{
  "best_metric": 1.707595944404602,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.3093289689034369,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006546644844517185,
      "grad_norm": 2.13655686378479,
      "learning_rate": 1e-05,
      "loss": 8.7017,
      "step": 1
    },
    {
      "epoch": 0.006546644844517185,
      "eval_loss": 2.8909528255462646,
      "eval_runtime": 26.1547,
      "eval_samples_per_second": 9.864,
      "eval_steps_per_second": 2.485,
      "step": 1
    },
    {
      "epoch": 0.01309328968903437,
      "grad_norm": 1.9584050178527832,
      "learning_rate": 2e-05,
      "loss": 8.7593,
      "step": 2
    },
    {
      "epoch": 0.019639934533551555,
      "grad_norm": 1.92738938331604,
      "learning_rate": 3e-05,
      "loss": 9.2191,
      "step": 3
    },
    {
      "epoch": 0.02618657937806874,
      "grad_norm": 1.802982211112976,
      "learning_rate": 4e-05,
      "loss": 8.9038,
      "step": 4
    },
    {
      "epoch": 0.03273322422258593,
      "grad_norm": 1.6323221921920776,
      "learning_rate": 5e-05,
      "loss": 9.051,
      "step": 5
    },
    {
      "epoch": 0.03927986906710311,
      "grad_norm": 1.6458439826965332,
      "learning_rate": 6e-05,
      "loss": 9.0919,
      "step": 6
    },
    {
      "epoch": 0.04582651391162029,
      "grad_norm": 1.7354305982589722,
      "learning_rate": 7e-05,
      "loss": 9.5186,
      "step": 7
    },
    {
      "epoch": 0.05237315875613748,
      "grad_norm": 1.9769362211227417,
      "learning_rate": 8e-05,
      "loss": 9.3298,
      "step": 8
    },
    {
      "epoch": 0.058919803600654665,
      "grad_norm": 1.8338909149169922,
      "learning_rate": 9e-05,
      "loss": 10.5342,
      "step": 9
    },
    {
      "epoch": 0.06546644844517185,
      "grad_norm": 1.890850305557251,
      "learning_rate": 0.0001,
      "loss": 8.1525,
      "step": 10
    },
    {
      "epoch": 0.07201309328968904,
      "grad_norm": 2.500300645828247,
      "learning_rate": 9.999316524962345e-05,
      "loss": 10.1409,
      "step": 11
    },
    {
      "epoch": 0.07855973813420622,
      "grad_norm": 2.272092342376709,
      "learning_rate": 9.997266286704631e-05,
      "loss": 10.0184,
      "step": 12
    },
    {
      "epoch": 0.0851063829787234,
      "grad_norm": 2.830733299255371,
      "learning_rate": 9.993849845741524e-05,
      "loss": 9.3191,
      "step": 13
    },
    {
      "epoch": 0.09165302782324058,
      "grad_norm": 3.044830322265625,
      "learning_rate": 9.989068136093873e-05,
      "loss": 8.7616,
      "step": 14
    },
    {
      "epoch": 0.09819967266775777,
      "grad_norm": 3.9173166751861572,
      "learning_rate": 9.98292246503335e-05,
      "loss": 9.8097,
      "step": 15
    },
    {
      "epoch": 0.10474631751227496,
      "grad_norm": 4.72042989730835,
      "learning_rate": 9.975414512725057e-05,
      "loss": 7.531,
      "step": 16
    },
    {
      "epoch": 0.11129296235679215,
      "grad_norm": 4.051272392272949,
      "learning_rate": 9.966546331768191e-05,
      "loss": 7.9077,
      "step": 17
    },
    {
      "epoch": 0.11783960720130933,
      "grad_norm": 3.737215042114258,
      "learning_rate": 9.956320346634876e-05,
      "loss": 9.7853,
      "step": 18
    },
    {
      "epoch": 0.12438625204582651,
      "grad_norm": 4.385887145996094,
      "learning_rate": 9.944739353007344e-05,
      "loss": 8.0269,
      "step": 19
    },
    {
      "epoch": 0.1309328968903437,
      "grad_norm": 3.413221597671509,
      "learning_rate": 9.931806517013612e-05,
      "loss": 8.4059,
      "step": 20
    },
    {
      "epoch": 0.13747954173486088,
      "grad_norm": 3.395505905151367,
      "learning_rate": 9.917525374361912e-05,
      "loss": 7.9983,
      "step": 21
    },
    {
      "epoch": 0.14402618657937807,
      "grad_norm": 3.604326009750366,
      "learning_rate": 9.901899829374047e-05,
      "loss": 7.5998,
      "step": 22
    },
    {
      "epoch": 0.15057283142389524,
      "grad_norm": 3.5756168365478516,
      "learning_rate": 9.884934153917997e-05,
      "loss": 8.0325,
      "step": 23
    },
    {
      "epoch": 0.15711947626841244,
      "grad_norm": 3.7345871925354004,
      "learning_rate": 9.86663298624003e-05,
      "loss": 7.4147,
      "step": 24
    },
    {
      "epoch": 0.16366612111292964,
      "grad_norm": 3.4550774097442627,
      "learning_rate": 9.847001329696653e-05,
      "loss": 7.8503,
      "step": 25
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 3.4074184894561768,
      "learning_rate": 9.826044551386744e-05,
      "loss": 6.5266,
      "step": 26
    },
    {
      "epoch": 0.176759410801964,
      "grad_norm": 3.3474767208099365,
      "learning_rate": 9.803768380684242e-05,
      "loss": 8.3397,
      "step": 27
    },
    {
      "epoch": 0.18330605564648117,
      "grad_norm": 3.2262606620788574,
      "learning_rate": 9.780178907671789e-05,
      "loss": 7.8356,
      "step": 28
    },
    {
      "epoch": 0.18985270049099837,
      "grad_norm": 3.6275148391723633,
      "learning_rate": 9.755282581475769e-05,
      "loss": 8.2803,
      "step": 29
    },
    {
      "epoch": 0.19639934533551553,
      "grad_norm": 3.781080722808838,
      "learning_rate": 9.729086208503174e-05,
      "loss": 8.1983,
      "step": 30
    },
    {
      "epoch": 0.20294599018003273,
      "grad_norm": 3.650963068008423,
      "learning_rate": 9.701596950580806e-05,
      "loss": 8.1172,
      "step": 31
    },
    {
      "epoch": 0.20949263502454993,
      "grad_norm": 4.347227573394775,
      "learning_rate": 9.672822322997305e-05,
      "loss": 9.1505,
      "step": 32
    },
    {
      "epoch": 0.2160392798690671,
      "grad_norm": 4.019195556640625,
      "learning_rate": 9.642770192448536e-05,
      "loss": 6.9112,
      "step": 33
    },
    {
      "epoch": 0.2225859247135843,
      "grad_norm": 4.088239669799805,
      "learning_rate": 9.611448774886924e-05,
      "loss": 7.8147,
      "step": 34
    },
    {
      "epoch": 0.22913256955810146,
      "grad_norm": 4.491867542266846,
      "learning_rate": 9.578866633275288e-05,
      "loss": 7.8134,
      "step": 35
    },
    {
      "epoch": 0.23567921440261866,
      "grad_norm": 4.543848514556885,
      "learning_rate": 9.545032675245813e-05,
      "loss": 6.7855,
      "step": 36
    },
    {
      "epoch": 0.24222585924713586,
      "grad_norm": 4.584990978240967,
      "learning_rate": 9.509956150664796e-05,
      "loss": 9.0826,
      "step": 37
    },
    {
      "epoch": 0.24877250409165302,
      "grad_norm": 5.8984479904174805,
      "learning_rate": 9.473646649103818e-05,
      "loss": 9.0719,
      "step": 38
    },
    {
      "epoch": 0.2553191489361702,
      "grad_norm": 13.235140800476074,
      "learning_rate": 9.43611409721806e-05,
      "loss": 7.4419,
      "step": 39
    },
    {
      "epoch": 0.2618657937806874,
      "grad_norm": 8.171952247619629,
      "learning_rate": 9.397368756032445e-05,
      "loss": 7.5932,
      "step": 40
    },
    {
      "epoch": 0.2684124386252046,
      "grad_norm": 5.803350448608398,
      "learning_rate": 9.357421218136386e-05,
      "loss": 7.1615,
      "step": 41
    },
    {
      "epoch": 0.27495908346972175,
      "grad_norm": 5.395884037017822,
      "learning_rate": 9.316282404787871e-05,
      "loss": 7.2399,
      "step": 42
    },
    {
      "epoch": 0.281505728314239,
      "grad_norm": 4.930261135101318,
      "learning_rate": 9.273963562927695e-05,
      "loss": 6.9501,
      "step": 43
    },
    {
      "epoch": 0.28805237315875615,
      "grad_norm": 4.3326826095581055,
      "learning_rate": 9.230476262104677e-05,
      "loss": 6.7461,
      "step": 44
    },
    {
      "epoch": 0.2945990180032733,
      "grad_norm": 3.443617820739746,
      "learning_rate": 9.185832391312644e-05,
      "loss": 7.678,
      "step": 45
    },
    {
      "epoch": 0.3011456628477905,
      "grad_norm": 3.4096052646636963,
      "learning_rate": 9.140044155740101e-05,
      "loss": 8.0259,
      "step": 46
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 3.3578953742980957,
      "learning_rate": 9.093124073433463e-05,
      "loss": 8.1257,
      "step": 47
    },
    {
      "epoch": 0.3142389525368249,
      "grad_norm": 3.4864139556884766,
      "learning_rate": 9.045084971874738e-05,
      "loss": 7.6784,
      "step": 48
    },
    {
      "epoch": 0.32078559738134205,
      "grad_norm": 3.0312955379486084,
      "learning_rate": 8.995939984474624e-05,
      "loss": 7.3808,
      "step": 49
    },
    {
      "epoch": 0.32733224222585927,
      "grad_norm": 3.1818759441375732,
      "learning_rate": 8.945702546981969e-05,
      "loss": 7.7415,
      "step": 50
    },
    {
      "epoch": 0.32733224222585927,
      "eval_loss": 1.9966988563537598,
      "eval_runtime": 26.5888,
      "eval_samples_per_second": 9.703,
      "eval_steps_per_second": 2.445,
      "step": 50
    },
    {
      "epoch": 0.33387888707037644,
      "grad_norm": 2.8749489784240723,
      "learning_rate": 8.894386393810563e-05,
      "loss": 7.5203,
      "step": 51
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 3.1056835651397705,
      "learning_rate": 8.842005554284296e-05,
      "loss": 6.7367,
      "step": 52
    },
    {
      "epoch": 0.3469721767594108,
      "grad_norm": 2.9506490230560303,
      "learning_rate": 8.788574348801675e-05,
      "loss": 8.1926,
      "step": 53
    },
    {
      "epoch": 0.353518821603928,
      "grad_norm": 2.955796003341675,
      "learning_rate": 8.73410738492077e-05,
      "loss": 7.9538,
      "step": 54
    },
    {
      "epoch": 0.36006546644844517,
      "grad_norm": 3.287029981613159,
      "learning_rate": 8.678619553365659e-05,
      "loss": 7.2937,
      "step": 55
    },
    {
      "epoch": 0.36661211129296234,
      "grad_norm": 3.352778911590576,
      "learning_rate": 8.622126023955446e-05,
      "loss": 6.9685,
      "step": 56
    },
    {
      "epoch": 0.37315875613747956,
      "grad_norm": 3.1772475242614746,
      "learning_rate": 8.564642241456986e-05,
      "loss": 8.2336,
      "step": 57
    },
    {
      "epoch": 0.37970540098199673,
      "grad_norm": 3.9878532886505127,
      "learning_rate": 8.506183921362443e-05,
      "loss": 7.6858,
      "step": 58
    },
    {
      "epoch": 0.3862520458265139,
      "grad_norm": 3.363985300064087,
      "learning_rate": 8.44676704559283e-05,
      "loss": 7.5955,
      "step": 59
    },
    {
      "epoch": 0.39279869067103107,
      "grad_norm": 3.496302366256714,
      "learning_rate": 8.386407858128706e-05,
      "loss": 8.1299,
      "step": 60
    },
    {
      "epoch": 0.3993453355155483,
      "grad_norm": 3.4857332706451416,
      "learning_rate": 8.32512286056924e-05,
      "loss": 6.8698,
      "step": 61
    },
    {
      "epoch": 0.40589198036006546,
      "grad_norm": 3.369488477706909,
      "learning_rate": 8.262928807620843e-05,
      "loss": 6.6206,
      "step": 62
    },
    {
      "epoch": 0.41243862520458263,
      "grad_norm": 4.747243404388428,
      "learning_rate": 8.199842702516583e-05,
      "loss": 8.7977,
      "step": 63
    },
    {
      "epoch": 0.41898527004909986,
      "grad_norm": 4.112165927886963,
      "learning_rate": 8.135881792367686e-05,
      "loss": 7.1464,
      "step": 64
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 4.025830268859863,
      "learning_rate": 8.07106356344834e-05,
      "loss": 7.6691,
      "step": 65
    },
    {
      "epoch": 0.4320785597381342,
      "grad_norm": 3.832612991333008,
      "learning_rate": 8.005405736415126e-05,
      "loss": 8.5289,
      "step": 66
    },
    {
      "epoch": 0.4386252045826514,
      "grad_norm": 3.802485466003418,
      "learning_rate": 7.938926261462366e-05,
      "loss": 8.9616,
      "step": 67
    },
    {
      "epoch": 0.4451718494271686,
      "grad_norm": 3.8989100456237793,
      "learning_rate": 7.871643313414718e-05,
      "loss": 9.464,
      "step": 68
    },
    {
      "epoch": 0.45171849427168576,
      "grad_norm": 5.011821269989014,
      "learning_rate": 7.803575286758364e-05,
      "loss": 8.5757,
      "step": 69
    },
    {
      "epoch": 0.4582651391162029,
      "grad_norm": 3.615846633911133,
      "learning_rate": 7.734740790612136e-05,
      "loss": 7.5682,
      "step": 70
    },
    {
      "epoch": 0.46481178396072015,
      "grad_norm": 3.701805830001831,
      "learning_rate": 7.66515864363997e-05,
      "loss": 5.8679,
      "step": 71
    },
    {
      "epoch": 0.4713584288052373,
      "grad_norm": 4.6807661056518555,
      "learning_rate": 7.594847868906076e-05,
      "loss": 8.063,
      "step": 72
    },
    {
      "epoch": 0.4779050736497545,
      "grad_norm": 3.485200881958008,
      "learning_rate": 7.52382768867422e-05,
      "loss": 7.4695,
      "step": 73
    },
    {
      "epoch": 0.4844517184942717,
      "grad_norm": 4.052950382232666,
      "learning_rate": 7.452117519152542e-05,
      "loss": 7.3515,
      "step": 74
    },
    {
      "epoch": 0.4909983633387889,
      "grad_norm": 4.987173557281494,
      "learning_rate": 7.379736965185368e-05,
      "loss": 8.0323,
      "step": 75
    },
    {
      "epoch": 0.49754500818330605,
      "grad_norm": 7.222501754760742,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.0476,
      "step": 76
    },
    {
      "epoch": 0.5040916530278232,
      "grad_norm": 4.9196391105651855,
      "learning_rate": 7.233044034264034e-05,
      "loss": 6.0554,
      "step": 77
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": 6.087227821350098,
      "learning_rate": 7.158771761692464e-05,
      "loss": 7.014,
      "step": 78
    },
    {
      "epoch": 0.5171849427168577,
      "grad_norm": 5.477786064147949,
      "learning_rate": 7.083909302476453e-05,
      "loss": 5.7058,
      "step": 79
    },
    {
      "epoch": 0.5237315875613748,
      "grad_norm": 4.938673496246338,
      "learning_rate": 7.008477123264848e-05,
      "loss": 6.7502,
      "step": 80
    },
    {
      "epoch": 0.530278232405892,
      "grad_norm": 4.272306442260742,
      "learning_rate": 6.932495846462261e-05,
      "loss": 6.1399,
      "step": 81
    },
    {
      "epoch": 0.5368248772504092,
      "grad_norm": 4.495388031005859,
      "learning_rate": 6.855986244591104e-05,
      "loss": 7.7363,
      "step": 82
    },
    {
      "epoch": 0.5433715220949263,
      "grad_norm": 3.923799991607666,
      "learning_rate": 6.778969234612584e-05,
      "loss": 6.6252,
      "step": 83
    },
    {
      "epoch": 0.5499181669394435,
      "grad_norm": 3.6672651767730713,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.3417,
      "step": 84
    },
    {
      "epoch": 0.5564648117839607,
      "grad_norm": 3.5456626415252686,
      "learning_rate": 6.623497346023418e-05,
      "loss": 7.1854,
      "step": 85
    },
    {
      "epoch": 0.563011456628478,
      "grad_norm": 3.7435622215270996,
      "learning_rate": 6.545084971874738e-05,
      "loss": 7.2925,
      "step": 86
    },
    {
      "epoch": 0.5695581014729951,
      "grad_norm": 3.5279946327209473,
      "learning_rate": 6.466250186922325e-05,
      "loss": 7.2659,
      "step": 87
    },
    {
      "epoch": 0.5761047463175123,
      "grad_norm": 3.6771609783172607,
      "learning_rate": 6.387014543809223e-05,
      "loss": 7.4484,
      "step": 88
    },
    {
      "epoch": 0.5826513911620295,
      "grad_norm": 3.353963613510132,
      "learning_rate": 6.307399704769099e-05,
      "loss": 7.9794,
      "step": 89
    },
    {
      "epoch": 0.5891980360065466,
      "grad_norm": 3.9403250217437744,
      "learning_rate": 6.227427435703997e-05,
      "loss": 7.3322,
      "step": 90
    },
    {
      "epoch": 0.5957446808510638,
      "grad_norm": 3.8401012420654297,
      "learning_rate": 6.147119600233758e-05,
      "loss": 7.652,
      "step": 91
    },
    {
      "epoch": 0.602291325695581,
      "grad_norm": 4.258552074432373,
      "learning_rate": 6.066498153718735e-05,
      "loss": 7.9399,
      "step": 92
    },
    {
      "epoch": 0.6088379705400983,
      "grad_norm": 3.9695637226104736,
      "learning_rate": 5.985585137257401e-05,
      "loss": 8.1528,
      "step": 93
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 3.8052663803100586,
      "learning_rate": 5.90440267166055e-05,
      "loss": 6.5618,
      "step": 94
    },
    {
      "epoch": 0.6219312602291326,
      "grad_norm": 3.762793779373169,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 7.6469,
      "step": 95
    },
    {
      "epoch": 0.6284779050736498,
      "grad_norm": 3.9404263496398926,
      "learning_rate": 5.74131823855921e-05,
      "loss": 6.7442,
      "step": 96
    },
    {
      "epoch": 0.6350245499181669,
      "grad_norm": 4.887734889984131,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 7.1085,
      "step": 97
    },
    {
      "epoch": 0.6415711947626841,
      "grad_norm": 3.9277775287628174,
      "learning_rate": 5.577423184847932e-05,
      "loss": 6.8363,
      "step": 98
    },
    {
      "epoch": 0.6481178396072013,
      "grad_norm": 4.448163986206055,
      "learning_rate": 5.495227651252315e-05,
      "loss": 8.6857,
      "step": 99
    },
    {
      "epoch": 0.6546644844517185,
      "grad_norm": 3.6832666397094727,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 7.4869,
      "step": 100
    },
    {
      "epoch": 0.6546644844517185,
      "eval_loss": 1.8067591190338135,
      "eval_runtime": 26.62,
      "eval_samples_per_second": 9.692,
      "eval_steps_per_second": 2.442,
      "step": 100
    },
    {
      "epoch": 0.6612111292962357,
      "grad_norm": 4.730930328369141,
      "learning_rate": 5.330452921628497e-05,
      "loss": 7.4568,
      "step": 101
    },
    {
      "epoch": 0.6677577741407529,
      "grad_norm": 4.326946258544922,
      "learning_rate": 5.247918773366112e-05,
      "loss": 7.3844,
      "step": 102
    },
    {
      "epoch": 0.67430441898527,
      "grad_norm": 4.695581436157227,
      "learning_rate": 5.165316846586541e-05,
      "loss": 7.6806,
      "step": 103
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 4.098830223083496,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 6.1029,
      "step": 104
    },
    {
      "epoch": 0.6873977086743044,
      "grad_norm": 3.6151130199432373,
      "learning_rate": 5e-05,
      "loss": 7.0723,
      "step": 105
    },
    {
      "epoch": 0.6939443535188216,
      "grad_norm": 4.55037784576416,
      "learning_rate": 4.917330276168208e-05,
      "loss": 7.8156,
      "step": 106
    },
    {
      "epoch": 0.7004909983633388,
      "grad_norm": 3.7407169342041016,
      "learning_rate": 4.834683153413459e-05,
      "loss": 7.0921,
      "step": 107
    },
    {
      "epoch": 0.707037643207856,
      "grad_norm": 4.5362372398376465,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 8.2413,
      "step": 108
    },
    {
      "epoch": 0.7135842880523732,
      "grad_norm": 5.322324275970459,
      "learning_rate": 4.669547078371504e-05,
      "loss": 8.7592,
      "step": 109
    },
    {
      "epoch": 0.7201309328968903,
      "grad_norm": 4.402954578399658,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 6.0549,
      "step": 110
    },
    {
      "epoch": 0.7266775777414075,
      "grad_norm": 4.464359283447266,
      "learning_rate": 4.504772348747687e-05,
      "loss": 6.4,
      "step": 111
    },
    {
      "epoch": 0.7332242225859247,
      "grad_norm": 4.0333333015441895,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 5.8919,
      "step": 112
    },
    {
      "epoch": 0.7397708674304418,
      "grad_norm": 4.631279468536377,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 6.8713,
      "step": 113
    },
    {
      "epoch": 0.7463175122749591,
      "grad_norm": 5.974384784698486,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 7.7068,
      "step": 114
    },
    {
      "epoch": 0.7528641571194763,
      "grad_norm": 3.7521626949310303,
      "learning_rate": 4.17702704859633e-05,
      "loss": 4.9648,
      "step": 115
    },
    {
      "epoch": 0.7594108019639935,
      "grad_norm": 3.2828927040100098,
      "learning_rate": 4.095597328339452e-05,
      "loss": 4.7717,
      "step": 116
    },
    {
      "epoch": 0.7659574468085106,
      "grad_norm": 3.8715484142303467,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 5.6252,
      "step": 117
    },
    {
      "epoch": 0.7725040916530278,
      "grad_norm": 3.565499782562256,
      "learning_rate": 3.933501846281267e-05,
      "loss": 5.0633,
      "step": 118
    },
    {
      "epoch": 0.779050736497545,
      "grad_norm": 4.171899318695068,
      "learning_rate": 3.852880399766243e-05,
      "loss": 5.5828,
      "step": 119
    },
    {
      "epoch": 0.7855973813420621,
      "grad_norm": 3.9527981281280518,
      "learning_rate": 3.772572564296005e-05,
      "loss": 6.6647,
      "step": 120
    },
    {
      "epoch": 0.7921440261865794,
      "grad_norm": 4.4996137619018555,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 6.0745,
      "step": 121
    },
    {
      "epoch": 0.7986906710310966,
      "grad_norm": 4.248125076293945,
      "learning_rate": 3.612985456190778e-05,
      "loss": 6.2583,
      "step": 122
    },
    {
      "epoch": 0.8052373158756138,
      "grad_norm": 4.524069309234619,
      "learning_rate": 3.533749813077677e-05,
      "loss": 7.6514,
      "step": 123
    },
    {
      "epoch": 0.8117839607201309,
      "grad_norm": 4.165614128112793,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 6.5651,
      "step": 124
    },
    {
      "epoch": 0.8183306055646481,
      "grad_norm": 3.7812719345092773,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 8.0841,
      "step": 125
    },
    {
      "epoch": 0.8248772504091653,
      "grad_norm": 3.8515408039093018,
      "learning_rate": 3.298534127791785e-05,
      "loss": 6.645,
      "step": 126
    },
    {
      "epoch": 0.8314238952536824,
      "grad_norm": 3.675601005554199,
      "learning_rate": 3.221030765387417e-05,
      "loss": 6.6233,
      "step": 127
    },
    {
      "epoch": 0.8379705400981997,
      "grad_norm": 4.184865474700928,
      "learning_rate": 3.144013755408895e-05,
      "loss": 6.5936,
      "step": 128
    },
    {
      "epoch": 0.8445171849427169,
      "grad_norm": 3.9905803203582764,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 6.6051,
      "step": 129
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 3.789395570755005,
      "learning_rate": 2.991522876735154e-05,
      "loss": 6.7708,
      "step": 130
    },
    {
      "epoch": 0.8576104746317512,
      "grad_norm": 4.332606792449951,
      "learning_rate": 2.916090697523549e-05,
      "loss": 6.4427,
      "step": 131
    },
    {
      "epoch": 0.8641571194762684,
      "grad_norm": 4.295475959777832,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 5.8015,
      "step": 132
    },
    {
      "epoch": 0.8707037643207856,
      "grad_norm": 4.5518388748168945,
      "learning_rate": 2.766955965735968e-05,
      "loss": 7.3128,
      "step": 133
    },
    {
      "epoch": 0.8772504091653028,
      "grad_norm": 4.419987201690674,
      "learning_rate": 2.693294185106562e-05,
      "loss": 7.4535,
      "step": 134
    },
    {
      "epoch": 0.88379705400982,
      "grad_norm": 4.4990458488464355,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 6.4881,
      "step": 135
    },
    {
      "epoch": 0.8903436988543372,
      "grad_norm": 5.589775085449219,
      "learning_rate": 2.547882480847461e-05,
      "loss": 6.4211,
      "step": 136
    },
    {
      "epoch": 0.8968903436988543,
      "grad_norm": 4.943357944488525,
      "learning_rate": 2.476172311325783e-05,
      "loss": 7.0361,
      "step": 137
    },
    {
      "epoch": 0.9034369885433715,
      "grad_norm": 5.135087490081787,
      "learning_rate": 2.405152131093926e-05,
      "loss": 8.3705,
      "step": 138
    },
    {
      "epoch": 0.9099836333878887,
      "grad_norm": 4.697838306427002,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 6.7204,
      "step": 139
    },
    {
      "epoch": 0.9165302782324058,
      "grad_norm": 4.322061538696289,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 8.1387,
      "step": 140
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 4.582803249359131,
      "learning_rate": 2.196424713241637e-05,
      "loss": 6.67,
      "step": 141
    },
    {
      "epoch": 0.9296235679214403,
      "grad_norm": 4.9027838706970215,
      "learning_rate": 2.128356686585282e-05,
      "loss": 7.9627,
      "step": 142
    },
    {
      "epoch": 0.9361702127659575,
      "grad_norm": 4.5149827003479,
      "learning_rate": 2.061073738537635e-05,
      "loss": 6.4197,
      "step": 143
    },
    {
      "epoch": 0.9427168576104746,
      "grad_norm": 5.014720916748047,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 7.2758,
      "step": 144
    },
    {
      "epoch": 0.9492635024549918,
      "grad_norm": 4.3964362144470215,
      "learning_rate": 1.928936436551661e-05,
      "loss": 6.5388,
      "step": 145
    },
    {
      "epoch": 0.955810147299509,
      "grad_norm": 4.305325508117676,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 7.328,
      "step": 146
    },
    {
      "epoch": 0.9623567921440261,
      "grad_norm": 4.718265056610107,
      "learning_rate": 1.800157297483417e-05,
      "loss": 8.0182,
      "step": 147
    },
    {
      "epoch": 0.9689034369885434,
      "grad_norm": 5.577218055725098,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 6.0249,
      "step": 148
    },
    {
      "epoch": 0.9754500818330606,
      "grad_norm": 4.048355579376221,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 5.1523,
      "step": 149
    },
    {
      "epoch": 0.9819967266775778,
      "grad_norm": 4.773705959320068,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 6.3304,
      "step": 150
    },
    {
      "epoch": 0.9819967266775778,
      "eval_loss": 1.7272412776947021,
      "eval_runtime": 26.6126,
      "eval_samples_per_second": 9.695,
      "eval_steps_per_second": 2.442,
      "step": 150
    },
    {
      "epoch": 0.9885433715220949,
      "grad_norm": 6.138540744781494,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.892,
      "step": 151
    },
    {
      "epoch": 0.9950900163666121,
      "grad_norm": 6.4282402992248535,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 7.8377,
      "step": 152
    },
    {
      "epoch": 1.0016366612111294,
      "grad_norm": 3.81278657913208,
      "learning_rate": 1.435357758543015e-05,
      "loss": 7.4905,
      "step": 153
    },
    {
      "epoch": 1.0081833060556464,
      "grad_norm": 2.5406858921051025,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 4.6057,
      "step": 154
    },
    {
      "epoch": 1.0147299509001637,
      "grad_norm": 2.9028007984161377,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.4537,
      "step": 155
    },
    {
      "epoch": 1.0212765957446808,
      "grad_norm": 3.118882417678833,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 4.9172,
      "step": 156
    },
    {
      "epoch": 1.027823240589198,
      "grad_norm": 3.208533525466919,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 4.9149,
      "step": 157
    },
    {
      "epoch": 1.034369885433715,
      "grad_norm": 3.223609209060669,
      "learning_rate": 1.157994445715706e-05,
      "loss": 4.4849,
      "step": 158
    },
    {
      "epoch": 1.0409165302782324,
      "grad_norm": 3.3039231300354004,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 5.4302,
      "step": 159
    },
    {
      "epoch": 1.0474631751227497,
      "grad_norm": 3.211501121520996,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 4.5527,
      "step": 160
    },
    {
      "epoch": 1.0540098199672667,
      "grad_norm": 3.086197853088379,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 6.0466,
      "step": 161
    },
    {
      "epoch": 1.060556464811784,
      "grad_norm": 3.426628589630127,
      "learning_rate": 9.549150281252633e-06,
      "loss": 5.1313,
      "step": 162
    },
    {
      "epoch": 1.067103109656301,
      "grad_norm": 3.1073648929595947,
      "learning_rate": 9.068759265665384e-06,
      "loss": 6.1329,
      "step": 163
    },
    {
      "epoch": 1.0736497545008183,
      "grad_norm": 3.4100728034973145,
      "learning_rate": 8.599558442598998e-06,
      "loss": 5.1229,
      "step": 164
    },
    {
      "epoch": 1.0801963993453354,
      "grad_norm": 3.2149834632873535,
      "learning_rate": 8.141676086873572e-06,
      "loss": 5.9124,
      "step": 165
    },
    {
      "epoch": 1.0867430441898527,
      "grad_norm": 3.5418028831481934,
      "learning_rate": 7.695237378953223e-06,
      "loss": 5.2255,
      "step": 166
    },
    {
      "epoch": 1.09328968903437,
      "grad_norm": 3.696084976196289,
      "learning_rate": 7.260364370723044e-06,
      "loss": 5.2342,
      "step": 167
    },
    {
      "epoch": 1.099836333878887,
      "grad_norm": 3.730708360671997,
      "learning_rate": 6.837175952121306e-06,
      "loss": 5.8431,
      "step": 168
    },
    {
      "epoch": 1.1063829787234043,
      "grad_norm": 3.742860794067383,
      "learning_rate": 6.425787818636131e-06,
      "loss": 4.6969,
      "step": 169
    },
    {
      "epoch": 1.1129296235679214,
      "grad_norm": 3.598694086074829,
      "learning_rate": 6.026312439675552e-06,
      "loss": 5.7549,
      "step": 170
    },
    {
      "epoch": 1.1194762684124386,
      "grad_norm": 3.616276502609253,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 5.1967,
      "step": 171
    },
    {
      "epoch": 1.1260229132569557,
      "grad_norm": 3.9378201961517334,
      "learning_rate": 5.263533508961827e-06,
      "loss": 5.3633,
      "step": 172
    },
    {
      "epoch": 1.132569558101473,
      "grad_norm": 3.6692304611206055,
      "learning_rate": 4.900438493352055e-06,
      "loss": 5.5043,
      "step": 173
    },
    {
      "epoch": 1.1391162029459903,
      "grad_norm": 3.8457822799682617,
      "learning_rate": 4.549673247541875e-06,
      "loss": 5.2013,
      "step": 174
    },
    {
      "epoch": 1.1456628477905073,
      "grad_norm": 4.024327754974365,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 5.0389,
      "step": 175
    },
    {
      "epoch": 1.1522094926350246,
      "grad_norm": 3.874255895614624,
      "learning_rate": 3.885512251130763e-06,
      "loss": 5.1974,
      "step": 176
    },
    {
      "epoch": 1.1587561374795416,
      "grad_norm": 4.2925920486450195,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 6.2995,
      "step": 177
    },
    {
      "epoch": 1.165302782324059,
      "grad_norm": 4.385538578033447,
      "learning_rate": 3.271776770026963e-06,
      "loss": 5.41,
      "step": 178
    },
    {
      "epoch": 1.171849427168576,
      "grad_norm": 4.054061412811279,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 5.9913,
      "step": 179
    },
    {
      "epoch": 1.1783960720130933,
      "grad_norm": 4.442548751831055,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 6.638,
      "step": 180
    },
    {
      "epoch": 1.1849427168576105,
      "grad_norm": 3.944359302520752,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 5.7074,
      "step": 181
    },
    {
      "epoch": 1.1914893617021276,
      "grad_norm": 4.47265625,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 6.308,
      "step": 182
    },
    {
      "epoch": 1.1980360065466449,
      "grad_norm": 4.241458892822266,
      "learning_rate": 1.962316193157593e-06,
      "loss": 5.8567,
      "step": 183
    },
    {
      "epoch": 1.204582651391162,
      "grad_norm": 4.236311435699463,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 6.3607,
      "step": 184
    },
    {
      "epoch": 1.2111292962356792,
      "grad_norm": 4.7658467292785645,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 6.0578,
      "step": 185
    },
    {
      "epoch": 1.2176759410801963,
      "grad_norm": 4.109133720397949,
      "learning_rate": 1.333670137599713e-06,
      "loss": 4.8959,
      "step": 186
    },
    {
      "epoch": 1.2242225859247136,
      "grad_norm": 4.497995853424072,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 5.8,
      "step": 187
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 4.011194229125977,
      "learning_rate": 9.810017062595322e-07,
      "loss": 4.6202,
      "step": 188
    },
    {
      "epoch": 1.237315875613748,
      "grad_norm": 4.688493251800537,
      "learning_rate": 8.247462563808817e-07,
      "loss": 5.2817,
      "step": 189
    },
    {
      "epoch": 1.2438625204582652,
      "grad_norm": 5.393063068389893,
      "learning_rate": 6.819348298638839e-07,
      "loss": 6.4032,
      "step": 190
    },
    {
      "epoch": 1.2504091653027825,
      "grad_norm": 5.186883449554443,
      "learning_rate": 5.526064699265753e-07,
      "loss": 6.4349,
      "step": 191
    },
    {
      "epoch": 1.2569558101472995,
      "grad_norm": 2.460773229598999,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.6578,
      "step": 192
    },
    {
      "epoch": 1.2635024549918166,
      "grad_norm": 3.062981128692627,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.9411,
      "step": 193
    },
    {
      "epoch": 1.2700490998363339,
      "grad_norm": 3.388601541519165,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.6924,
      "step": 194
    },
    {
      "epoch": 1.2765957446808511,
      "grad_norm": 3.355208158493042,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.653,
      "step": 195
    },
    {
      "epoch": 1.2831423895253682,
      "grad_norm": 3.150766611099243,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.5391,
      "step": 196
    },
    {
      "epoch": 1.2896890343698855,
      "grad_norm": 3.37044095993042,
      "learning_rate": 6.150154258476315e-08,
      "loss": 5.5631,
      "step": 197
    },
    {
      "epoch": 1.2962356792144027,
      "grad_norm": 3.706301212310791,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 5.4415,
      "step": 198
    },
    {
      "epoch": 1.3027823240589198,
      "grad_norm": 3.9071426391601562,
      "learning_rate": 6.834750376549792e-09,
      "loss": 5.121,
      "step": 199
    },
    {
      "epoch": 1.3093289689034369,
      "grad_norm": 3.6013264656066895,
      "learning_rate": 0.0,
      "loss": 6.7051,
      "step": 200
    },
    {
      "epoch": 1.3093289689034369,
      "eval_loss": 1.707595944404602,
      "eval_runtime": 26.5748,
      "eval_samples_per_second": 9.708,
      "eval_steps_per_second": 2.446,
      "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": 5.09187566075904e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}