{
  "best_metric": 1.7381101846694946,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.14828544949026876,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007414272474513438,
      "grad_norm": 0.6124485731124878,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.7852,
      "step": 1
    },
    {
      "epoch": 0.0007414272474513438,
      "eval_loss": 2.6430418491363525,
      "eval_runtime": 80.388,
      "eval_samples_per_second": 28.263,
      "eval_steps_per_second": 14.131,
      "step": 1
    },
    {
      "epoch": 0.0014828544949026877,
      "grad_norm": 0.5803061723709106,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.6473,
      "step": 2
    },
    {
      "epoch": 0.0022242817423540314,
      "grad_norm": 0.6363569498062134,
      "learning_rate": 1e-05,
      "loss": 1.7952,
      "step": 3
    },
    {
      "epoch": 0.0029657089898053753,
      "grad_norm": 0.6355988383293152,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.7877,
      "step": 4
    },
    {
      "epoch": 0.0037071362372567192,
      "grad_norm": 0.6372725367546082,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.8128,
      "step": 5
    },
    {
      "epoch": 0.004448563484708063,
      "grad_norm": 0.6257951259613037,
      "learning_rate": 2e-05,
      "loss": 1.8193,
      "step": 6
    },
    {
      "epoch": 0.005189990732159407,
      "grad_norm": 0.5836296081542969,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 1.8082,
      "step": 7
    },
    {
      "epoch": 0.005931417979610751,
      "grad_norm": 0.5540176630020142,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 1.8171,
      "step": 8
    },
    {
      "epoch": 0.006672845227062094,
      "grad_norm": 0.5469110012054443,
      "learning_rate": 3e-05,
      "loss": 1.7839,
      "step": 9
    },
    {
      "epoch": 0.0074142724745134385,
      "grad_norm": 0.5312670469284058,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 1.8795,
      "step": 10
    },
    {
      "epoch": 0.008155699721964782,
      "grad_norm": 0.5280024409294128,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 1.7759,
      "step": 11
    },
    {
      "epoch": 0.008897126969416125,
      "grad_norm": 0.6129993796348572,
      "learning_rate": 4e-05,
      "loss": 1.7813,
      "step": 12
    },
    {
      "epoch": 0.00963855421686747,
      "grad_norm": 0.6406863331794739,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 1.6631,
      "step": 13
    },
    {
      "epoch": 0.010379981464318814,
      "grad_norm": 0.7365619540214539,
      "learning_rate": 4.666666666666667e-05,
      "loss": 1.8269,
      "step": 14
    },
    {
      "epoch": 0.011121408711770158,
      "grad_norm": 0.713071882724762,
      "learning_rate": 5e-05,
      "loss": 1.7048,
      "step": 15
    },
    {
      "epoch": 0.011862835959221501,
      "grad_norm": 0.6798648238182068,
      "learning_rate": 5.333333333333333e-05,
      "loss": 1.8568,
      "step": 16
    },
    {
      "epoch": 0.012604263206672845,
      "grad_norm": 0.6837913393974304,
      "learning_rate": 5.666666666666667e-05,
      "loss": 1.8505,
      "step": 17
    },
    {
      "epoch": 0.013345690454124188,
      "grad_norm": 0.6939889788627625,
      "learning_rate": 6e-05,
      "loss": 1.7763,
      "step": 18
    },
    {
      "epoch": 0.014087117701575533,
      "grad_norm": 0.8109320402145386,
      "learning_rate": 6.333333333333333e-05,
      "loss": 1.7803,
      "step": 19
    },
    {
      "epoch": 0.014828544949026877,
      "grad_norm": 0.7811878323554993,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.8487,
      "step": 20
    },
    {
      "epoch": 0.01556997219647822,
      "grad_norm": 0.7671340703964233,
      "learning_rate": 7e-05,
      "loss": 1.83,
      "step": 21
    },
    {
      "epoch": 0.016311399443929564,
      "grad_norm": 0.8294042944908142,
      "learning_rate": 7.333333333333333e-05,
      "loss": 1.8335,
      "step": 22
    },
    {
      "epoch": 0.017052826691380907,
      "grad_norm": 0.8303661346435547,
      "learning_rate": 7.666666666666667e-05,
      "loss": 1.7366,
      "step": 23
    },
    {
      "epoch": 0.01779425393883225,
      "grad_norm": 0.8452559113502502,
      "learning_rate": 8e-05,
      "loss": 1.667,
      "step": 24
    },
    {
      "epoch": 0.018535681186283594,
      "grad_norm": 0.9195181727409363,
      "learning_rate": 8.333333333333334e-05,
      "loss": 1.7515,
      "step": 25
    },
    {
      "epoch": 0.01927710843373494,
      "grad_norm": 1.0009349584579468,
      "learning_rate": 8.666666666666667e-05,
      "loss": 1.8507,
      "step": 26
    },
    {
      "epoch": 0.020018535681186285,
      "grad_norm": 1.005056619644165,
      "learning_rate": 9e-05,
      "loss": 1.8436,
      "step": 27
    },
    {
      "epoch": 0.02075996292863763,
      "grad_norm": 1.0935652256011963,
      "learning_rate": 9.333333333333334e-05,
      "loss": 1.8048,
      "step": 28
    },
    {
      "epoch": 0.021501390176088972,
      "grad_norm": 1.1234335899353027,
      "learning_rate": 9.666666666666667e-05,
      "loss": 1.7684,
      "step": 29
    },
    {
      "epoch": 0.022242817423540315,
      "grad_norm": 1.1144816875457764,
      "learning_rate": 0.0001,
      "loss": 1.768,
      "step": 30
    },
    {
      "epoch": 0.02298424467099166,
      "grad_norm": 1.286267876625061,
      "learning_rate": 9.999146252290264e-05,
      "loss": 1.7945,
      "step": 31
    },
    {
      "epoch": 0.023725671918443002,
      "grad_norm": 1.2776055335998535,
      "learning_rate": 9.996585300715116e-05,
      "loss": 1.8183,
      "step": 32
    },
    {
      "epoch": 0.024467099165894346,
      "grad_norm": 1.4702638387680054,
      "learning_rate": 9.99231801983717e-05,
      "loss": 1.9644,
      "step": 33
    },
    {
      "epoch": 0.02520852641334569,
      "grad_norm": 1.4706964492797852,
      "learning_rate": 9.986345866928941e-05,
      "loss": 1.7956,
      "step": 34
    },
    {
      "epoch": 0.025949953660797033,
      "grad_norm": 1.4871596097946167,
      "learning_rate": 9.978670881475172e-05,
      "loss": 1.9165,
      "step": 35
    },
    {
      "epoch": 0.026691380908248376,
      "grad_norm": 1.6169719696044922,
      "learning_rate": 9.96929568447637e-05,
      "loss": 1.8174,
      "step": 36
    },
    {
      "epoch": 0.027432808155699723,
      "grad_norm": 1.7516857385635376,
      "learning_rate": 9.958223477553714e-05,
      "loss": 1.8462,
      "step": 37
    },
    {
      "epoch": 0.028174235403151067,
      "grad_norm": 2.038487672805786,
      "learning_rate": 9.94545804185573e-05,
      "loss": 2.0841,
      "step": 38
    },
    {
      "epoch": 0.02891566265060241,
      "grad_norm": 1.951445460319519,
      "learning_rate": 9.931003736767013e-05,
      "loss": 2.0004,
      "step": 39
    },
    {
      "epoch": 0.029657089898053754,
      "grad_norm": 2.489302396774292,
      "learning_rate": 9.91486549841951e-05,
      "loss": 2.0845,
      "step": 40
    },
    {
      "epoch": 0.030398517145505097,
      "grad_norm": 2.4680185317993164,
      "learning_rate": 9.89704883800683e-05,
      "loss": 1.9711,
      "step": 41
    },
    {
      "epoch": 0.03113994439295644,
      "grad_norm": 2.4111521244049072,
      "learning_rate": 9.877559839902184e-05,
      "loss": 1.6936,
      "step": 42
    },
    {
      "epoch": 0.03188137164040779,
      "grad_norm": 3.0472605228424072,
      "learning_rate": 9.85640515958057e-05,
      "loss": 2.2948,
      "step": 43
    },
    {
      "epoch": 0.03262279888785913,
      "grad_norm": 2.4441635608673096,
      "learning_rate": 9.833592021345937e-05,
      "loss": 1.696,
      "step": 44
    },
    {
      "epoch": 0.033364226135310475,
      "grad_norm": 2.5454494953155518,
      "learning_rate": 9.809128215864097e-05,
      "loss": 1.8259,
      "step": 45
    },
    {
      "epoch": 0.034105653382761815,
      "grad_norm": 4.027335166931152,
      "learning_rate": 9.783022097502204e-05,
      "loss": 1.8416,
      "step": 46
    },
    {
      "epoch": 0.03484708063021316,
      "grad_norm": 3.3237364292144775,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3515,
      "step": 47
    },
    {
      "epoch": 0.0355885078776645,
      "grad_norm": 3.3497061729431152,
      "learning_rate": 9.725919140804099e-05,
      "loss": 2.1897,
      "step": 48
    },
    {
      "epoch": 0.03632993512511585,
      "grad_norm": 3.8192644119262695,
      "learning_rate": 9.694941803075283e-05,
      "loss": 2.014,
      "step": 49
    },
    {
      "epoch": 0.03707136237256719,
      "grad_norm": 5.863681316375732,
      "learning_rate": 9.662361147021779e-05,
      "loss": 2.4335,
      "step": 50
    },
    {
      "epoch": 0.03707136237256719,
      "eval_loss": 2.0708811283111572,
      "eval_runtime": 80.3453,
      "eval_samples_per_second": 28.278,
      "eval_steps_per_second": 14.139,
      "step": 50
    },
    {
      "epoch": 0.037812789620018536,
      "grad_norm": 1.7871085405349731,
      "learning_rate": 9.628188298907782e-05,
      "loss": 1.9307,
      "step": 51
    },
    {
      "epoch": 0.03855421686746988,
      "grad_norm": 1.4125314950942993,
      "learning_rate": 9.592434928729616e-05,
      "loss": 1.8907,
      "step": 52
    },
    {
      "epoch": 0.03929564411492122,
      "grad_norm": 1.0121806859970093,
      "learning_rate": 9.555113246230442e-05,
      "loss": 1.7131,
      "step": 53
    },
    {
      "epoch": 0.04003707136237257,
      "grad_norm": 0.7524218559265137,
      "learning_rate": 9.516235996730645e-05,
      "loss": 1.7604,
      "step": 54
    },
    {
      "epoch": 0.04077849860982391,
      "grad_norm": 0.5749344825744629,
      "learning_rate": 9.475816456775313e-05,
      "loss": 1.7245,
      "step": 55
    },
    {
      "epoch": 0.04151992585727526,
      "grad_norm": 0.5453425645828247,
      "learning_rate": 9.43386842960031e-05,
      "loss": 1.7206,
      "step": 56
    },
    {
      "epoch": 0.0422613531047266,
      "grad_norm": 0.5486029982566833,
      "learning_rate": 9.39040624041849e-05,
      "loss": 1.6954,
      "step": 57
    },
    {
      "epoch": 0.043002780352177944,
      "grad_norm": 0.5450177788734436,
      "learning_rate": 9.345444731527642e-05,
      "loss": 1.7471,
      "step": 58
    },
    {
      "epoch": 0.043744207599629284,
      "grad_norm": 0.5128135681152344,
      "learning_rate": 9.298999257241863e-05,
      "loss": 1.7346,
      "step": 59
    },
    {
      "epoch": 0.04448563484708063,
      "grad_norm": 0.516810417175293,
      "learning_rate": 9.251085678648072e-05,
      "loss": 1.7397,
      "step": 60
    },
    {
      "epoch": 0.04522706209453197,
      "grad_norm": 0.49133753776550293,
      "learning_rate": 9.201720358189464e-05,
      "loss": 1.6186,
      "step": 61
    },
    {
      "epoch": 0.04596848934198332,
      "grad_norm": 0.519758403301239,
      "learning_rate": 9.150920154077754e-05,
      "loss": 1.7109,
      "step": 62
    },
    {
      "epoch": 0.046709916589434665,
      "grad_norm": 0.5506690144538879,
      "learning_rate": 9.098702414536107e-05,
      "loss": 1.6649,
      "step": 63
    },
    {
      "epoch": 0.047451343836886005,
      "grad_norm": 0.5608676075935364,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.7661,
      "step": 64
    },
    {
      "epoch": 0.04819277108433735,
      "grad_norm": 0.5808055996894836,
      "learning_rate": 8.9900861364012e-05,
      "loss": 1.9087,
      "step": 65
    },
    {
      "epoch": 0.04893419833178869,
      "grad_norm": 0.5688429474830627,
      "learning_rate": 8.933724690167417e-05,
      "loss": 1.8975,
      "step": 66
    },
    {
      "epoch": 0.04967562557924004,
      "grad_norm": 0.5601984858512878,
      "learning_rate": 8.876019880555649e-05,
      "loss": 1.7305,
      "step": 67
    },
    {
      "epoch": 0.05041705282669138,
      "grad_norm": 0.5874759554862976,
      "learning_rate": 8.816991413705516e-05,
      "loss": 1.5907,
      "step": 68
    },
    {
      "epoch": 0.051158480074142726,
      "grad_norm": 0.6670622825622559,
      "learning_rate": 8.756659447784368e-05,
      "loss": 1.781,
      "step": 69
    },
    {
      "epoch": 0.051899907321594066,
      "grad_norm": 0.6576986908912659,
      "learning_rate": 8.695044586103296e-05,
      "loss": 1.63,
      "step": 70
    },
    {
      "epoch": 0.05264133456904541,
      "grad_norm": 0.7194452881813049,
      "learning_rate": 8.632167870081121e-05,
      "loss": 1.6655,
      "step": 71
    },
    {
      "epoch": 0.05338276181649675,
      "grad_norm": 0.7311973571777344,
      "learning_rate": 8.568050772058762e-05,
      "loss": 1.7641,
      "step": 72
    },
    {
      "epoch": 0.0541241890639481,
      "grad_norm": 0.7574784755706787,
      "learning_rate": 8.502715187966455e-05,
      "loss": 1.7437,
      "step": 73
    },
    {
      "epoch": 0.05486561631139945,
      "grad_norm": 0.7845456004142761,
      "learning_rate": 8.436183429846313e-05,
      "loss": 1.4992,
      "step": 74
    },
    {
      "epoch": 0.05560704355885079,
      "grad_norm": 0.8040277361869812,
      "learning_rate": 8.368478218232787e-05,
      "loss": 1.7136,
      "step": 75
    },
    {
      "epoch": 0.056348470806302134,
      "grad_norm": 0.888521134853363,
      "learning_rate": 8.299622674393614e-05,
      "loss": 1.7331,
      "step": 76
    },
    {
      "epoch": 0.057089898053753474,
      "grad_norm": 0.9198666214942932,
      "learning_rate": 8.229640312433937e-05,
      "loss": 1.7389,
      "step": 77
    },
    {
      "epoch": 0.05783132530120482,
      "grad_norm": 0.9883677959442139,
      "learning_rate": 8.158555031266254e-05,
      "loss": 1.7816,
      "step": 78
    },
    {
      "epoch": 0.05857275254865616,
      "grad_norm": 1.129840612411499,
      "learning_rate": 8.086391106448965e-05,
      "loss": 1.8353,
      "step": 79
    },
    {
      "epoch": 0.05931417979610751,
      "grad_norm": 1.0864940881729126,
      "learning_rate": 8.013173181896283e-05,
      "loss": 1.8058,
      "step": 80
    },
    {
      "epoch": 0.06005560704355885,
      "grad_norm": 1.2116289138793945,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.7599,
      "step": 81
    },
    {
      "epoch": 0.060797034291010195,
      "grad_norm": 1.3380213975906372,
      "learning_rate": 7.863675700402526e-05,
      "loss": 1.7537,
      "step": 82
    },
    {
      "epoch": 0.06153846153846154,
      "grad_norm": 1.3464603424072266,
      "learning_rate": 7.787447196714427e-05,
      "loss": 1.9293,
      "step": 83
    },
    {
      "epoch": 0.06227988878591288,
      "grad_norm": 1.5085811614990234,
      "learning_rate": 7.710266782362247e-05,
      "loss": 1.911,
      "step": 84
    },
    {
      "epoch": 0.06302131603336422,
      "grad_norm": 1.618571162223816,
      "learning_rate": 7.63216081438678e-05,
      "loss": 2.0127,
      "step": 85
    },
    {
      "epoch": 0.06376274328081558,
      "grad_norm": 1.5068186521530151,
      "learning_rate": 7.553155965904535e-05,
      "loss": 1.7679,
      "step": 86
    },
    {
      "epoch": 0.06450417052826692,
      "grad_norm": 1.789281964302063,
      "learning_rate": 7.473279216998895e-05,
      "loss": 1.8173,
      "step": 87
    },
    {
      "epoch": 0.06524559777571826,
      "grad_norm": 1.8799840211868286,
      "learning_rate": 7.392557845506432e-05,
      "loss": 1.8448,
      "step": 88
    },
    {
      "epoch": 0.0659870250231696,
      "grad_norm": 1.6632227897644043,
      "learning_rate": 7.311019417701566e-05,
      "loss": 1.7062,
      "step": 89
    },
    {
      "epoch": 0.06672845227062095,
      "grad_norm": 2.1564183235168457,
      "learning_rate": 7.228691778882693e-05,
      "loss": 1.7266,
      "step": 90
    },
    {
      "epoch": 0.06746987951807229,
      "grad_norm": 2.1212780475616455,
      "learning_rate": 7.145603043863045e-05,
      "loss": 1.6921,
      "step": 91
    },
    {
      "epoch": 0.06821130676552363,
      "grad_norm": 2.2401561737060547,
      "learning_rate": 7.061781587369519e-05,
      "loss": 1.7909,
      "step": 92
    },
    {
      "epoch": 0.06895273401297497,
      "grad_norm": 2.520695686340332,
      "learning_rate": 6.977256034352712e-05,
      "loss": 1.8632,
      "step": 93
    },
    {
      "epoch": 0.06969416126042632,
      "grad_norm": 2.3248720169067383,
      "learning_rate": 6.892055250211552e-05,
      "loss": 1.7475,
      "step": 94
    },
    {
      "epoch": 0.07043558850787766,
      "grad_norm": 2.5022664070129395,
      "learning_rate": 6.806208330935766e-05,
      "loss": 1.7445,
      "step": 95
    },
    {
      "epoch": 0.071177015755329,
      "grad_norm": 2.5968821048736572,
      "learning_rate": 6.719744593169641e-05,
      "loss": 1.8706,
      "step": 96
    },
    {
      "epoch": 0.07191844300278036,
      "grad_norm": 3.383408308029175,
      "learning_rate": 6.632693564200416e-05,
      "loss": 2.3247,
      "step": 97
    },
    {
      "epoch": 0.0726598702502317,
      "grad_norm": 3.1737587451934814,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.1752,
      "step": 98
    },
    {
      "epoch": 0.07340129749768304,
      "grad_norm": 4.253147125244141,
      "learning_rate": 6.456948734446624e-05,
      "loss": 2.4434,
      "step": 99
    },
    {
      "epoch": 0.07414272474513438,
      "grad_norm": 5.350001335144043,
      "learning_rate": 6.368314950360415e-05,
      "loss": 2.4461,
      "step": 100
    },
    {
      "epoch": 0.07414272474513438,
      "eval_loss": 1.8864513635635376,
      "eval_runtime": 80.2739,
      "eval_samples_per_second": 28.303,
      "eval_steps_per_second": 14.152,
      "step": 100
    },
    {
      "epoch": 0.07488415199258573,
      "grad_norm": 0.8407055139541626,
      "learning_rate": 6.279213887972179e-05,
      "loss": 1.6659,
      "step": 101
    },
    {
      "epoch": 0.07562557924003707,
      "grad_norm": 0.8665005564689636,
      "learning_rate": 6.189675975213094e-05,
      "loss": 1.7601,
      "step": 102
    },
    {
      "epoch": 0.07636700648748841,
      "grad_norm": 0.7515180706977844,
      "learning_rate": 6.099731789198344e-05,
      "loss": 1.7242,
      "step": 103
    },
    {
      "epoch": 0.07710843373493977,
      "grad_norm": 0.7238166928291321,
      "learning_rate": 6.009412045785051e-05,
      "loss": 1.7741,
      "step": 104
    },
    {
      "epoch": 0.0778498609823911,
      "grad_norm": 0.620429277420044,
      "learning_rate": 5.918747589082853e-05,
      "loss": 1.6914,
      "step": 105
    },
    {
      "epoch": 0.07859128822984245,
      "grad_norm": 0.5298355221748352,
      "learning_rate": 5.82776938092065e-05,
      "loss": 1.6122,
      "step": 106
    },
    {
      "epoch": 0.07933271547729379,
      "grad_norm": 0.4990004599094391,
      "learning_rate": 5.736508490273188e-05,
      "loss": 1.6041,
      "step": 107
    },
    {
      "epoch": 0.08007414272474514,
      "grad_norm": 0.4943569004535675,
      "learning_rate": 5.644996082651017e-05,
      "loss": 1.5963,
      "step": 108
    },
    {
      "epoch": 0.08081556997219648,
      "grad_norm": 0.5157232284545898,
      "learning_rate": 5.553263409457504e-05,
      "loss": 1.5643,
      "step": 109
    },
    {
      "epoch": 0.08155699721964782,
      "grad_norm": 0.5312955379486084,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 1.6662,
      "step": 110
    },
    {
      "epoch": 0.08229842446709916,
      "grad_norm": 0.6033543348312378,
      "learning_rate": 5.3692626373743706e-05,
      "loss": 1.8139,
      "step": 111
    },
    {
      "epoch": 0.08303985171455051,
      "grad_norm": 0.5611836910247803,
      "learning_rate": 5.27705737457985e-05,
      "loss": 1.7621,
      "step": 112
    },
    {
      "epoch": 0.08378127896200185,
      "grad_norm": 0.5755486488342285,
      "learning_rate": 5.184757496945726e-05,
      "loss": 1.7926,
      "step": 113
    },
    {
      "epoch": 0.0845227062094532,
      "grad_norm": 0.5679196715354919,
      "learning_rate": 5.092394524795649e-05,
      "loss": 1.7033,
      "step": 114
    },
    {
      "epoch": 0.08526413345690455,
      "grad_norm": 0.5803284049034119,
      "learning_rate": 5e-05,
      "loss": 1.6124,
      "step": 115
    },
    {
      "epoch": 0.08600556070435589,
      "grad_norm": 0.6112229228019714,
      "learning_rate": 4.907605475204352e-05,
      "loss": 1.7132,
      "step": 116
    },
    {
      "epoch": 0.08674698795180723,
      "grad_norm": 0.6114505529403687,
      "learning_rate": 4.8152425030542766e-05,
      "loss": 1.8163,
      "step": 117
    },
    {
      "epoch": 0.08748841519925857,
      "grad_norm": 0.611460268497467,
      "learning_rate": 4.72294262542015e-05,
      "loss": 1.7346,
      "step": 118
    },
    {
      "epoch": 0.08822984244670992,
      "grad_norm": 0.6732529401779175,
      "learning_rate": 4.6307373626256306e-05,
      "loss": 1.7252,
      "step": 119
    },
    {
      "epoch": 0.08897126969416126,
      "grad_norm": 0.6358098983764648,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 1.6471,
      "step": 120
    },
    {
      "epoch": 0.0897126969416126,
      "grad_norm": 0.6858165264129639,
      "learning_rate": 4.446736590542497e-05,
      "loss": 1.7277,
      "step": 121
    },
    {
      "epoch": 0.09045412418906394,
      "grad_norm": 0.7065533399581909,
      "learning_rate": 4.3550039173489845e-05,
      "loss": 1.6971,
      "step": 122
    },
    {
      "epoch": 0.0911955514365153,
      "grad_norm": 0.7156745195388794,
      "learning_rate": 4.2634915097268115e-05,
      "loss": 1.6511,
      "step": 123
    },
    {
      "epoch": 0.09193697868396664,
      "grad_norm": 0.7486314177513123,
      "learning_rate": 4.1722306190793495e-05,
      "loss": 1.7122,
      "step": 124
    },
    {
      "epoch": 0.09267840593141798,
      "grad_norm": 0.77154940366745,
      "learning_rate": 4.0812524109171476e-05,
      "loss": 1.6197,
      "step": 125
    },
    {
      "epoch": 0.09341983317886933,
      "grad_norm": 0.8462448120117188,
      "learning_rate": 3.99058795421495e-05,
      "loss": 1.6343,
      "step": 126
    },
    {
      "epoch": 0.09416126042632067,
      "grad_norm": 0.8176885843276978,
      "learning_rate": 3.9002682108016585e-05,
      "loss": 1.595,
      "step": 127
    },
    {
      "epoch": 0.09490268767377201,
      "grad_norm": 0.9120206236839294,
      "learning_rate": 3.8103240247869075e-05,
      "loss": 1.6338,
      "step": 128
    },
    {
      "epoch": 0.09564411492122335,
      "grad_norm": 1.0192011594772339,
      "learning_rate": 3.720786112027822e-05,
      "loss": 1.7273,
      "step": 129
    },
    {
      "epoch": 0.0963855421686747,
      "grad_norm": 1.0697766542434692,
      "learning_rate": 3.631685049639586e-05,
      "loss": 1.74,
      "step": 130
    },
    {
      "epoch": 0.09712696941612604,
      "grad_norm": 1.1573799848556519,
      "learning_rate": 3.543051265553377e-05,
      "loss": 1.6338,
      "step": 131
    },
    {
      "epoch": 0.09786839666357738,
      "grad_norm": 1.1837265491485596,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.6193,
      "step": 132
    },
    {
      "epoch": 0.09860982391102872,
      "grad_norm": 1.4880319833755493,
      "learning_rate": 3.367306435799584e-05,
      "loss": 1.7338,
      "step": 133
    },
    {
      "epoch": 0.09935125115848008,
      "grad_norm": 1.383131742477417,
      "learning_rate": 3.2802554068303596e-05,
      "loss": 1.7442,
      "step": 134
    },
    {
      "epoch": 0.10009267840593142,
      "grad_norm": 1.4564555883407593,
      "learning_rate": 3.1937916690642356e-05,
      "loss": 1.7425,
      "step": 135
    },
    {
      "epoch": 0.10083410565338276,
      "grad_norm": 1.4855623245239258,
      "learning_rate": 3.107944749788449e-05,
      "loss": 1.8507,
      "step": 136
    },
    {
      "epoch": 0.10157553290083411,
      "grad_norm": 1.6391143798828125,
      "learning_rate": 3.0227439656472877e-05,
      "loss": 1.6847,
      "step": 137
    },
    {
      "epoch": 0.10231696014828545,
      "grad_norm": 1.664311408996582,
      "learning_rate": 2.9382184126304834e-05,
      "loss": 1.9334,
      "step": 138
    },
    {
      "epoch": 0.10305838739573679,
      "grad_norm": 1.4228198528289795,
      "learning_rate": 2.8543969561369556e-05,
      "loss": 1.4721,
      "step": 139
    },
    {
      "epoch": 0.10379981464318813,
      "grad_norm": 1.847296118736267,
      "learning_rate": 2.771308221117309e-05,
      "loss": 1.8911,
      "step": 140
    },
    {
      "epoch": 0.10454124189063949,
      "grad_norm": 1.9254475831985474,
      "learning_rate": 2.688980582298435e-05,
      "loss": 2.0965,
      "step": 141
    },
    {
      "epoch": 0.10528266913809083,
      "grad_norm": 1.9739066362380981,
      "learning_rate": 2.607442154493568e-05,
      "loss": 1.7287,
      "step": 142
    },
    {
      "epoch": 0.10602409638554217,
      "grad_norm": 2.1684720516204834,
      "learning_rate": 2.5267207830011068e-05,
      "loss": 1.556,
      "step": 143
    },
    {
      "epoch": 0.1067655236329935,
      "grad_norm": 2.355837345123291,
      "learning_rate": 2.446844034095466e-05,
      "loss": 1.9661,
      "step": 144
    },
    {
      "epoch": 0.10750695088044486,
      "grad_norm": 2.715618371963501,
      "learning_rate": 2.3678391856132204e-05,
      "loss": 2.0255,
      "step": 145
    },
    {
      "epoch": 0.1082483781278962,
      "grad_norm": 2.5407278537750244,
      "learning_rate": 2.2897332176377528e-05,
      "loss": 1.7645,
      "step": 146
    },
    {
      "epoch": 0.10898980537534754,
      "grad_norm": 2.632134437561035,
      "learning_rate": 2.2125528032855724e-05,
      "loss": 1.9769,
      "step": 147
    },
    {
      "epoch": 0.1097312326227989,
      "grad_norm": 2.989884376525879,
      "learning_rate": 2.136324299597474e-05,
      "loss": 2.0118,
      "step": 148
    },
    {
      "epoch": 0.11047265987025023,
      "grad_norm": 2.9037551879882812,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.8668,
      "step": 149
    },
    {
      "epoch": 0.11121408711770157,
      "grad_norm": 4.2353739738464355,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 2.3665,
      "step": 150
    },
    {
      "epoch": 0.11121408711770157,
      "eval_loss": 1.7685366868972778,
      "eval_runtime": 80.2767,
      "eval_samples_per_second": 28.302,
      "eval_steps_per_second": 14.151,
      "step": 150
    },
    {
      "epoch": 0.11195551436515291,
      "grad_norm": 0.47507578134536743,
      "learning_rate": 1.9136088935510362e-05,
      "loss": 1.651,
      "step": 151
    },
    {
      "epoch": 0.11269694161260427,
      "grad_norm": 0.46119046211242676,
      "learning_rate": 1.8414449687337464e-05,
      "loss": 1.6019,
      "step": 152
    },
    {
      "epoch": 0.11343836886005561,
      "grad_norm": 0.46028992533683777,
      "learning_rate": 1.7703596875660645e-05,
      "loss": 1.5145,
      "step": 153
    },
    {
      "epoch": 0.11417979610750695,
      "grad_norm": 0.48683980107307434,
      "learning_rate": 1.700377325606388e-05,
      "loss": 1.6048,
      "step": 154
    },
    {
      "epoch": 0.11492122335495829,
      "grad_norm": 0.4676201343536377,
      "learning_rate": 1.631521781767214e-05,
      "loss": 1.514,
      "step": 155
    },
    {
      "epoch": 0.11566265060240964,
      "grad_norm": 0.5439676642417908,
      "learning_rate": 1.5638165701536868e-05,
      "loss": 1.6788,
      "step": 156
    },
    {
      "epoch": 0.11640407784986098,
      "grad_norm": 0.4836425185203552,
      "learning_rate": 1.4972848120335453e-05,
      "loss": 1.5205,
      "step": 157
    },
    {
      "epoch": 0.11714550509731232,
      "grad_norm": 0.5258874893188477,
      "learning_rate": 1.4319492279412388e-05,
      "loss": 1.6609,
      "step": 158
    },
    {
      "epoch": 0.11788693234476368,
      "grad_norm": 0.5179567337036133,
      "learning_rate": 1.3678321299188801e-05,
      "loss": 1.6209,
      "step": 159
    },
    {
      "epoch": 0.11862835959221502,
      "grad_norm": 0.5643657445907593,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 1.6721,
      "step": 160
    },
    {
      "epoch": 0.11936978683966636,
      "grad_norm": 0.56312096118927,
      "learning_rate": 1.2433405522156332e-05,
      "loss": 1.6607,
      "step": 161
    },
    {
      "epoch": 0.1201112140871177,
      "grad_norm": 0.5756739974021912,
      "learning_rate": 1.183008586294485e-05,
      "loss": 1.6286,
      "step": 162
    },
    {
      "epoch": 0.12085264133456905,
      "grad_norm": 0.5807479023933411,
      "learning_rate": 1.1239801194443506e-05,
      "loss": 1.6405,
      "step": 163
    },
    {
      "epoch": 0.12159406858202039,
      "grad_norm": 0.6560777425765991,
      "learning_rate": 1.066275309832584e-05,
      "loss": 1.8103,
      "step": 164
    },
    {
      "epoch": 0.12233549582947173,
      "grad_norm": 0.6384656429290771,
      "learning_rate": 1.0099138635988026e-05,
      "loss": 1.8201,
      "step": 165
    },
    {
      "epoch": 0.12307692307692308,
      "grad_norm": 0.6288926005363464,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.6129,
      "step": 166
    },
    {
      "epoch": 0.12381835032437442,
      "grad_norm": 0.698594331741333,
      "learning_rate": 9.012975854638949e-06,
      "loss": 1.6471,
      "step": 167
    },
    {
      "epoch": 0.12455977757182576,
      "grad_norm": 0.7008000016212463,
      "learning_rate": 8.490798459222476e-06,
      "loss": 1.6478,
      "step": 168
    },
    {
      "epoch": 0.12530120481927712,
      "grad_norm": 0.7610568404197693,
      "learning_rate": 7.982796418105371e-06,
      "loss": 1.752,
      "step": 169
    },
    {
      "epoch": 0.12604263206672844,
      "grad_norm": 0.7720162868499756,
      "learning_rate": 7.489143213519301e-06,
      "loss": 1.7412,
      "step": 170
    },
    {
      "epoch": 0.1267840593141798,
      "grad_norm": 0.7579473853111267,
      "learning_rate": 7.010007427581378e-06,
      "loss": 1.6226,
      "step": 171
    },
    {
      "epoch": 0.12752548656163115,
      "grad_norm": 0.79985511302948,
      "learning_rate": 6.5455526847235825e-06,
      "loss": 1.7923,
      "step": 172
    },
    {
      "epoch": 0.12826691380908248,
      "grad_norm": 0.7975371479988098,
      "learning_rate": 6.0959375958151045e-06,
      "loss": 1.7456,
      "step": 173
    },
    {
      "epoch": 0.12900834105653383,
      "grad_norm": 0.8902814984321594,
      "learning_rate": 5.6613157039969055e-06,
      "loss": 1.6546,
      "step": 174
    },
    {
      "epoch": 0.12974976830398516,
      "grad_norm": 0.9001719355583191,
      "learning_rate": 5.241835432246889e-06,
      "loss": 1.7422,
      "step": 175
    },
    {
      "epoch": 0.1304911955514365,
      "grad_norm": 0.914344310760498,
      "learning_rate": 4.837640032693558e-06,
      "loss": 1.6568,
      "step": 176
    },
    {
      "epoch": 0.13123262279888787,
      "grad_norm": 0.9830794930458069,
      "learning_rate": 4.448867537695578e-06,
      "loss": 1.6425,
      "step": 177
    },
    {
      "epoch": 0.1319740500463392,
      "grad_norm": 1.0287317037582397,
      "learning_rate": 4.075650712703849e-06,
      "loss": 1.6279,
      "step": 178
    },
    {
      "epoch": 0.13271547729379055,
      "grad_norm": 1.1703566312789917,
      "learning_rate": 3.71811701092219e-06,
      "loss": 1.7118,
      "step": 179
    },
    {
      "epoch": 0.1334569045412419,
      "grad_norm": 1.1219487190246582,
      "learning_rate": 3.376388529782215e-06,
      "loss": 1.8348,
      "step": 180
    },
    {
      "epoch": 0.13419833178869323,
      "grad_norm": 1.178866982460022,
      "learning_rate": 3.0505819692471792e-06,
      "loss": 1.8233,
      "step": 181
    },
    {
      "epoch": 0.13493975903614458,
      "grad_norm": 1.215016484260559,
      "learning_rate": 2.7408085919590264e-06,
      "loss": 1.5181,
      "step": 182
    },
    {
      "epoch": 0.13568118628359593,
      "grad_norm": 1.243989109992981,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.884,
      "step": 183
    },
    {
      "epoch": 0.13642261353104726,
      "grad_norm": 1.4585505723953247,
      "learning_rate": 2.1697790249779636e-06,
      "loss": 1.7382,
      "step": 184
    },
    {
      "epoch": 0.1371640407784986,
      "grad_norm": 1.507137656211853,
      "learning_rate": 1.908717841359048e-06,
      "loss": 1.8559,
      "step": 185
    },
    {
      "epoch": 0.13790546802594994,
      "grad_norm": 1.4388847351074219,
      "learning_rate": 1.6640797865406288e-06,
      "loss": 1.6133,
      "step": 186
    },
    {
      "epoch": 0.1386468952734013,
      "grad_norm": 1.6415081024169922,
      "learning_rate": 1.4359484041943038e-06,
      "loss": 1.7556,
      "step": 187
    },
    {
      "epoch": 0.13938832252085265,
      "grad_norm": 1.722662329673767,
      "learning_rate": 1.2244016009781701e-06,
      "loss": 1.7759,
      "step": 188
    },
    {
      "epoch": 0.14012974976830397,
      "grad_norm": 1.832926630973816,
      "learning_rate": 1.0295116199317057e-06,
      "loss": 1.8388,
      "step": 189
    },
    {
      "epoch": 0.14087117701575533,
      "grad_norm": 1.967829704284668,
      "learning_rate": 8.513450158049108e-07,
      "loss": 1.8099,
      "step": 190
    },
    {
      "epoch": 0.14161260426320668,
      "grad_norm": 1.9063888788223267,
      "learning_rate": 6.899626323298713e-07,
      "loss": 1.9254,
      "step": 191
    },
    {
      "epoch": 0.142354031510658,
      "grad_norm": 2.3432540893554688,
      "learning_rate": 5.454195814427021e-07,
      "loss": 1.9337,
      "step": 192
    },
    {
      "epoch": 0.14309545875810936,
      "grad_norm": 2.5970942974090576,
      "learning_rate": 4.177652244628627e-07,
      "loss": 1.7244,
      "step": 193
    },
    {
      "epoch": 0.14383688600556072,
      "grad_norm": 2.4058358669281006,
      "learning_rate": 3.0704315523631953e-07,
      "loss": 2.1262,
      "step": 194
    },
    {
      "epoch": 0.14457831325301204,
      "grad_norm": 2.3678603172302246,
      "learning_rate": 2.1329118524827662e-07,
      "loss": 1.8319,
      "step": 195
    },
    {
      "epoch": 0.1453197405004634,
      "grad_norm": 2.7229273319244385,
      "learning_rate": 1.3654133071059893e-07,
      "loss": 1.8925,
      "step": 196
    },
    {
      "epoch": 0.14606116774791475,
      "grad_norm": 2.4889092445373535,
      "learning_rate": 7.681980162830282e-08,
      "loss": 1.841,
      "step": 197
    },
    {
      "epoch": 0.14680259499536608,
      "grad_norm": 5.052427768707275,
      "learning_rate": 3.4146992848854695e-08,
      "loss": 1.9927,
      "step": 198
    },
    {
      "epoch": 0.14754402224281743,
      "grad_norm": 6.398392200469971,
      "learning_rate": 8.537477097364522e-09,
      "loss": 2.1303,
      "step": 199
    },
    {
      "epoch": 0.14828544949026876,
      "grad_norm": 5.143312931060791,
      "learning_rate": 0.0,
      "loss": 2.7409,
      "step": 200
    },
    {
      "epoch": 0.14828544949026876,
      "eval_loss": 1.7381101846694946,
      "eval_runtime": 80.3286,
      "eval_samples_per_second": 28.284,
      "eval_steps_per_second": 14.142,
      "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": 5.4428946137088e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}