{
  "best_metric": 1.6095012426376343,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.37243947858473,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00186219739292365,
      "grad_norm": 1.178178310394287,
      "learning_rate": 5e-06,
      "loss": 2.6211,
      "step": 1
    },
    {
      "epoch": 0.00186219739292365,
      "eval_loss": 3.3070647716522217,
      "eval_runtime": 21.7175,
      "eval_samples_per_second": 41.672,
      "eval_steps_per_second": 20.859,
      "step": 1
    },
    {
      "epoch": 0.0037243947858473,
      "grad_norm": 2.1356124877929688,
      "learning_rate": 1e-05,
      "loss": 2.7551,
      "step": 2
    },
    {
      "epoch": 0.00558659217877095,
      "grad_norm": 1.9746735095977783,
      "learning_rate": 1.5e-05,
      "loss": 3.0001,
      "step": 3
    },
    {
      "epoch": 0.0074487895716946,
      "grad_norm": 1.9692213535308838,
      "learning_rate": 2e-05,
      "loss": 2.8061,
      "step": 4
    },
    {
      "epoch": 0.00931098696461825,
      "grad_norm": 2.3928768634796143,
      "learning_rate": 2.5e-05,
      "loss": 2.6685,
      "step": 5
    },
    {
      "epoch": 0.0111731843575419,
      "grad_norm": 2.2253520488739014,
      "learning_rate": 3e-05,
      "loss": 2.6104,
      "step": 6
    },
    {
      "epoch": 0.01303538175046555,
      "grad_norm": 2.323080539703369,
      "learning_rate": 3.5e-05,
      "loss": 2.689,
      "step": 7
    },
    {
      "epoch": 0.0148975791433892,
      "grad_norm": 2.3765766620635986,
      "learning_rate": 4e-05,
      "loss": 2.9615,
      "step": 8
    },
    {
      "epoch": 0.01675977653631285,
      "grad_norm": 2.05714750289917,
      "learning_rate": 4.5e-05,
      "loss": 2.723,
      "step": 9
    },
    {
      "epoch": 0.0186219739292365,
      "grad_norm": 2.683811902999878,
      "learning_rate": 5e-05,
      "loss": 2.5645,
      "step": 10
    },
    {
      "epoch": 0.020484171322160148,
      "grad_norm": 2.434372663497925,
      "learning_rate": 5.500000000000001e-05,
      "loss": 2.6508,
      "step": 11
    },
    {
      "epoch": 0.0223463687150838,
      "grad_norm": 2.7041714191436768,
      "learning_rate": 6e-05,
      "loss": 2.5503,
      "step": 12
    },
    {
      "epoch": 0.024208566108007448,
      "grad_norm": 2.5691897869110107,
      "learning_rate": 6.500000000000001e-05,
      "loss": 2.6562,
      "step": 13
    },
    {
      "epoch": 0.0260707635009311,
      "grad_norm": 3.0065274238586426,
      "learning_rate": 7e-05,
      "loss": 2.6278,
      "step": 14
    },
    {
      "epoch": 0.027932960893854747,
      "grad_norm": 2.199162244796753,
      "learning_rate": 7.500000000000001e-05,
      "loss": 2.2827,
      "step": 15
    },
    {
      "epoch": 0.0297951582867784,
      "grad_norm": 2.762291431427002,
      "learning_rate": 8e-05,
      "loss": 2.2615,
      "step": 16
    },
    {
      "epoch": 0.03165735567970205,
      "grad_norm": 2.3539557456970215,
      "learning_rate": 8.5e-05,
      "loss": 2.3974,
      "step": 17
    },
    {
      "epoch": 0.0335195530726257,
      "grad_norm": 2.7445669174194336,
      "learning_rate": 9e-05,
      "loss": 2.2605,
      "step": 18
    },
    {
      "epoch": 0.035381750465549346,
      "grad_norm": 2.828781843185425,
      "learning_rate": 9.5e-05,
      "loss": 2.4178,
      "step": 19
    },
    {
      "epoch": 0.037243947858473,
      "grad_norm": 2.554708480834961,
      "learning_rate": 0.0001,
      "loss": 2.1424,
      "step": 20
    },
    {
      "epoch": 0.03910614525139665,
      "grad_norm": 2.6451961994171143,
      "learning_rate": 9.999238475781957e-05,
      "loss": 2.1344,
      "step": 21
    },
    {
      "epoch": 0.040968342644320296,
      "grad_norm": 2.7068910598754883,
      "learning_rate": 9.99695413509548e-05,
      "loss": 2.2556,
      "step": 22
    },
    {
      "epoch": 0.04283054003724395,
      "grad_norm": 2.764464855194092,
      "learning_rate": 9.99314767377287e-05,
      "loss": 2.3873,
      "step": 23
    },
    {
      "epoch": 0.0446927374301676,
      "grad_norm": 2.798506021499634,
      "learning_rate": 9.987820251299122e-05,
      "loss": 2.2367,
      "step": 24
    },
    {
      "epoch": 0.04655493482309125,
      "grad_norm": 2.565629243850708,
      "learning_rate": 9.980973490458728e-05,
      "loss": 2.4074,
      "step": 25
    },
    {
      "epoch": 0.048417132216014895,
      "grad_norm": 2.5067367553710938,
      "learning_rate": 9.972609476841367e-05,
      "loss": 2.0703,
      "step": 26
    },
    {
      "epoch": 0.05027932960893855,
      "grad_norm": 3.635667562484741,
      "learning_rate": 9.962730758206611e-05,
      "loss": 2.1721,
      "step": 27
    },
    {
      "epoch": 0.0521415270018622,
      "grad_norm": 3.774707317352295,
      "learning_rate": 9.951340343707852e-05,
      "loss": 2.1216,
      "step": 28
    },
    {
      "epoch": 0.054003724394785846,
      "grad_norm": 2.8153574466705322,
      "learning_rate": 9.938441702975689e-05,
      "loss": 2.2958,
      "step": 29
    },
    {
      "epoch": 0.055865921787709494,
      "grad_norm": 3.2285168170928955,
      "learning_rate": 9.924038765061042e-05,
      "loss": 2.3483,
      "step": 30
    },
    {
      "epoch": 0.05772811918063315,
      "grad_norm": 4.216028690338135,
      "learning_rate": 9.908135917238321e-05,
      "loss": 2.566,
      "step": 31
    },
    {
      "epoch": 0.0595903165735568,
      "grad_norm": 3.771660804748535,
      "learning_rate": 9.890738003669029e-05,
      "loss": 2.1817,
      "step": 32
    },
    {
      "epoch": 0.061452513966480445,
      "grad_norm": 3.534721851348877,
      "learning_rate": 9.871850323926177e-05,
      "loss": 2.2499,
      "step": 33
    },
    {
      "epoch": 0.0633147113594041,
      "grad_norm": 2.9690186977386475,
      "learning_rate": 9.851478631379982e-05,
      "loss": 2.243,
      "step": 34
    },
    {
      "epoch": 0.06517690875232775,
      "grad_norm": 3.739225149154663,
      "learning_rate": 9.829629131445342e-05,
      "loss": 2.3059,
      "step": 35
    },
    {
      "epoch": 0.0670391061452514,
      "grad_norm": 3.2151126861572266,
      "learning_rate": 9.806308479691595e-05,
      "loss": 2.3284,
      "step": 36
    },
    {
      "epoch": 0.06890130353817504,
      "grad_norm": 3.2973902225494385,
      "learning_rate": 9.781523779815179e-05,
      "loss": 2.8754,
      "step": 37
    },
    {
      "epoch": 0.07076350093109869,
      "grad_norm": 3.0244390964508057,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.6087,
      "step": 38
    },
    {
      "epoch": 0.07262569832402235,
      "grad_norm": 3.9037907123565674,
      "learning_rate": 9.727592877996585e-05,
      "loss": 2.3161,
      "step": 39
    },
    {
      "epoch": 0.074487895716946,
      "grad_norm": 3.0399837493896484,
      "learning_rate": 9.698463103929542e-05,
      "loss": 2.9208,
      "step": 40
    },
    {
      "epoch": 0.07635009310986965,
      "grad_norm": 2.6149749755859375,
      "learning_rate": 9.667902132486009e-05,
      "loss": 2.7519,
      "step": 41
    },
    {
      "epoch": 0.0782122905027933,
      "grad_norm": 2.8330459594726562,
      "learning_rate": 9.635919272833938e-05,
      "loss": 3.002,
      "step": 42
    },
    {
      "epoch": 0.08007448789571694,
      "grad_norm": 3.024914264678955,
      "learning_rate": 9.602524267262203e-05,
      "loss": 2.7255,
      "step": 43
    },
    {
      "epoch": 0.08193668528864059,
      "grad_norm": 3.103067398071289,
      "learning_rate": 9.567727288213005e-05,
      "loss": 2.9374,
      "step": 44
    },
    {
      "epoch": 0.08379888268156424,
      "grad_norm": 4.024307727813721,
      "learning_rate": 9.53153893518325e-05,
      "loss": 2.4808,
      "step": 45
    },
    {
      "epoch": 0.0856610800744879,
      "grad_norm": 3.7541255950927734,
      "learning_rate": 9.493970231495835e-05,
      "loss": 2.9799,
      "step": 46
    },
    {
      "epoch": 0.08752327746741155,
      "grad_norm": 3.815789222717285,
      "learning_rate": 9.45503262094184e-05,
      "loss": 3.0052,
      "step": 47
    },
    {
      "epoch": 0.0893854748603352,
      "grad_norm": 4.527659893035889,
      "learning_rate": 9.414737964294636e-05,
      "loss": 3.3865,
      "step": 48
    },
    {
      "epoch": 0.09124767225325885,
      "grad_norm": 4.477816581726074,
      "learning_rate": 9.373098535696979e-05,
      "loss": 3.0642,
      "step": 49
    },
    {
      "epoch": 0.0931098696461825,
      "grad_norm": 5.919867515563965,
      "learning_rate": 9.330127018922194e-05,
      "loss": 3.1839,
      "step": 50
    },
    {
      "epoch": 0.0931098696461825,
      "eval_loss": 2.5658836364746094,
      "eval_runtime": 21.709,
      "eval_samples_per_second": 41.688,
      "eval_steps_per_second": 20.867,
      "step": 50
    },
    {
      "epoch": 0.09497206703910614,
      "grad_norm": 4.072330474853516,
      "learning_rate": 9.285836503510562e-05,
      "loss": 2.6279,
      "step": 51
    },
    {
      "epoch": 0.09683426443202979,
      "grad_norm": 3.412520408630371,
      "learning_rate": 9.24024048078213e-05,
      "loss": 2.712,
      "step": 52
    },
    {
      "epoch": 0.09869646182495345,
      "grad_norm": 2.8926405906677246,
      "learning_rate": 9.193352839727121e-05,
      "loss": 2.6263,
      "step": 53
    },
    {
      "epoch": 0.1005586592178771,
      "grad_norm": 2.1868202686309814,
      "learning_rate": 9.145187862775209e-05,
      "loss": 2.3611,
      "step": 54
    },
    {
      "epoch": 0.10242085661080075,
      "grad_norm": 1.9672703742980957,
      "learning_rate": 9.09576022144496e-05,
      "loss": 2.1171,
      "step": 55
    },
    {
      "epoch": 0.1042830540037244,
      "grad_norm": 1.7502118349075317,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.0406,
      "step": 56
    },
    {
      "epoch": 0.10614525139664804,
      "grad_norm": 1.7855284214019775,
      "learning_rate": 8.993177550236464e-05,
      "loss": 2.2114,
      "step": 57
    },
    {
      "epoch": 0.10800744878957169,
      "grad_norm": 1.6733700037002563,
      "learning_rate": 8.940053768033609e-05,
      "loss": 2.1213,
      "step": 58
    },
    {
      "epoch": 0.10986964618249534,
      "grad_norm": 2.0783803462982178,
      "learning_rate": 8.885729807284856e-05,
      "loss": 2.0092,
      "step": 59
    },
    {
      "epoch": 0.11173184357541899,
      "grad_norm": 1.8364993333816528,
      "learning_rate": 8.83022221559489e-05,
      "loss": 2.0301,
      "step": 60
    },
    {
      "epoch": 0.11359404096834265,
      "grad_norm": 2.194591999053955,
      "learning_rate": 8.773547901113862e-05,
      "loss": 1.9808,
      "step": 61
    },
    {
      "epoch": 0.1154562383612663,
      "grad_norm": 2.308046579360962,
      "learning_rate": 8.715724127386972e-05,
      "loss": 2.0264,
      "step": 62
    },
    {
      "epoch": 0.11731843575418995,
      "grad_norm": 2.148824691772461,
      "learning_rate": 8.656768508095853e-05,
      "loss": 1.8808,
      "step": 63
    },
    {
      "epoch": 0.1191806331471136,
      "grad_norm": 2.341665029525757,
      "learning_rate": 8.596699001693255e-05,
      "loss": 2.1458,
      "step": 64
    },
    {
      "epoch": 0.12104283054003724,
      "grad_norm": 1.7752113342285156,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.6025,
      "step": 65
    },
    {
      "epoch": 0.12290502793296089,
      "grad_norm": 1.866705060005188,
      "learning_rate": 8.473291852294987e-05,
      "loss": 1.8797,
      "step": 66
    },
    {
      "epoch": 0.12476722532588454,
      "grad_norm": 2.4019017219543457,
      "learning_rate": 8.409991800312493e-05,
      "loss": 1.6268,
      "step": 67
    },
    {
      "epoch": 0.1266294227188082,
      "grad_norm": 2.2571306228637695,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.688,
      "step": 68
    },
    {
      "epoch": 0.12849162011173185,
      "grad_norm": 2.4326064586639404,
      "learning_rate": 8.280295144952536e-05,
      "loss": 1.7,
      "step": 69
    },
    {
      "epoch": 0.1303538175046555,
      "grad_norm": 1.9197702407836914,
      "learning_rate": 8.213938048432697e-05,
      "loss": 1.6002,
      "step": 70
    },
    {
      "epoch": 0.13221601489757914,
      "grad_norm": 2.502194881439209,
      "learning_rate": 8.146601955249188e-05,
      "loss": 1.6403,
      "step": 71
    },
    {
      "epoch": 0.1340782122905028,
      "grad_norm": 2.363044261932373,
      "learning_rate": 8.07830737662829e-05,
      "loss": 1.6535,
      "step": 72
    },
    {
      "epoch": 0.13594040968342644,
      "grad_norm": 2.375981569290161,
      "learning_rate": 8.009075115760243e-05,
      "loss": 1.6231,
      "step": 73
    },
    {
      "epoch": 0.1378026070763501,
      "grad_norm": 3.1262547969818115,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.7436,
      "step": 74
    },
    {
      "epoch": 0.13966480446927373,
      "grad_norm": 2.7820394039154053,
      "learning_rate": 7.86788218175523e-05,
      "loss": 1.8905,
      "step": 75
    },
    {
      "epoch": 0.14152700186219738,
      "grad_norm": 2.7952208518981934,
      "learning_rate": 7.795964517353735e-05,
      "loss": 1.485,
      "step": 76
    },
    {
      "epoch": 0.14338919925512103,
      "grad_norm": 2.5299875736236572,
      "learning_rate": 7.723195175075136e-05,
      "loss": 1.7068,
      "step": 77
    },
    {
      "epoch": 0.1452513966480447,
      "grad_norm": 2.388131618499756,
      "learning_rate": 7.649596321166024e-05,
      "loss": 1.702,
      "step": 78
    },
    {
      "epoch": 0.14711359404096835,
      "grad_norm": 2.770792245864868,
      "learning_rate": 7.575190374550272e-05,
      "loss": 1.7674,
      "step": 79
    },
    {
      "epoch": 0.148975791433892,
      "grad_norm": 3.067246675491333,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.6699,
      "step": 80
    },
    {
      "epoch": 0.15083798882681565,
      "grad_norm": 2.8982298374176025,
      "learning_rate": 7.424048101231686e-05,
      "loss": 1.6871,
      "step": 81
    },
    {
      "epoch": 0.1527001862197393,
      "grad_norm": 3.0862808227539062,
      "learning_rate": 7.347357813929454e-05,
      "loss": 1.6892,
      "step": 82
    },
    {
      "epoch": 0.15456238361266295,
      "grad_norm": 2.5842466354370117,
      "learning_rate": 7.269952498697734e-05,
      "loss": 1.5902,
      "step": 83
    },
    {
      "epoch": 0.1564245810055866,
      "grad_norm": 2.8635683059692383,
      "learning_rate": 7.191855733945387e-05,
      "loss": 1.599,
      "step": 84
    },
    {
      "epoch": 0.15828677839851024,
      "grad_norm": 3.0610272884368896,
      "learning_rate": 7.113091308703498e-05,
      "loss": 1.6147,
      "step": 85
    },
    {
      "epoch": 0.1601489757914339,
      "grad_norm": 2.9081881046295166,
      "learning_rate": 7.033683215379002e-05,
      "loss": 1.8976,
      "step": 86
    },
    {
      "epoch": 0.16201117318435754,
      "grad_norm": 2.963331460952759,
      "learning_rate": 6.953655642446368e-05,
      "loss": 2.3989,
      "step": 87
    },
    {
      "epoch": 0.16387337057728119,
      "grad_norm": 3.06968092918396,
      "learning_rate": 6.873032967079561e-05,
      "loss": 1.677,
      "step": 88
    },
    {
      "epoch": 0.16573556797020483,
      "grad_norm": 2.6805789470672607,
      "learning_rate": 6.7918397477265e-05,
      "loss": 2.3708,
      "step": 89
    },
    {
      "epoch": 0.16759776536312848,
      "grad_norm": 2.7528600692749023,
      "learning_rate": 6.710100716628344e-05,
      "loss": 2.8791,
      "step": 90
    },
    {
      "epoch": 0.16945996275605213,
      "grad_norm": 2.7579615116119385,
      "learning_rate": 6.627840772285784e-05,
      "loss": 2.9369,
      "step": 91
    },
    {
      "epoch": 0.1713221601489758,
      "grad_norm": 2.825394630432129,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.9602,
      "step": 92
    },
    {
      "epoch": 0.17318435754189945,
      "grad_norm": 2.8151633739471436,
      "learning_rate": 6.461858523613684e-05,
      "loss": 2.6122,
      "step": 93
    },
    {
      "epoch": 0.1750465549348231,
      "grad_norm": 2.9391984939575195,
      "learning_rate": 6.378186779084995e-05,
      "loss": 2.7571,
      "step": 94
    },
    {
      "epoch": 0.17690875232774675,
      "grad_norm": 2.8012423515319824,
      "learning_rate": 6.294095225512603e-05,
      "loss": 2.4203,
      "step": 95
    },
    {
      "epoch": 0.1787709497206704,
      "grad_norm": 3.2889254093170166,
      "learning_rate": 6.209609477998338e-05,
      "loss": 2.263,
      "step": 96
    },
    {
      "epoch": 0.18063314711359404,
      "grad_norm": 3.5561633110046387,
      "learning_rate": 6.124755271719325e-05,
      "loss": 2.6601,
      "step": 97
    },
    {
      "epoch": 0.1824953445065177,
      "grad_norm": 3.6981186866760254,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 3.0337,
      "step": 98
    },
    {
      "epoch": 0.18435754189944134,
      "grad_norm": 3.9297776222229004,
      "learning_rate": 5.9540449768827246e-05,
      "loss": 2.9557,
      "step": 99
    },
    {
      "epoch": 0.186219739292365,
      "grad_norm": 5.433001518249512,
      "learning_rate": 5.868240888334653e-05,
      "loss": 3.6159,
      "step": 100
    },
    {
      "epoch": 0.186219739292365,
      "eval_loss": 2.111635208129883,
      "eval_runtime": 21.7221,
      "eval_samples_per_second": 41.663,
      "eval_steps_per_second": 20.854,
      "step": 100
    },
    {
      "epoch": 0.18808193668528864,
      "grad_norm": 2.3293561935424805,
      "learning_rate": 5.782172325201155e-05,
      "loss": 2.4448,
      "step": 101
    },
    {
      "epoch": 0.18994413407821228,
      "grad_norm": 2.2285311222076416,
      "learning_rate": 5.695865504800327e-05,
      "loss": 2.4747,
      "step": 102
    },
    {
      "epoch": 0.19180633147113593,
      "grad_norm": 2.032970666885376,
      "learning_rate": 5.6093467170257374e-05,
      "loss": 2.5569,
      "step": 103
    },
    {
      "epoch": 0.19366852886405958,
      "grad_norm": 1.9000310897827148,
      "learning_rate": 5.522642316338268e-05,
      "loss": 2.2785,
      "step": 104
    },
    {
      "epoch": 0.19553072625698323,
      "grad_norm": 2.129242181777954,
      "learning_rate": 5.435778713738292e-05,
      "loss": 1.8737,
      "step": 105
    },
    {
      "epoch": 0.1973929236499069,
      "grad_norm": 1.6860032081604004,
      "learning_rate": 5.348782368720626e-05,
      "loss": 1.8208,
      "step": 106
    },
    {
      "epoch": 0.19925512104283055,
      "grad_norm": 1.7974435091018677,
      "learning_rate": 5.26167978121472e-05,
      "loss": 1.6613,
      "step": 107
    },
    {
      "epoch": 0.2011173184357542,
      "grad_norm": 1.80308198928833,
      "learning_rate": 5.174497483512506e-05,
      "loss": 1.9122,
      "step": 108
    },
    {
      "epoch": 0.20297951582867785,
      "grad_norm": 1.9534099102020264,
      "learning_rate": 5.0872620321864185e-05,
      "loss": 1.6929,
      "step": 109
    },
    {
      "epoch": 0.2048417132216015,
      "grad_norm": 1.9603544473648071,
      "learning_rate": 5e-05,
      "loss": 1.7386,
      "step": 110
    },
    {
      "epoch": 0.20670391061452514,
      "grad_norm": 1.875958800315857,
      "learning_rate": 4.912737967813583e-05,
      "loss": 1.675,
      "step": 111
    },
    {
      "epoch": 0.2085661080074488,
      "grad_norm": 2.435465097427368,
      "learning_rate": 4.825502516487497e-05,
      "loss": 1.6366,
      "step": 112
    },
    {
      "epoch": 0.21042830540037244,
      "grad_norm": 2.0310440063476562,
      "learning_rate": 4.738320218785281e-05,
      "loss": 1.5549,
      "step": 113
    },
    {
      "epoch": 0.2122905027932961,
      "grad_norm": 2.066591501235962,
      "learning_rate": 4.6512176312793736e-05,
      "loss": 1.6041,
      "step": 114
    },
    {
      "epoch": 0.21415270018621974,
      "grad_norm": 2.1647772789001465,
      "learning_rate": 4.564221286261709e-05,
      "loss": 1.571,
      "step": 115
    },
    {
      "epoch": 0.21601489757914338,
      "grad_norm": 2.1143767833709717,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.3452,
      "step": 116
    },
    {
      "epoch": 0.21787709497206703,
      "grad_norm": 2.4085280895233154,
      "learning_rate": 4.390653282974264e-05,
      "loss": 1.5415,
      "step": 117
    },
    {
      "epoch": 0.21973929236499068,
      "grad_norm": 2.171875238418579,
      "learning_rate": 4.3041344951996746e-05,
      "loss": 1.2254,
      "step": 118
    },
    {
      "epoch": 0.22160148975791433,
      "grad_norm": 2.8386919498443604,
      "learning_rate": 4.2178276747988446e-05,
      "loss": 1.4564,
      "step": 119
    },
    {
      "epoch": 0.22346368715083798,
      "grad_norm": 2.413219928741455,
      "learning_rate": 4.131759111665349e-05,
      "loss": 1.2426,
      "step": 120
    },
    {
      "epoch": 0.22532588454376165,
      "grad_norm": 2.3246631622314453,
      "learning_rate": 4.045955023117276e-05,
      "loss": 1.1519,
      "step": 121
    },
    {
      "epoch": 0.2271880819366853,
      "grad_norm": 3.1036009788513184,
      "learning_rate": 3.960441545911204e-05,
      "loss": 1.3616,
      "step": 122
    },
    {
      "epoch": 0.22905027932960895,
      "grad_norm": 2.8557889461517334,
      "learning_rate": 3.875244728280676e-05,
      "loss": 1.2942,
      "step": 123
    },
    {
      "epoch": 0.2309124767225326,
      "grad_norm": 2.639146089553833,
      "learning_rate": 3.790390522001662e-05,
      "loss": 1.341,
      "step": 124
    },
    {
      "epoch": 0.23277467411545624,
      "grad_norm": 2.424593210220337,
      "learning_rate": 3.705904774487396e-05,
      "loss": 1.5446,
      "step": 125
    },
    {
      "epoch": 0.2346368715083799,
      "grad_norm": 2.4945600032806396,
      "learning_rate": 3.6218132209150045e-05,
      "loss": 1.2111,
      "step": 126
    },
    {
      "epoch": 0.23649906890130354,
      "grad_norm": 3.035121440887451,
      "learning_rate": 3.5381414763863166e-05,
      "loss": 1.2049,
      "step": 127
    },
    {
      "epoch": 0.2383612662942272,
      "grad_norm": 3.434382438659668,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.301,
      "step": 128
    },
    {
      "epoch": 0.24022346368715083,
      "grad_norm": 2.730940341949463,
      "learning_rate": 3.372159227714218e-05,
      "loss": 1.2729,
      "step": 129
    },
    {
      "epoch": 0.24208566108007448,
      "grad_norm": 3.3356800079345703,
      "learning_rate": 3.289899283371657e-05,
      "loss": 1.3789,
      "step": 130
    },
    {
      "epoch": 0.24394785847299813,
      "grad_norm": 3.292501926422119,
      "learning_rate": 3.2081602522734986e-05,
      "loss": 1.4141,
      "step": 131
    },
    {
      "epoch": 0.24581005586592178,
      "grad_norm": 3.2377631664276123,
      "learning_rate": 3.12696703292044e-05,
      "loss": 1.2394,
      "step": 132
    },
    {
      "epoch": 0.24767225325884543,
      "grad_norm": 3.1759567260742188,
      "learning_rate": 3.046344357553632e-05,
      "loss": 1.2913,
      "step": 133
    },
    {
      "epoch": 0.24953445065176907,
      "grad_norm": 2.8026719093322754,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 1.2144,
      "step": 134
    },
    {
      "epoch": 0.25139664804469275,
      "grad_norm": 2.286036968231201,
      "learning_rate": 2.886908691296504e-05,
      "loss": 1.3749,
      "step": 135
    },
    {
      "epoch": 0.2532588454376164,
      "grad_norm": 1.9794367551803589,
      "learning_rate": 2.8081442660546125e-05,
      "loss": 1.7901,
      "step": 136
    },
    {
      "epoch": 0.25512104283054005,
      "grad_norm": 2.388559103012085,
      "learning_rate": 2.7300475013022663e-05,
      "loss": 2.1486,
      "step": 137
    },
    {
      "epoch": 0.2569832402234637,
      "grad_norm": 2.6771240234375,
      "learning_rate": 2.6526421860705473e-05,
      "loss": 1.4815,
      "step": 138
    },
    {
      "epoch": 0.25884543761638734,
      "grad_norm": 2.727090358734131,
      "learning_rate": 2.575951898768315e-05,
      "loss": 2.4409,
      "step": 139
    },
    {
      "epoch": 0.260707635009311,
      "grad_norm": 3.0475521087646484,
      "learning_rate": 2.500000000000001e-05,
      "loss": 3.0248,
      "step": 140
    },
    {
      "epoch": 0.26256983240223464,
      "grad_norm": 2.9400222301483154,
      "learning_rate": 2.4248096254497288e-05,
      "loss": 2.5705,
      "step": 141
    },
    {
      "epoch": 0.2644320297951583,
      "grad_norm": 3.1902108192443848,
      "learning_rate": 2.350403678833976e-05,
      "loss": 2.8102,
      "step": 142
    },
    {
      "epoch": 0.26629422718808193,
      "grad_norm": 3.4387946128845215,
      "learning_rate": 2.2768048249248648e-05,
      "loss": 2.7094,
      "step": 143
    },
    {
      "epoch": 0.2681564245810056,
      "grad_norm": 3.343712329864502,
      "learning_rate": 2.2040354826462668e-05,
      "loss": 2.5977,
      "step": 144
    },
    {
      "epoch": 0.27001862197392923,
      "grad_norm": 3.5541982650756836,
      "learning_rate": 2.132117818244771e-05,
      "loss": 2.6829,
      "step": 145
    },
    {
      "epoch": 0.2718808193668529,
      "grad_norm": 3.7226669788360596,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.748,
      "step": 146
    },
    {
      "epoch": 0.2737430167597765,
      "grad_norm": 3.643068790435791,
      "learning_rate": 1.9909248842397584e-05,
      "loss": 2.8173,
      "step": 147
    },
    {
      "epoch": 0.2756052141527002,
      "grad_norm": 4.142628192901611,
      "learning_rate": 1.9216926233717085e-05,
      "loss": 2.7241,
      "step": 148
    },
    {
      "epoch": 0.2774674115456238,
      "grad_norm": 4.557801246643066,
      "learning_rate": 1.8533980447508137e-05,
      "loss": 2.8083,
      "step": 149
    },
    {
      "epoch": 0.27932960893854747,
      "grad_norm": 5.352628707885742,
      "learning_rate": 1.7860619515673033e-05,
      "loss": 3.1628,
      "step": 150
    },
    {
      "epoch": 0.27932960893854747,
      "eval_loss": 1.7213300466537476,
      "eval_runtime": 21.823,
      "eval_samples_per_second": 41.47,
      "eval_steps_per_second": 20.758,
      "step": 150
    },
    {
      "epoch": 0.2811918063314711,
      "grad_norm": 3.0417568683624268,
      "learning_rate": 1.7197048550474643e-05,
      "loss": 2.3803,
      "step": 151
    },
    {
      "epoch": 0.28305400372439476,
      "grad_norm": 3.4583489894866943,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 2.3051,
      "step": 152
    },
    {
      "epoch": 0.2849162011173184,
      "grad_norm": 3.545855760574341,
      "learning_rate": 1.5900081996875083e-05,
      "loss": 2.4062,
      "step": 153
    },
    {
      "epoch": 0.28677839851024206,
      "grad_norm": 3.136542797088623,
      "learning_rate": 1.526708147705013e-05,
      "loss": 2.1766,
      "step": 154
    },
    {
      "epoch": 0.2886405959031657,
      "grad_norm": 2.3954520225524902,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 1.86,
      "step": 155
    },
    {
      "epoch": 0.2905027932960894,
      "grad_norm": 2.5641558170318604,
      "learning_rate": 1.4033009983067452e-05,
      "loss": 1.776,
      "step": 156
    },
    {
      "epoch": 0.29236499068901306,
      "grad_norm": 3.0906219482421875,
      "learning_rate": 1.3432314919041478e-05,
      "loss": 1.5075,
      "step": 157
    },
    {
      "epoch": 0.2942271880819367,
      "grad_norm": 2.7268283367156982,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 1.7205,
      "step": 158
    },
    {
      "epoch": 0.29608938547486036,
      "grad_norm": 2.318650722503662,
      "learning_rate": 1.22645209888614e-05,
      "loss": 1.5238,
      "step": 159
    },
    {
      "epoch": 0.297951582867784,
      "grad_norm": 2.666076898574829,
      "learning_rate": 1.1697777844051105e-05,
      "loss": 1.4581,
      "step": 160
    },
    {
      "epoch": 0.29981378026070765,
      "grad_norm": 2.5508248805999756,
      "learning_rate": 1.1142701927151456e-05,
      "loss": 1.3903,
      "step": 161
    },
    {
      "epoch": 0.3016759776536313,
      "grad_norm": 2.377032518386841,
      "learning_rate": 1.0599462319663905e-05,
      "loss": 1.6382,
      "step": 162
    },
    {
      "epoch": 0.30353817504655495,
      "grad_norm": 2.2218053340911865,
      "learning_rate": 1.006822449763537e-05,
      "loss": 1.3799,
      "step": 163
    },
    {
      "epoch": 0.3054003724394786,
      "grad_norm": 2.3433680534362793,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.3963,
      "step": 164
    },
    {
      "epoch": 0.30726256983240224,
      "grad_norm": 2.080559730529785,
      "learning_rate": 9.042397785550405e-06,
      "loss": 1.5325,
      "step": 165
    },
    {
      "epoch": 0.3091247672253259,
      "grad_norm": 2.0785741806030273,
      "learning_rate": 8.548121372247918e-06,
      "loss": 1.3357,
      "step": 166
    },
    {
      "epoch": 0.31098696461824954,
      "grad_norm": 2.0232858657836914,
      "learning_rate": 8.066471602728803e-06,
      "loss": 1.1811,
      "step": 167
    },
    {
      "epoch": 0.3128491620111732,
      "grad_norm": 2.474153757095337,
      "learning_rate": 7.597595192178702e-06,
      "loss": 1.3566,
      "step": 168
    },
    {
      "epoch": 0.31471135940409684,
      "grad_norm": 2.2368853092193604,
      "learning_rate": 7.1416349648943894e-06,
      "loss": 1.2236,
      "step": 169
    },
    {
      "epoch": 0.3165735567970205,
      "grad_norm": 2.4082045555114746,
      "learning_rate": 6.698729810778065e-06,
      "loss": 1.2531,
      "step": 170
    },
    {
      "epoch": 0.31843575418994413,
      "grad_norm": 2.352832078933716,
      "learning_rate": 6.269014643030213e-06,
      "loss": 1.2055,
      "step": 171
    },
    {
      "epoch": 0.3202979515828678,
      "grad_norm": 2.19207763671875,
      "learning_rate": 5.852620357053651e-06,
      "loss": 1.0017,
      "step": 172
    },
    {
      "epoch": 0.3221601489757914,
      "grad_norm": 2.4302475452423096,
      "learning_rate": 5.449673790581611e-06,
      "loss": 1.1596,
      "step": 173
    },
    {
      "epoch": 0.3240223463687151,
      "grad_norm": 2.7753303050994873,
      "learning_rate": 5.060297685041659e-06,
      "loss": 1.0642,
      "step": 174
    },
    {
      "epoch": 0.3258845437616387,
      "grad_norm": 2.511533737182617,
      "learning_rate": 4.684610648167503e-06,
      "loss": 1.1777,
      "step": 175
    },
    {
      "epoch": 0.32774674115456237,
      "grad_norm": 2.498866081237793,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.1822,
      "step": 176
    },
    {
      "epoch": 0.329608938547486,
      "grad_norm": 2.347407102584839,
      "learning_rate": 3.974757327377981e-06,
      "loss": 0.8817,
      "step": 177
    },
    {
      "epoch": 0.33147113594040967,
      "grad_norm": 2.721827268600464,
      "learning_rate": 3.6408072716606346e-06,
      "loss": 1.0618,
      "step": 178
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 2.5452587604522705,
      "learning_rate": 3.3209786751399187e-06,
      "loss": 1.0357,
      "step": 179
    },
    {
      "epoch": 0.33519553072625696,
      "grad_norm": 2.477797269821167,
      "learning_rate": 3.0153689607045845e-06,
      "loss": 1.155,
      "step": 180
    },
    {
      "epoch": 0.3370577281191806,
      "grad_norm": 2.593416929244995,
      "learning_rate": 2.724071220034158e-06,
      "loss": 1.1718,
      "step": 181
    },
    {
      "epoch": 0.33891992551210426,
      "grad_norm": 2.054661512374878,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.1298,
      "step": 182
    },
    {
      "epoch": 0.3407821229050279,
      "grad_norm": 2.7391607761383057,
      "learning_rate": 2.1847622018482283e-06,
      "loss": 0.9452,
      "step": 183
    },
    {
      "epoch": 0.3426443202979516,
      "grad_norm": 2.3026256561279297,
      "learning_rate": 1.9369152030840556e-06,
      "loss": 0.9141,
      "step": 184
    },
    {
      "epoch": 0.34450651769087526,
      "grad_norm": 2.6749353408813477,
      "learning_rate": 1.70370868554659e-06,
      "loss": 1.0283,
      "step": 185
    },
    {
      "epoch": 0.3463687150837989,
      "grad_norm": 2.3529865741729736,
      "learning_rate": 1.4852136862001764e-06,
      "loss": 1.211,
      "step": 186
    },
    {
      "epoch": 0.34823091247672255,
      "grad_norm": 2.1153926849365234,
      "learning_rate": 1.2814967607382432e-06,
      "loss": 1.7651,
      "step": 187
    },
    {
      "epoch": 0.3500931098696462,
      "grad_norm": 2.1270318031311035,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 1.6019,
      "step": 188
    },
    {
      "epoch": 0.35195530726256985,
      "grad_norm": 2.39420485496521,
      "learning_rate": 9.186408276168013e-07,
      "loss": 2.1101,
      "step": 189
    },
    {
      "epoch": 0.3538175046554935,
      "grad_norm": 2.7849035263061523,
      "learning_rate": 7.596123493895991e-07,
      "loss": 2.3423,
      "step": 190
    },
    {
      "epoch": 0.35567970204841715,
      "grad_norm": 2.613570213317871,
      "learning_rate": 6.15582970243117e-07,
      "loss": 2.3859,
      "step": 191
    },
    {
      "epoch": 0.3575418994413408,
      "grad_norm": 2.6046884059906006,
      "learning_rate": 4.865965629214819e-07,
      "loss": 2.4552,
      "step": 192
    },
    {
      "epoch": 0.35940409683426444,
      "grad_norm": 2.9097020626068115,
      "learning_rate": 3.7269241793390085e-07,
      "loss": 2.5263,
      "step": 193
    },
    {
      "epoch": 0.3612662942271881,
      "grad_norm": 3.0416860580444336,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 2.4687,
      "step": 194
    },
    {
      "epoch": 0.36312849162011174,
      "grad_norm": 3.228081464767456,
      "learning_rate": 1.9026509541272275e-07,
      "loss": 2.5028,
      "step": 195
    },
    {
      "epoch": 0.3649906890130354,
      "grad_norm": 3.299856662750244,
      "learning_rate": 1.2179748700879012e-07,
      "loss": 2.4265,
      "step": 196
    },
    {
      "epoch": 0.36685288640595903,
      "grad_norm": 3.205040693283081,
      "learning_rate": 6.852326227130834e-08,
      "loss": 2.3468,
      "step": 197
    },
    {
      "epoch": 0.3687150837988827,
      "grad_norm": 3.343839168548584,
      "learning_rate": 3.04586490452119e-08,
      "loss": 2.9085,
      "step": 198
    },
    {
      "epoch": 0.37057728119180633,
      "grad_norm": 3.3727385997772217,
      "learning_rate": 7.615242180436522e-09,
      "loss": 2.5751,
      "step": 199
    },
    {
      "epoch": 0.37243947858473,
      "grad_norm": 5.750079154968262,
      "learning_rate": 0.0,
      "loss": 2.8956,
      "step": 200
    },
    {
      "epoch": 0.37243947858473,
      "eval_loss": 1.6095012426376343,
      "eval_runtime": 21.7692,
      "eval_samples_per_second": 41.573,
      "eval_steps_per_second": 20.809,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.033891434233856e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}