{
  "best_metric": 10.745776176452637,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.043215211754537596,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000216076058772688,
      "grad_norm": 0.200728178024292,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 10.8463,
      "step": 1
    },
    {
      "epoch": 0.000216076058772688,
      "eval_loss": 10.84037971496582,
      "eval_runtime": 27.2248,
      "eval_samples_per_second": 286.32,
      "eval_steps_per_second": 143.178,
      "step": 1
    },
    {
      "epoch": 0.000432152117545376,
      "grad_norm": 0.23658832907676697,
      "learning_rate": 6.666666666666667e-06,
      "loss": 10.8472,
      "step": 2
    },
    {
      "epoch": 0.0006482281763180639,
      "grad_norm": 0.20040327310562134,
      "learning_rate": 1e-05,
      "loss": 10.8415,
      "step": 3
    },
    {
      "epoch": 0.000864304235090752,
      "grad_norm": 0.2336120903491974,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 10.828,
      "step": 4
    },
    {
      "epoch": 0.00108038029386344,
      "grad_norm": 0.20422425866127014,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 10.8437,
      "step": 5
    },
    {
      "epoch": 0.0012964563526361278,
      "grad_norm": 0.2417161911725998,
      "learning_rate": 2e-05,
      "loss": 10.8422,
      "step": 6
    },
    {
      "epoch": 0.001512532411408816,
      "grad_norm": 0.22463224828243256,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 10.8167,
      "step": 7
    },
    {
      "epoch": 0.001728608470181504,
      "grad_norm": 0.22320792078971863,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 10.8394,
      "step": 8
    },
    {
      "epoch": 0.0019446845289541918,
      "grad_norm": 0.22033357620239258,
      "learning_rate": 3e-05,
      "loss": 10.8351,
      "step": 9
    },
    {
      "epoch": 0.00216076058772688,
      "grad_norm": 0.20691785216331482,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 10.8433,
      "step": 10
    },
    {
      "epoch": 0.0023768366464995676,
      "grad_norm": 0.21809300780296326,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 10.8364,
      "step": 11
    },
    {
      "epoch": 0.0025929127052722557,
      "grad_norm": 0.2736707031726837,
      "learning_rate": 4e-05,
      "loss": 10.8364,
      "step": 12
    },
    {
      "epoch": 0.0028089887640449437,
      "grad_norm": 0.24762022495269775,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 10.8266,
      "step": 13
    },
    {
      "epoch": 0.003025064822817632,
      "grad_norm": 0.24276190996170044,
      "learning_rate": 4.666666666666667e-05,
      "loss": 10.8375,
      "step": 14
    },
    {
      "epoch": 0.00324114088159032,
      "grad_norm": 0.23069171607494354,
      "learning_rate": 5e-05,
      "loss": 10.8393,
      "step": 15
    },
    {
      "epoch": 0.003457216940363008,
      "grad_norm": 0.24690602719783783,
      "learning_rate": 5.333333333333333e-05,
      "loss": 10.8348,
      "step": 16
    },
    {
      "epoch": 0.003673292999135696,
      "grad_norm": 0.23763756453990936,
      "learning_rate": 5.666666666666667e-05,
      "loss": 10.8294,
      "step": 17
    },
    {
      "epoch": 0.0038893690579083835,
      "grad_norm": 0.23151656985282898,
      "learning_rate": 6e-05,
      "loss": 10.8329,
      "step": 18
    },
    {
      "epoch": 0.004105445116681072,
      "grad_norm": 0.23155605792999268,
      "learning_rate": 6.333333333333333e-05,
      "loss": 10.8299,
      "step": 19
    },
    {
      "epoch": 0.00432152117545376,
      "grad_norm": 0.2375253140926361,
      "learning_rate": 6.666666666666667e-05,
      "loss": 10.8228,
      "step": 20
    },
    {
      "epoch": 0.004537597234226448,
      "grad_norm": 0.31918394565582275,
      "learning_rate": 7e-05,
      "loss": 10.8364,
      "step": 21
    },
    {
      "epoch": 0.004753673292999135,
      "grad_norm": 0.2590275704860687,
      "learning_rate": 7.333333333333333e-05,
      "loss": 10.8279,
      "step": 22
    },
    {
      "epoch": 0.004969749351771823,
      "grad_norm": 0.3064877986907959,
      "learning_rate": 7.666666666666667e-05,
      "loss": 10.8388,
      "step": 23
    },
    {
      "epoch": 0.005185825410544511,
      "grad_norm": 0.3065643310546875,
      "learning_rate": 8e-05,
      "loss": 10.8247,
      "step": 24
    },
    {
      "epoch": 0.0054019014693171994,
      "grad_norm": 0.2851259708404541,
      "learning_rate": 8.333333333333334e-05,
      "loss": 10.8244,
      "step": 25
    },
    {
      "epoch": 0.0056179775280898875,
      "grad_norm": 0.30654624104499817,
      "learning_rate": 8.666666666666667e-05,
      "loss": 10.8216,
      "step": 26
    },
    {
      "epoch": 0.0058340535868625755,
      "grad_norm": 0.2787514328956604,
      "learning_rate": 9e-05,
      "loss": 10.8265,
      "step": 27
    },
    {
      "epoch": 0.006050129645635264,
      "grad_norm": 0.314116895198822,
      "learning_rate": 9.333333333333334e-05,
      "loss": 10.8232,
      "step": 28
    },
    {
      "epoch": 0.006266205704407952,
      "grad_norm": 0.28672200441360474,
      "learning_rate": 9.666666666666667e-05,
      "loss": 10.8145,
      "step": 29
    },
    {
      "epoch": 0.00648228176318064,
      "grad_norm": 0.27016451954841614,
      "learning_rate": 0.0001,
      "loss": 10.8084,
      "step": 30
    },
    {
      "epoch": 0.006698357821953328,
      "grad_norm": 0.2747277617454529,
      "learning_rate": 9.999146252290264e-05,
      "loss": 10.81,
      "step": 31
    },
    {
      "epoch": 0.006914433880726016,
      "grad_norm": 0.29148223996162415,
      "learning_rate": 9.996585300715116e-05,
      "loss": 10.813,
      "step": 32
    },
    {
      "epoch": 0.007130509939498704,
      "grad_norm": 0.3096551299095154,
      "learning_rate": 9.99231801983717e-05,
      "loss": 10.802,
      "step": 33
    },
    {
      "epoch": 0.007346585998271392,
      "grad_norm": 0.3342970609664917,
      "learning_rate": 9.986345866928941e-05,
      "loss": 10.8117,
      "step": 34
    },
    {
      "epoch": 0.00756266205704408,
      "grad_norm": 0.32004502415657043,
      "learning_rate": 9.978670881475172e-05,
      "loss": 10.7982,
      "step": 35
    },
    {
      "epoch": 0.007778738115816767,
      "grad_norm": 0.3316425383090973,
      "learning_rate": 9.96929568447637e-05,
      "loss": 10.8037,
      "step": 36
    },
    {
      "epoch": 0.007994814174589455,
      "grad_norm": 0.32472503185272217,
      "learning_rate": 9.958223477553714e-05,
      "loss": 10.7961,
      "step": 37
    },
    {
      "epoch": 0.008210890233362144,
      "grad_norm": 0.3180193603038788,
      "learning_rate": 9.94545804185573e-05,
      "loss": 10.7925,
      "step": 38
    },
    {
      "epoch": 0.008426966292134831,
      "grad_norm": 0.34710487723350525,
      "learning_rate": 9.931003736767013e-05,
      "loss": 10.7994,
      "step": 39
    },
    {
      "epoch": 0.00864304235090752,
      "grad_norm": 0.34599560499191284,
      "learning_rate": 9.91486549841951e-05,
      "loss": 10.7997,
      "step": 40
    },
    {
      "epoch": 0.008859118409680207,
      "grad_norm": 0.4090355634689331,
      "learning_rate": 9.89704883800683e-05,
      "loss": 10.7918,
      "step": 41
    },
    {
      "epoch": 0.009075194468452896,
      "grad_norm": 0.38032764196395874,
      "learning_rate": 9.877559839902184e-05,
      "loss": 10.7877,
      "step": 42
    },
    {
      "epoch": 0.009291270527225583,
      "grad_norm": 0.3663756847381592,
      "learning_rate": 9.85640515958057e-05,
      "loss": 10.7845,
      "step": 43
    },
    {
      "epoch": 0.00950734658599827,
      "grad_norm": 0.38913989067077637,
      "learning_rate": 9.833592021345937e-05,
      "loss": 10.7775,
      "step": 44
    },
    {
      "epoch": 0.00972342264477096,
      "grad_norm": 0.4066934585571289,
      "learning_rate": 9.809128215864097e-05,
      "loss": 10.7805,
      "step": 45
    },
    {
      "epoch": 0.009939498703543647,
      "grad_norm": 0.40702107548713684,
      "learning_rate": 9.783022097502204e-05,
      "loss": 10.766,
      "step": 46
    },
    {
      "epoch": 0.010155574762316336,
      "grad_norm": 0.5095567107200623,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.7686,
      "step": 47
    },
    {
      "epoch": 0.010371650821089023,
      "grad_norm": 0.5398594737052917,
      "learning_rate": 9.725919140804099e-05,
      "loss": 10.7674,
      "step": 48
    },
    {
      "epoch": 0.010587726879861712,
      "grad_norm": 0.5338243246078491,
      "learning_rate": 9.694941803075283e-05,
      "loss": 10.7631,
      "step": 49
    },
    {
      "epoch": 0.010803802938634399,
      "grad_norm": 0.7691417932510376,
      "learning_rate": 9.662361147021779e-05,
      "loss": 10.7645,
      "step": 50
    },
    {
      "epoch": 0.010803802938634399,
      "eval_loss": 10.766061782836914,
      "eval_runtime": 27.2467,
      "eval_samples_per_second": 286.089,
      "eval_steps_per_second": 143.063,
      "step": 50
    },
    {
      "epoch": 0.011019878997407088,
      "grad_norm": 0.31955963373184204,
      "learning_rate": 9.628188298907782e-05,
      "loss": 10.7833,
      "step": 51
    },
    {
      "epoch": 0.011235955056179775,
      "grad_norm": 0.2600715458393097,
      "learning_rate": 9.592434928729616e-05,
      "loss": 10.7863,
      "step": 52
    },
    {
      "epoch": 0.011452031114952464,
      "grad_norm": 0.25826898217201233,
      "learning_rate": 9.555113246230442e-05,
      "loss": 10.7734,
      "step": 53
    },
    {
      "epoch": 0.011668107173725151,
      "grad_norm": 0.33920153975486755,
      "learning_rate": 9.516235996730645e-05,
      "loss": 10.7821,
      "step": 54
    },
    {
      "epoch": 0.01188418323249784,
      "grad_norm": 0.2760711908340454,
      "learning_rate": 9.475816456775313e-05,
      "loss": 10.7771,
      "step": 55
    },
    {
      "epoch": 0.012100259291270527,
      "grad_norm": 0.2651015818119049,
      "learning_rate": 9.43386842960031e-05,
      "loss": 10.7812,
      "step": 56
    },
    {
      "epoch": 0.012316335350043216,
      "grad_norm": 0.3452089726924896,
      "learning_rate": 9.39040624041849e-05,
      "loss": 10.7577,
      "step": 57
    },
    {
      "epoch": 0.012532411408815903,
      "grad_norm": 0.24794746935367584,
      "learning_rate": 9.345444731527642e-05,
      "loss": 10.7618,
      "step": 58
    },
    {
      "epoch": 0.01274848746758859,
      "grad_norm": 0.26710182428359985,
      "learning_rate": 9.298999257241863e-05,
      "loss": 10.7622,
      "step": 59
    },
    {
      "epoch": 0.01296456352636128,
      "grad_norm": 0.2514508068561554,
      "learning_rate": 9.251085678648072e-05,
      "loss": 10.7781,
      "step": 60
    },
    {
      "epoch": 0.013180639585133967,
      "grad_norm": 0.2650161385536194,
      "learning_rate": 9.201720358189464e-05,
      "loss": 10.7779,
      "step": 61
    },
    {
      "epoch": 0.013396715643906655,
      "grad_norm": 0.2767448127269745,
      "learning_rate": 9.150920154077754e-05,
      "loss": 10.7757,
      "step": 62
    },
    {
      "epoch": 0.013612791702679343,
      "grad_norm": 0.29482796788215637,
      "learning_rate": 9.098702414536107e-05,
      "loss": 10.771,
      "step": 63
    },
    {
      "epoch": 0.013828867761452032,
      "grad_norm": 0.274749755859375,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.7587,
      "step": 64
    },
    {
      "epoch": 0.014044943820224719,
      "grad_norm": 0.21438376605510712,
      "learning_rate": 8.9900861364012e-05,
      "loss": 10.773,
      "step": 65
    },
    {
      "epoch": 0.014261019878997408,
      "grad_norm": 0.2738189995288849,
      "learning_rate": 8.933724690167417e-05,
      "loss": 10.7707,
      "step": 66
    },
    {
      "epoch": 0.014477095937770095,
      "grad_norm": 0.2180059552192688,
      "learning_rate": 8.876019880555649e-05,
      "loss": 10.7649,
      "step": 67
    },
    {
      "epoch": 0.014693171996542784,
      "grad_norm": 0.2755647301673889,
      "learning_rate": 8.816991413705516e-05,
      "loss": 10.7556,
      "step": 68
    },
    {
      "epoch": 0.014909248055315471,
      "grad_norm": 0.2628243863582611,
      "learning_rate": 8.756659447784368e-05,
      "loss": 10.7485,
      "step": 69
    },
    {
      "epoch": 0.01512532411408816,
      "grad_norm": 0.27653780579566956,
      "learning_rate": 8.695044586103296e-05,
      "loss": 10.7742,
      "step": 70
    },
    {
      "epoch": 0.015341400172860847,
      "grad_norm": 0.292253315448761,
      "learning_rate": 8.632167870081121e-05,
      "loss": 10.7542,
      "step": 71
    },
    {
      "epoch": 0.015557476231633534,
      "grad_norm": 0.21869610249996185,
      "learning_rate": 8.568050772058762e-05,
      "loss": 10.7724,
      "step": 72
    },
    {
      "epoch": 0.015773552290406223,
      "grad_norm": 0.20132863521575928,
      "learning_rate": 8.502715187966455e-05,
      "loss": 10.7621,
      "step": 73
    },
    {
      "epoch": 0.01598962834917891,
      "grad_norm": 0.2435745745897293,
      "learning_rate": 8.436183429846313e-05,
      "loss": 10.7661,
      "step": 74
    },
    {
      "epoch": 0.016205704407951597,
      "grad_norm": 0.23909196257591248,
      "learning_rate": 8.368478218232787e-05,
      "loss": 10.7694,
      "step": 75
    },
    {
      "epoch": 0.016421780466724288,
      "grad_norm": 0.22482334077358246,
      "learning_rate": 8.299622674393614e-05,
      "loss": 10.7642,
      "step": 76
    },
    {
      "epoch": 0.016637856525496975,
      "grad_norm": 0.2314310371875763,
      "learning_rate": 8.229640312433937e-05,
      "loss": 10.7584,
      "step": 77
    },
    {
      "epoch": 0.016853932584269662,
      "grad_norm": 0.23524075746536255,
      "learning_rate": 8.158555031266254e-05,
      "loss": 10.7632,
      "step": 78
    },
    {
      "epoch": 0.01707000864304235,
      "grad_norm": 0.20763476192951202,
      "learning_rate": 8.086391106448965e-05,
      "loss": 10.7533,
      "step": 79
    },
    {
      "epoch": 0.01728608470181504,
      "grad_norm": 0.1953965127468109,
      "learning_rate": 8.013173181896283e-05,
      "loss": 10.7633,
      "step": 80
    },
    {
      "epoch": 0.017502160760587727,
      "grad_norm": 0.1644563525915146,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.764,
      "step": 81
    },
    {
      "epoch": 0.017718236819360415,
      "grad_norm": 0.2148687094449997,
      "learning_rate": 7.863675700402526e-05,
      "loss": 10.7566,
      "step": 82
    },
    {
      "epoch": 0.017934312878133102,
      "grad_norm": 0.2654571831226349,
      "learning_rate": 7.787447196714427e-05,
      "loss": 10.7444,
      "step": 83
    },
    {
      "epoch": 0.018150388936905792,
      "grad_norm": 0.231702521443367,
      "learning_rate": 7.710266782362247e-05,
      "loss": 10.753,
      "step": 84
    },
    {
      "epoch": 0.01836646499567848,
      "grad_norm": 0.26304373145103455,
      "learning_rate": 7.63216081438678e-05,
      "loss": 10.7559,
      "step": 85
    },
    {
      "epoch": 0.018582541054451167,
      "grad_norm": 0.19173169136047363,
      "learning_rate": 7.553155965904535e-05,
      "loss": 10.7675,
      "step": 86
    },
    {
      "epoch": 0.018798617113223854,
      "grad_norm": 0.2556131184101105,
      "learning_rate": 7.473279216998895e-05,
      "loss": 10.7589,
      "step": 87
    },
    {
      "epoch": 0.01901469317199654,
      "grad_norm": 0.22893981635570526,
      "learning_rate": 7.392557845506432e-05,
      "loss": 10.7599,
      "step": 88
    },
    {
      "epoch": 0.019230769230769232,
      "grad_norm": 0.21988166868686676,
      "learning_rate": 7.311019417701566e-05,
      "loss": 10.7564,
      "step": 89
    },
    {
      "epoch": 0.01944684528954192,
      "grad_norm": 0.24298734962940216,
      "learning_rate": 7.228691778882693e-05,
      "loss": 10.7532,
      "step": 90
    },
    {
      "epoch": 0.019662921348314606,
      "grad_norm": 0.21422868967056274,
      "learning_rate": 7.145603043863045e-05,
      "loss": 10.7588,
      "step": 91
    },
    {
      "epoch": 0.019878997407087293,
      "grad_norm": 0.18029530346393585,
      "learning_rate": 7.061781587369519e-05,
      "loss": 10.7508,
      "step": 92
    },
    {
      "epoch": 0.020095073465859984,
      "grad_norm": 0.29323095083236694,
      "learning_rate": 6.977256034352712e-05,
      "loss": 10.7534,
      "step": 93
    },
    {
      "epoch": 0.02031114952463267,
      "grad_norm": 0.24313682317733765,
      "learning_rate": 6.892055250211552e-05,
      "loss": 10.753,
      "step": 94
    },
    {
      "epoch": 0.02052722558340536,
      "grad_norm": 0.30919376015663147,
      "learning_rate": 6.806208330935766e-05,
      "loss": 10.7502,
      "step": 95
    },
    {
      "epoch": 0.020743301642178046,
      "grad_norm": 0.3091332018375397,
      "learning_rate": 6.719744593169641e-05,
      "loss": 10.7492,
      "step": 96
    },
    {
      "epoch": 0.020959377700950736,
      "grad_norm": 0.31597936153411865,
      "learning_rate": 6.632693564200416e-05,
      "loss": 10.7353,
      "step": 97
    },
    {
      "epoch": 0.021175453759723423,
      "grad_norm": 0.41740843653678894,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.753,
      "step": 98
    },
    {
      "epoch": 0.02139152981849611,
      "grad_norm": 0.3725014328956604,
      "learning_rate": 6.456948734446624e-05,
      "loss": 10.72,
      "step": 99
    },
    {
      "epoch": 0.021607605877268798,
      "grad_norm": 0.6240119934082031,
      "learning_rate": 6.368314950360415e-05,
      "loss": 10.7379,
      "step": 100
    },
    {
      "epoch": 0.021607605877268798,
      "eval_loss": 10.751439094543457,
      "eval_runtime": 27.1534,
      "eval_samples_per_second": 287.073,
      "eval_steps_per_second": 143.555,
      "step": 100
    },
    {
      "epoch": 0.021823681936041485,
      "grad_norm": 0.21080626547336578,
      "learning_rate": 6.279213887972179e-05,
      "loss": 10.7706,
      "step": 101
    },
    {
      "epoch": 0.022039757994814176,
      "grad_norm": 0.20775455236434937,
      "learning_rate": 6.189675975213094e-05,
      "loss": 10.7617,
      "step": 102
    },
    {
      "epoch": 0.022255834053586863,
      "grad_norm": 0.1883164793252945,
      "learning_rate": 6.099731789198344e-05,
      "loss": 10.7622,
      "step": 103
    },
    {
      "epoch": 0.02247191011235955,
      "grad_norm": 0.22989457845687866,
      "learning_rate": 6.009412045785051e-05,
      "loss": 10.767,
      "step": 104
    },
    {
      "epoch": 0.022687986171132237,
      "grad_norm": 0.21413500607013702,
      "learning_rate": 5.918747589082853e-05,
      "loss": 10.7754,
      "step": 105
    },
    {
      "epoch": 0.022904062229904928,
      "grad_norm": 0.20524542033672333,
      "learning_rate": 5.82776938092065e-05,
      "loss": 10.7728,
      "step": 106
    },
    {
      "epoch": 0.023120138288677615,
      "grad_norm": 0.276962012052536,
      "learning_rate": 5.736508490273188e-05,
      "loss": 10.7504,
      "step": 107
    },
    {
      "epoch": 0.023336214347450302,
      "grad_norm": 0.2452148199081421,
      "learning_rate": 5.644996082651017e-05,
      "loss": 10.7616,
      "step": 108
    },
    {
      "epoch": 0.02355229040622299,
      "grad_norm": 0.25284409523010254,
      "learning_rate": 5.553263409457504e-05,
      "loss": 10.7604,
      "step": 109
    },
    {
      "epoch": 0.02376836646499568,
      "grad_norm": 0.247440367937088,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 10.7641,
      "step": 110
    },
    {
      "epoch": 0.023984442523768367,
      "grad_norm": 0.18067599833011627,
      "learning_rate": 5.3692626373743706e-05,
      "loss": 10.7746,
      "step": 111
    },
    {
      "epoch": 0.024200518582541054,
      "grad_norm": 0.19137042760849,
      "learning_rate": 5.27705737457985e-05,
      "loss": 10.7714,
      "step": 112
    },
    {
      "epoch": 0.02441659464131374,
      "grad_norm": 0.2149994820356369,
      "learning_rate": 5.184757496945726e-05,
      "loss": 10.7733,
      "step": 113
    },
    {
      "epoch": 0.024632670700086432,
      "grad_norm": 0.20906798541545868,
      "learning_rate": 5.092394524795649e-05,
      "loss": 10.7726,
      "step": 114
    },
    {
      "epoch": 0.02484874675885912,
      "grad_norm": 0.24666734039783478,
      "learning_rate": 5e-05,
      "loss": 10.7536,
      "step": 115
    },
    {
      "epoch": 0.025064822817631807,
      "grad_norm": 0.23656940460205078,
      "learning_rate": 4.907605475204352e-05,
      "loss": 10.7602,
      "step": 116
    },
    {
      "epoch": 0.025280898876404494,
      "grad_norm": 0.25157082080841064,
      "learning_rate": 4.8152425030542766e-05,
      "loss": 10.7619,
      "step": 117
    },
    {
      "epoch": 0.02549697493517718,
      "grad_norm": 0.2460283637046814,
      "learning_rate": 4.72294262542015e-05,
      "loss": 10.7613,
      "step": 118
    },
    {
      "epoch": 0.02571305099394987,
      "grad_norm": 0.1980646699666977,
      "learning_rate": 4.6307373626256306e-05,
      "loss": 10.7639,
      "step": 119
    },
    {
      "epoch": 0.02592912705272256,
      "grad_norm": 0.20079004764556885,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 10.7587,
      "step": 120
    },
    {
      "epoch": 0.026145203111495246,
      "grad_norm": 0.22045651078224182,
      "learning_rate": 4.446736590542497e-05,
      "loss": 10.7678,
      "step": 121
    },
    {
      "epoch": 0.026361279170267933,
      "grad_norm": 0.22736740112304688,
      "learning_rate": 4.3550039173489845e-05,
      "loss": 10.7542,
      "step": 122
    },
    {
      "epoch": 0.026577355229040624,
      "grad_norm": 0.17214606702327728,
      "learning_rate": 4.2634915097268115e-05,
      "loss": 10.7608,
      "step": 123
    },
    {
      "epoch": 0.02679343128781331,
      "grad_norm": 0.18513911962509155,
      "learning_rate": 4.1722306190793495e-05,
      "loss": 10.7495,
      "step": 124
    },
    {
      "epoch": 0.027009507346585998,
      "grad_norm": 0.2161649465560913,
      "learning_rate": 4.0812524109171476e-05,
      "loss": 10.7578,
      "step": 125
    },
    {
      "epoch": 0.027225583405358685,
      "grad_norm": 0.2477547526359558,
      "learning_rate": 3.99058795421495e-05,
      "loss": 10.7566,
      "step": 126
    },
    {
      "epoch": 0.027441659464131376,
      "grad_norm": 0.19045643508434296,
      "learning_rate": 3.9002682108016585e-05,
      "loss": 10.753,
      "step": 127
    },
    {
      "epoch": 0.027657735522904063,
      "grad_norm": 0.2026197761297226,
      "learning_rate": 3.8103240247869075e-05,
      "loss": 10.7643,
      "step": 128
    },
    {
      "epoch": 0.02787381158167675,
      "grad_norm": 0.23443067073822021,
      "learning_rate": 3.720786112027822e-05,
      "loss": 10.7545,
      "step": 129
    },
    {
      "epoch": 0.028089887640449437,
      "grad_norm": 0.21837399899959564,
      "learning_rate": 3.631685049639586e-05,
      "loss": 10.7466,
      "step": 130
    },
    {
      "epoch": 0.028305963699222125,
      "grad_norm": 0.228903129696846,
      "learning_rate": 3.543051265553377e-05,
      "loss": 10.7553,
      "step": 131
    },
    {
      "epoch": 0.028522039757994815,
      "grad_norm": 0.22999686002731323,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.7365,
      "step": 132
    },
    {
      "epoch": 0.028738115816767502,
      "grad_norm": 0.2587832808494568,
      "learning_rate": 3.367306435799584e-05,
      "loss": 10.7495,
      "step": 133
    },
    {
      "epoch": 0.02895419187554019,
      "grad_norm": 0.2272672802209854,
      "learning_rate": 3.2802554068303596e-05,
      "loss": 10.7485,
      "step": 134
    },
    {
      "epoch": 0.029170267934312877,
      "grad_norm": 0.1977323740720749,
      "learning_rate": 3.1937916690642356e-05,
      "loss": 10.7478,
      "step": 135
    },
    {
      "epoch": 0.029386343993085567,
      "grad_norm": 0.20933398604393005,
      "learning_rate": 3.107944749788449e-05,
      "loss": 10.759,
      "step": 136
    },
    {
      "epoch": 0.029602420051858255,
      "grad_norm": 0.2519747316837311,
      "learning_rate": 3.0227439656472877e-05,
      "loss": 10.7503,
      "step": 137
    },
    {
      "epoch": 0.029818496110630942,
      "grad_norm": 0.2388688176870346,
      "learning_rate": 2.9382184126304834e-05,
      "loss": 10.7351,
      "step": 138
    },
    {
      "epoch": 0.03003457216940363,
      "grad_norm": 0.29849356412887573,
      "learning_rate": 2.8543969561369556e-05,
      "loss": 10.7364,
      "step": 139
    },
    {
      "epoch": 0.03025064822817632,
      "grad_norm": 0.2356027066707611,
      "learning_rate": 2.771308221117309e-05,
      "loss": 10.7399,
      "step": 140
    },
    {
      "epoch": 0.030466724286949007,
      "grad_norm": 0.25474485754966736,
      "learning_rate": 2.688980582298435e-05,
      "loss": 10.74,
      "step": 141
    },
    {
      "epoch": 0.030682800345721694,
      "grad_norm": 0.23734347522258759,
      "learning_rate": 2.607442154493568e-05,
      "loss": 10.7393,
      "step": 142
    },
    {
      "epoch": 0.03089887640449438,
      "grad_norm": 0.28821271657943726,
      "learning_rate": 2.5267207830011068e-05,
      "loss": 10.7336,
      "step": 143
    },
    {
      "epoch": 0.03111495246326707,
      "grad_norm": 0.2773192822933197,
      "learning_rate": 2.446844034095466e-05,
      "loss": 10.7414,
      "step": 144
    },
    {
      "epoch": 0.031331028522039756,
      "grad_norm": 0.3964485824108124,
      "learning_rate": 2.3678391856132204e-05,
      "loss": 10.742,
      "step": 145
    },
    {
      "epoch": 0.031547104580812446,
      "grad_norm": 0.33878329396247864,
      "learning_rate": 2.2897332176377528e-05,
      "loss": 10.7489,
      "step": 146
    },
    {
      "epoch": 0.03176318063958514,
      "grad_norm": 0.3486817479133606,
      "learning_rate": 2.2125528032855724e-05,
      "loss": 10.7379,
      "step": 147
    },
    {
      "epoch": 0.03197925669835782,
      "grad_norm": 0.38555988669395447,
      "learning_rate": 2.136324299597474e-05,
      "loss": 10.7374,
      "step": 148
    },
    {
      "epoch": 0.03219533275713051,
      "grad_norm": 0.521436333656311,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.7291,
      "step": 149
    },
    {
      "epoch": 0.032411408815903195,
      "grad_norm": 0.8618770241737366,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 10.7151,
      "step": 150
    },
    {
      "epoch": 0.032411408815903195,
      "eval_loss": 10.746877670288086,
      "eval_runtime": 27.2327,
      "eval_samples_per_second": 286.237,
      "eval_steps_per_second": 143.137,
      "step": 150
    },
    {
      "epoch": 0.032627484874675886,
      "grad_norm": 0.2572970688343048,
      "learning_rate": 1.9136088935510362e-05,
      "loss": 10.7659,
      "step": 151
    },
    {
      "epoch": 0.032843560933448576,
      "grad_norm": 0.23141218721866608,
      "learning_rate": 1.8414449687337464e-05,
      "loss": 10.7649,
      "step": 152
    },
    {
      "epoch": 0.03305963699222126,
      "grad_norm": 0.2419702559709549,
      "learning_rate": 1.7703596875660645e-05,
      "loss": 10.7615,
      "step": 153
    },
    {
      "epoch": 0.03327571305099395,
      "grad_norm": 0.19218285381793976,
      "learning_rate": 1.700377325606388e-05,
      "loss": 10.7609,
      "step": 154
    },
    {
      "epoch": 0.03349178910976664,
      "grad_norm": 0.2367192953824997,
      "learning_rate": 1.631521781767214e-05,
      "loss": 10.7647,
      "step": 155
    },
    {
      "epoch": 0.033707865168539325,
      "grad_norm": 0.21695514023303986,
      "learning_rate": 1.5638165701536868e-05,
      "loss": 10.765,
      "step": 156
    },
    {
      "epoch": 0.033923941227312016,
      "grad_norm": 0.2253107875585556,
      "learning_rate": 1.4972848120335453e-05,
      "loss": 10.7699,
      "step": 157
    },
    {
      "epoch": 0.0341400172860847,
      "grad_norm": 0.19992785155773163,
      "learning_rate": 1.4319492279412388e-05,
      "loss": 10.7592,
      "step": 158
    },
    {
      "epoch": 0.03435609334485739,
      "grad_norm": 0.21194936335086823,
      "learning_rate": 1.3678321299188801e-05,
      "loss": 10.7629,
      "step": 159
    },
    {
      "epoch": 0.03457216940363008,
      "grad_norm": 0.26007580757141113,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 10.7648,
      "step": 160
    },
    {
      "epoch": 0.034788245462402764,
      "grad_norm": 0.22751381993293762,
      "learning_rate": 1.2433405522156332e-05,
      "loss": 10.7603,
      "step": 161
    },
    {
      "epoch": 0.035004321521175455,
      "grad_norm": 0.2545979619026184,
      "learning_rate": 1.183008586294485e-05,
      "loss": 10.7617,
      "step": 162
    },
    {
      "epoch": 0.03522039757994814,
      "grad_norm": 0.20070402324199677,
      "learning_rate": 1.1239801194443506e-05,
      "loss": 10.7547,
      "step": 163
    },
    {
      "epoch": 0.03543647363872083,
      "grad_norm": 0.20730213820934296,
      "learning_rate": 1.066275309832584e-05,
      "loss": 10.761,
      "step": 164
    },
    {
      "epoch": 0.03565254969749352,
      "grad_norm": 0.2222391963005066,
      "learning_rate": 1.0099138635988026e-05,
      "loss": 10.7707,
      "step": 165
    },
    {
      "epoch": 0.035868625756266204,
      "grad_norm": 0.21017976105213165,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.7582,
      "step": 166
    },
    {
      "epoch": 0.036084701815038894,
      "grad_norm": 0.21349570155143738,
      "learning_rate": 9.012975854638949e-06,
      "loss": 10.7534,
      "step": 167
    },
    {
      "epoch": 0.036300777873811585,
      "grad_norm": 0.19311493635177612,
      "learning_rate": 8.490798459222476e-06,
      "loss": 10.7536,
      "step": 168
    },
    {
      "epoch": 0.03651685393258427,
      "grad_norm": 0.18329201638698578,
      "learning_rate": 7.982796418105371e-06,
      "loss": 10.7595,
      "step": 169
    },
    {
      "epoch": 0.03673292999135696,
      "grad_norm": 0.24617359042167664,
      "learning_rate": 7.489143213519301e-06,
      "loss": 10.7562,
      "step": 170
    },
    {
      "epoch": 0.03694900605012964,
      "grad_norm": 0.24574460089206696,
      "learning_rate": 7.010007427581378e-06,
      "loss": 10.7496,
      "step": 171
    },
    {
      "epoch": 0.037165082108902334,
      "grad_norm": 0.2094903290271759,
      "learning_rate": 6.5455526847235825e-06,
      "loss": 10.7454,
      "step": 172
    },
    {
      "epoch": 0.037381158167675024,
      "grad_norm": 0.23481670022010803,
      "learning_rate": 6.0959375958151045e-06,
      "loss": 10.7503,
      "step": 173
    },
    {
      "epoch": 0.03759723422644771,
      "grad_norm": 0.20529882609844208,
      "learning_rate": 5.6613157039969055e-06,
      "loss": 10.7551,
      "step": 174
    },
    {
      "epoch": 0.0378133102852204,
      "grad_norm": 0.2087268829345703,
      "learning_rate": 5.241835432246889e-06,
      "loss": 10.7512,
      "step": 175
    },
    {
      "epoch": 0.03802938634399308,
      "grad_norm": 0.24301747977733612,
      "learning_rate": 4.837640032693558e-06,
      "loss": 10.7587,
      "step": 176
    },
    {
      "epoch": 0.03824546240276577,
      "grad_norm": 0.24005451798439026,
      "learning_rate": 4.448867537695578e-06,
      "loss": 10.7615,
      "step": 177
    },
    {
      "epoch": 0.038461538461538464,
      "grad_norm": 0.2205115705728531,
      "learning_rate": 4.075650712703849e-06,
      "loss": 10.7405,
      "step": 178
    },
    {
      "epoch": 0.03867761452031115,
      "grad_norm": 0.22657622396945953,
      "learning_rate": 3.71811701092219e-06,
      "loss": 10.7544,
      "step": 179
    },
    {
      "epoch": 0.03889369057908384,
      "grad_norm": 0.24251677095890045,
      "learning_rate": 3.376388529782215e-06,
      "loss": 10.7535,
      "step": 180
    },
    {
      "epoch": 0.03910976663785653,
      "grad_norm": 0.22308377921581268,
      "learning_rate": 3.0505819692471792e-06,
      "loss": 10.7536,
      "step": 181
    },
    {
      "epoch": 0.03932584269662921,
      "grad_norm": 0.2320922166109085,
      "learning_rate": 2.7408085919590264e-06,
      "loss": 10.7482,
      "step": 182
    },
    {
      "epoch": 0.0395419187554019,
      "grad_norm": 0.23464159667491913,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.7496,
      "step": 183
    },
    {
      "epoch": 0.03975799481417459,
      "grad_norm": 0.21937261521816254,
      "learning_rate": 2.1697790249779636e-06,
      "loss": 10.7502,
      "step": 184
    },
    {
      "epoch": 0.03997407087294728,
      "grad_norm": 0.23580560088157654,
      "learning_rate": 1.908717841359048e-06,
      "loss": 10.7532,
      "step": 185
    },
    {
      "epoch": 0.04019014693171997,
      "grad_norm": 0.2424219399690628,
      "learning_rate": 1.6640797865406288e-06,
      "loss": 10.745,
      "step": 186
    },
    {
      "epoch": 0.04040622299049265,
      "grad_norm": 0.2548362612724304,
      "learning_rate": 1.4359484041943038e-06,
      "loss": 10.7458,
      "step": 187
    },
    {
      "epoch": 0.04062229904926534,
      "grad_norm": 0.31250232458114624,
      "learning_rate": 1.2244016009781701e-06,
      "loss": 10.744,
      "step": 188
    },
    {
      "epoch": 0.040838375108038026,
      "grad_norm": 0.2745141088962555,
      "learning_rate": 1.0295116199317057e-06,
      "loss": 10.7386,
      "step": 189
    },
    {
      "epoch": 0.04105445116681072,
      "grad_norm": 0.2804180383682251,
      "learning_rate": 8.513450158049108e-07,
      "loss": 10.7314,
      "step": 190
    },
    {
      "epoch": 0.04127052722558341,
      "grad_norm": 0.2679109573364258,
      "learning_rate": 6.899626323298713e-07,
      "loss": 10.7439,
      "step": 191
    },
    {
      "epoch": 0.04148660328435609,
      "grad_norm": 0.22815419733524323,
      "learning_rate": 5.454195814427021e-07,
      "loss": 10.7384,
      "step": 192
    },
    {
      "epoch": 0.04170267934312878,
      "grad_norm": 0.2827376127243042,
      "learning_rate": 4.177652244628627e-07,
      "loss": 10.7385,
      "step": 193
    },
    {
      "epoch": 0.04191875540190147,
      "grad_norm": 0.3146594762802124,
      "learning_rate": 3.0704315523631953e-07,
      "loss": 10.7451,
      "step": 194
    },
    {
      "epoch": 0.042134831460674156,
      "grad_norm": 0.30587294697761536,
      "learning_rate": 2.1329118524827662e-07,
      "loss": 10.7321,
      "step": 195
    },
    {
      "epoch": 0.04235090751944685,
      "grad_norm": 0.26626572012901306,
      "learning_rate": 1.3654133071059893e-07,
      "loss": 10.7262,
      "step": 196
    },
    {
      "epoch": 0.04256698357821953,
      "grad_norm": 0.3111208975315094,
      "learning_rate": 7.681980162830282e-08,
      "loss": 10.7239,
      "step": 197
    },
    {
      "epoch": 0.04278305963699222,
      "grad_norm": 0.4671022891998291,
      "learning_rate": 3.4146992848854695e-08,
      "loss": 10.7375,
      "step": 198
    },
    {
      "epoch": 0.04299913569576491,
      "grad_norm": 0.520592987537384,
      "learning_rate": 8.537477097364522e-09,
      "loss": 10.7301,
      "step": 199
    },
    {
      "epoch": 0.043215211754537596,
      "grad_norm": 0.5407875776290894,
      "learning_rate": 0.0,
      "loss": 10.6871,
      "step": 200
    },
    {
      "epoch": 0.043215211754537596,
      "eval_loss": 10.745776176452637,
      "eval_runtime": 27.2162,
      "eval_samples_per_second": 286.41,
      "eval_steps_per_second": 143.223,
      "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": 139523405119488.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}