": 8.983684210526316e-05,
      "loss": 0.7615,
      "step": 31
    },
    {
      "epoch": 0.0515713134568896,
      "grad_norm": 0.20455282926559448,
      "learning_rate": 8.930526315789474e-05,
      "loss": 0.7787,
      "step": 32
    },
    {
      "epoch": 0.0531829170024174,
      "grad_norm": 0.16608926653862,
      "learning_rate": 8.877368421052632e-05,
      "loss": 0.7066,
      "step": 33
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 0.18552589416503906,
      "learning_rate": 8.82421052631579e-05,
      "loss": 0.7656,
      "step": 34
    },
    {
      "epoch": 0.05640612409347301,
      "grad_norm": 0.20650997757911682,
      "learning_rate": 8.771052631578948e-05,
      "loss": 0.6993,
      "step": 35
    },
    {
      "epoch": 0.05801772763900081,
      "grad_norm": 0.2687518298625946,
      "learning_rate": 8.717894736842105e-05,
      "loss": 0.6904,
      "step": 36
    },
    {
      "epoch": 0.05962933118452861,
      "grad_norm": 0.34716078639030457,
      "learning_rate": 8.664736842105263e-05,
      "loss": 0.7152,
      "step": 37
    },
    {
      "epoch": 0.06124093473005641,
      "grad_norm": 0.19605892896652222,
      "learning_rate": 8.61157894736842e-05,
      "loss": 0.6811,
      "step": 38
    },
    {
      "epoch": 0.06285253827558421,
      "grad_norm": 0.1753331869840622,
      "learning_rate": 8.55842105263158e-05,
      "loss": 0.713,
      "step": 39
    },
    {
      "epoch": 0.064464141821112,
      "grad_norm": 0.18865598738193512,
      "learning_rate": 8.505263157894737e-05,
      "loss": 0.7027,
      "step": 40
    },
    {
      "epoch": 0.06607574536663981,
      "grad_norm": 0.2409752905368805,
      "learning_rate": 8.452105263157896e-05,
      "loss": 0.7385,
      "step": 41
    },
    {
      "epoch": 0.06768734891216761,
      "grad_norm": 0.2615523934364319,
      "learning_rate": 8.398947368421053e-05,
      "loss": 0.6761,
      "step": 42
    },
    {
      "epoch": 0.0692989524576954,
      "grad_norm": 0.3525382876396179,
      "learning_rate": 8.345789473684211e-05,
      "loss": 0.7173,
      "step": 43
    },
    {
      "epoch": 0.07091055600322321,
      "grad_norm": 0.23152434825897217,
      "learning_rate": 8.292631578947368e-05,
      "loss": 0.6942,
      "step": 44
    },
    {
      "epoch": 0.072522159548751,
      "grad_norm": 0.19304005801677704,
      "learning_rate": 8.239473684210526e-05,
      "loss": 0.711,
      "step": 45
    },
    {
      "epoch": 0.07413376309427881,
      "grad_norm": 0.20851251482963562,
      "learning_rate": 8.186315789473683e-05,
      "loss": 0.6829,
      "step": 46
    },
    {
      "epoch": 0.0757453666398066,
      "grad_norm": 0.24184094369411469,
      "learning_rate": 8.133157894736842e-05,
      "loss": 0.7245,
      "step": 47
    },
    {
      "epoch": 0.07735697018533441,
      "grad_norm": 0.27612966299057007,
      "learning_rate": 8.080000000000001e-05,
      "loss": 0.6987,
      "step": 48
    },
    {
      "epoch": 0.0789685737308622,
      "grad_norm": 0.337848961353302,
      "learning_rate": 8.026842105263159e-05,
      "loss": 0.6726,
      "step": 49
    },
    {
      "epoch": 0.08058017727639001,
      "grad_norm": 0.52480149269104,
      "learning_rate": 7.973684210526316e-05,
      "loss": 0.6298,
      "step": 50
    },
    {
      "epoch": 0.08058017727639001,
      "eval_loss": 0.6599936485290527,
      "eval_runtime": 135.8696,
      "eval_samples_per_second": 61.515,
      "eval_steps_per_second": 1.928,
      "step": 50
    },
    {
      "epoch": 0.0821917808219178,
      "grad_norm": 0.19816854596138,
      "learning_rate": 7.920526315789474e-05,
      "loss": 0.7042,
      "step": 51
    },
    {
      "epoch": 0.08380338436744561,
      "grad_norm": 0.23714330792427063,
      "learning_rate": 7.867368421052631e-05,
      "loss": 0.6846,
      "step": 52
    },
    {
      "epoch": 0.0854149879129734,
      "grad_norm": 0.2528240382671356,
      "learning_rate": 7.814210526315789e-05,
      "loss": 0.668,
      "step": 53
    },
    {
      "epoch": 0.08702659145850121,
      "grad_norm": 0.29058927297592163,
      "learning_rate": 7.761052631578946e-05,
      "loss": 0.6611,
      "step": 54
    },
    {
      "epoch": 0.088638195004029,
      "grad_norm": 0.36324694752693176,
      "learning_rate": 7.707894736842105e-05,
      "loss": 0.6686,
      "step": 55
    },
    {
      "epoch": 0.09024979854955681,
      "grad_norm": 0.4872949719429016,
      "learning_rate": 7.654736842105264e-05,
      "loss": 0.6209,
      "step": 56
    },
    {
      "epoch": 0.09186140209508462,
      "grad_norm": 0.21689631044864655,
      "learning_rate": 7.601578947368422e-05,
      "loss": 0.6876,
      "step": 57
    },
    {
      "epoch": 0.09347300564061241,
      "grad_norm": 0.23922550678253174,
      "learning_rate": 7.548421052631579e-05,
      "loss": 0.7083,
      "step": 58
    },
    {
      "epoch": 0.09508460918614021,
      "grad_norm": 0.26386719942092896,
      "learning_rate": 7.495263157894737e-05,
      "loss": 0.6284,
      "step": 59
    },
    {
      "epoch": 0.09669621273166801,
      "grad_norm": 0.3023104667663574,
      "learning_rate": 7.442105263157894e-05,
      "loss": 0.5894,
      "step": 60
    },
    {
      "epoch": 0.09830781627719581,
      "grad_norm": 0.36962202191352844,
      "learning_rate": 7.388947368421053e-05,
      "loss": 0.6337,
      "step": 61
    },
    {
      "epoch": 0.0999194198227236,
      "grad_norm": 0.47718724608421326,
      "learning_rate": 7.335789473684211e-05,
      "loss": 0.5995,
      "step": 62
    },
    {
      "epoch": 0.10153102336825141,
      "grad_norm": 0.26407232880592346,
      "learning_rate": 7.282631578947368e-05,
      "loss": 0.6403,
      "step": 63
    },
    {
      "epoch": 0.1031426269137792,
      "grad_norm": 0.23950377106666565,
      "learning_rate": 7.229473684210527e-05,
      "loss": 0.6717,
      "step": 64
    },
    {
      "epoch": 0.10475423045930701,
      "grad_norm": 0.2847888767719269,
      "learning_rate": 7.176315789473685e-05,
      "loss": 0.6053,
      "step": 65
    },
    {
      "epoch": 0.1063658340048348,
      "grad_norm": 0.30500850081443787,
      "learning_rate": 7.123157894736842e-05,
      "loss": 0.6456,
      "step": 66
    },
    {
      "epoch": 0.10797743755036261,
      "grad_norm": 0.35994887351989746,
      "learning_rate": 7.07e-05,
      "loss": 0.5984,
      "step": 67
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 0.45406046509742737,
      "learning_rate": 7.016842105263159e-05,
      "loss": 0.5379,
      "step": 68
    },
    {
      "epoch": 0.11120064464141821,
      "grad_norm": 0.36862146854400635,
      "learning_rate": 6.963684210526316e-05,
      "loss": 0.6479,
      "step": 69
    },
    {
      "epoch": 0.11281224818694602,
      "grad_norm": 0.2420656830072403,
      "learning_rate": 6.910526315789474e-05,
      "loss": 0.6218,
      "step": 70
    },
    {
      "epoch": 0.11442385173247381,
      "grad_norm": 0.29916492104530334,
      "learning_rate": 6.857368421052631e-05,
      "loss": 0.6343,
      "step": 71
    },
    {
      "epoch": 0.11603545527800162,
      "grad_norm": 0.33885183930397034,
      "learning_rate": 6.80421052631579e-05,
      "loss": 0.6124,
      "step": 72
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.38673973083496094,
      "learning_rate": 6.751052631578948e-05,
      "loss": 0.5623,
      "step": 73
    },
    {
      "epoch": 0.11925866236905722,
      "grad_norm": 0.4571478068828583,
      "learning_rate": 6.697894736842105e-05,
      "loss": 0.5642,
      "step": 74
    },
    {
      "epoch": 0.12087026591458501,
      "grad_norm": 0.6491063237190247,
      "learning_rate": 6.644736842105264e-05,
      "loss": 0.4955,
      "step": 75
    },
    {
      "epoch": 0.12248186946011282,
      "grad_norm": 0.24757473170757294,
      "learning_rate": 6.591578947368422e-05,
      "loss": 0.6216,
      "step": 76
    },
    {
      "epoch": 0.12409347300564061,
      "grad_norm": 0.3110860288143158,
      "learning_rate": 6.538421052631579e-05,
      "loss": 0.6069,
      "step": 77
    },
    {
      "epoch": 0.12570507655116842,
      "grad_norm": 0.3447476923465729,
      "learning_rate": 6.485263157894737e-05,
      "loss": 0.5747,
      "step": 78
    },
    {
      "epoch": 0.12731668009669622,
      "grad_norm": 0.3997146785259247,
      "learning_rate": 6.432105263157894e-05,
      "loss": 0.5514,
      "step": 79
    },
    {
      "epoch": 0.128928283642224,
      "grad_norm": 0.4768727421760559,
      "learning_rate": 6.378947368421053e-05,
      "loss": 0.5112,
      "step": 80
    },
    {
      "epoch": 0.1305398871877518,
      "grad_norm": 0.6860184073448181,
      "learning_rate": 6.32578947368421e-05,
      "loss": 0.5494,
      "step": 81
    },
    {
      "epoch": 0.13215149073327961,
      "grad_norm": 0.28173205256462097,
      "learning_rate": 6.27263157894737e-05,
      "loss": 0.6339,
      "step": 82
    },
    {
      "epoch": 0.13376309427880742,
      "grad_norm": 0.30443888902664185,
      "learning_rate": 6.219473684210527e-05,
      "loss": 0.6196,
      "step": 83
    },
    {
      "epoch": 0.13537469782433523,
      "grad_norm": 0.3652026653289795,
      "learning_rate": 6.166315789473685e-05,
      "loss": 0.588,
      "step": 84
    },
    {
      "epoch": 0.136986301369863,
      "grad_norm": 0.3979891240596771,
      "learning_rate": 6.113157894736842e-05,
      "loss": 0.546,
      "step": 85
    },
    {
      "epoch": 0.1385979049153908,
      "grad_norm": 0.4760202169418335,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 0.5122,
      "step": 86
    },
    {
      "epoch": 0.14020950846091862,
      "grad_norm": 0.6287760138511658,
      "learning_rate": 6.006842105263158e-05,
      "loss": 0.4889,
      "step": 87
    },
    {
      "epoch": 0.14182111200644643,
      "grad_norm": 0.32981643080711365,
      "learning_rate": 5.953684210526315e-05,
      "loss": 0.5776,
      "step": 88
    },
    {
      "epoch": 0.1434327155519742,
      "grad_norm": 0.3730184733867645,
      "learning_rate": 5.900526315789474e-05,
      "loss": 0.6509,
      "step": 89
    },
    {
      "epoch": 0.145044319097502,
      "grad_norm": 0.3809673488140106,
      "learning_rate": 5.847368421052632e-05,
      "loss": 0.5744,
      "step": 90
    },
    {
      "epoch": 0.14665592264302982,
      "grad_norm": 0.411126047372818,
      "learning_rate": 5.79421052631579e-05,
      "loss": 0.5536,
      "step": 91
    },
    {
      "epoch": 0.14826752618855762,
      "grad_norm": 0.49867916107177734,
      "learning_rate": 5.7410526315789475e-05,
      "loss": 0.4925,
      "step": 92
    },
    {
      "epoch": 0.1498791297340854,
      "grad_norm": 0.593421995639801,
      "learning_rate": 5.687894736842105e-05,
      "loss": 0.4427,
      "step": 93
    },
    {
      "epoch": 0.1514907332796132,
      "grad_norm": 0.39515072107315063,
      "learning_rate": 5.6347368421052625e-05,
      "loss": 0.5233,
      "step": 94
    },
    {
      "epoch": 0.15310233682514102,
      "grad_norm": 0.33038094639778137,
      "learning_rate": 5.5815789473684214e-05,
      "loss": 0.5878,
      "step": 95
    },
    {
      "epoch": 0.15471394037066882,
      "grad_norm": 0.4073917865753174,
      "learning_rate": 5.5284210526315796e-05,
      "loss": 0.578,
      "step": 96
    },
    {
      "epoch": 0.1563255439161966,
      "grad_norm": 0.4969066083431244,
      "learning_rate": 5.475263157894737e-05,
      "loss": 0.5495,
      "step": 97
    },
    {
      "epoch": 0.1579371474617244,
      "grad_norm": 0.48956212401390076,
      "learning_rate": 5.422105263157895e-05,
      "loss": 0.4926,
      "step": 98
    },
    {
      "epoch": 0.15954875100725222,
      "grad_norm": 0.5708914995193481,
      "learning_rate": 5.368947368421053e-05,
      "loss": 0.4043,
      "step": 99
    },
    {
      "epoch": 0.16116035455278002,
      "grad_norm": 0.8494532108306885,
      "learning_rate": 5.3157894736842104e-05,
      "loss": 0.3979,
      "step": 100
    },
    {
      "epoch": 0.16116035455278002,
      "eval_loss": 0.4972478449344635,
      "eval_runtime": 134.4776,
      "eval_samples_per_second": 62.152,
      "eval_steps_per_second": 1.948,
      "step": 100
    },
    {
      "epoch": 0.16277195809830783,
      "grad_norm": 0.2953222393989563,
      "learning_rate": 5.262631578947368e-05,
      "loss": 0.5638,
      "step": 101
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 0.4076739549636841,
      "learning_rate": 5.209473684210527e-05,
      "loss": 0.5997,
      "step": 102
    },
    {
      "epoch": 0.16599516518936341,
      "grad_norm": 0.46156543493270874,
      "learning_rate": 5.1563157894736844e-05,
      "loss": 0.5238,
      "step": 103
    },
    {
      "epoch": 0.16760676873489122,
      "grad_norm": 0.5150144696235657,
      "learning_rate": 5.1031578947368426e-05,
      "loss": 0.4905,
      "step": 104
    },
    {
      "epoch": 0.16921837228041903,
      "grad_norm": 0.6315293312072754,
      "learning_rate": 5.05e-05,
      "loss": 0.4555,
      "step": 105
    },
    {
      "epoch": 0.1708299758259468,
      "grad_norm": 0.9211300015449524,
      "learning_rate": 4.9968421052631576e-05,
      "loss": 0.3999,
      "step": 106
    },
    {
      "epoch": 0.1724415793714746,
      "grad_norm": 0.36812564730644226,
      "learning_rate": 4.943684210526316e-05,
      "loss": 0.5902,
      "step": 107
    },
    {
      "epoch": 0.17405318291700242,
      "grad_norm": 0.38568469882011414,
      "learning_rate": 4.890526315789474e-05,
      "loss": 0.5742,
      "step": 108
    },
    {
      "epoch": 0.17566478646253023,
      "grad_norm": 0.4459795653820038,
      "learning_rate": 4.8373684210526316e-05,
      "loss": 0.5317,
      "step": 109
    },
    {
      "epoch": 0.177276390008058,
      "grad_norm": 0.5016929507255554,
      "learning_rate": 4.784210526315789e-05,
      "loss": 0.4765,
      "step": 110
    },
    {
      "epoch": 0.1788879935535858,
      "grad_norm": 0.6309992671012878,
      "learning_rate": 4.731052631578947e-05,
      "loss": 0.442,
      "step": 111
    },
    {
      "epoch": 0.18049959709911362,
      "grad_norm": 0.7072840332984924,
      "learning_rate": 4.6778947368421055e-05,
      "loss": 0.4028,
      "step": 112
    },
    {
      "epoch": 0.18211120064464142,
      "grad_norm": 0.38350537419319153,
      "learning_rate": 4.624736842105263e-05,
      "loss": 0.5352,
      "step": 113
    },
    {
      "epoch": 0.18372280419016923,
      "grad_norm": 0.4897186756134033,
      "learning_rate": 4.571578947368421e-05,
      "loss": 0.6138,
      "step": 114
    },
    {
      "epoch": 0.185334407735697,
      "grad_norm": 0.5892842411994934,
      "learning_rate": 4.518421052631579e-05,
      "loss": 0.5777,
      "step": 115
    },
    {
      "epoch": 0.18694601128122482,
      "grad_norm": 0.5452648401260376,
      "learning_rate": 4.465263157894737e-05,
      "loss": 0.475,
      "step": 116
    },
    {
      "epoch": 0.18855761482675262,
      "grad_norm": 0.6141505837440491,
      "learning_rate": 4.412105263157895e-05,
      "loss": 0.4451,
      "step": 117
    },
    {
      "epoch": 0.19016921837228043,
      "grad_norm": 0.6963237524032593,
      "learning_rate": 4.358947368421053e-05,
      "loss": 0.332,
      "step": 118
    },
    {
      "epoch": 0.1917808219178082,
      "grad_norm": 0.512565016746521,
      "learning_rate": 4.30578947368421e-05,
      "loss": 0.5548,
      "step": 119
    },
    {
      "epoch": 0.19339242546333602,
      "grad_norm": 0.39101463556289673,
      "learning_rate": 4.2526315789473685e-05,
      "loss": 0.5824,
      "step": 120
    },
    {
      "epoch": 0.19500402900886382,
      "grad_norm": 0.4487123489379883,
      "learning_rate": 4.199473684210527e-05,
      "loss": 0.5225,
      "step": 121
    },
    {
      "epoch": 0.19661563255439163,
      "grad_norm": 0.5278434753417969,
      "learning_rate": 4.146315789473684e-05,
      "loss": 0.4837,
      "step": 122
    },
    {
      "epoch": 0.1982272360999194,
      "grad_norm": 0.6028698086738586,
      "learning_rate": 4.093157894736842e-05,
      "loss": 0.4294,
      "step": 123
    },
    {
      "epoch": 0.1998388396454472,
      "grad_norm": 0.8378256559371948,
      "learning_rate": 4.0400000000000006e-05,
      "loss": 0.379,
      "step": 124
    },
    {
      "epoch": 0.20145044319097502,
      "grad_norm": 1.1504991054534912,
      "learning_rate": 3.986842105263158e-05,
      "loss": 0.3705,
      "step": 125
    },
    {
      "epoch": 0.20306204673650283,
      "grad_norm": 0.3160516023635864,
      "learning_rate": 3.933684210526316e-05,
      "loss": 0.5496,
      "step": 126
    },
    {
      "epoch": 0.20467365028203063,
      "grad_norm": 0.43820229172706604,
      "learning_rate": 3.880526315789473e-05,
      "loss": 0.5012,
      "step": 127
    },
    {
      "epoch": 0.2062852538275584,
      "grad_norm": 0.4975963532924652,
      "learning_rate": 3.827368421052632e-05,
      "loss": 0.5217,
      "step": 128
    },
    {
      "epoch": 0.20789685737308622,
      "grad_norm": 0.5185565948486328,
      "learning_rate": 3.7742105263157896e-05,
      "loss": 0.428,
      "step": 129
    },
    {
      "epoch": 0.20950846091861403,
      "grad_norm": 0.6332120895385742,
      "learning_rate": 3.721052631578947e-05,
      "loss": 0.3511,
      "step": 130
    },
    {
      "epoch": 0.21112006446414183,
      "grad_norm": 0.7808757424354553,
      "learning_rate": 3.6678947368421054e-05,
      "loss": 0.2946,
      "step": 131
    },
    {
      "epoch": 0.2127316680096696,
      "grad_norm": 0.4457676112651825,
      "learning_rate": 3.6147368421052636e-05,
      "loss": 0.5695,
      "step": 132
    },
    {
      "epoch": 0.21434327155519742,
      "grad_norm": 0.5079824328422546,
      "learning_rate": 3.561578947368421e-05,
      "loss": 0.5236,
      "step": 133
    },
    {
      "epoch": 0.21595487510072522,
      "grad_norm": 0.5814209580421448,
      "learning_rate": 3.508421052631579e-05,
      "loss": 0.4928,
      "step": 134
    },
    {
      "epoch": 0.21756647864625303,
      "grad_norm": 0.5811838507652283,
      "learning_rate": 3.455263157894737e-05,
      "loss": 0.423,
      "step": 135
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 0.6463267207145691,
      "learning_rate": 3.402105263157895e-05,
      "loss": 0.3671,
      "step": 136
    },
    {
      "epoch": 0.22078968573730862,
      "grad_norm": 0.825130820274353,
      "learning_rate": 3.3489473684210526e-05,
      "loss": 0.2856,
      "step": 137
    },
    {
      "epoch": 0.22240128928283642,
      "grad_norm": 0.45928215980529785,
      "learning_rate": 3.295789473684211e-05,
      "loss": 0.5312,
      "step": 138
    },
    {
      "epoch": 0.22401289282836423,
      "grad_norm": 0.4779850244522095,
      "learning_rate": 3.242631578947368e-05,
      "loss": 0.5565,
      "step": 139
    },
    {
      "epoch": 0.22562449637389204,
      "grad_norm": 0.5173392295837402,
      "learning_rate": 3.1894736842105265e-05,
      "loss": 0.4862,
      "step": 140
    },
    {
      "epoch": 0.22723609991941982,
      "grad_norm": 0.6725038290023804,
      "learning_rate": 3.136315789473685e-05,
      "loss": 0.4738,
      "step": 141
    },
    {
      "epoch": 0.22884770346494762,
      "grad_norm": 0.6613323092460632,
      "learning_rate": 3.083157894736842e-05,
      "loss": 0.3635,
      "step": 142
    },
    {
      "epoch": 0.23045930701047543,
      "grad_norm": 0.726622998714447,
      "learning_rate": 3.0299999999999998e-05,
      "loss": 0.255,
      "step": 143
    },
    {
      "epoch": 0.23207091055600323,
      "grad_norm": 0.5666455626487732,
      "learning_rate": 2.9768421052631577e-05,
      "loss": 0.3874,
      "step": 144
    },
    {
      "epoch": 0.233682514101531,
      "grad_norm": 0.459031343460083,
      "learning_rate": 2.923684210526316e-05,
      "loss": 0.5113,
      "step": 145
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.5253135561943054,
      "learning_rate": 2.8705263157894737e-05,
      "loss": 0.4246,
      "step": 146
    },
    {
      "epoch": 0.23690572119258663,
      "grad_norm": 0.6632147431373596,
      "learning_rate": 2.8173684210526313e-05,
      "loss": 0.4323,
      "step": 147
    },
    {
      "epoch": 0.23851732473811443,
      "grad_norm": 0.749000072479248,
      "learning_rate": 2.7642105263157898e-05,
      "loss": 0.3744,
      "step": 148
    },
    {
      "epoch": 0.2401289282836422,
      "grad_norm": 0.7715749740600586,
      "learning_rate": 2.7110526315789473e-05,
      "loss": 0.257,
      "step": 149
    },
    {
      "epoch": 0.24174053182917002,
      "grad_norm": 1.1665058135986328,
      "learning_rate": 2.6578947368421052e-05,
      "loss": 0.2613,
      "step": 150
    },
    {
      "epoch": 0.24174053182917002,
      "eval_loss": 0.3877754807472229,
      "eval_runtime": 136.1771,
      "eval_samples_per_second": 61.376,
      "eval_steps_per_second": 1.924,
      "step": 150
    },
    {
      "epoch": 0.24335213537469783,
      "grad_norm": 0.40307724475860596,
      "learning_rate": 2.6047368421052634e-05,
      "loss": 0.5158,
      "step": 151
    },
    {
      "epoch": 0.24496373892022563,
      "grad_norm": 0.47922101616859436,
      "learning_rate": 2.5515789473684213e-05,
      "loss": 0.5136,
      "step": 152
    },
    {
      "epoch": 0.2465753424657534,
      "grad_norm": 0.5659621357917786,
      "learning_rate": 2.4984210526315788e-05,
      "loss": 0.3903,
      "step": 153
    },
    {
      "epoch": 0.24818694601128122,
      "grad_norm": 0.6030014753341675,
      "learning_rate": 2.445263157894737e-05,
      "loss": 0.3398,
      "step": 154
    },
    {
      "epoch": 0.24979854955680902,
      "grad_norm": 0.6882525086402893,
      "learning_rate": 2.3921052631578946e-05,
      "loss": 0.2735,
      "step": 155
    },
    {
      "epoch": 0.25141015310233683,
      "grad_norm": 0.9658460021018982,
      "learning_rate": 2.3389473684210528e-05,
      "loss": 0.2586,
      "step": 156
    },
    {
      "epoch": 0.25302175664786464,
      "grad_norm": 0.5281833410263062,
      "learning_rate": 2.2857894736842106e-05,
      "loss": 0.5193,
      "step": 157
    },
    {
      "epoch": 0.25463336019339244,
      "grad_norm": 0.6106149554252625,
      "learning_rate": 2.2326315789473685e-05,
      "loss": 0.4958,
      "step": 158
    },
    {
      "epoch": 0.25624496373892025,
      "grad_norm": 0.7677047252655029,
      "learning_rate": 2.1794736842105264e-05,
      "loss": 0.4363,
      "step": 159
    },
    {
      "epoch": 0.257856567284448,
      "grad_norm": 0.655020534992218,
      "learning_rate": 2.1263157894736842e-05,
      "loss": 0.3346,
      "step": 160
    },
    {
      "epoch": 0.2594681708299758,
      "grad_norm": 0.7973310351371765,
      "learning_rate": 2.073157894736842e-05,
      "loss": 0.3355,
      "step": 161
    },
    {
      "epoch": 0.2610797743755036,
      "grad_norm": 0.9603102207183838,
      "learning_rate": 2.0200000000000003e-05,
      "loss": 0.2597,
      "step": 162
    },
    {
      "epoch": 0.2626913779210314,
      "grad_norm": 0.5082424879074097,
      "learning_rate": 1.966842105263158e-05,
      "loss": 0.4856,
      "step": 163
    },
    {
      "epoch": 0.26430298146655923,
      "grad_norm": 0.47313550114631653,
      "learning_rate": 1.913684210526316e-05,
      "loss": 0.521,
      "step": 164
    },
    {
      "epoch": 0.26591458501208703,
      "grad_norm": 0.5528334379196167,
      "learning_rate": 1.8605263157894736e-05,
      "loss": 0.4273,
      "step": 165
    },
    {
      "epoch": 0.26752618855761484,
      "grad_norm": 0.6417384743690491,
      "learning_rate": 1.8073684210526318e-05,
      "loss": 0.377,
      "step": 166
    },
    {
      "epoch": 0.26913779210314265,
      "grad_norm": 0.7294735908508301,
      "learning_rate": 1.7542105263157897e-05,
      "loss": 0.3155,
      "step": 167
    },
    {
      "epoch": 0.27074939564867045,
      "grad_norm": 0.8516286611557007,
      "learning_rate": 1.7010526315789475e-05,
      "loss": 0.2491,
      "step": 168
    },
    {
      "epoch": 0.2723609991941982,
      "grad_norm": 0.5871580839157104,
      "learning_rate": 1.6478947368421054e-05,
      "loss": 0.408,
      "step": 169
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 0.5080535411834717,
      "learning_rate": 1.5947368421052633e-05,
      "loss": 0.5006,
      "step": 170
    },
    {
      "epoch": 0.2755842062852538,
      "grad_norm": 0.572222113609314,
      "learning_rate": 1.541578947368421e-05,
      "loss": 0.4072,
      "step": 171
    },
    {
      "epoch": 0.2771958098307816,
      "grad_norm": 0.6648744940757751,
      "learning_rate": 1.4884210526315788e-05,
      "loss": 0.3782,
      "step": 172
    },
    {
      "epoch": 0.27880741337630943,
      "grad_norm": 0.7872058153152466,
      "learning_rate": 1.4352631578947369e-05,
      "loss": 0.3245,
      "step": 173
    },
    {
      "epoch": 0.28041901692183724,
      "grad_norm": 0.8955470323562622,
      "learning_rate": 1.3821052631578949e-05,
      "loss": 0.2479,
      "step": 174
    },
    {
      "epoch": 0.28203062046736505,
      "grad_norm": 1.2318774461746216,
      "learning_rate": 1.3289473684210526e-05,
      "loss": 0.2326,
      "step": 175
    },
    {
      "epoch": 0.28364222401289285,
      "grad_norm": 0.4534154534339905,
      "learning_rate": 1.2757894736842106e-05,
      "loss": 0.4958,
      "step": 176
    },
    {
      "epoch": 0.2852538275584206,
      "grad_norm": 0.5369049906730652,
      "learning_rate": 1.2226315789473685e-05,
      "loss": 0.4478,
      "step": 177
    },
    {
      "epoch": 0.2868654311039484,
      "grad_norm": 0.6685469746589661,
      "learning_rate": 1.1694736842105264e-05,
      "loss": 0.3742,
      "step": 178
    },
    {
      "epoch": 0.2884770346494762,
      "grad_norm": 0.666946530342102,
      "learning_rate": 1.1163157894736842e-05,
      "loss": 0.3135,
      "step": 179
    },
    {
      "epoch": 0.290088638195004,
      "grad_norm": 0.7749336361885071,
      "learning_rate": 1.0631578947368421e-05,
      "loss": 0.2429,
      "step": 180
    },
    {
      "epoch": 0.29170024174053183,
      "grad_norm": 1.028951644897461,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 0.2296,
      "step": 181
    },
    {
      "epoch": 0.29331184528605964,
      "grad_norm": 0.46223780512809753,
      "learning_rate": 9.56842105263158e-06,
      "loss": 0.4958,
      "step": 182
    },
    {
      "epoch": 0.29492344883158744,
      "grad_norm": 0.518415093421936,
      "learning_rate": 9.036842105263159e-06,
      "loss": 0.4638,
      "step": 183
    },
    {
      "epoch": 0.29653505237711525,
      "grad_norm": 0.6299134492874146,
      "learning_rate": 8.505263157894738e-06,
      "loss": 0.3941,
      "step": 184
    },
    {
      "epoch": 0.29814665592264306,
      "grad_norm": 0.6578340530395508,
      "learning_rate": 7.973684210526316e-06,
      "loss": 0.3176,
      "step": 185
    },
    {
      "epoch": 0.2997582594681708,
      "grad_norm": 0.8014324903488159,
      "learning_rate": 7.442105263157894e-06,
      "loss": 0.2383,
      "step": 186
    },
    {
      "epoch": 0.3013698630136986,
      "grad_norm": 0.9844911694526672,
      "learning_rate": 6.9105263157894745e-06,
      "loss": 0.219,
      "step": 187
    },
    {
      "epoch": 0.3029814665592264,
      "grad_norm": 0.5260629653930664,
      "learning_rate": 6.378947368421053e-06,
      "loss": 0.4447,
      "step": 188
    },
    {
      "epoch": 0.3045930701047542,
      "grad_norm": 0.5315889120101929,
      "learning_rate": 5.847368421052632e-06,
      "loss": 0.4569,
      "step": 189
    },
    {
      "epoch": 0.30620467365028203,
      "grad_norm": 0.6365461349487305,
      "learning_rate": 5.315789473684211e-06,
      "loss": 0.4326,
      "step": 190
    },
    {
      "epoch": 0.30781627719580984,
      "grad_norm": 0.658093273639679,
      "learning_rate": 4.78421052631579e-06,
      "loss": 0.33,
      "step": 191
    },
    {
      "epoch": 0.30942788074133765,
      "grad_norm": 0.7685239911079407,
      "learning_rate": 4.252631578947369e-06,
      "loss": 0.2842,
      "step": 192
    },
    {
      "epoch": 0.31103948428686545,
      "grad_norm": 0.8007004261016846,
      "learning_rate": 3.721052631578947e-06,
      "loss": 0.1996,
      "step": 193
    },
    {
      "epoch": 0.3126510878323932,
      "grad_norm": 0.6490435004234314,
      "learning_rate": 3.1894736842105266e-06,
      "loss": 0.3936,
      "step": 194
    },
    {
      "epoch": 0.314262691377921,
      "grad_norm": 0.5127969980239868,
      "learning_rate": 2.6578947368421053e-06,
      "loss": 0.4567,
      "step": 195
    },
    {
      "epoch": 0.3158742949234488,
      "grad_norm": 0.6253898739814758,
      "learning_rate": 2.1263157894736844e-06,
      "loss": 0.4243,
      "step": 196
    },
    {
      "epoch": 0.3174858984689766,
      "grad_norm": 0.6929444074630737,
      "learning_rate": 1.5947368421052633e-06,
      "loss": 0.3584,
      "step": 197
    },
    {
      "epoch": 0.31909750201450443,
      "grad_norm": 0.8625342845916748,
      "learning_rate": 1.0631578947368422e-06,
      "loss": 0.29,
      "step": 198
    },
    {
      "epoch": 0.32070910556003224,
      "grad_norm": 0.9054241180419922,
      "learning_rate": 5.315789473684211e-07,
      "loss": 0.2209,
      "step": 199
    },
    {
      "epoch": 0.32232070910556004,
      "grad_norm": 1.2763004302978516,
      "learning_rate": 0.0,
      "loss": 0.2273,
      "step": 200
    },
    {
      "epoch": 0.32232070910556004,
      "eval_loss": 0.33371686935424805,
      "eval_runtime": 137.2091,
      "eval_samples_per_second": 60.914,
      "eval_steps_per_second": 1.909,
      "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": 4.4879212787531776e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}