{
  "best_metric": 1.6125844717025757,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.5810276679841897,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007905138339920948,
      "grad_norm": 1.2275149822235107,
      "learning_rate": 1e-05,
      "loss": 1.6629,
      "step": 1
    },
    {
      "epoch": 0.007905138339920948,
      "eval_loss": 1.9508434534072876,
      "eval_runtime": 4.796,
      "eval_samples_per_second": 44.412,
      "eval_steps_per_second": 11.259,
      "step": 1
    },
    {
      "epoch": 0.015810276679841896,
      "grad_norm": 1.312914252281189,
      "learning_rate": 2e-05,
      "loss": 1.7673,
      "step": 2
    },
    {
      "epoch": 0.023715415019762844,
      "grad_norm": 1.3910435438156128,
      "learning_rate": 3e-05,
      "loss": 1.8517,
      "step": 3
    },
    {
      "epoch": 0.03162055335968379,
      "grad_norm": 1.2056818008422852,
      "learning_rate": 4e-05,
      "loss": 1.8878,
      "step": 4
    },
    {
      "epoch": 0.039525691699604744,
      "grad_norm": 1.0080242156982422,
      "learning_rate": 5e-05,
      "loss": 1.7955,
      "step": 5
    },
    {
      "epoch": 0.04743083003952569,
      "grad_norm": 1.912818193435669,
      "learning_rate": 6e-05,
      "loss": 1.8229,
      "step": 6
    },
    {
      "epoch": 0.05533596837944664,
      "grad_norm": 1.671106219291687,
      "learning_rate": 7e-05,
      "loss": 1.835,
      "step": 7
    },
    {
      "epoch": 0.06324110671936758,
      "grad_norm": 1.3343749046325684,
      "learning_rate": 8e-05,
      "loss": 1.877,
      "step": 8
    },
    {
      "epoch": 0.07114624505928854,
      "grad_norm": 1.4020658731460571,
      "learning_rate": 9e-05,
      "loss": 1.7761,
      "step": 9
    },
    {
      "epoch": 0.07905138339920949,
      "grad_norm": 1.0474133491516113,
      "learning_rate": 0.0001,
      "loss": 1.6915,
      "step": 10
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 1.050320029258728,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.8034,
      "step": 11
    },
    {
      "epoch": 0.09486166007905138,
      "grad_norm": 0.9719471335411072,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8651,
      "step": 12
    },
    {
      "epoch": 0.10276679841897234,
      "grad_norm": 1.0876976251602173,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.8264,
      "step": 13
    },
    {
      "epoch": 0.11067193675889328,
      "grad_norm": 0.9134715795516968,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.7016,
      "step": 14
    },
    {
      "epoch": 0.11857707509881422,
      "grad_norm": 1.3684470653533936,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.7452,
      "step": 15
    },
    {
      "epoch": 0.12648221343873517,
      "grad_norm": 0.82088303565979,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.7485,
      "step": 16
    },
    {
      "epoch": 0.13438735177865613,
      "grad_norm": 1.0163344144821167,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.8706,
      "step": 17
    },
    {
      "epoch": 0.1422924901185771,
      "grad_norm": 1.0233978033065796,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.8125,
      "step": 18
    },
    {
      "epoch": 0.15019762845849802,
      "grad_norm": 1.0371999740600586,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.7027,
      "step": 19
    },
    {
      "epoch": 0.15810276679841898,
      "grad_norm": 0.9143949747085571,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6941,
      "step": 20
    },
    {
      "epoch": 0.16600790513833993,
      "grad_norm": 0.9650872349739075,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.7789,
      "step": 21
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.9396098256111145,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.6984,
      "step": 22
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.9005189538002014,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.8407,
      "step": 23
    },
    {
      "epoch": 0.18972332015810275,
      "grad_norm": 0.9187804460525513,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.7712,
      "step": 24
    },
    {
      "epoch": 0.1976284584980237,
      "grad_norm": 1.1316477060317993,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.7481,
      "step": 25
    },
    {
      "epoch": 0.20553359683794467,
      "grad_norm": 0.9262543320655823,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.7299,
      "step": 26
    },
    {
      "epoch": 0.2134387351778656,
      "grad_norm": 1.292194128036499,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.9521,
      "step": 27
    },
    {
      "epoch": 0.22134387351778656,
      "grad_norm": 1.153373122215271,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.8683,
      "step": 28
    },
    {
      "epoch": 0.22924901185770752,
      "grad_norm": 1.2291576862335205,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.8653,
      "step": 29
    },
    {
      "epoch": 0.23715415019762845,
      "grad_norm": 1.186466932296753,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.7617,
      "step": 30
    },
    {
      "epoch": 0.2450592885375494,
      "grad_norm": 1.8452413082122803,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.009,
      "step": 31
    },
    {
      "epoch": 0.25296442687747034,
      "grad_norm": 2.0136189460754395,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.707,
      "step": 32
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 1.5197522640228271,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.5445,
      "step": 33
    },
    {
      "epoch": 0.26877470355731226,
      "grad_norm": 0.8320451378822327,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.6573,
      "step": 34
    },
    {
      "epoch": 0.2766798418972332,
      "grad_norm": 0.8250799775123596,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.7422,
      "step": 35
    },
    {
      "epoch": 0.2845849802371542,
      "grad_norm": 0.8387137055397034,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.7918,
      "step": 36
    },
    {
      "epoch": 0.2924901185770751,
      "grad_norm": 0.7751452326774597,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.5503,
      "step": 37
    },
    {
      "epoch": 0.30039525691699603,
      "grad_norm": 0.8183238506317139,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5502,
      "step": 38
    },
    {
      "epoch": 0.308300395256917,
      "grad_norm": 0.8241856098175049,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6668,
      "step": 39
    },
    {
      "epoch": 0.31620553359683795,
      "grad_norm": 0.7364861369132996,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.7508,
      "step": 40
    },
    {
      "epoch": 0.3241106719367589,
      "grad_norm": 0.7785086035728455,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.7751,
      "step": 41
    },
    {
      "epoch": 0.33201581027667987,
      "grad_norm": 0.7593626379966736,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.7295,
      "step": 42
    },
    {
      "epoch": 0.33992094861660077,
      "grad_norm": 0.9561973214149475,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.7577,
      "step": 43
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.8719379305839539,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.7155,
      "step": 44
    },
    {
      "epoch": 0.3557312252964427,
      "grad_norm": 0.7628946304321289,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.6977,
      "step": 45
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.8044358491897583,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.8296,
      "step": 46
    },
    {
      "epoch": 0.3715415019762846,
      "grad_norm": 0.7543927431106567,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.6795,
      "step": 47
    },
    {
      "epoch": 0.3794466403162055,
      "grad_norm": 0.7223030924797058,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5737,
      "step": 48
    },
    {
      "epoch": 0.38735177865612647,
      "grad_norm": 0.8203468918800354,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.6581,
      "step": 49
    },
    {
      "epoch": 0.3952569169960474,
      "grad_norm": 0.8775282502174377,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.8492,
      "step": 50
    },
    {
      "epoch": 0.3952569169960474,
      "eval_loss": 1.6608927249908447,
      "eval_runtime": 4.8136,
      "eval_samples_per_second": 44.249,
      "eval_steps_per_second": 11.218,
      "step": 50
    },
    {
      "epoch": 0.4031620553359684,
      "grad_norm": 0.8258354067802429,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.7065,
      "step": 51
    },
    {
      "epoch": 0.41106719367588934,
      "grad_norm": 0.7727131843566895,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.7414,
      "step": 52
    },
    {
      "epoch": 0.4189723320158103,
      "grad_norm": 0.767753541469574,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.6911,
      "step": 53
    },
    {
      "epoch": 0.4268774703557312,
      "grad_norm": 0.896858274936676,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.8343,
      "step": 54
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 0.8155654072761536,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7718,
      "step": 55
    },
    {
      "epoch": 0.4426877470355731,
      "grad_norm": 0.8962050080299377,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7042,
      "step": 56
    },
    {
      "epoch": 0.4505928853754941,
      "grad_norm": 0.8581175208091736,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.7504,
      "step": 57
    },
    {
      "epoch": 0.45849802371541504,
      "grad_norm": 0.948732852935791,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.8929,
      "step": 58
    },
    {
      "epoch": 0.466403162055336,
      "grad_norm": 0.9872562289237976,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.7411,
      "step": 59
    },
    {
      "epoch": 0.4743083003952569,
      "grad_norm": 1.1096147298812866,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7791,
      "step": 60
    },
    {
      "epoch": 0.48221343873517786,
      "grad_norm": 1.0001335144042969,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.7769,
      "step": 61
    },
    {
      "epoch": 0.4901185770750988,
      "grad_norm": 1.14747953414917,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.8926,
      "step": 62
    },
    {
      "epoch": 0.4980237154150198,
      "grad_norm": 0.8204768300056458,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4519,
      "step": 63
    },
    {
      "epoch": 0.5059288537549407,
      "grad_norm": 0.7378841042518616,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.5685,
      "step": 64
    },
    {
      "epoch": 0.5138339920948617,
      "grad_norm": 0.7584918141365051,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6277,
      "step": 65
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.6478469967842102,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.6634,
      "step": 66
    },
    {
      "epoch": 0.5296442687747036,
      "grad_norm": 0.621213436126709,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.7036,
      "step": 67
    },
    {
      "epoch": 0.5375494071146245,
      "grad_norm": 0.6001406311988831,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.7276,
      "step": 68
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.6318312287330627,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.7278,
      "step": 69
    },
    {
      "epoch": 0.5533596837944664,
      "grad_norm": 0.6229626536369324,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6364,
      "step": 70
    },
    {
      "epoch": 0.5612648221343873,
      "grad_norm": 0.6349075436592102,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.6318,
      "step": 71
    },
    {
      "epoch": 0.5691699604743083,
      "grad_norm": 0.6748497486114502,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.5987,
      "step": 72
    },
    {
      "epoch": 0.5770750988142292,
      "grad_norm": 0.7020136117935181,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.7664,
      "step": 73
    },
    {
      "epoch": 0.5849802371541502,
      "grad_norm": 0.6632049083709717,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.6788,
      "step": 74
    },
    {
      "epoch": 0.5928853754940712,
      "grad_norm": 0.6933974623680115,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.7321,
      "step": 75
    },
    {
      "epoch": 0.6007905138339921,
      "grad_norm": 0.6252143979072571,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6356,
      "step": 76
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 0.6765616536140442,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.6761,
      "step": 77
    },
    {
      "epoch": 0.616600790513834,
      "grad_norm": 0.6760627627372742,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7556,
      "step": 78
    },
    {
      "epoch": 0.6245059288537549,
      "grad_norm": 0.7049459218978882,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.7011,
      "step": 79
    },
    {
      "epoch": 0.6324110671936759,
      "grad_norm": 0.7250329852104187,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.8144,
      "step": 80
    },
    {
      "epoch": 0.6403162055335968,
      "grad_norm": 0.7103596329689026,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.6611,
      "step": 81
    },
    {
      "epoch": 0.6482213438735178,
      "grad_norm": 0.7164954543113708,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.6305,
      "step": 82
    },
    {
      "epoch": 0.6561264822134387,
      "grad_norm": 0.7123523950576782,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.6529,
      "step": 83
    },
    {
      "epoch": 0.6640316205533597,
      "grad_norm": 0.7113423943519592,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.6658,
      "step": 84
    },
    {
      "epoch": 0.6719367588932806,
      "grad_norm": 0.7906448841094971,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.7731,
      "step": 85
    },
    {
      "epoch": 0.6798418972332015,
      "grad_norm": 0.8516830801963806,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.755,
      "step": 86
    },
    {
      "epoch": 0.6877470355731226,
      "grad_norm": 0.8049620985984802,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.7237,
      "step": 87
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.8168877363204956,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.7243,
      "step": 88
    },
    {
      "epoch": 0.7035573122529645,
      "grad_norm": 0.8963014483451843,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.6904,
      "step": 89
    },
    {
      "epoch": 0.7114624505928854,
      "grad_norm": 0.9513223171234131,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.7241,
      "step": 90
    },
    {
      "epoch": 0.7193675889328063,
      "grad_norm": 0.8429911136627197,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.7412,
      "step": 91
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.9856913089752197,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.7932,
      "step": 92
    },
    {
      "epoch": 0.7351778656126482,
      "grad_norm": 1.2035722732543945,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.7209,
      "step": 93
    },
    {
      "epoch": 0.7430830039525692,
      "grad_norm": 0.7656598091125488,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.6114,
      "step": 94
    },
    {
      "epoch": 0.7509881422924901,
      "grad_norm": 0.7776647806167603,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.5802,
      "step": 95
    },
    {
      "epoch": 0.758893280632411,
      "grad_norm": 0.6927933096885681,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.6236,
      "step": 96
    },
    {
      "epoch": 0.766798418972332,
      "grad_norm": 0.6408959031105042,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.6836,
      "step": 97
    },
    {
      "epoch": 0.7747035573122529,
      "grad_norm": 0.6113343834877014,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.5781,
      "step": 98
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 0.6380818486213684,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.648,
      "step": 99
    },
    {
      "epoch": 0.7905138339920948,
      "grad_norm": 0.6411934494972229,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.6435,
      "step": 100
    },
    {
      "epoch": 0.7905138339920948,
      "eval_loss": 1.644432544708252,
      "eval_runtime": 4.8149,
      "eval_samples_per_second": 44.238,
      "eval_steps_per_second": 11.215,
      "step": 100
    },
    {
      "epoch": 0.7984189723320159,
      "grad_norm": 0.6788752675056458,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.7212,
      "step": 101
    },
    {
      "epoch": 0.8063241106719368,
      "grad_norm": 0.5979511737823486,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.5423,
      "step": 102
    },
    {
      "epoch": 0.8142292490118577,
      "grad_norm": 0.6435640454292297,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5651,
      "step": 103
    },
    {
      "epoch": 0.8221343873517787,
      "grad_norm": 0.6285018920898438,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5639,
      "step": 104
    },
    {
      "epoch": 0.8300395256916996,
      "grad_norm": 0.6168222427368164,
      "learning_rate": 5e-05,
      "loss": 1.5667,
      "step": 105
    },
    {
      "epoch": 0.8379446640316206,
      "grad_norm": 0.6518228650093079,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.7141,
      "step": 106
    },
    {
      "epoch": 0.8458498023715415,
      "grad_norm": 0.7004503011703491,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.765,
      "step": 107
    },
    {
      "epoch": 0.8537549407114624,
      "grad_norm": 0.6650953888893127,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.5443,
      "step": 108
    },
    {
      "epoch": 0.8616600790513834,
      "grad_norm": 0.7480284571647644,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.7513,
      "step": 109
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 0.7257048487663269,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.7061,
      "step": 110
    },
    {
      "epoch": 0.8774703557312253,
      "grad_norm": 0.7339716553688049,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.6675,
      "step": 111
    },
    {
      "epoch": 0.8853754940711462,
      "grad_norm": 0.7776533365249634,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.8499,
      "step": 112
    },
    {
      "epoch": 0.8932806324110671,
      "grad_norm": 0.7437097430229187,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.8209,
      "step": 113
    },
    {
      "epoch": 0.9011857707509882,
      "grad_norm": 0.7620865106582642,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.745,
      "step": 114
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.8105087280273438,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.7531,
      "step": 115
    },
    {
      "epoch": 0.9169960474308301,
      "grad_norm": 0.8046197295188904,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.7271,
      "step": 116
    },
    {
      "epoch": 0.924901185770751,
      "grad_norm": 0.8412743806838989,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.8155,
      "step": 117
    },
    {
      "epoch": 0.932806324110672,
      "grad_norm": 0.838920533657074,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.8482,
      "step": 118
    },
    {
      "epoch": 0.9407114624505929,
      "grad_norm": 0.8457604050636292,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7529,
      "step": 119
    },
    {
      "epoch": 0.9486166007905138,
      "grad_norm": 0.9167115092277527,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.6983,
      "step": 120
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 0.9630374312400818,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.7011,
      "step": 121
    },
    {
      "epoch": 0.9644268774703557,
      "grad_norm": 0.9056899547576904,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.6826,
      "step": 122
    },
    {
      "epoch": 0.9723320158102767,
      "grad_norm": 1.020810604095459,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.6647,
      "step": 123
    },
    {
      "epoch": 0.9802371541501976,
      "grad_norm": 1.327959418296814,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.848,
      "step": 124
    },
    {
      "epoch": 0.9881422924901185,
      "grad_norm": 0.6786101460456848,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.5836,
      "step": 125
    },
    {
      "epoch": 0.9960474308300395,
      "grad_norm": 0.7688708305358887,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7463,
      "step": 126
    },
    {
      "epoch": 1.0039525691699605,
      "grad_norm": 1.5957648754119873,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.6306,
      "step": 127
    },
    {
      "epoch": 1.0118577075098814,
      "grad_norm": 0.5522214770317078,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.532,
      "step": 128
    },
    {
      "epoch": 1.0197628458498025,
      "grad_norm": 0.5481433868408203,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4873,
      "step": 129
    },
    {
      "epoch": 1.0276679841897234,
      "grad_norm": 0.669899046421051,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.689,
      "step": 130
    },
    {
      "epoch": 1.0355731225296443,
      "grad_norm": 0.5601217150688171,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.4955,
      "step": 131
    },
    {
      "epoch": 1.0434782608695652,
      "grad_norm": 0.5741488933563232,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.577,
      "step": 132
    },
    {
      "epoch": 1.051383399209486,
      "grad_norm": 0.5481381416320801,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.5231,
      "step": 133
    },
    {
      "epoch": 1.0592885375494072,
      "grad_norm": 0.5438144207000732,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4542,
      "step": 134
    },
    {
      "epoch": 1.0671936758893281,
      "grad_norm": 0.5592533349990845,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.6158,
      "step": 135
    },
    {
      "epoch": 1.075098814229249,
      "grad_norm": 0.601783812046051,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.5103,
      "step": 136
    },
    {
      "epoch": 1.08300395256917,
      "grad_norm": 0.6082609295845032,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.5273,
      "step": 137
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.6211904287338257,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.5463,
      "step": 138
    },
    {
      "epoch": 1.098814229249012,
      "grad_norm": 0.5987580418586731,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5708,
      "step": 139
    },
    {
      "epoch": 1.1067193675889329,
      "grad_norm": 0.6398665904998779,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.6496,
      "step": 140
    },
    {
      "epoch": 1.1146245059288538,
      "grad_norm": 0.6528900265693665,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.5757,
      "step": 141
    },
    {
      "epoch": 1.1225296442687747,
      "grad_norm": 0.6666259169578552,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.6528,
      "step": 142
    },
    {
      "epoch": 1.1304347826086956,
      "grad_norm": 0.7091323137283325,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.6874,
      "step": 143
    },
    {
      "epoch": 1.1383399209486167,
      "grad_norm": 0.7314757108688354,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.7092,
      "step": 144
    },
    {
      "epoch": 1.1462450592885376,
      "grad_norm": 0.6564605832099915,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.465,
      "step": 145
    },
    {
      "epoch": 1.1541501976284585,
      "grad_norm": 0.6789471507072449,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.5696,
      "step": 146
    },
    {
      "epoch": 1.1620553359683794,
      "grad_norm": 0.6962265372276306,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.5224,
      "step": 147
    },
    {
      "epoch": 1.1699604743083003,
      "grad_norm": 0.736736536026001,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.6225,
      "step": 148
    },
    {
      "epoch": 1.1778656126482214,
      "grad_norm": 0.7357221245765686,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.5452,
      "step": 149
    },
    {
      "epoch": 1.1857707509881423,
      "grad_norm": 0.7465088367462158,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.6068,
      "step": 150
    },
    {
      "epoch": 1.1857707509881423,
      "eval_loss": 1.6133692264556885,
      "eval_runtime": 4.8741,
      "eval_samples_per_second": 43.7,
      "eval_steps_per_second": 11.079,
      "step": 150
    },
    {
      "epoch": 1.1936758893280632,
      "grad_norm": 0.7584455013275146,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.5367,
      "step": 151
    },
    {
      "epoch": 1.2015810276679841,
      "grad_norm": 0.8176235556602478,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.7604,
      "step": 152
    },
    {
      "epoch": 1.2094861660079053,
      "grad_norm": 0.8447201251983643,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.567,
      "step": 153
    },
    {
      "epoch": 1.2173913043478262,
      "grad_norm": 0.9284098744392395,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.6537,
      "step": 154
    },
    {
      "epoch": 1.225296442687747,
      "grad_norm": 0.9046016931533813,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.6645,
      "step": 155
    },
    {
      "epoch": 1.233201581027668,
      "grad_norm": 0.9470634460449219,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.6331,
      "step": 156
    },
    {
      "epoch": 1.2411067193675889,
      "grad_norm": 1.0975604057312012,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.6016,
      "step": 157
    },
    {
      "epoch": 1.2490118577075098,
      "grad_norm": 0.8554549813270569,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.6293,
      "step": 158
    },
    {
      "epoch": 1.256916996047431,
      "grad_norm": 0.5330277681350708,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.4049,
      "step": 159
    },
    {
      "epoch": 1.2648221343873518,
      "grad_norm": 0.5859397053718567,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4133,
      "step": 160
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.6284499764442444,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5504,
      "step": 161
    },
    {
      "epoch": 1.2806324110671936,
      "grad_norm": 0.668923556804657,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.6168,
      "step": 162
    },
    {
      "epoch": 1.2885375494071147,
      "grad_norm": 0.654583215713501,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.5011,
      "step": 163
    },
    {
      "epoch": 1.2964426877470356,
      "grad_norm": 0.6803271770477295,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.6328,
      "step": 164
    },
    {
      "epoch": 1.3043478260869565,
      "grad_norm": 0.6967841982841492,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.524,
      "step": 165
    },
    {
      "epoch": 1.3122529644268774,
      "grad_norm": 0.6923108696937561,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.6087,
      "step": 166
    },
    {
      "epoch": 1.3201581027667983,
      "grad_norm": 0.6514832973480225,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.6023,
      "step": 167
    },
    {
      "epoch": 1.3280632411067192,
      "grad_norm": 0.6509909629821777,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5239,
      "step": 168
    },
    {
      "epoch": 1.3359683794466404,
      "grad_norm": 0.6328653693199158,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5371,
      "step": 169
    },
    {
      "epoch": 1.3438735177865613,
      "grad_norm": 0.6849666833877563,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.5266,
      "step": 170
    },
    {
      "epoch": 1.3517786561264822,
      "grad_norm": 0.675087034702301,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.5834,
      "step": 171
    },
    {
      "epoch": 1.359683794466403,
      "grad_norm": 0.6388532519340515,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.5514,
      "step": 172
    },
    {
      "epoch": 1.3675889328063242,
      "grad_norm": 0.6572067737579346,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.5335,
      "step": 173
    },
    {
      "epoch": 1.3754940711462451,
      "grad_norm": 0.6312106251716614,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4,
      "step": 174
    },
    {
      "epoch": 1.383399209486166,
      "grad_norm": 0.6779483556747437,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.6211,
      "step": 175
    },
    {
      "epoch": 1.391304347826087,
      "grad_norm": 0.7150636911392212,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.5703,
      "step": 176
    },
    {
      "epoch": 1.3992094861660078,
      "grad_norm": 0.7181360721588135,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.5645,
      "step": 177
    },
    {
      "epoch": 1.4071146245059287,
      "grad_norm": 0.7471019625663757,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.7797,
      "step": 178
    },
    {
      "epoch": 1.4150197628458498,
      "grad_norm": 0.7033824920654297,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.4998,
      "step": 179
    },
    {
      "epoch": 1.4229249011857708,
      "grad_norm": 0.7901676893234253,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.7054,
      "step": 180
    },
    {
      "epoch": 1.4308300395256917,
      "grad_norm": 0.7970178723335266,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.6345,
      "step": 181
    },
    {
      "epoch": 1.4387351778656128,
      "grad_norm": 0.821295440196991,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.5289,
      "step": 182
    },
    {
      "epoch": 1.4466403162055337,
      "grad_norm": 0.8141527771949768,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.4823,
      "step": 183
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 0.909451425075531,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.6517,
      "step": 184
    },
    {
      "epoch": 1.4624505928853755,
      "grad_norm": 0.9176106452941895,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.6951,
      "step": 185
    },
    {
      "epoch": 1.4703557312252964,
      "grad_norm": 0.9050935506820679,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.555,
      "step": 186
    },
    {
      "epoch": 1.4782608695652173,
      "grad_norm": 0.9574251174926758,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.4957,
      "step": 187
    },
    {
      "epoch": 1.4861660079051384,
      "grad_norm": 1.1122318506240845,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.6216,
      "step": 188
    },
    {
      "epoch": 1.4940711462450593,
      "grad_norm": 0.8216851353645325,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.5164,
      "step": 189
    },
    {
      "epoch": 1.5019762845849802,
      "grad_norm": 0.5152871012687683,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.4198,
      "step": 190
    },
    {
      "epoch": 1.5098814229249014,
      "grad_norm": 0.5294923782348633,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.4342,
      "step": 191
    },
    {
      "epoch": 1.5177865612648223,
      "grad_norm": 0.5930154323577881,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.6225,
      "step": 192
    },
    {
      "epoch": 1.5256916996047432,
      "grad_norm": 0.5818495750427246,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.5699,
      "step": 193
    },
    {
      "epoch": 1.533596837944664,
      "grad_norm": 0.5709719061851501,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.4472,
      "step": 194
    },
    {
      "epoch": 1.541501976284585,
      "grad_norm": 0.6492523550987244,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.6734,
      "step": 195
    },
    {
      "epoch": 1.5494071146245059,
      "grad_norm": 0.583744466304779,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.4661,
      "step": 196
    },
    {
      "epoch": 1.5573122529644268,
      "grad_norm": 0.5916106700897217,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.5221,
      "step": 197
    },
    {
      "epoch": 1.5652173913043477,
      "grad_norm": 0.5946465134620667,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.4913,
      "step": 198
    },
    {
      "epoch": 1.5731225296442688,
      "grad_norm": 0.5889039039611816,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.4059,
      "step": 199
    },
    {
      "epoch": 1.5810276679841897,
      "grad_norm": 0.6330969333648682,
      "learning_rate": 0.0,
      "loss": 1.5537,
      "step": 200
    },
    {
      "epoch": 1.5810276679841897,
      "eval_loss": 1.6125844717025757,
      "eval_runtime": 4.8243,
      "eval_samples_per_second": 44.151,
      "eval_steps_per_second": 11.193,
      "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": 2.0127866433306624e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}