{
  "best_metric": 3.3320932388305664,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.37261294829995345,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0018630647414997672,
      "grad_norm": 1.2902519702911377,
      "learning_rate": 5e-06,
      "loss": 3.6391,
      "step": 1
    },
    {
      "epoch": 0.0018630647414997672,
      "eval_loss": 5.516392230987549,
      "eval_runtime": 32.1445,
      "eval_samples_per_second": 28.154,
      "eval_steps_per_second": 14.093,
      "step": 1
    },
    {
      "epoch": 0.0037261294829995344,
      "grad_norm": 1.502539873123169,
      "learning_rate": 1e-05,
      "loss": 4.285,
      "step": 2
    },
    {
      "epoch": 0.0055891942244993015,
      "grad_norm": 1.6728174686431885,
      "learning_rate": 1.5e-05,
      "loss": 4.3833,
      "step": 3
    },
    {
      "epoch": 0.007452258965999069,
      "grad_norm": 1.809103012084961,
      "learning_rate": 2e-05,
      "loss": 4.5885,
      "step": 4
    },
    {
      "epoch": 0.009315323707498836,
      "grad_norm": 1.8019917011260986,
      "learning_rate": 2.5e-05,
      "loss": 4.6271,
      "step": 5
    },
    {
      "epoch": 0.011178388448998603,
      "grad_norm": 1.8416504859924316,
      "learning_rate": 3e-05,
      "loss": 4.6321,
      "step": 6
    },
    {
      "epoch": 0.01304145319049837,
      "grad_norm": 1.9281914234161377,
      "learning_rate": 3.5e-05,
      "loss": 4.3472,
      "step": 7
    },
    {
      "epoch": 0.014904517931998137,
      "grad_norm": 2.1244757175445557,
      "learning_rate": 4e-05,
      "loss": 4.8937,
      "step": 8
    },
    {
      "epoch": 0.016767582673497903,
      "grad_norm": 1.9028316736221313,
      "learning_rate": 4.5e-05,
      "loss": 5.0692,
      "step": 9
    },
    {
      "epoch": 0.018630647414997672,
      "grad_norm": 2.2906155586242676,
      "learning_rate": 5e-05,
      "loss": 4.3372,
      "step": 10
    },
    {
      "epoch": 0.020493712156497437,
      "grad_norm": 2.0029475688934326,
      "learning_rate": 5.500000000000001e-05,
      "loss": 4.4519,
      "step": 11
    },
    {
      "epoch": 0.022356776897997206,
      "grad_norm": 1.9513241052627563,
      "learning_rate": 6e-05,
      "loss": 4.3201,
      "step": 12
    },
    {
      "epoch": 0.02421984163949697,
      "grad_norm": 1.7343864440917969,
      "learning_rate": 6.500000000000001e-05,
      "loss": 4.0463,
      "step": 13
    },
    {
      "epoch": 0.02608290638099674,
      "grad_norm": 1.9337594509124756,
      "learning_rate": 7e-05,
      "loss": 4.0561,
      "step": 14
    },
    {
      "epoch": 0.027945971122496506,
      "grad_norm": 1.6187396049499512,
      "learning_rate": 7.500000000000001e-05,
      "loss": 3.5715,
      "step": 15
    },
    {
      "epoch": 0.029809035863996275,
      "grad_norm": 1.5559985637664795,
      "learning_rate": 8e-05,
      "loss": 4.0944,
      "step": 16
    },
    {
      "epoch": 0.031672100605496044,
      "grad_norm": 1.6261332035064697,
      "learning_rate": 8.5e-05,
      "loss": 3.9513,
      "step": 17
    },
    {
      "epoch": 0.033535165346995806,
      "grad_norm": 1.8501555919647217,
      "learning_rate": 9e-05,
      "loss": 3.9689,
      "step": 18
    },
    {
      "epoch": 0.035398230088495575,
      "grad_norm": 2.036672592163086,
      "learning_rate": 9.5e-05,
      "loss": 3.9789,
      "step": 19
    },
    {
      "epoch": 0.037261294829995344,
      "grad_norm": 2.047884702682495,
      "learning_rate": 0.0001,
      "loss": 3.663,
      "step": 20
    },
    {
      "epoch": 0.03912435957149511,
      "grad_norm": 1.90353262424469,
      "learning_rate": 9.999238475781957e-05,
      "loss": 3.5892,
      "step": 21
    },
    {
      "epoch": 0.040987424312994875,
      "grad_norm": 1.863972783088684,
      "learning_rate": 9.99695413509548e-05,
      "loss": 3.9369,
      "step": 22
    },
    {
      "epoch": 0.04285048905449464,
      "grad_norm": 1.4511252641677856,
      "learning_rate": 9.99314767377287e-05,
      "loss": 3.1318,
      "step": 23
    },
    {
      "epoch": 0.04471355379599441,
      "grad_norm": 1.4355757236480713,
      "learning_rate": 9.987820251299122e-05,
      "loss": 3.983,
      "step": 24
    },
    {
      "epoch": 0.04657661853749418,
      "grad_norm": 1.4540237188339233,
      "learning_rate": 9.980973490458728e-05,
      "loss": 4.1114,
      "step": 25
    },
    {
      "epoch": 0.04843968327899394,
      "grad_norm": 1.5578112602233887,
      "learning_rate": 9.972609476841367e-05,
      "loss": 3.8437,
      "step": 26
    },
    {
      "epoch": 0.05030274802049371,
      "grad_norm": 1.8616338968276978,
      "learning_rate": 9.962730758206611e-05,
      "loss": 3.516,
      "step": 27
    },
    {
      "epoch": 0.05216581276199348,
      "grad_norm": 1.6210098266601562,
      "learning_rate": 9.951340343707852e-05,
      "loss": 3.3105,
      "step": 28
    },
    {
      "epoch": 0.05402887750349324,
      "grad_norm": 1.6511797904968262,
      "learning_rate": 9.938441702975689e-05,
      "loss": 3.9478,
      "step": 29
    },
    {
      "epoch": 0.05589194224499301,
      "grad_norm": 1.6636693477630615,
      "learning_rate": 9.924038765061042e-05,
      "loss": 3.6779,
      "step": 30
    },
    {
      "epoch": 0.05775500698649278,
      "grad_norm": 1.4905117750167847,
      "learning_rate": 9.908135917238321e-05,
      "loss": 3.2574,
      "step": 31
    },
    {
      "epoch": 0.05961807172799255,
      "grad_norm": 1.6063334941864014,
      "learning_rate": 9.890738003669029e-05,
      "loss": 3.9851,
      "step": 32
    },
    {
      "epoch": 0.06148113646949231,
      "grad_norm": 1.7283437252044678,
      "learning_rate": 9.871850323926177e-05,
      "loss": 3.705,
      "step": 33
    },
    {
      "epoch": 0.06334420121099209,
      "grad_norm": 1.5343276262283325,
      "learning_rate": 9.851478631379982e-05,
      "loss": 3.5288,
      "step": 34
    },
    {
      "epoch": 0.06520726595249185,
      "grad_norm": 1.7014079093933105,
      "learning_rate": 9.829629131445342e-05,
      "loss": 3.8288,
      "step": 35
    },
    {
      "epoch": 0.06707033069399161,
      "grad_norm": 1.7944095134735107,
      "learning_rate": 9.806308479691595e-05,
      "loss": 3.8894,
      "step": 36
    },
    {
      "epoch": 0.06893339543549139,
      "grad_norm": 1.5314217805862427,
      "learning_rate": 9.781523779815179e-05,
      "loss": 3.2699,
      "step": 37
    },
    {
      "epoch": 0.07079646017699115,
      "grad_norm": 1.6168136596679688,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.8024,
      "step": 38
    },
    {
      "epoch": 0.07265952491849091,
      "grad_norm": 1.3477504253387451,
      "learning_rate": 9.727592877996585e-05,
      "loss": 3.0357,
      "step": 39
    },
    {
      "epoch": 0.07452258965999069,
      "grad_norm": 2.0698187351226807,
      "learning_rate": 9.698463103929542e-05,
      "loss": 3.7317,
      "step": 40
    },
    {
      "epoch": 0.07638565440149045,
      "grad_norm": 1.6671048402786255,
      "learning_rate": 9.667902132486009e-05,
      "loss": 3.4916,
      "step": 41
    },
    {
      "epoch": 0.07824871914299023,
      "grad_norm": 1.8396328687667847,
      "learning_rate": 9.635919272833938e-05,
      "loss": 3.5342,
      "step": 42
    },
    {
      "epoch": 0.08011178388448999,
      "grad_norm": 1.5536775588989258,
      "learning_rate": 9.602524267262203e-05,
      "loss": 3.8207,
      "step": 43
    },
    {
      "epoch": 0.08197484862598975,
      "grad_norm": 1.5929099321365356,
      "learning_rate": 9.567727288213005e-05,
      "loss": 3.8526,
      "step": 44
    },
    {
      "epoch": 0.08383791336748952,
      "grad_norm": 2.140622854232788,
      "learning_rate": 9.53153893518325e-05,
      "loss": 3.9536,
      "step": 45
    },
    {
      "epoch": 0.08570097810898929,
      "grad_norm": 2.204415798187256,
      "learning_rate": 9.493970231495835e-05,
      "loss": 3.8141,
      "step": 46
    },
    {
      "epoch": 0.08756404285048905,
      "grad_norm": 2.16359806060791,
      "learning_rate": 9.45503262094184e-05,
      "loss": 4.0383,
      "step": 47
    },
    {
      "epoch": 0.08942710759198882,
      "grad_norm": 2.2646803855895996,
      "learning_rate": 9.414737964294636e-05,
      "loss": 3.9917,
      "step": 48
    },
    {
      "epoch": 0.09129017233348859,
      "grad_norm": 2.4855337142944336,
      "learning_rate": 9.373098535696979e-05,
      "loss": 3.8354,
      "step": 49
    },
    {
      "epoch": 0.09315323707498836,
      "grad_norm": 2.349766254425049,
      "learning_rate": 9.330127018922194e-05,
      "loss": 3.9016,
      "step": 50
    },
    {
      "epoch": 0.09315323707498836,
      "eval_loss": 3.6060919761657715,
      "eval_runtime": 32.1835,
      "eval_samples_per_second": 28.12,
      "eval_steps_per_second": 14.076,
      "step": 50
    },
    {
      "epoch": 0.09501630181648812,
      "grad_norm": 2.1820530891418457,
      "learning_rate": 9.285836503510562e-05,
      "loss": 2.9137,
      "step": 51
    },
    {
      "epoch": 0.09687936655798789,
      "grad_norm": 2.1806118488311768,
      "learning_rate": 9.24024048078213e-05,
      "loss": 3.1755,
      "step": 52
    },
    {
      "epoch": 0.09874243129948766,
      "grad_norm": 1.8996943235397339,
      "learning_rate": 9.193352839727121e-05,
      "loss": 3.3476,
      "step": 53
    },
    {
      "epoch": 0.10060549604098742,
      "grad_norm": 1.8744741678237915,
      "learning_rate": 9.145187862775209e-05,
      "loss": 3.196,
      "step": 54
    },
    {
      "epoch": 0.10246856078248719,
      "grad_norm": 1.70234215259552,
      "learning_rate": 9.09576022144496e-05,
      "loss": 3.3386,
      "step": 55
    },
    {
      "epoch": 0.10433162552398696,
      "grad_norm": 1.7539249658584595,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.4391,
      "step": 56
    },
    {
      "epoch": 0.10619469026548672,
      "grad_norm": 1.4793843030929565,
      "learning_rate": 8.993177550236464e-05,
      "loss": 3.9207,
      "step": 57
    },
    {
      "epoch": 0.10805775500698649,
      "grad_norm": 1.1978991031646729,
      "learning_rate": 8.940053768033609e-05,
      "loss": 3.2176,
      "step": 58
    },
    {
      "epoch": 0.10992081974848626,
      "grad_norm": 1.1151164770126343,
      "learning_rate": 8.885729807284856e-05,
      "loss": 3.0396,
      "step": 59
    },
    {
      "epoch": 0.11178388448998602,
      "grad_norm": 1.0803781747817993,
      "learning_rate": 8.83022221559489e-05,
      "loss": 2.9356,
      "step": 60
    },
    {
      "epoch": 0.1136469492314858,
      "grad_norm": 1.1919324398040771,
      "learning_rate": 8.773547901113862e-05,
      "loss": 3.0124,
      "step": 61
    },
    {
      "epoch": 0.11551001397298556,
      "grad_norm": 1.0512971878051758,
      "learning_rate": 8.715724127386972e-05,
      "loss": 3.2844,
      "step": 62
    },
    {
      "epoch": 0.11737307871448532,
      "grad_norm": 1.3854892253875732,
      "learning_rate": 8.656768508095853e-05,
      "loss": 3.3558,
      "step": 63
    },
    {
      "epoch": 0.1192361434559851,
      "grad_norm": 1.2489628791809082,
      "learning_rate": 8.596699001693255e-05,
      "loss": 3.2082,
      "step": 64
    },
    {
      "epoch": 0.12109920819748486,
      "grad_norm": 1.0986685752868652,
      "learning_rate": 8.535533905932738e-05,
      "loss": 3.0937,
      "step": 65
    },
    {
      "epoch": 0.12296227293898462,
      "grad_norm": 1.2437379360198975,
      "learning_rate": 8.473291852294987e-05,
      "loss": 3.2174,
      "step": 66
    },
    {
      "epoch": 0.1248253376804844,
      "grad_norm": 1.1477092504501343,
      "learning_rate": 8.409991800312493e-05,
      "loss": 3.2221,
      "step": 67
    },
    {
      "epoch": 0.12668840242198418,
      "grad_norm": 1.4317594766616821,
      "learning_rate": 8.345653031794292e-05,
      "loss": 3.2315,
      "step": 68
    },
    {
      "epoch": 0.12855146716348392,
      "grad_norm": 1.2513046264648438,
      "learning_rate": 8.280295144952536e-05,
      "loss": 3.4117,
      "step": 69
    },
    {
      "epoch": 0.1304145319049837,
      "grad_norm": 1.2725088596343994,
      "learning_rate": 8.213938048432697e-05,
      "loss": 3.4154,
      "step": 70
    },
    {
      "epoch": 0.13227759664648348,
      "grad_norm": 1.301317572593689,
      "learning_rate": 8.146601955249188e-05,
      "loss": 3.2894,
      "step": 71
    },
    {
      "epoch": 0.13414066138798322,
      "grad_norm": 1.2711513042449951,
      "learning_rate": 8.07830737662829e-05,
      "loss": 3.352,
      "step": 72
    },
    {
      "epoch": 0.136003726129483,
      "grad_norm": 1.2225326299667358,
      "learning_rate": 8.009075115760243e-05,
      "loss": 3.2548,
      "step": 73
    },
    {
      "epoch": 0.13786679087098277,
      "grad_norm": 1.2163617610931396,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.5588,
      "step": 74
    },
    {
      "epoch": 0.13972985561248252,
      "grad_norm": 2.666508436203003,
      "learning_rate": 7.86788218175523e-05,
      "loss": 3.3621,
      "step": 75
    },
    {
      "epoch": 0.1415929203539823,
      "grad_norm": 1.1624727249145508,
      "learning_rate": 7.795964517353735e-05,
      "loss": 3.3753,
      "step": 76
    },
    {
      "epoch": 0.14345598509548207,
      "grad_norm": 1.2207733392715454,
      "learning_rate": 7.723195175075136e-05,
      "loss": 3.0322,
      "step": 77
    },
    {
      "epoch": 0.14531904983698182,
      "grad_norm": 1.153766393661499,
      "learning_rate": 7.649596321166024e-05,
      "loss": 3.3144,
      "step": 78
    },
    {
      "epoch": 0.1471821145784816,
      "grad_norm": 1.0764729976654053,
      "learning_rate": 7.575190374550272e-05,
      "loss": 2.8701,
      "step": 79
    },
    {
      "epoch": 0.14904517931998137,
      "grad_norm": 1.2744287252426147,
      "learning_rate": 7.500000000000001e-05,
      "loss": 3.5407,
      "step": 80
    },
    {
      "epoch": 0.15090824406148112,
      "grad_norm": 1.4500547647476196,
      "learning_rate": 7.424048101231686e-05,
      "loss": 3.3808,
      "step": 81
    },
    {
      "epoch": 0.1527713088029809,
      "grad_norm": 2.205737352371216,
      "learning_rate": 7.347357813929454e-05,
      "loss": 3.2512,
      "step": 82
    },
    {
      "epoch": 0.15463437354448067,
      "grad_norm": 1.1358613967895508,
      "learning_rate": 7.269952498697734e-05,
      "loss": 3.4559,
      "step": 83
    },
    {
      "epoch": 0.15649743828598045,
      "grad_norm": 1.6348944902420044,
      "learning_rate": 7.191855733945387e-05,
      "loss": 3.9512,
      "step": 84
    },
    {
      "epoch": 0.1583605030274802,
      "grad_norm": 1.229777455329895,
      "learning_rate": 7.113091308703498e-05,
      "loss": 2.9064,
      "step": 85
    },
    {
      "epoch": 0.16022356776897997,
      "grad_norm": 1.1657277345657349,
      "learning_rate": 7.033683215379002e-05,
      "loss": 3.0757,
      "step": 86
    },
    {
      "epoch": 0.16208663251047975,
      "grad_norm": 1.4670532941818237,
      "learning_rate": 6.953655642446368e-05,
      "loss": 3.4218,
      "step": 87
    },
    {
      "epoch": 0.1639496972519795,
      "grad_norm": 1.5080535411834717,
      "learning_rate": 6.873032967079561e-05,
      "loss": 3.3886,
      "step": 88
    },
    {
      "epoch": 0.16581276199347927,
      "grad_norm": 1.4212788343429565,
      "learning_rate": 6.7918397477265e-05,
      "loss": 3.6124,
      "step": 89
    },
    {
      "epoch": 0.16767582673497905,
      "grad_norm": 1.553883671760559,
      "learning_rate": 6.710100716628344e-05,
      "loss": 3.7264,
      "step": 90
    },
    {
      "epoch": 0.1695388914764788,
      "grad_norm": 1.4848854541778564,
      "learning_rate": 6.627840772285784e-05,
      "loss": 3.1516,
      "step": 91
    },
    {
      "epoch": 0.17140195621797857,
      "grad_norm": 1.2597936391830444,
      "learning_rate": 6.545084971874738e-05,
      "loss": 3.3816,
      "step": 92
    },
    {
      "epoch": 0.17326502095947835,
      "grad_norm": 1.4741157293319702,
      "learning_rate": 6.461858523613684e-05,
      "loss": 3.4027,
      "step": 93
    },
    {
      "epoch": 0.1751280857009781,
      "grad_norm": 1.5933159589767456,
      "learning_rate": 6.378186779084995e-05,
      "loss": 3.4609,
      "step": 94
    },
    {
      "epoch": 0.17699115044247787,
      "grad_norm": 1.6153674125671387,
      "learning_rate": 6.294095225512603e-05,
      "loss": 3.7559,
      "step": 95
    },
    {
      "epoch": 0.17885421518397765,
      "grad_norm": 1.5263915061950684,
      "learning_rate": 6.209609477998338e-05,
      "loss": 3.5021,
      "step": 96
    },
    {
      "epoch": 0.1807172799254774,
      "grad_norm": 2.1421830654144287,
      "learning_rate": 6.124755271719325e-05,
      "loss": 4.2071,
      "step": 97
    },
    {
      "epoch": 0.18258034466697717,
      "grad_norm": 1.6600545644760132,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 3.4893,
      "step": 98
    },
    {
      "epoch": 0.18444340940847695,
      "grad_norm": 2.4432666301727295,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 3.889,
      "step": 99
    },
    {
      "epoch": 0.18630647414997673,
      "grad_norm": 3.4110183715820312,
      "learning_rate": 5.868240888334653e-05,
      "loss": 4.2759,
      "step": 100
    },
    {
      "epoch": 0.18630647414997673,
      "eval_loss": 3.3999626636505127,
      "eval_runtime": 32.1259,
      "eval_samples_per_second": 28.17,
      "eval_steps_per_second": 14.101,
      "step": 100
    },
    {
      "epoch": 0.18816953889147647,
      "grad_norm": 1.373574137687683,
      "learning_rate": 5.782172325201155e-05,
      "loss": 3.0844,
      "step": 101
    },
    {
      "epoch": 0.19003260363297625,
      "grad_norm": 1.2612931728363037,
      "learning_rate": 5.695865504800327e-05,
      "loss": 3.4433,
      "step": 102
    },
    {
      "epoch": 0.19189566837447602,
      "grad_norm": 1.346120834350586,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 3.3105,
      "step": 103
    },
    {
      "epoch": 0.19375873311597577,
      "grad_norm": 1.3103049993515015,
      "learning_rate": 5.522642316338268e-05,
      "loss": 3.7383,
      "step": 104
    },
    {
      "epoch": 0.19562179785747555,
      "grad_norm": 1.2798504829406738,
      "learning_rate": 5.435778713738292e-05,
      "loss": 3.0195,
      "step": 105
    },
    {
      "epoch": 0.19748486259897532,
      "grad_norm": 1.2848749160766602,
      "learning_rate": 5.348782368720626e-05,
      "loss": 2.9208,
      "step": 106
    },
    {
      "epoch": 0.19934792734047507,
      "grad_norm": 1.1921271085739136,
      "learning_rate": 5.26167978121472e-05,
      "loss": 3.238,
      "step": 107
    },
    {
      "epoch": 0.20121099208197485,
      "grad_norm": 1.2079963684082031,
      "learning_rate": 5.174497483512506e-05,
      "loss": 3.2653,
      "step": 108
    },
    {
      "epoch": 0.20307405682347462,
      "grad_norm": 1.150623083114624,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 3.3047,
      "step": 109
    },
    {
      "epoch": 0.20493712156497437,
      "grad_norm": 1.12302565574646,
      "learning_rate": 5e-05,
      "loss": 2.9365,
      "step": 110
    },
    {
      "epoch": 0.20680018630647415,
      "grad_norm": 0.8901563882827759,
      "learning_rate": 4.912737967813583e-05,
      "loss": 2.5866,
      "step": 111
    },
    {
      "epoch": 0.20866325104797392,
      "grad_norm": 1.5094679594039917,
      "learning_rate": 4.825502516487497e-05,
      "loss": 3.5975,
      "step": 112
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.1921519041061401,
      "learning_rate": 4.738320218785281e-05,
      "loss": 3.0366,
      "step": 113
    },
    {
      "epoch": 0.21238938053097345,
      "grad_norm": 1.2302180528640747,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 3.1477,
      "step": 114
    },
    {
      "epoch": 0.21425244527247322,
      "grad_norm": 0.9722185730934143,
      "learning_rate": 4.564221286261709e-05,
      "loss": 2.9138,
      "step": 115
    },
    {
      "epoch": 0.21611551001397297,
      "grad_norm": 1.2122098207473755,
      "learning_rate": 4.477357683661734e-05,
      "loss": 3.3952,
      "step": 116
    },
    {
      "epoch": 0.21797857475547275,
      "grad_norm": 1.0260534286499023,
      "learning_rate": 4.390653282974264e-05,
      "loss": 3.0228,
      "step": 117
    },
    {
      "epoch": 0.21984163949697252,
      "grad_norm": 1.134804606437683,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 3.2426,
      "step": 118
    },
    {
      "epoch": 0.2217047042384723,
      "grad_norm": 1.2314471006393433,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 3.1156,
      "step": 119
    },
    {
      "epoch": 0.22356776897997205,
      "grad_norm": 1.9310802221298218,
      "learning_rate": 4.131759111665349e-05,
      "loss": 3.2605,
      "step": 120
    },
    {
      "epoch": 0.22543083372147182,
      "grad_norm": 1.244099497795105,
      "learning_rate": 4.045955023117276e-05,
      "loss": 3.3355,
      "step": 121
    },
    {
      "epoch": 0.2272938984629716,
      "grad_norm": 1.1614549160003662,
      "learning_rate": 3.960441545911204e-05,
      "loss": 2.9461,
      "step": 122
    },
    {
      "epoch": 0.22915696320447135,
      "grad_norm": 1.1840636730194092,
      "learning_rate": 3.875244728280676e-05,
      "loss": 3.1649,
      "step": 123
    },
    {
      "epoch": 0.23102002794597112,
      "grad_norm": 1.1917698383331299,
      "learning_rate": 3.790390522001662e-05,
      "loss": 3.2872,
      "step": 124
    },
    {
      "epoch": 0.2328830926874709,
      "grad_norm": 1.3176980018615723,
      "learning_rate": 3.705904774487396e-05,
      "loss": 3.4799,
      "step": 125
    },
    {
      "epoch": 0.23474615742897065,
      "grad_norm": 1.1151647567749023,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 2.9211,
      "step": 126
    },
    {
      "epoch": 0.23660922217047042,
      "grad_norm": 1.0221425294876099,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 2.8233,
      "step": 127
    },
    {
      "epoch": 0.2384722869119702,
      "grad_norm": 1.9823639392852783,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.0263,
      "step": 128
    },
    {
      "epoch": 0.24033535165346995,
      "grad_norm": 1.2429327964782715,
      "learning_rate": 3.372159227714218e-05,
      "loss": 3.3876,
      "step": 129
    },
    {
      "epoch": 0.24219841639496972,
      "grad_norm": 1.3118374347686768,
      "learning_rate": 3.289899283371657e-05,
      "loss": 3.3716,
      "step": 130
    },
    {
      "epoch": 0.2440614811364695,
      "grad_norm": 1.3129647970199585,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 3.2332,
      "step": 131
    },
    {
      "epoch": 0.24592454587796925,
      "grad_norm": 1.2996779680252075,
      "learning_rate": 3.12696703292044e-05,
      "loss": 3.4357,
      "step": 132
    },
    {
      "epoch": 0.24778761061946902,
      "grad_norm": 1.2652643918991089,
      "learning_rate": 3.046344357553632e-05,
      "loss": 3.3381,
      "step": 133
    },
    {
      "epoch": 0.2496506753609688,
      "grad_norm": 1.2161061763763428,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 3.4872,
      "step": 134
    },
    {
      "epoch": 0.2515137401024686,
      "grad_norm": 1.251867651939392,
      "learning_rate": 2.886908691296504e-05,
      "loss": 3.2131,
      "step": 135
    },
    {
      "epoch": 0.25337680484396835,
      "grad_norm": 1.3443360328674316,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 3.1788,
      "step": 136
    },
    {
      "epoch": 0.25523986958546807,
      "grad_norm": 1.188568115234375,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 3.4101,
      "step": 137
    },
    {
      "epoch": 0.25710293432696785,
      "grad_norm": 1.5125073194503784,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 3.6851,
      "step": 138
    },
    {
      "epoch": 0.2589659990684676,
      "grad_norm": 1.3027081489562988,
      "learning_rate": 2.575951898768315e-05,
      "loss": 3.3351,
      "step": 139
    },
    {
      "epoch": 0.2608290638099674,
      "grad_norm": 1.2359957695007324,
      "learning_rate": 2.500000000000001e-05,
      "loss": 3.2338,
      "step": 140
    },
    {
      "epoch": 0.2626921285514672,
      "grad_norm": 1.3322489261627197,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 2.8686,
      "step": 141
    },
    {
      "epoch": 0.26455519329296695,
      "grad_norm": 2.0788941383361816,
      "learning_rate": 2.350403678833976e-05,
      "loss": 3.6646,
      "step": 142
    },
    {
      "epoch": 0.26641825803446667,
      "grad_norm": 1.3164119720458984,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 3.4545,
      "step": 143
    },
    {
      "epoch": 0.26828132277596645,
      "grad_norm": 1.652212381362915,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 3.5121,
      "step": 144
    },
    {
      "epoch": 0.2701443875174662,
      "grad_norm": 1.5663607120513916,
      "learning_rate": 2.132117818244771e-05,
      "loss": 3.7725,
      "step": 145
    },
    {
      "epoch": 0.272007452258966,
      "grad_norm": 1.5914528369903564,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.7862,
      "step": 146
    },
    {
      "epoch": 0.2738705170004658,
      "grad_norm": 1.9115759134292603,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 4.0243,
      "step": 147
    },
    {
      "epoch": 0.27573358174196555,
      "grad_norm": 1.703169584274292,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 3.3129,
      "step": 148
    },
    {
      "epoch": 0.2775966464834653,
      "grad_norm": 1.8172022104263306,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 3.9249,
      "step": 149
    },
    {
      "epoch": 0.27945971122496505,
      "grad_norm": 2.922093391418457,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 4.2185,
      "step": 150
    },
    {
      "epoch": 0.27945971122496505,
      "eval_loss": 3.3465943336486816,
      "eval_runtime": 32.1695,
      "eval_samples_per_second": 28.132,
      "eval_steps_per_second": 14.082,
      "step": 150
    },
    {
      "epoch": 0.2813227759664648,
      "grad_norm": 1.0994431972503662,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 2.7072,
      "step": 151
    },
    {
      "epoch": 0.2831858407079646,
      "grad_norm": 1.0657951831817627,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 3.1659,
      "step": 152
    },
    {
      "epoch": 0.2850489054494644,
      "grad_norm": 1.3122929334640503,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 3.1423,
      "step": 153
    },
    {
      "epoch": 0.28691197019096415,
      "grad_norm": 1.0391207933425903,
      "learning_rate": 1.526708147705013e-05,
      "loss": 2.8446,
      "step": 154
    },
    {
      "epoch": 0.2887750349324639,
      "grad_norm": 1.1318707466125488,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 3.1408,
      "step": 155
    },
    {
      "epoch": 0.29063809967396365,
      "grad_norm": 1.162157416343689,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 2.852,
      "step": 156
    },
    {
      "epoch": 0.2925011644154634,
      "grad_norm": 1.4153108596801758,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 3.2088,
      "step": 157
    },
    {
      "epoch": 0.2943642291569632,
      "grad_norm": 1.2663054466247559,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 3.3267,
      "step": 158
    },
    {
      "epoch": 0.296227293898463,
      "grad_norm": 1.0257394313812256,
      "learning_rate": 1.22645209888614e-05,
      "loss": 3.1659,
      "step": 159
    },
    {
      "epoch": 0.29809035863996275,
      "grad_norm": 1.4004676342010498,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 3.1246,
      "step": 160
    },
    {
      "epoch": 0.2999534233814625,
      "grad_norm": 1.103905439376831,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 3.2662,
      "step": 161
    },
    {
      "epoch": 0.30181648812296225,
      "grad_norm": 1.0684690475463867,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 3.0609,
      "step": 162
    },
    {
      "epoch": 0.303679552864462,
      "grad_norm": 1.1447290182113647,
      "learning_rate": 1.006822449763537e-05,
      "loss": 2.9505,
      "step": 163
    },
    {
      "epoch": 0.3055426176059618,
      "grad_norm": 1.2237478494644165,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.2494,
      "step": 164
    },
    {
      "epoch": 0.3074056823474616,
      "grad_norm": 1.0637820959091187,
      "learning_rate": 9.042397785550405e-06,
      "loss": 2.9076,
      "step": 165
    },
    {
      "epoch": 0.30926874708896135,
      "grad_norm": 1.2343932390213013,
      "learning_rate": 8.548121372247918e-06,
      "loss": 3.3097,
      "step": 166
    },
    {
      "epoch": 0.3111318118304611,
      "grad_norm": 1.0267237424850464,
      "learning_rate": 8.066471602728803e-06,
      "loss": 2.92,
      "step": 167
    },
    {
      "epoch": 0.3129948765719609,
      "grad_norm": 1.1801772117614746,
      "learning_rate": 7.597595192178702e-06,
      "loss": 3.0281,
      "step": 168
    },
    {
      "epoch": 0.3148579413134606,
      "grad_norm": 1.3864883184432983,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 3.2964,
      "step": 169
    },
    {
      "epoch": 0.3167210060549604,
      "grad_norm": 1.212416410446167,
      "learning_rate": 6.698729810778065e-06,
      "loss": 3.1213,
      "step": 170
    },
    {
      "epoch": 0.3185840707964602,
      "grad_norm": 1.1211936473846436,
      "learning_rate": 6.269014643030213e-06,
      "loss": 2.8045,
      "step": 171
    },
    {
      "epoch": 0.32044713553795995,
      "grad_norm": 1.1482990980148315,
      "learning_rate": 5.852620357053651e-06,
      "loss": 3.0468,
      "step": 172
    },
    {
      "epoch": 0.3223102002794597,
      "grad_norm": 1.3345028162002563,
      "learning_rate": 5.449673790581611e-06,
      "loss": 2.9481,
      "step": 173
    },
    {
      "epoch": 0.3241732650209595,
      "grad_norm": 1.0090028047561646,
      "learning_rate": 5.060297685041659e-06,
      "loss": 3.0541,
      "step": 174
    },
    {
      "epoch": 0.3260363297624592,
      "grad_norm": 1.2291425466537476,
      "learning_rate": 4.684610648167503e-06,
      "loss": 3.4524,
      "step": 175
    },
    {
      "epoch": 0.327899394503959,
      "grad_norm": 1.4188092947006226,
      "learning_rate": 4.322727117869951e-06,
      "loss": 3.387,
      "step": 176
    },
    {
      "epoch": 0.32976245924545877,
      "grad_norm": 1.2818740606307983,
      "learning_rate": 3.974757327377981e-06,
      "loss": 3.5364,
      "step": 177
    },
    {
      "epoch": 0.33162552398695855,
      "grad_norm": 1.3271952867507935,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 3.1821,
      "step": 178
    },
    {
      "epoch": 0.3334885887284583,
      "grad_norm": 1.2895276546478271,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 3.0432,
      "step": 179
    },
    {
      "epoch": 0.3353516534699581,
      "grad_norm": 1.2879928350448608,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 3.367,
      "step": 180
    },
    {
      "epoch": 0.3372147182114579,
      "grad_norm": 1.3287880420684814,
      "learning_rate": 2.724071220034158e-06,
      "loss": 3.2295,
      "step": 181
    },
    {
      "epoch": 0.3390777829529576,
      "grad_norm": 1.314154863357544,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 3.3232,
      "step": 182
    },
    {
      "epoch": 0.34094084769445737,
      "grad_norm": 1.3569366931915283,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 2.9797,
      "step": 183
    },
    {
      "epoch": 0.34280391243595715,
      "grad_norm": 1.5291014909744263,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 3.2835,
      "step": 184
    },
    {
      "epoch": 0.3446669771774569,
      "grad_norm": 1.3267407417297363,
      "learning_rate": 1.70370868554659e-06,
      "loss": 3.1526,
      "step": 185
    },
    {
      "epoch": 0.3465300419189567,
      "grad_norm": 1.36220383644104,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 3.7465,
      "step": 186
    },
    {
      "epoch": 0.3483931066604565,
      "grad_norm": 1.2978960275650024,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 3.7277,
      "step": 187
    },
    {
      "epoch": 0.3502561714019562,
      "grad_norm": 1.3940683603286743,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 3.4685,
      "step": 188
    },
    {
      "epoch": 0.35211923614345597,
      "grad_norm": 1.4953304529190063,
      "learning_rate": 9.186408276168013e-07,
      "loss": 3.5472,
      "step": 189
    },
    {
      "epoch": 0.35398230088495575,
      "grad_norm": 1.4648510217666626,
      "learning_rate": 7.596123493895991e-07,
      "loss": 3.3383,
      "step": 190
    },
    {
      "epoch": 0.3558453656264555,
      "grad_norm": 1.5643545389175415,
      "learning_rate": 6.15582970243117e-07,
      "loss": 3.4377,
      "step": 191
    },
    {
      "epoch": 0.3577084303679553,
      "grad_norm": 1.4430068731307983,
      "learning_rate": 4.865965629214819e-07,
      "loss": 3.6607,
      "step": 192
    },
    {
      "epoch": 0.3595714951094551,
      "grad_norm": 1.5247834920883179,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 3.8584,
      "step": 193
    },
    {
      "epoch": 0.3614345598509548,
      "grad_norm": 1.5962194204330444,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 3.6967,
      "step": 194
    },
    {
      "epoch": 0.36329762459245457,
      "grad_norm": 3.414898633956909,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 3.7934,
      "step": 195
    },
    {
      "epoch": 0.36516068933395435,
      "grad_norm": 1.7514005899429321,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 3.9039,
      "step": 196
    },
    {
      "epoch": 0.3670237540754541,
      "grad_norm": 1.91195809841156,
      "learning_rate": 6.852326227130834e-08,
      "loss": 3.604,
      "step": 197
    },
    {
      "epoch": 0.3688868188169539,
      "grad_norm": 2.1014304161071777,
      "learning_rate": 3.04586490452119e-08,
      "loss": 3.6654,
      "step": 198
    },
    {
      "epoch": 0.3707498835584537,
      "grad_norm": 2.118112564086914,
      "learning_rate": 7.615242180436522e-09,
      "loss": 3.8243,
      "step": 199
    },
    {
      "epoch": 0.37261294829995345,
      "grad_norm": 3.0300862789154053,
      "learning_rate": 0.0,
      "loss": 3.6882,
      "step": 200
    },
    {
      "epoch": 0.37261294829995345,
      "eval_loss": 3.3320932388305664,
      "eval_runtime": 32.1296,
      "eval_samples_per_second": 28.167,
      "eval_steps_per_second": 14.099,
      "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
}