{
  "best_metric": 0.33944934606552124,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 0.025076273665733274,
  "eval_steps": 50,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016717515777155514,
      "grad_norm": 66.1699447631836,
      "learning_rate": 5e-06,
      "loss": 20.8217,
      "step": 1
    },
    {
      "epoch": 0.00016717515777155514,
      "eval_loss": 4.830644607543945,
      "eval_runtime": 1176.2555,
      "eval_samples_per_second": 8.565,
      "eval_steps_per_second": 4.283,
      "step": 1
    },
    {
      "epoch": 0.00033435031554311027,
      "grad_norm": 55.28761672973633,
      "learning_rate": 1e-05,
      "loss": 22.0512,
      "step": 2
    },
    {
      "epoch": 0.0005015254733146654,
      "grad_norm": 34.258522033691406,
      "learning_rate": 1.5e-05,
      "loss": 16.6346,
      "step": 3
    },
    {
      "epoch": 0.0006687006310862205,
      "grad_norm": 35.852420806884766,
      "learning_rate": 2e-05,
      "loss": 17.7986,
      "step": 4
    },
    {
      "epoch": 0.0008358757888577757,
      "grad_norm": 44.334144592285156,
      "learning_rate": 2.5e-05,
      "loss": 18.5218,
      "step": 5
    },
    {
      "epoch": 0.0010030509466293308,
      "grad_norm": 39.2244987487793,
      "learning_rate": 3e-05,
      "loss": 15.099,
      "step": 6
    },
    {
      "epoch": 0.001170226104400886,
      "grad_norm": 31.828834533691406,
      "learning_rate": 3.5e-05,
      "loss": 14.2471,
      "step": 7
    },
    {
      "epoch": 0.001337401262172441,
      "grad_norm": 31.544771194458008,
      "learning_rate": 4e-05,
      "loss": 15.0025,
      "step": 8
    },
    {
      "epoch": 0.0015045764199439963,
      "grad_norm": 38.78912353515625,
      "learning_rate": 4.5e-05,
      "loss": 12.5948,
      "step": 9
    },
    {
      "epoch": 0.0016717515777155514,
      "grad_norm": 31.890718460083008,
      "learning_rate": 5e-05,
      "loss": 9.837,
      "step": 10
    },
    {
      "epoch": 0.0018389267354871066,
      "grad_norm": 1420.156005859375,
      "learning_rate": 5.500000000000001e-05,
      "loss": 9.1239,
      "step": 11
    },
    {
      "epoch": 0.0020061018932586616,
      "grad_norm": 25.306236267089844,
      "learning_rate": 6e-05,
      "loss": 7.4396,
      "step": 12
    },
    {
      "epoch": 0.002173277051030217,
      "grad_norm": 26.349756240844727,
      "learning_rate": 6.500000000000001e-05,
      "loss": 8.3412,
      "step": 13
    },
    {
      "epoch": 0.002340452208801772,
      "grad_norm": 21.180355072021484,
      "learning_rate": 7e-05,
      "loss": 6.5257,
      "step": 14
    },
    {
      "epoch": 0.0025076273665733274,
      "grad_norm": 23.34146499633789,
      "learning_rate": 7.500000000000001e-05,
      "loss": 5.9087,
      "step": 15
    },
    {
      "epoch": 0.002674802524344882,
      "grad_norm": 22.092315673828125,
      "learning_rate": 8e-05,
      "loss": 5.759,
      "step": 16
    },
    {
      "epoch": 0.0028419776821164374,
      "grad_norm": 20.397918701171875,
      "learning_rate": 8.5e-05,
      "loss": 4.5725,
      "step": 17
    },
    {
      "epoch": 0.0030091528398879927,
      "grad_norm": 18.830806732177734,
      "learning_rate": 9e-05,
      "loss": 4.2111,
      "step": 18
    },
    {
      "epoch": 0.003176327997659548,
      "grad_norm": 15.320545196533203,
      "learning_rate": 9.5e-05,
      "loss": 3.3252,
      "step": 19
    },
    {
      "epoch": 0.0033435031554311027,
      "grad_norm": 21.95885467529297,
      "learning_rate": 0.0001,
      "loss": 4.4532,
      "step": 20
    },
    {
      "epoch": 0.003510678313202658,
      "grad_norm": 15.07732105255127,
      "learning_rate": 9.999238475781957e-05,
      "loss": 2.7677,
      "step": 21
    },
    {
      "epoch": 0.003677853470974213,
      "grad_norm": 14.574703216552734,
      "learning_rate": 9.99695413509548e-05,
      "loss": 3.1061,
      "step": 22
    },
    {
      "epoch": 0.0038450286287457685,
      "grad_norm": 16.069921493530273,
      "learning_rate": 9.99314767377287e-05,
      "loss": 2.7588,
      "step": 23
    },
    {
      "epoch": 0.004012203786517323,
      "grad_norm": 13.70803451538086,
      "learning_rate": 9.987820251299122e-05,
      "loss": 2.7997,
      "step": 24
    },
    {
      "epoch": 0.004179378944288879,
      "grad_norm": 10.898150444030762,
      "learning_rate": 9.980973490458728e-05,
      "loss": 2.7861,
      "step": 25
    },
    {
      "epoch": 0.004346554102060434,
      "grad_norm": 11.45754623413086,
      "learning_rate": 9.972609476841367e-05,
      "loss": 1.7007,
      "step": 26
    },
    {
      "epoch": 0.004513729259831989,
      "grad_norm": 11.79631519317627,
      "learning_rate": 9.962730758206611e-05,
      "loss": 2.3383,
      "step": 27
    },
    {
      "epoch": 0.004680904417603544,
      "grad_norm": 11.452874183654785,
      "learning_rate": 9.951340343707852e-05,
      "loss": 2.8686,
      "step": 28
    },
    {
      "epoch": 0.004848079575375099,
      "grad_norm": 11.521985054016113,
      "learning_rate": 9.938441702975689e-05,
      "loss": 2.8974,
      "step": 29
    },
    {
      "epoch": 0.005015254733146655,
      "grad_norm": 8.12588882446289,
      "learning_rate": 9.924038765061042e-05,
      "loss": 1.7916,
      "step": 30
    },
    {
      "epoch": 0.0051824298909182095,
      "grad_norm": 8.71818733215332,
      "learning_rate": 9.908135917238321e-05,
      "loss": 1.7343,
      "step": 31
    },
    {
      "epoch": 0.005349605048689764,
      "grad_norm": 7.344757080078125,
      "learning_rate": 9.890738003669029e-05,
      "loss": 2.0697,
      "step": 32
    },
    {
      "epoch": 0.00551678020646132,
      "grad_norm": 7.359161376953125,
      "learning_rate": 9.871850323926177e-05,
      "loss": 1.9041,
      "step": 33
    },
    {
      "epoch": 0.005683955364232875,
      "grad_norm": 6.432816982269287,
      "learning_rate": 9.851478631379982e-05,
      "loss": 1.5344,
      "step": 34
    },
    {
      "epoch": 0.0058511305220044305,
      "grad_norm": 7.933882713317871,
      "learning_rate": 9.829629131445342e-05,
      "loss": 1.6259,
      "step": 35
    },
    {
      "epoch": 0.006018305679775985,
      "grad_norm": 11.05882740020752,
      "learning_rate": 9.806308479691595e-05,
      "loss": 2.3703,
      "step": 36
    },
    {
      "epoch": 0.00618548083754754,
      "grad_norm": 6.093107223510742,
      "learning_rate": 9.781523779815179e-05,
      "loss": 1.5933,
      "step": 37
    },
    {
      "epoch": 0.006352655995319096,
      "grad_norm": 7.78998327255249,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.0594,
      "step": 38
    },
    {
      "epoch": 0.006519831153090651,
      "grad_norm": 12.237889289855957,
      "learning_rate": 9.727592877996585e-05,
      "loss": 2.5323,
      "step": 39
    },
    {
      "epoch": 0.0066870063108622054,
      "grad_norm": 6.168281078338623,
      "learning_rate": 9.698463103929542e-05,
      "loss": 2.0113,
      "step": 40
    },
    {
      "epoch": 0.006854181468633761,
      "grad_norm": 6.7144293785095215,
      "learning_rate": 9.667902132486009e-05,
      "loss": 1.621,
      "step": 41
    },
    {
      "epoch": 0.007021356626405316,
      "grad_norm": 6.263976573944092,
      "learning_rate": 9.635919272833938e-05,
      "loss": 1.9262,
      "step": 42
    },
    {
      "epoch": 0.007188531784176872,
      "grad_norm": 5.80394172668457,
      "learning_rate": 9.602524267262203e-05,
      "loss": 2.3022,
      "step": 43
    },
    {
      "epoch": 0.007355706941948426,
      "grad_norm": 5.633117198944092,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.4775,
      "step": 44
    },
    {
      "epoch": 0.007522882099719981,
      "grad_norm": 9.90259075164795,
      "learning_rate": 9.53153893518325e-05,
      "loss": 2.639,
      "step": 45
    },
    {
      "epoch": 0.007690057257491537,
      "grad_norm": 6.790462970733643,
      "learning_rate": 9.493970231495835e-05,
      "loss": 2.1576,
      "step": 46
    },
    {
      "epoch": 0.007857232415263092,
      "grad_norm": 7.405107498168945,
      "learning_rate": 9.45503262094184e-05,
      "loss": 2.4812,
      "step": 47
    },
    {
      "epoch": 0.008024407573034647,
      "grad_norm": 5.291300296783447,
      "learning_rate": 9.414737964294636e-05,
      "loss": 1.6901,
      "step": 48
    },
    {
      "epoch": 0.008191582730806201,
      "grad_norm": 7.042820453643799,
      "learning_rate": 9.373098535696979e-05,
      "loss": 2.4124,
      "step": 49
    },
    {
      "epoch": 0.008358757888577758,
      "grad_norm": 6.7682905197143555,
      "learning_rate": 9.330127018922194e-05,
      "loss": 2.4717,
      "step": 50
    },
    {
      "epoch": 0.008358757888577758,
      "eval_loss": 0.6559868454933167,
      "eval_runtime": 1183.0601,
      "eval_samples_per_second": 8.516,
      "eval_steps_per_second": 4.258,
      "step": 50
    },
    {
      "epoch": 0.008525933046349313,
      "grad_norm": 89.21569061279297,
      "learning_rate": 9.285836503510562e-05,
      "loss": 7.3617,
      "step": 51
    },
    {
      "epoch": 0.008693108204120868,
      "grad_norm": 18.17483901977539,
      "learning_rate": 9.24024048078213e-05,
      "loss": 2.9755,
      "step": 52
    },
    {
      "epoch": 0.008860283361892422,
      "grad_norm": 19.885055541992188,
      "learning_rate": 9.193352839727121e-05,
      "loss": 3.7002,
      "step": 53
    },
    {
      "epoch": 0.009027458519663977,
      "grad_norm": 9.078715324401855,
      "learning_rate": 9.145187862775209e-05,
      "loss": 3.1884,
      "step": 54
    },
    {
      "epoch": 0.009194633677435534,
      "grad_norm": 11.488103866577148,
      "learning_rate": 9.09576022144496e-05,
      "loss": 2.6635,
      "step": 55
    },
    {
      "epoch": 0.009361808835207088,
      "grad_norm": 15.700331687927246,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.4526,
      "step": 56
    },
    {
      "epoch": 0.009528983992978643,
      "grad_norm": 6.886660575866699,
      "learning_rate": 8.993177550236464e-05,
      "loss": 2.0983,
      "step": 57
    },
    {
      "epoch": 0.009696159150750198,
      "grad_norm": 7.425085067749023,
      "learning_rate": 8.940053768033609e-05,
      "loss": 2.3903,
      "step": 58
    },
    {
      "epoch": 0.009863334308521753,
      "grad_norm": 7.20428991317749,
      "learning_rate": 8.885729807284856e-05,
      "loss": 1.9328,
      "step": 59
    },
    {
      "epoch": 0.01003050946629331,
      "grad_norm": 7.445702075958252,
      "learning_rate": 8.83022221559489e-05,
      "loss": 2.2298,
      "step": 60
    },
    {
      "epoch": 0.010197684624064864,
      "grad_norm": 6.846135139465332,
      "learning_rate": 8.773547901113862e-05,
      "loss": 2.3124,
      "step": 61
    },
    {
      "epoch": 0.010364859781836419,
      "grad_norm": 10.974035263061523,
      "learning_rate": 8.715724127386972e-05,
      "loss": 2.0219,
      "step": 62
    },
    {
      "epoch": 0.010532034939607974,
      "grad_norm": 4.6794843673706055,
      "learning_rate": 8.656768508095853e-05,
      "loss": 1.4291,
      "step": 63
    },
    {
      "epoch": 0.010699210097379529,
      "grad_norm": 4.862498760223389,
      "learning_rate": 8.596699001693255e-05,
      "loss": 1.5775,
      "step": 64
    },
    {
      "epoch": 0.010866385255151085,
      "grad_norm": 6.960198402404785,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.8942,
      "step": 65
    },
    {
      "epoch": 0.01103356041292264,
      "grad_norm": 5.895086765289307,
      "learning_rate": 8.473291852294987e-05,
      "loss": 1.4487,
      "step": 66
    },
    {
      "epoch": 0.011200735570694195,
      "grad_norm": 5.19366979598999,
      "learning_rate": 8.409991800312493e-05,
      "loss": 1.3874,
      "step": 67
    },
    {
      "epoch": 0.01136791072846575,
      "grad_norm": 5.545513153076172,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.8828,
      "step": 68
    },
    {
      "epoch": 0.011535085886237305,
      "grad_norm": 4.446866035461426,
      "learning_rate": 8.280295144952536e-05,
      "loss": 1.2592,
      "step": 69
    },
    {
      "epoch": 0.011702261044008861,
      "grad_norm": 4.8530659675598145,
      "learning_rate": 8.213938048432697e-05,
      "loss": 1.5734,
      "step": 70
    },
    {
      "epoch": 0.011869436201780416,
      "grad_norm": 6.671334266662598,
      "learning_rate": 8.146601955249188e-05,
      "loss": 1.2566,
      "step": 71
    },
    {
      "epoch": 0.01203661135955197,
      "grad_norm": 5.107539176940918,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.1086,
      "step": 72
    },
    {
      "epoch": 0.012203786517323525,
      "grad_norm": 5.993882656097412,
      "learning_rate": 8.009075115760243e-05,
      "loss": 1.3573,
      "step": 73
    },
    {
      "epoch": 0.01237096167509508,
      "grad_norm": 4.740096569061279,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3498,
      "step": 74
    },
    {
      "epoch": 0.012538136832866637,
      "grad_norm": 6.381992340087891,
      "learning_rate": 7.86788218175523e-05,
      "loss": 1.2942,
      "step": 75
    },
    {
      "epoch": 0.012705311990638192,
      "grad_norm": 4.995385646820068,
      "learning_rate": 7.795964517353735e-05,
      "loss": 1.205,
      "step": 76
    },
    {
      "epoch": 0.012872487148409746,
      "grad_norm": 4.214781284332275,
      "learning_rate": 7.723195175075136e-05,
      "loss": 1.1073,
      "step": 77
    },
    {
      "epoch": 0.013039662306181301,
      "grad_norm": 4.874910831451416,
      "learning_rate": 7.649596321166024e-05,
      "loss": 1.3798,
      "step": 78
    },
    {
      "epoch": 0.013206837463952856,
      "grad_norm": 5.055243492126465,
      "learning_rate": 7.575190374550272e-05,
      "loss": 1.2199,
      "step": 79
    },
    {
      "epoch": 0.013374012621724411,
      "grad_norm": 4.44370698928833,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.1394,
      "step": 80
    },
    {
      "epoch": 0.013541187779495967,
      "grad_norm": 5.529434680938721,
      "learning_rate": 7.424048101231686e-05,
      "loss": 1.2294,
      "step": 81
    },
    {
      "epoch": 0.013708362937267522,
      "grad_norm": 6.484023094177246,
      "learning_rate": 7.347357813929454e-05,
      "loss": 1.5444,
      "step": 82
    },
    {
      "epoch": 0.013875538095039077,
      "grad_norm": 6.249424934387207,
      "learning_rate": 7.269952498697734e-05,
      "loss": 1.5945,
      "step": 83
    },
    {
      "epoch": 0.014042713252810632,
      "grad_norm": 3.867103099822998,
      "learning_rate": 7.191855733945387e-05,
      "loss": 1.0598,
      "step": 84
    },
    {
      "epoch": 0.014209888410582187,
      "grad_norm": 7.214300632476807,
      "learning_rate": 7.113091308703498e-05,
      "loss": 1.549,
      "step": 85
    },
    {
      "epoch": 0.014377063568353743,
      "grad_norm": 7.799088478088379,
      "learning_rate": 7.033683215379002e-05,
      "loss": 1.3431,
      "step": 86
    },
    {
      "epoch": 0.014544238726125298,
      "grad_norm": 4.070684432983398,
      "learning_rate": 6.953655642446368e-05,
      "loss": 0.9098,
      "step": 87
    },
    {
      "epoch": 0.014711413883896853,
      "grad_norm": 7.896481990814209,
      "learning_rate": 6.873032967079561e-05,
      "loss": 2.171,
      "step": 88
    },
    {
      "epoch": 0.014878589041668408,
      "grad_norm": 6.196599960327148,
      "learning_rate": 6.7918397477265e-05,
      "loss": 2.0469,
      "step": 89
    },
    {
      "epoch": 0.015045764199439962,
      "grad_norm": 4.253558158874512,
      "learning_rate": 6.710100716628344e-05,
      "loss": 1.7969,
      "step": 90
    },
    {
      "epoch": 0.015212939357211519,
      "grad_norm": 4.0346784591674805,
      "learning_rate": 6.627840772285784e-05,
      "loss": 1.5023,
      "step": 91
    },
    {
      "epoch": 0.015380114514983074,
      "grad_norm": 4.52296257019043,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.6998,
      "step": 92
    },
    {
      "epoch": 0.015547289672754629,
      "grad_norm": 4.166619777679443,
      "learning_rate": 6.461858523613684e-05,
      "loss": 1.2437,
      "step": 93
    },
    {
      "epoch": 0.015714464830526183,
      "grad_norm": 3.6152470111846924,
      "learning_rate": 6.378186779084995e-05,
      "loss": 0.9291,
      "step": 94
    },
    {
      "epoch": 0.015881639988297738,
      "grad_norm": 5.102316379547119,
      "learning_rate": 6.294095225512603e-05,
      "loss": 1.4993,
      "step": 95
    },
    {
      "epoch": 0.016048815146069293,
      "grad_norm": 3.7798969745635986,
      "learning_rate": 6.209609477998338e-05,
      "loss": 1.4346,
      "step": 96
    },
    {
      "epoch": 0.016215990303840848,
      "grad_norm": 6.013340473175049,
      "learning_rate": 6.124755271719325e-05,
      "loss": 2.4886,
      "step": 97
    },
    {
      "epoch": 0.016383165461612403,
      "grad_norm": 4.5874924659729,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 1.4179,
      "step": 98
    },
    {
      "epoch": 0.01655034061938396,
      "grad_norm": 3.6912167072296143,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 1.276,
      "step": 99
    },
    {
      "epoch": 0.016717515777155516,
      "grad_norm": 5.08207368850708,
      "learning_rate": 5.868240888334653e-05,
      "loss": 2.1187,
      "step": 100
    },
    {
      "epoch": 0.016717515777155516,
      "eval_loss": 0.3840712308883667,
      "eval_runtime": 1180.8021,
      "eval_samples_per_second": 8.532,
      "eval_steps_per_second": 4.267,
      "step": 100
    },
    {
      "epoch": 0.01688469093492707,
      "grad_norm": 8.497836112976074,
      "learning_rate": 5.782172325201155e-05,
      "loss": 1.1939,
      "step": 101
    },
    {
      "epoch": 0.017051866092698625,
      "grad_norm": 7.477349758148193,
      "learning_rate": 5.695865504800327e-05,
      "loss": 2.0787,
      "step": 102
    },
    {
      "epoch": 0.01721904125047018,
      "grad_norm": 7.172429084777832,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 1.6864,
      "step": 103
    },
    {
      "epoch": 0.017386216408241735,
      "grad_norm": 8.384211540222168,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.8249,
      "step": 104
    },
    {
      "epoch": 0.01755339156601329,
      "grad_norm": 5.726856231689453,
      "learning_rate": 5.435778713738292e-05,
      "loss": 1.5703,
      "step": 105
    },
    {
      "epoch": 0.017720566723784845,
      "grad_norm": 6.198292255401611,
      "learning_rate": 5.348782368720626e-05,
      "loss": 2.1273,
      "step": 106
    },
    {
      "epoch": 0.0178877418815564,
      "grad_norm": 4.357252597808838,
      "learning_rate": 5.26167978121472e-05,
      "loss": 1.1151,
      "step": 107
    },
    {
      "epoch": 0.018054917039327954,
      "grad_norm": 4.520287036895752,
      "learning_rate": 5.174497483512506e-05,
      "loss": 1.7011,
      "step": 108
    },
    {
      "epoch": 0.018222092197099513,
      "grad_norm": 3.671816349029541,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 1.0517,
      "step": 109
    },
    {
      "epoch": 0.018389267354871067,
      "grad_norm": 5.582248687744141,
      "learning_rate": 5e-05,
      "loss": 1.6454,
      "step": 110
    },
    {
      "epoch": 0.018556442512642622,
      "grad_norm": 4.559932708740234,
      "learning_rate": 4.912737967813583e-05,
      "loss": 1.6997,
      "step": 111
    },
    {
      "epoch": 0.018723617670414177,
      "grad_norm": 4.87907600402832,
      "learning_rate": 4.825502516487497e-05,
      "loss": 1.6121,
      "step": 112
    },
    {
      "epoch": 0.018890792828185732,
      "grad_norm": 5.985311508178711,
      "learning_rate": 4.738320218785281e-05,
      "loss": 1.894,
      "step": 113
    },
    {
      "epoch": 0.019057967985957287,
      "grad_norm": 5.563694477081299,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 1.4096,
      "step": 114
    },
    {
      "epoch": 0.01922514314372884,
      "grad_norm": 4.012616157531738,
      "learning_rate": 4.564221286261709e-05,
      "loss": 1.1671,
      "step": 115
    },
    {
      "epoch": 0.019392318301500396,
      "grad_norm": 5.445400238037109,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.1218,
      "step": 116
    },
    {
      "epoch": 0.01955949345927195,
      "grad_norm": 4.193072319030762,
      "learning_rate": 4.390653282974264e-05,
      "loss": 1.1577,
      "step": 117
    },
    {
      "epoch": 0.019726668617043506,
      "grad_norm": 4.184262752532959,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 1.3176,
      "step": 118
    },
    {
      "epoch": 0.019893843774815064,
      "grad_norm": 3.5578956604003906,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 1.0705,
      "step": 119
    },
    {
      "epoch": 0.02006101893258662,
      "grad_norm": 4.154637336730957,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.6145,
      "step": 120
    },
    {
      "epoch": 0.020228194090358174,
      "grad_norm": 8.214539527893066,
      "learning_rate": 4.045955023117276e-05,
      "loss": 1.3255,
      "step": 121
    },
    {
      "epoch": 0.02039536924812973,
      "grad_norm": 3.9702048301696777,
      "learning_rate": 3.960441545911204e-05,
      "loss": 1.3089,
      "step": 122
    },
    {
      "epoch": 0.020562544405901283,
      "grad_norm": 4.168085098266602,
      "learning_rate": 3.875244728280676e-05,
      "loss": 1.2921,
      "step": 123
    },
    {
      "epoch": 0.020729719563672838,
      "grad_norm": 4.158751487731934,
      "learning_rate": 3.790390522001662e-05,
      "loss": 1.1961,
      "step": 124
    },
    {
      "epoch": 0.020896894721444393,
      "grad_norm": 3.0978500843048096,
      "learning_rate": 3.705904774487396e-05,
      "loss": 1.027,
      "step": 125
    },
    {
      "epoch": 0.021064069879215948,
      "grad_norm": 4.032132625579834,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 1.1975,
      "step": 126
    },
    {
      "epoch": 0.021231245036987503,
      "grad_norm": 3.2967450618743896,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 1.1724,
      "step": 127
    },
    {
      "epoch": 0.021398420194759057,
      "grad_norm": 4.184244632720947,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.0997,
      "step": 128
    },
    {
      "epoch": 0.021565595352530612,
      "grad_norm": 3.61553955078125,
      "learning_rate": 3.372159227714218e-05,
      "loss": 1.2168,
      "step": 129
    },
    {
      "epoch": 0.02173277051030217,
      "grad_norm": 2.7184441089630127,
      "learning_rate": 3.289899283371657e-05,
      "loss": 0.9683,
      "step": 130
    },
    {
      "epoch": 0.021899945668073725,
      "grad_norm": 2.7383294105529785,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 0.7704,
      "step": 131
    },
    {
      "epoch": 0.02206712082584528,
      "grad_norm": 3.5409862995147705,
      "learning_rate": 3.12696703292044e-05,
      "loss": 0.9742,
      "step": 132
    },
    {
      "epoch": 0.022234295983616835,
      "grad_norm": 2.9273834228515625,
      "learning_rate": 3.046344357553632e-05,
      "loss": 1.0346,
      "step": 133
    },
    {
      "epoch": 0.02240147114138839,
      "grad_norm": 3.1921045780181885,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 1.1151,
      "step": 134
    },
    {
      "epoch": 0.022568646299159945,
      "grad_norm": 3.4800987243652344,
      "learning_rate": 2.886908691296504e-05,
      "loss": 1.0467,
      "step": 135
    },
    {
      "epoch": 0.0227358214569315,
      "grad_norm": 3.3189215660095215,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 1.1769,
      "step": 136
    },
    {
      "epoch": 0.022902996614703054,
      "grad_norm": 3.6555469036102295,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 1.1048,
      "step": 137
    },
    {
      "epoch": 0.02307017177247461,
      "grad_norm": 3.6582908630371094,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 1.1974,
      "step": 138
    },
    {
      "epoch": 0.023237346930246164,
      "grad_norm": 2.606125593185425,
      "learning_rate": 2.575951898768315e-05,
      "loss": 0.7513,
      "step": 139
    },
    {
      "epoch": 0.023404522088017722,
      "grad_norm": 3.5903327465057373,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.2009,
      "step": 140
    },
    {
      "epoch": 0.023571697245789277,
      "grad_norm": 4.257547855377197,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 1.7713,
      "step": 141
    },
    {
      "epoch": 0.02373887240356083,
      "grad_norm": 3.8268728256225586,
      "learning_rate": 2.350403678833976e-05,
      "loss": 1.3315,
      "step": 142
    },
    {
      "epoch": 0.023906047561332387,
      "grad_norm": 3.9802815914154053,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 1.531,
      "step": 143
    },
    {
      "epoch": 0.02407322271910394,
      "grad_norm": 3.5218703746795654,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 0.9738,
      "step": 144
    },
    {
      "epoch": 0.024240397876875496,
      "grad_norm": 3.5325393676757812,
      "learning_rate": 2.132117818244771e-05,
      "loss": 1.1883,
      "step": 145
    },
    {
      "epoch": 0.02440757303464705,
      "grad_norm": 2.8327746391296387,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1112,
      "step": 146
    },
    {
      "epoch": 0.024574748192418606,
      "grad_norm": 3.83736252784729,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 1.6377,
      "step": 147
    },
    {
      "epoch": 0.02474192335019016,
      "grad_norm": 3.671001672744751,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 1.3769,
      "step": 148
    },
    {
      "epoch": 0.024909098507961715,
      "grad_norm": 5.089761257171631,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 2.1632,
      "step": 149
    },
    {
      "epoch": 0.025076273665733274,
      "grad_norm": 11.317509651184082,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 1.8115,
      "step": 150
    },
    {
      "epoch": 0.025076273665733274,
      "eval_loss": 0.33944934606552124,
      "eval_runtime": 1179.9193,
      "eval_samples_per_second": 8.539,
      "eval_steps_per_second": 4.27,
      "step": 150
    }
  ],
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.4929696483311616e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}