{
  "best_metric": 0.08097303658723831,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.4260249554367201,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0071301247771836,
      "grad_norm": 4.860265731811523,
      "learning_rate": 1e-05,
      "loss": 2.2642,
      "step": 1
    },
    {
      "epoch": 0.0071301247771836,
      "eval_loss": 3.368955373764038,
      "eval_runtime": 17.5564,
      "eval_samples_per_second": 13.442,
      "eval_steps_per_second": 3.361,
      "step": 1
    },
    {
      "epoch": 0.0142602495543672,
      "grad_norm": 4.241779327392578,
      "learning_rate": 2e-05,
      "loss": 2.4353,
      "step": 2
    },
    {
      "epoch": 0.0213903743315508,
      "grad_norm": 5.996744632720947,
      "learning_rate": 3e-05,
      "loss": 2.4873,
      "step": 3
    },
    {
      "epoch": 0.0285204991087344,
      "grad_norm": 4.913257122039795,
      "learning_rate": 4e-05,
      "loss": 2.3505,
      "step": 4
    },
    {
      "epoch": 0.035650623885918005,
      "grad_norm": 3.3362646102905273,
      "learning_rate": 5e-05,
      "loss": 2.137,
      "step": 5
    },
    {
      "epoch": 0.0427807486631016,
      "grad_norm": 3.666640281677246,
      "learning_rate": 6e-05,
      "loss": 1.7876,
      "step": 6
    },
    {
      "epoch": 0.049910873440285206,
      "grad_norm": 6.828310012817383,
      "learning_rate": 7e-05,
      "loss": 1.755,
      "step": 7
    },
    {
      "epoch": 0.0570409982174688,
      "grad_norm": 7.097423553466797,
      "learning_rate": 8e-05,
      "loss": 1.853,
      "step": 8
    },
    {
      "epoch": 0.06417112299465241,
      "grad_norm": 5.807572364807129,
      "learning_rate": 9e-05,
      "loss": 1.7299,
      "step": 9
    },
    {
      "epoch": 0.07130124777183601,
      "grad_norm": 4.308953285217285,
      "learning_rate": 0.0001,
      "loss": 1.1758,
      "step": 10
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 8.310755729675293,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5666,
      "step": 11
    },
    {
      "epoch": 0.0855614973262032,
      "grad_norm": 4.563036918640137,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.3147,
      "step": 12
    },
    {
      "epoch": 0.09269162210338681,
      "grad_norm": 5.149623394012451,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.0886,
      "step": 13
    },
    {
      "epoch": 0.09982174688057041,
      "grad_norm": 5.882086753845215,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.386,
      "step": 14
    },
    {
      "epoch": 0.10695187165775401,
      "grad_norm": 12.239624977111816,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.2012,
      "step": 15
    },
    {
      "epoch": 0.1140819964349376,
      "grad_norm": 7.540602207183838,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.2995,
      "step": 16
    },
    {
      "epoch": 0.12121212121212122,
      "grad_norm": 5.4989423751831055,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1511,
      "step": 17
    },
    {
      "epoch": 0.12834224598930483,
      "grad_norm": 6.216000556945801,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.8327,
      "step": 18
    },
    {
      "epoch": 0.1354723707664884,
      "grad_norm": 4.877588748931885,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.6784,
      "step": 19
    },
    {
      "epoch": 0.14260249554367202,
      "grad_norm": 10.093413352966309,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.0459,
      "step": 20
    },
    {
      "epoch": 0.1497326203208556,
      "grad_norm": 7.529651165008545,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4024,
      "step": 21
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 15.220786094665527,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3263,
      "step": 22
    },
    {
      "epoch": 0.16399286987522282,
      "grad_norm": 9.120606422424316,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.2663,
      "step": 23
    },
    {
      "epoch": 0.1711229946524064,
      "grad_norm": 7.058254718780518,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.3212,
      "step": 24
    },
    {
      "epoch": 0.17825311942959002,
      "grad_norm": 9.098106384277344,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6655,
      "step": 25
    },
    {
      "epoch": 0.18538324420677363,
      "grad_norm": 6.505765914916992,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.1041,
      "step": 26
    },
    {
      "epoch": 0.1925133689839572,
      "grad_norm": 3.545391321182251,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.7916,
      "step": 27
    },
    {
      "epoch": 0.19964349376114082,
      "grad_norm": 4.667784214019775,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.8798,
      "step": 28
    },
    {
      "epoch": 0.20677361853832443,
      "grad_norm": 15.075148582458496,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.6309,
      "step": 29
    },
    {
      "epoch": 0.21390374331550802,
      "grad_norm": 16.655683517456055,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.8559,
      "step": 30
    },
    {
      "epoch": 0.22103386809269163,
      "grad_norm": 7.557066440582275,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.3669,
      "step": 31
    },
    {
      "epoch": 0.2281639928698752,
      "grad_norm": 16.793487548828125,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.2009,
      "step": 32
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 7.860363483428955,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2864,
      "step": 33
    },
    {
      "epoch": 0.24242424242424243,
      "grad_norm": 6.455472946166992,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0194,
      "step": 34
    },
    {
      "epoch": 0.24955436720142601,
      "grad_norm": 5.507026672363281,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.7551,
      "step": 35
    },
    {
      "epoch": 0.25668449197860965,
      "grad_norm": 17.20283317565918,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.9312,
      "step": 36
    },
    {
      "epoch": 0.2638146167557932,
      "grad_norm": 14.8767728805542,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.2403,
      "step": 37
    },
    {
      "epoch": 0.2709447415329768,
      "grad_norm": 9.64128303527832,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.6086,
      "step": 38
    },
    {
      "epoch": 0.27807486631016043,
      "grad_norm": 4.818795680999756,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.8585,
      "step": 39
    },
    {
      "epoch": 0.28520499108734404,
      "grad_norm": 3.329906940460205,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6471,
      "step": 40
    },
    {
      "epoch": 0.29233511586452765,
      "grad_norm": 3.4146223068237305,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5584,
      "step": 41
    },
    {
      "epoch": 0.2994652406417112,
      "grad_norm": 2.892646551132202,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.541,
      "step": 42
    },
    {
      "epoch": 0.3065953654188948,
      "grad_norm": 6.643548488616943,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.7923,
      "step": 43
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 3.204444646835327,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.5552,
      "step": 44
    },
    {
      "epoch": 0.32085561497326204,
      "grad_norm": 2.1902480125427246,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.373,
      "step": 45
    },
    {
      "epoch": 0.32798573975044565,
      "grad_norm": 3.2742395401000977,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.5669,
      "step": 46
    },
    {
      "epoch": 0.33511586452762926,
      "grad_norm": 2.8352434635162354,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.5118,
      "step": 47
    },
    {
      "epoch": 0.3422459893048128,
      "grad_norm": 3.1886863708496094,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.4386,
      "step": 48
    },
    {
      "epoch": 0.3493761140819964,
      "grad_norm": 2.8913755416870117,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.2077,
      "step": 49
    },
    {
      "epoch": 0.35650623885918004,
      "grad_norm": 4.413907527923584,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.2928,
      "step": 50
    },
    {
      "epoch": 0.35650623885918004,
      "eval_loss": 0.6329964995384216,
      "eval_runtime": 17.9761,
      "eval_samples_per_second": 13.129,
      "eval_steps_per_second": 3.282,
      "step": 50
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 3.1114044189453125,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.2523,
      "step": 51
    },
    {
      "epoch": 0.37076648841354726,
      "grad_norm": 6.861588954925537,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.3154,
      "step": 52
    },
    {
      "epoch": 0.3778966131907308,
      "grad_norm": 4.95829963684082,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.2599,
      "step": 53
    },
    {
      "epoch": 0.3850267379679144,
      "grad_norm": 2.8263261318206787,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.3866,
      "step": 54
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 4.474704742431641,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.4625,
      "step": 55
    },
    {
      "epoch": 0.39928698752228164,
      "grad_norm": 5.040213108062744,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.3389,
      "step": 56
    },
    {
      "epoch": 0.40641711229946526,
      "grad_norm": 3.870736837387085,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3838,
      "step": 57
    },
    {
      "epoch": 0.41354723707664887,
      "grad_norm": 4.120948314666748,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.5858,
      "step": 58
    },
    {
      "epoch": 0.4206773618538324,
      "grad_norm": 3.038480758666992,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3756,
      "step": 59
    },
    {
      "epoch": 0.42780748663101603,
      "grad_norm": 4.331995487213135,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.3357,
      "step": 60
    },
    {
      "epoch": 0.43493761140819964,
      "grad_norm": 2.101313829421997,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.1877,
      "step": 61
    },
    {
      "epoch": 0.44206773618538325,
      "grad_norm": 2.31194806098938,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.2805,
      "step": 62
    },
    {
      "epoch": 0.44919786096256686,
      "grad_norm": 14.086769104003906,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.7347,
      "step": 63
    },
    {
      "epoch": 0.4563279857397504,
      "grad_norm": 4.054104804992676,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.5074,
      "step": 64
    },
    {
      "epoch": 0.46345811051693403,
      "grad_norm": 5.280353546142578,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.4249,
      "step": 65
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 6.034688949584961,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.5558,
      "step": 66
    },
    {
      "epoch": 0.47771836007130125,
      "grad_norm": 5.80851411819458,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.438,
      "step": 67
    },
    {
      "epoch": 0.48484848484848486,
      "grad_norm": 6.945198059082031,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3322,
      "step": 68
    },
    {
      "epoch": 0.4919786096256685,
      "grad_norm": 4.005733966827393,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3609,
      "step": 69
    },
    {
      "epoch": 0.49910873440285203,
      "grad_norm": 6.704404354095459,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.5501,
      "step": 70
    },
    {
      "epoch": 0.5062388591800356,
      "grad_norm": 13.100396156311035,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.0206,
      "step": 71
    },
    {
      "epoch": 0.5133689839572193,
      "grad_norm": 10.754317283630371,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.8247,
      "step": 72
    },
    {
      "epoch": 0.5204991087344029,
      "grad_norm": 8.525388717651367,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4684,
      "step": 73
    },
    {
      "epoch": 0.5276292335115864,
      "grad_norm": 5.457864761352539,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.9784,
      "step": 74
    },
    {
      "epoch": 0.5347593582887701,
      "grad_norm": 3.7644906044006348,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5458,
      "step": 75
    },
    {
      "epoch": 0.5418894830659536,
      "grad_norm": 3.4349308013916016,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.5357,
      "step": 76
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 1.9708448648452759,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3842,
      "step": 77
    },
    {
      "epoch": 0.5561497326203209,
      "grad_norm": 3.3483309745788574,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.4531,
      "step": 78
    },
    {
      "epoch": 0.5632798573975044,
      "grad_norm": 3.8980767726898193,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.3537,
      "step": 79
    },
    {
      "epoch": 0.5704099821746881,
      "grad_norm": 2.0954179763793945,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.2575,
      "step": 80
    },
    {
      "epoch": 0.5775401069518716,
      "grad_norm": 1.6207815408706665,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.2062,
      "step": 81
    },
    {
      "epoch": 0.5846702317290553,
      "grad_norm": 3.7690038681030273,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.2722,
      "step": 82
    },
    {
      "epoch": 0.5918003565062389,
      "grad_norm": 1.8468037843704224,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.1658,
      "step": 83
    },
    {
      "epoch": 0.5989304812834224,
      "grad_norm": 3.562955141067505,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.1585,
      "step": 84
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 2.0755581855773926,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.1578,
      "step": 85
    },
    {
      "epoch": 0.6131907308377896,
      "grad_norm": 1.4795328378677368,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.0802,
      "step": 86
    },
    {
      "epoch": 0.6203208556149733,
      "grad_norm": 2.3597075939178467,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.2977,
      "step": 87
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 3.4502038955688477,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.3013,
      "step": 88
    },
    {
      "epoch": 0.6345811051693404,
      "grad_norm": 2.782360792160034,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.2053,
      "step": 89
    },
    {
      "epoch": 0.6417112299465241,
      "grad_norm": 2.4194414615631104,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.2017,
      "step": 90
    },
    {
      "epoch": 0.6488413547237076,
      "grad_norm": 1.4441661834716797,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.0997,
      "step": 91
    },
    {
      "epoch": 0.6559714795008913,
      "grad_norm": 1.8876862525939941,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1564,
      "step": 92
    },
    {
      "epoch": 0.6631016042780749,
      "grad_norm": 3.6038668155670166,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.2264,
      "step": 93
    },
    {
      "epoch": 0.6702317290552585,
      "grad_norm": 4.474193572998047,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3177,
      "step": 94
    },
    {
      "epoch": 0.6773618538324421,
      "grad_norm": 4.002146244049072,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.3284,
      "step": 95
    },
    {
      "epoch": 0.6844919786096256,
      "grad_norm": 3.826871871948242,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.1862,
      "step": 96
    },
    {
      "epoch": 0.6916221033868093,
      "grad_norm": 1.5918521881103516,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2408,
      "step": 97
    },
    {
      "epoch": 0.6987522281639929,
      "grad_norm": 2.2593703269958496,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3048,
      "step": 98
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 2.6590259075164795,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.2294,
      "step": 99
    },
    {
      "epoch": 0.7130124777183601,
      "grad_norm": 2.615184783935547,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.2174,
      "step": 100
    },
    {
      "epoch": 0.7130124777183601,
      "eval_loss": 0.32512158155441284,
      "eval_runtime": 17.9967,
      "eval_samples_per_second": 13.114,
      "eval_steps_per_second": 3.278,
      "step": 100
    },
    {
      "epoch": 0.7201426024955436,
      "grad_norm": 4.077305316925049,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.2662,
      "step": 101
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 4.493444919586182,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.2757,
      "step": 102
    },
    {
      "epoch": 0.7344028520499108,
      "grad_norm": 1.2807857990264893,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.1086,
      "step": 103
    },
    {
      "epoch": 0.7415329768270945,
      "grad_norm": 1.7022565603256226,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.2093,
      "step": 104
    },
    {
      "epoch": 0.7486631016042781,
      "grad_norm": 3.4729650020599365,
      "learning_rate": 5e-05,
      "loss": 0.4945,
      "step": 105
    },
    {
      "epoch": 0.7557932263814616,
      "grad_norm": 9.623605728149414,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.8191,
      "step": 106
    },
    {
      "epoch": 0.7629233511586453,
      "grad_norm": 6.82785177230835,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.6385,
      "step": 107
    },
    {
      "epoch": 0.7700534759358288,
      "grad_norm": 3.6015989780426025,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.509,
      "step": 108
    },
    {
      "epoch": 0.7771836007130125,
      "grad_norm": 4.353756904602051,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4556,
      "step": 109
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 4.157517433166504,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.3514,
      "step": 110
    },
    {
      "epoch": 0.7914438502673797,
      "grad_norm": 1.487565040588379,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.2651,
      "step": 111
    },
    {
      "epoch": 0.7985739750445633,
      "grad_norm": 1.8361451625823975,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.214,
      "step": 112
    },
    {
      "epoch": 0.8057040998217468,
      "grad_norm": 2.8729050159454346,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.1751,
      "step": 113
    },
    {
      "epoch": 0.8128342245989305,
      "grad_norm": 2.9342260360717773,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.1975,
      "step": 114
    },
    {
      "epoch": 0.8199643493761141,
      "grad_norm": 1.1883810758590698,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.1764,
      "step": 115
    },
    {
      "epoch": 0.8270944741532977,
      "grad_norm": 2.3309926986694336,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.2418,
      "step": 116
    },
    {
      "epoch": 0.8342245989304813,
      "grad_norm": 2.0552427768707275,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.1506,
      "step": 117
    },
    {
      "epoch": 0.8413547237076648,
      "grad_norm": 1.2462787628173828,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.1211,
      "step": 118
    },
    {
      "epoch": 0.8484848484848485,
      "grad_norm": 1.4371592998504639,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.2315,
      "step": 119
    },
    {
      "epoch": 0.8556149732620321,
      "grad_norm": 1.6682649850845337,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.211,
      "step": 120
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 2.4928805828094482,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1682,
      "step": 121
    },
    {
      "epoch": 0.8698752228163993,
      "grad_norm": 1.7620739936828613,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.1171,
      "step": 122
    },
    {
      "epoch": 0.8770053475935828,
      "grad_norm": 1.4409494400024414,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.2122,
      "step": 123
    },
    {
      "epoch": 0.8841354723707665,
      "grad_norm": 3.6312010288238525,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1687,
      "step": 124
    },
    {
      "epoch": 0.8912655971479501,
      "grad_norm": 1.3037965297698975,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.1059,
      "step": 125
    },
    {
      "epoch": 0.8983957219251337,
      "grad_norm": 1.5927499532699585,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.1325,
      "step": 126
    },
    {
      "epoch": 0.9055258467023173,
      "grad_norm": 1.1492993831634521,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.0601,
      "step": 127
    },
    {
      "epoch": 0.9126559714795008,
      "grad_norm": 1.1006429195404053,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.1772,
      "step": 128
    },
    {
      "epoch": 0.9197860962566845,
      "grad_norm": 2.085655927658081,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.1353,
      "step": 129
    },
    {
      "epoch": 0.9269162210338681,
      "grad_norm": 3.305060625076294,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.2093,
      "step": 130
    },
    {
      "epoch": 0.9340463458110517,
      "grad_norm": 1.5962597131729126,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.1222,
      "step": 131
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 1.1807433366775513,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.1408,
      "step": 132
    },
    {
      "epoch": 0.948306595365419,
      "grad_norm": 1.820581316947937,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.2907,
      "step": 133
    },
    {
      "epoch": 0.9554367201426025,
      "grad_norm": 2.320713758468628,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.1864,
      "step": 134
    },
    {
      "epoch": 0.9625668449197861,
      "grad_norm": 2.3209404945373535,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3219,
      "step": 135
    },
    {
      "epoch": 0.9696969696969697,
      "grad_norm": 2.7532081604003906,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.3793,
      "step": 136
    },
    {
      "epoch": 0.9768270944741533,
      "grad_norm": 1.9822131395339966,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1778,
      "step": 137
    },
    {
      "epoch": 0.983957219251337,
      "grad_norm": 1.895337700843811,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.2372,
      "step": 138
    },
    {
      "epoch": 0.9910873440285205,
      "grad_norm": 1.6673623323440552,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.1409,
      "step": 139
    },
    {
      "epoch": 0.9982174688057041,
      "grad_norm": 2.0436630249023438,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.3324,
      "step": 140
    },
    {
      "epoch": 1.0053475935828877,
      "grad_norm": 5.202945232391357,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.6321,
      "step": 141
    },
    {
      "epoch": 1.0124777183600713,
      "grad_norm": 1.9828206300735474,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.2057,
      "step": 142
    },
    {
      "epoch": 1.0196078431372548,
      "grad_norm": 1.4824261665344238,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2002,
      "step": 143
    },
    {
      "epoch": 1.0267379679144386,
      "grad_norm": 1.8293224573135376,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.2948,
      "step": 144
    },
    {
      "epoch": 1.0338680926916222,
      "grad_norm": 1.3274524211883545,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2465,
      "step": 145
    },
    {
      "epoch": 1.0409982174688057,
      "grad_norm": 2.0107944011688232,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.3174,
      "step": 146
    },
    {
      "epoch": 1.0481283422459893,
      "grad_norm": 2.3455100059509277,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2253,
      "step": 147
    },
    {
      "epoch": 1.0552584670231728,
      "grad_norm": 2.172856569290161,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.1358,
      "step": 148
    },
    {
      "epoch": 1.0623885918003566,
      "grad_norm": 1.3267221450805664,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0817,
      "step": 149
    },
    {
      "epoch": 1.0695187165775402,
      "grad_norm": 0.9412680268287659,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0696,
      "step": 150
    },
    {
      "epoch": 1.0695187165775402,
      "eval_loss": 0.11750635504722595,
      "eval_runtime": 17.9532,
      "eval_samples_per_second": 13.145,
      "eval_steps_per_second": 3.286,
      "step": 150
    },
    {
      "epoch": 1.0766488413547237,
      "grad_norm": 1.2027430534362793,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.1208,
      "step": 151
    },
    {
      "epoch": 1.0837789661319073,
      "grad_norm": 0.8935040831565857,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.0642,
      "step": 152
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.5809460282325745,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0771,
      "step": 153
    },
    {
      "epoch": 1.0980392156862746,
      "grad_norm": 0.721865713596344,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.1294,
      "step": 154
    },
    {
      "epoch": 1.1051693404634582,
      "grad_norm": 0.9808659553527832,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.1029,
      "step": 155
    },
    {
      "epoch": 1.1122994652406417,
      "grad_norm": 1.3903586864471436,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0838,
      "step": 156
    },
    {
      "epoch": 1.1194295900178253,
      "grad_norm": 0.488554447889328,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.0659,
      "step": 157
    },
    {
      "epoch": 1.1265597147950088,
      "grad_norm": 1.0842068195343018,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.1556,
      "step": 158
    },
    {
      "epoch": 1.1336898395721926,
      "grad_norm": 0.8102473616600037,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1055,
      "step": 159
    },
    {
      "epoch": 1.1408199643493762,
      "grad_norm": 1.008832335472107,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.1441,
      "step": 160
    },
    {
      "epoch": 1.1479500891265597,
      "grad_norm": 0.5591416954994202,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.078,
      "step": 161
    },
    {
      "epoch": 1.1550802139037433,
      "grad_norm": 0.49514836072921753,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0577,
      "step": 162
    },
    {
      "epoch": 1.1622103386809268,
      "grad_norm": 1.054601788520813,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.1102,
      "step": 163
    },
    {
      "epoch": 1.1693404634581106,
      "grad_norm": 1.8332737684249878,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.1187,
      "step": 164
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 1.0433210134506226,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1209,
      "step": 165
    },
    {
      "epoch": 1.1836007130124777,
      "grad_norm": 1.4950175285339355,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1011,
      "step": 166
    },
    {
      "epoch": 1.1907308377896613,
      "grad_norm": 1.3030064105987549,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.1125,
      "step": 167
    },
    {
      "epoch": 1.1978609625668448,
      "grad_norm": 1.0447028875350952,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.1956,
      "step": 168
    },
    {
      "epoch": 1.2049910873440286,
      "grad_norm": 0.8809400200843811,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.1583,
      "step": 169
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 0.5815510749816895,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.0629,
      "step": 170
    },
    {
      "epoch": 1.2192513368983957,
      "grad_norm": 0.6817440986633301,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.1159,
      "step": 171
    },
    {
      "epoch": 1.2263814616755793,
      "grad_norm": 0.912253737449646,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.1104,
      "step": 172
    },
    {
      "epoch": 1.2335115864527628,
      "grad_norm": 1.0698485374450684,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.1236,
      "step": 173
    },
    {
      "epoch": 1.2406417112299466,
      "grad_norm": 1.0547733306884766,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.0854,
      "step": 174
    },
    {
      "epoch": 1.2477718360071302,
      "grad_norm": 1.7927907705307007,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3529,
      "step": 175
    },
    {
      "epoch": 1.2549019607843137,
      "grad_norm": 1.203934669494629,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1879,
      "step": 176
    },
    {
      "epoch": 1.2620320855614973,
      "grad_norm": 0.8924445509910583,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.052,
      "step": 177
    },
    {
      "epoch": 1.2691622103386808,
      "grad_norm": 0.706015408039093,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0991,
      "step": 178
    },
    {
      "epoch": 1.2762923351158646,
      "grad_norm": 0.6402453780174255,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0539,
      "step": 179
    },
    {
      "epoch": 1.2834224598930482,
      "grad_norm": 0.37091413140296936,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.0289,
      "step": 180
    },
    {
      "epoch": 1.2905525846702317,
      "grad_norm": 0.5855924487113953,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1113,
      "step": 181
    },
    {
      "epoch": 1.2976827094474153,
      "grad_norm": 0.7205154299736023,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0795,
      "step": 182
    },
    {
      "epoch": 1.3048128342245988,
      "grad_norm": 0.4117750823497772,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0398,
      "step": 183
    },
    {
      "epoch": 1.3119429590017826,
      "grad_norm": 0.39361798763275146,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0461,
      "step": 184
    },
    {
      "epoch": 1.3190730837789661,
      "grad_norm": 0.41086381673812866,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.0415,
      "step": 185
    },
    {
      "epoch": 1.3262032085561497,
      "grad_norm": 0.4956147372722626,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.0867,
      "step": 186
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.4040631353855133,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0381,
      "step": 187
    },
    {
      "epoch": 1.3404634581105168,
      "grad_norm": 0.2972346544265747,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.0252,
      "step": 188
    },
    {
      "epoch": 1.3475935828877006,
      "grad_norm": 0.5021019577980042,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1145,
      "step": 189
    },
    {
      "epoch": 1.3547237076648841,
      "grad_norm": 0.4569206237792969,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0782,
      "step": 190
    },
    {
      "epoch": 1.3618538324420677,
      "grad_norm": 0.40431860089302063,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.0194,
      "step": 191
    },
    {
      "epoch": 1.3689839572192513,
      "grad_norm": 0.45667824149131775,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0572,
      "step": 192
    },
    {
      "epoch": 1.3761140819964348,
      "grad_norm": 0.5554242730140686,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0774,
      "step": 193
    },
    {
      "epoch": 1.3832442067736186,
      "grad_norm": 0.49616220593452454,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0599,
      "step": 194
    },
    {
      "epoch": 1.3903743315508021,
      "grad_norm": 0.5026013255119324,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0821,
      "step": 195
    },
    {
      "epoch": 1.3975044563279857,
      "grad_norm": 0.32124361395835876,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.0363,
      "step": 196
    },
    {
      "epoch": 1.4046345811051695,
      "grad_norm": 0.4889766573905945,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0667,
      "step": 197
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 0.5499719381332397,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0915,
      "step": 198
    },
    {
      "epoch": 1.4188948306595366,
      "grad_norm": 0.5564050674438477,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0995,
      "step": 199
    },
    {
      "epoch": 1.4260249554367201,
      "grad_norm": 0.5922126770019531,
      "learning_rate": 0.0,
      "loss": 0.069,
      "step": 200
    },
    {
      "epoch": 1.4260249554367201,
      "eval_loss": 0.08097303658723831,
      "eval_runtime": 17.9734,
      "eval_samples_per_second": 13.131,
      "eval_steps_per_second": 3.283,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 3.01514151551828e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}