{
  "best_metric": 13.871726989746094,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.07982438635002993,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003991219317501497,
      "grad_norm": 1226.1240234375,
      "learning_rate": 1e-05,
      "loss": 38.9391,
      "step": 1
    },
    {
      "epoch": 0.0003991219317501497,
      "eval_loss": 202.20130920410156,
      "eval_runtime": 14.865,
      "eval_samples_per_second": 283.887,
      "eval_steps_per_second": 70.972,
      "step": 1
    },
    {
      "epoch": 0.0007982438635002994,
      "grad_norm": 1416.827392578125,
      "learning_rate": 2e-05,
      "loss": 64.1441,
      "step": 2
    },
    {
      "epoch": 0.001197365795250449,
      "grad_norm": 1256.3453369140625,
      "learning_rate": 3e-05,
      "loss": 91.2296,
      "step": 3
    },
    {
      "epoch": 0.0015964877270005988,
      "grad_norm": 1671.798095703125,
      "learning_rate": 4e-05,
      "loss": 119.4075,
      "step": 4
    },
    {
      "epoch": 0.001995609658750748,
      "grad_norm": 1328.7642822265625,
      "learning_rate": 5e-05,
      "loss": 139.5309,
      "step": 5
    },
    {
      "epoch": 0.002394731590500898,
      "grad_norm": 1983.4398193359375,
      "learning_rate": 6e-05,
      "loss": 160.3559,
      "step": 6
    },
    {
      "epoch": 0.0027938535222510478,
      "grad_norm": 2041.7457275390625,
      "learning_rate": 7e-05,
      "loss": 168.7198,
      "step": 7
    },
    {
      "epoch": 0.0031929754540011976,
      "grad_norm": 2199.233154296875,
      "learning_rate": 8e-05,
      "loss": 192.3734,
      "step": 8
    },
    {
      "epoch": 0.003592097385751347,
      "grad_norm": 2561.250732421875,
      "learning_rate": 9e-05,
      "loss": 252.5804,
      "step": 9
    },
    {
      "epoch": 0.003991219317501496,
      "grad_norm": 2527.688232421875,
      "learning_rate": 0.0001,
      "loss": 296.5642,
      "step": 10
    },
    {
      "epoch": 0.004390341249251646,
      "grad_norm": 2493.8369140625,
      "learning_rate": 9.999316524962345e-05,
      "loss": 271.4316,
      "step": 11
    },
    {
      "epoch": 0.004789463181001796,
      "grad_norm": 3381.2294921875,
      "learning_rate": 9.997266286704631e-05,
      "loss": 327.654,
      "step": 12
    },
    {
      "epoch": 0.005188585112751946,
      "grad_norm": 2970.509033203125,
      "learning_rate": 9.993849845741524e-05,
      "loss": 333.6062,
      "step": 13
    },
    {
      "epoch": 0.0055877070445020955,
      "grad_norm": 4172.30712890625,
      "learning_rate": 9.989068136093873e-05,
      "loss": 463.3643,
      "step": 14
    },
    {
      "epoch": 0.005986828976252245,
      "grad_norm": 3153.865966796875,
      "learning_rate": 9.98292246503335e-05,
      "loss": 397.7449,
      "step": 15
    },
    {
      "epoch": 0.006385950908002395,
      "grad_norm": 4419.95166015625,
      "learning_rate": 9.975414512725057e-05,
      "loss": 441.5753,
      "step": 16
    },
    {
      "epoch": 0.006785072839752544,
      "grad_norm": 4235.5966796875,
      "learning_rate": 9.966546331768191e-05,
      "loss": 396.0317,
      "step": 17
    },
    {
      "epoch": 0.007184194771502694,
      "grad_norm": 12731.900390625,
      "learning_rate": 9.956320346634876e-05,
      "loss": 640.484,
      "step": 18
    },
    {
      "epoch": 0.007583316703252844,
      "grad_norm": 9764.1474609375,
      "learning_rate": 9.944739353007344e-05,
      "loss": 467.8669,
      "step": 19
    },
    {
      "epoch": 0.007982438635002993,
      "grad_norm": 9968.888671875,
      "learning_rate": 9.931806517013612e-05,
      "loss": 525.7101,
      "step": 20
    },
    {
      "epoch": 0.008381560566753142,
      "grad_norm": 47287.6640625,
      "learning_rate": 9.917525374361912e-05,
      "loss": 742.3817,
      "step": 21
    },
    {
      "epoch": 0.008780682498503292,
      "grad_norm": 21406.8828125,
      "learning_rate": 9.901899829374047e-05,
      "loss": 605.5703,
      "step": 22
    },
    {
      "epoch": 0.009179804430253442,
      "grad_norm": 88159.1796875,
      "learning_rate": 9.884934153917997e-05,
      "loss": 600.4219,
      "step": 23
    },
    {
      "epoch": 0.009578926362003592,
      "grad_norm": 85608.625,
      "learning_rate": 9.86663298624003e-05,
      "loss": 688.0475,
      "step": 24
    },
    {
      "epoch": 0.009978048293753742,
      "grad_norm": 116664.5390625,
      "learning_rate": 9.847001329696653e-05,
      "loss": 593.6304,
      "step": 25
    },
    {
      "epoch": 0.010377170225503891,
      "grad_norm": 362271.0,
      "learning_rate": 9.826044551386744e-05,
      "loss": 539.4735,
      "step": 26
    },
    {
      "epoch": 0.010776292157254041,
      "grad_norm": 830614.625,
      "learning_rate": 9.803768380684242e-05,
      "loss": 635.4505,
      "step": 27
    },
    {
      "epoch": 0.011175414089004191,
      "grad_norm": 432272.625,
      "learning_rate": 9.780178907671789e-05,
      "loss": 480.1909,
      "step": 28
    },
    {
      "epoch": 0.01157453602075434,
      "grad_norm": 483271.3125,
      "learning_rate": 9.755282581475769e-05,
      "loss": 525.3255,
      "step": 29
    },
    {
      "epoch": 0.01197365795250449,
      "grad_norm": 639318.5625,
      "learning_rate": 9.729086208503174e-05,
      "loss": 507.821,
      "step": 30
    },
    {
      "epoch": 0.01237277988425464,
      "grad_norm": 228757.046875,
      "learning_rate": 9.701596950580806e-05,
      "loss": 282.2744,
      "step": 31
    },
    {
      "epoch": 0.01277190181600479,
      "grad_norm": 368029.1875,
      "learning_rate": 9.672822322997305e-05,
      "loss": 201.4605,
      "step": 32
    },
    {
      "epoch": 0.013171023747754938,
      "grad_norm": 229454.859375,
      "learning_rate": 9.642770192448536e-05,
      "loss": 153.6841,
      "step": 33
    },
    {
      "epoch": 0.013570145679505088,
      "grad_norm": 122546.921875,
      "learning_rate": 9.611448774886924e-05,
      "loss": 211.8427,
      "step": 34
    },
    {
      "epoch": 0.013969267611255238,
      "grad_norm": 286743.6875,
      "learning_rate": 9.578866633275288e-05,
      "loss": 219.0722,
      "step": 35
    },
    {
      "epoch": 0.014368389543005388,
      "grad_norm": 166404.265625,
      "learning_rate": 9.545032675245813e-05,
      "loss": 238.9741,
      "step": 36
    },
    {
      "epoch": 0.014767511474755537,
      "grad_norm": 126706.3359375,
      "learning_rate": 9.509956150664796e-05,
      "loss": 205.2873,
      "step": 37
    },
    {
      "epoch": 0.015166633406505687,
      "grad_norm": 148975.078125,
      "learning_rate": 9.473646649103818e-05,
      "loss": 198.7447,
      "step": 38
    },
    {
      "epoch": 0.015565755338255837,
      "grad_norm": 121823.96875,
      "learning_rate": 9.43611409721806e-05,
      "loss": 197.3551,
      "step": 39
    },
    {
      "epoch": 0.015964877270005985,
      "grad_norm": 172075.890625,
      "learning_rate": 9.397368756032445e-05,
      "loss": 157.7947,
      "step": 40
    },
    {
      "epoch": 0.016363999201756137,
      "grad_norm": 256489.21875,
      "learning_rate": 9.357421218136386e-05,
      "loss": 201.5538,
      "step": 41
    },
    {
      "epoch": 0.016763121133506285,
      "grad_norm": 156741.5,
      "learning_rate": 9.316282404787871e-05,
      "loss": 186.1655,
      "step": 42
    },
    {
      "epoch": 0.017162243065256436,
      "grad_norm": 220616.15625,
      "learning_rate": 9.273963562927695e-05,
      "loss": 255.3266,
      "step": 43
    },
    {
      "epoch": 0.017561364997006584,
      "grad_norm": 353515.8125,
      "learning_rate": 9.230476262104677e-05,
      "loss": 237.0643,
      "step": 44
    },
    {
      "epoch": 0.017960486928756736,
      "grad_norm": 287432.84375,
      "learning_rate": 9.185832391312644e-05,
      "loss": 208.0767,
      "step": 45
    },
    {
      "epoch": 0.018359608860506884,
      "grad_norm": 328080.34375,
      "learning_rate": 9.140044155740101e-05,
      "loss": 215.2941,
      "step": 46
    },
    {
      "epoch": 0.018758730792257036,
      "grad_norm": 177177.578125,
      "learning_rate": 9.093124073433463e-05,
      "loss": 192.6448,
      "step": 47
    },
    {
      "epoch": 0.019157852724007184,
      "grad_norm": 261426.3125,
      "learning_rate": 9.045084971874738e-05,
      "loss": 180.7107,
      "step": 48
    },
    {
      "epoch": 0.019556974655757335,
      "grad_norm": 223036.84375,
      "learning_rate": 8.995939984474624e-05,
      "loss": 200.2887,
      "step": 49
    },
    {
      "epoch": 0.019956096587507483,
      "grad_norm": 132772.1875,
      "learning_rate": 8.945702546981969e-05,
      "loss": 188.221,
      "step": 50
    },
    {
      "epoch": 0.019956096587507483,
      "eval_loss": 56.96770477294922,
      "eval_runtime": 14.8903,
      "eval_samples_per_second": 283.407,
      "eval_steps_per_second": 70.852,
      "step": 50
    },
    {
      "epoch": 0.020355218519257635,
      "grad_norm": 49609.2890625,
      "learning_rate": 8.894386393810563e-05,
      "loss": 285.5734,
      "step": 51
    },
    {
      "epoch": 0.020754340451007783,
      "grad_norm": 48169.328125,
      "learning_rate": 8.842005554284296e-05,
      "loss": 228.8088,
      "step": 52
    },
    {
      "epoch": 0.02115346238275793,
      "grad_norm": 52463.15625,
      "learning_rate": 8.788574348801675e-05,
      "loss": 180.6322,
      "step": 53
    },
    {
      "epoch": 0.021552584314508082,
      "grad_norm": 32927.296875,
      "learning_rate": 8.73410738492077e-05,
      "loss": 95.8114,
      "step": 54
    },
    {
      "epoch": 0.02195170624625823,
      "grad_norm": 42245.27734375,
      "learning_rate": 8.678619553365659e-05,
      "loss": 82.339,
      "step": 55
    },
    {
      "epoch": 0.022350828178008382,
      "grad_norm": 21761.7265625,
      "learning_rate": 8.622126023955446e-05,
      "loss": 81.9877,
      "step": 56
    },
    {
      "epoch": 0.02274995010975853,
      "grad_norm": 28905.279296875,
      "learning_rate": 8.564642241456986e-05,
      "loss": 81.3431,
      "step": 57
    },
    {
      "epoch": 0.02314907204150868,
      "grad_norm": 40905.703125,
      "learning_rate": 8.506183921362443e-05,
      "loss": 100.6945,
      "step": 58
    },
    {
      "epoch": 0.02354819397325883,
      "grad_norm": 54149.60546875,
      "learning_rate": 8.44676704559283e-05,
      "loss": 104.6599,
      "step": 59
    },
    {
      "epoch": 0.02394731590500898,
      "grad_norm": 72751.4140625,
      "learning_rate": 8.386407858128706e-05,
      "loss": 128.9418,
      "step": 60
    },
    {
      "epoch": 0.02434643783675913,
      "grad_norm": 52937.75390625,
      "learning_rate": 8.32512286056924e-05,
      "loss": 125.6734,
      "step": 61
    },
    {
      "epoch": 0.02474555976850928,
      "grad_norm": 165529.0,
      "learning_rate": 8.262928807620843e-05,
      "loss": 152.1745,
      "step": 62
    },
    {
      "epoch": 0.02514468170025943,
      "grad_norm": 43565.19140625,
      "learning_rate": 8.199842702516583e-05,
      "loss": 146.1279,
      "step": 63
    },
    {
      "epoch": 0.02554380363200958,
      "grad_norm": 92479.96875,
      "learning_rate": 8.135881792367686e-05,
      "loss": 157.8514,
      "step": 64
    },
    {
      "epoch": 0.02594292556375973,
      "grad_norm": 75337.59375,
      "learning_rate": 8.07106356344834e-05,
      "loss": 137.6836,
      "step": 65
    },
    {
      "epoch": 0.026342047495509877,
      "grad_norm": 31915.263671875,
      "learning_rate": 8.005405736415126e-05,
      "loss": 140.0353,
      "step": 66
    },
    {
      "epoch": 0.026741169427260028,
      "grad_norm": 79302.203125,
      "learning_rate": 7.938926261462366e-05,
      "loss": 150.151,
      "step": 67
    },
    {
      "epoch": 0.027140291359010176,
      "grad_norm": 58931.72265625,
      "learning_rate": 7.871643313414718e-05,
      "loss": 161.0658,
      "step": 68
    },
    {
      "epoch": 0.027539413290760328,
      "grad_norm": 58412.02734375,
      "learning_rate": 7.803575286758364e-05,
      "loss": 163.3198,
      "step": 69
    },
    {
      "epoch": 0.027938535222510476,
      "grad_norm": 43485.09765625,
      "learning_rate": 7.734740790612136e-05,
      "loss": 141.5732,
      "step": 70
    },
    {
      "epoch": 0.028337657154260627,
      "grad_norm": 51471.87890625,
      "learning_rate": 7.66515864363997e-05,
      "loss": 150.5095,
      "step": 71
    },
    {
      "epoch": 0.028736779086010775,
      "grad_norm": 134108.828125,
      "learning_rate": 7.594847868906076e-05,
      "loss": 195.731,
      "step": 72
    },
    {
      "epoch": 0.029135901017760927,
      "grad_norm": 76437.6796875,
      "learning_rate": 7.52382768867422e-05,
      "loss": 183.1471,
      "step": 73
    },
    {
      "epoch": 0.029535022949511075,
      "grad_norm": 56783.12890625,
      "learning_rate": 7.452117519152542e-05,
      "loss": 148.0741,
      "step": 74
    },
    {
      "epoch": 0.029934144881261227,
      "grad_norm": 54941.4921875,
      "learning_rate": 7.379736965185368e-05,
      "loss": 133.0357,
      "step": 75
    },
    {
      "epoch": 0.030333266813011375,
      "grad_norm": 105390.1796875,
      "learning_rate": 7.30670581489344e-05,
      "loss": 134.0874,
      "step": 76
    },
    {
      "epoch": 0.030732388744761526,
      "grad_norm": 59304.33203125,
      "learning_rate": 7.233044034264034e-05,
      "loss": 108.7234,
      "step": 77
    },
    {
      "epoch": 0.031131510676511674,
      "grad_norm": 61507.59765625,
      "learning_rate": 7.158771761692464e-05,
      "loss": 97.8852,
      "step": 78
    },
    {
      "epoch": 0.031530632608261826,
      "grad_norm": 72501.4609375,
      "learning_rate": 7.083909302476453e-05,
      "loss": 81.9072,
      "step": 79
    },
    {
      "epoch": 0.03192975454001197,
      "grad_norm": 63124.64453125,
      "learning_rate": 7.008477123264848e-05,
      "loss": 87.3968,
      "step": 80
    },
    {
      "epoch": 0.03232887647176212,
      "grad_norm": 88091.9375,
      "learning_rate": 6.932495846462261e-05,
      "loss": 69.6634,
      "step": 81
    },
    {
      "epoch": 0.03272799840351227,
      "grad_norm": 85914.1953125,
      "learning_rate": 6.855986244591104e-05,
      "loss": 105.5995,
      "step": 82
    },
    {
      "epoch": 0.033127120335262425,
      "grad_norm": 87051.7890625,
      "learning_rate": 6.778969234612584e-05,
      "loss": 105.501,
      "step": 83
    },
    {
      "epoch": 0.03352624226701257,
      "grad_norm": 82317.3671875,
      "learning_rate": 6.701465872208216e-05,
      "loss": 116.337,
      "step": 84
    },
    {
      "epoch": 0.03392536419876272,
      "grad_norm": 77026.0703125,
      "learning_rate": 6.623497346023418e-05,
      "loss": 98.7354,
      "step": 85
    },
    {
      "epoch": 0.03432448613051287,
      "grad_norm": 48951.1015625,
      "learning_rate": 6.545084971874738e-05,
      "loss": 98.0408,
      "step": 86
    },
    {
      "epoch": 0.034723608062263024,
      "grad_norm": 51775.6484375,
      "learning_rate": 6.466250186922325e-05,
      "loss": 92.755,
      "step": 87
    },
    {
      "epoch": 0.03512272999401317,
      "grad_norm": 65015.9765625,
      "learning_rate": 6.387014543809223e-05,
      "loss": 81.3799,
      "step": 88
    },
    {
      "epoch": 0.03552185192576332,
      "grad_norm": 58333.90234375,
      "learning_rate": 6.307399704769099e-05,
      "loss": 97.5375,
      "step": 89
    },
    {
      "epoch": 0.03592097385751347,
      "grad_norm": 131453.96875,
      "learning_rate": 6.227427435703997e-05,
      "loss": 100.9756,
      "step": 90
    },
    {
      "epoch": 0.03632009578926362,
      "grad_norm": 122682.8828125,
      "learning_rate": 6.147119600233758e-05,
      "loss": 109.7902,
      "step": 91
    },
    {
      "epoch": 0.03671921772101377,
      "grad_norm": 60680.98828125,
      "learning_rate": 6.066498153718735e-05,
      "loss": 103.3138,
      "step": 92
    },
    {
      "epoch": 0.03711833965276392,
      "grad_norm": 34939.5546875,
      "learning_rate": 5.985585137257401e-05,
      "loss": 96.6016,
      "step": 93
    },
    {
      "epoch": 0.03751746158451407,
      "grad_norm": 236591.03125,
      "learning_rate": 5.90440267166055e-05,
      "loss": 130.7618,
      "step": 94
    },
    {
      "epoch": 0.037916583516264216,
      "grad_norm": 92131.3125,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 137.2894,
      "step": 95
    },
    {
      "epoch": 0.03831570544801437,
      "grad_norm": 28281.841796875,
      "learning_rate": 5.74131823855921e-05,
      "loss": 88.6467,
      "step": 96
    },
    {
      "epoch": 0.03871482737976452,
      "grad_norm": 115048.9765625,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 74.7532,
      "step": 97
    },
    {
      "epoch": 0.03911394931151467,
      "grad_norm": 44177.62109375,
      "learning_rate": 5.577423184847932e-05,
      "loss": 79.8893,
      "step": 98
    },
    {
      "epoch": 0.039513071243264815,
      "grad_norm": 31357.87890625,
      "learning_rate": 5.495227651252315e-05,
      "loss": 73.405,
      "step": 99
    },
    {
      "epoch": 0.039912193175014966,
      "grad_norm": 29447.419921875,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 103.3852,
      "step": 100
    },
    {
      "epoch": 0.039912193175014966,
      "eval_loss": 24.225305557250977,
      "eval_runtime": 14.8796,
      "eval_samples_per_second": 283.61,
      "eval_steps_per_second": 70.902,
      "step": 100
    },
    {
      "epoch": 0.04031131510676512,
      "grad_norm": 15173.3427734375,
      "learning_rate": 5.330452921628497e-05,
      "loss": 93.8994,
      "step": 101
    },
    {
      "epoch": 0.04071043703851527,
      "grad_norm": 19899.49609375,
      "learning_rate": 5.247918773366112e-05,
      "loss": 89.582,
      "step": 102
    },
    {
      "epoch": 0.041109558970265414,
      "grad_norm": 25137.298828125,
      "learning_rate": 5.165316846586541e-05,
      "loss": 72.3924,
      "step": 103
    },
    {
      "epoch": 0.041508680902015566,
      "grad_norm": 86265.0078125,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 58.8551,
      "step": 104
    },
    {
      "epoch": 0.04190780283376572,
      "grad_norm": 6620.14453125,
      "learning_rate": 5e-05,
      "loss": 57.1258,
      "step": 105
    },
    {
      "epoch": 0.04230692476551586,
      "grad_norm": 3978.382568359375,
      "learning_rate": 4.917330276168208e-05,
      "loss": 53.8962,
      "step": 106
    },
    {
      "epoch": 0.04270604669726601,
      "grad_norm": 8527.5732421875,
      "learning_rate": 4.834683153413459e-05,
      "loss": 55.932,
      "step": 107
    },
    {
      "epoch": 0.043105168629016165,
      "grad_norm": 6914.259765625,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 57.8621,
      "step": 108
    },
    {
      "epoch": 0.043504290560766316,
      "grad_norm": 20578.068359375,
      "learning_rate": 4.669547078371504e-05,
      "loss": 62.593,
      "step": 109
    },
    {
      "epoch": 0.04390341249251646,
      "grad_norm": 14920.2265625,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 68.1114,
      "step": 110
    },
    {
      "epoch": 0.04430253442426661,
      "grad_norm": 41930.08984375,
      "learning_rate": 4.504772348747687e-05,
      "loss": 68.2349,
      "step": 111
    },
    {
      "epoch": 0.044701656356016764,
      "grad_norm": 24784.23046875,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 61.8663,
      "step": 112
    },
    {
      "epoch": 0.045100778287766916,
      "grad_norm": 6655.806640625,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 58.0144,
      "step": 113
    },
    {
      "epoch": 0.04549990021951706,
      "grad_norm": 11499.470703125,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 59.2773,
      "step": 114
    },
    {
      "epoch": 0.04589902215126721,
      "grad_norm": 15461.00390625,
      "learning_rate": 4.17702704859633e-05,
      "loss": 62.8515,
      "step": 115
    },
    {
      "epoch": 0.04629814408301736,
      "grad_norm": 18039.513671875,
      "learning_rate": 4.095597328339452e-05,
      "loss": 61.4904,
      "step": 116
    },
    {
      "epoch": 0.046697266014767515,
      "grad_norm": 83189.375,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 76.6094,
      "step": 117
    },
    {
      "epoch": 0.04709638794651766,
      "grad_norm": 9933.810546875,
      "learning_rate": 3.933501846281267e-05,
      "loss": 57.0005,
      "step": 118
    },
    {
      "epoch": 0.04749550987826781,
      "grad_norm": 50326.92578125,
      "learning_rate": 3.852880399766243e-05,
      "loss": 66.0802,
      "step": 119
    },
    {
      "epoch": 0.04789463181001796,
      "grad_norm": 17718.609375,
      "learning_rate": 3.772572564296005e-05,
      "loss": 57.6921,
      "step": 120
    },
    {
      "epoch": 0.04829375374176811,
      "grad_norm": 30671.853515625,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 48.0764,
      "step": 121
    },
    {
      "epoch": 0.04869287567351826,
      "grad_norm": 62980.0859375,
      "learning_rate": 3.612985456190778e-05,
      "loss": 52.5336,
      "step": 122
    },
    {
      "epoch": 0.04909199760526841,
      "grad_norm": 36415.1015625,
      "learning_rate": 3.533749813077677e-05,
      "loss": 57.152,
      "step": 123
    },
    {
      "epoch": 0.04949111953701856,
      "grad_norm": 53034.4296875,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 53.2671,
      "step": 124
    },
    {
      "epoch": 0.049890241468768706,
      "grad_norm": 41480.47265625,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 59.2209,
      "step": 125
    },
    {
      "epoch": 0.05028936340051886,
      "grad_norm": 52756.0390625,
      "learning_rate": 3.298534127791785e-05,
      "loss": 59.0244,
      "step": 126
    },
    {
      "epoch": 0.05068848533226901,
      "grad_norm": 58621.4609375,
      "learning_rate": 3.221030765387417e-05,
      "loss": 62.4222,
      "step": 127
    },
    {
      "epoch": 0.05108760726401916,
      "grad_norm": 34545.16015625,
      "learning_rate": 3.144013755408895e-05,
      "loss": 56.1123,
      "step": 128
    },
    {
      "epoch": 0.051486729195769305,
      "grad_norm": 57221.70703125,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 56.5079,
      "step": 129
    },
    {
      "epoch": 0.05188585112751946,
      "grad_norm": 33589.03515625,
      "learning_rate": 2.991522876735154e-05,
      "loss": 61.1767,
      "step": 130
    },
    {
      "epoch": 0.05228497305926961,
      "grad_norm": 34117.0234375,
      "learning_rate": 2.916090697523549e-05,
      "loss": 52.1322,
      "step": 131
    },
    {
      "epoch": 0.05268409499101975,
      "grad_norm": 50370.43359375,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 58.2877,
      "step": 132
    },
    {
      "epoch": 0.053083216922769905,
      "grad_norm": 56908.3359375,
      "learning_rate": 2.766955965735968e-05,
      "loss": 60.1616,
      "step": 133
    },
    {
      "epoch": 0.053482338854520056,
      "grad_norm": 48866.3984375,
      "learning_rate": 2.693294185106562e-05,
      "loss": 58.2007,
      "step": 134
    },
    {
      "epoch": 0.05388146078627021,
      "grad_norm": 39187.26171875,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 58.7294,
      "step": 135
    },
    {
      "epoch": 0.05428058271802035,
      "grad_norm": 26180.66015625,
      "learning_rate": 2.547882480847461e-05,
      "loss": 67.5557,
      "step": 136
    },
    {
      "epoch": 0.054679704649770504,
      "grad_norm": 213279.390625,
      "learning_rate": 2.476172311325783e-05,
      "loss": 69.8095,
      "step": 137
    },
    {
      "epoch": 0.055078826581520655,
      "grad_norm": 72242.5625,
      "learning_rate": 2.405152131093926e-05,
      "loss": 75.8229,
      "step": 138
    },
    {
      "epoch": 0.05547794851327081,
      "grad_norm": 69482.5546875,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 63.0787,
      "step": 139
    },
    {
      "epoch": 0.05587707044502095,
      "grad_norm": 62033.5390625,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 58.8768,
      "step": 140
    },
    {
      "epoch": 0.0562761923767711,
      "grad_norm": 29750.419921875,
      "learning_rate": 2.196424713241637e-05,
      "loss": 72.2296,
      "step": 141
    },
    {
      "epoch": 0.056675314308521255,
      "grad_norm": 49227.828125,
      "learning_rate": 2.128356686585282e-05,
      "loss": 77.8471,
      "step": 142
    },
    {
      "epoch": 0.057074436240271406,
      "grad_norm": 88195.8671875,
      "learning_rate": 2.061073738537635e-05,
      "loss": 65.1452,
      "step": 143
    },
    {
      "epoch": 0.05747355817202155,
      "grad_norm": 70083.7421875,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 51.9114,
      "step": 144
    },
    {
      "epoch": 0.0578726801037717,
      "grad_norm": 32487.505859375,
      "learning_rate": 1.928936436551661e-05,
      "loss": 55.1692,
      "step": 145
    },
    {
      "epoch": 0.058271802035521854,
      "grad_norm": 46871.6484375,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 51.7074,
      "step": 146
    },
    {
      "epoch": 0.058670923967272,
      "grad_norm": 60043.66015625,
      "learning_rate": 1.800157297483417e-05,
      "loss": 61.5072,
      "step": 147
    },
    {
      "epoch": 0.05907004589902215,
      "grad_norm": 52521.32421875,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 57.5619,
      "step": 148
    },
    {
      "epoch": 0.0594691678307723,
      "grad_norm": 50972.00390625,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 61.5767,
      "step": 149
    },
    {
      "epoch": 0.05986828976252245,
      "grad_norm": 49810.4375,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 80.1615,
      "step": 150
    },
    {
      "epoch": 0.05986828976252245,
      "eval_loss": 14.149837493896484,
      "eval_runtime": 14.8445,
      "eval_samples_per_second": 284.281,
      "eval_steps_per_second": 71.07,
      "step": 150
    },
    {
      "epoch": 0.0602674116942726,
      "grad_norm": 5319.2880859375,
      "learning_rate": 1.553232954407171e-05,
      "loss": 47.2432,
      "step": 151
    },
    {
      "epoch": 0.06066653362602275,
      "grad_norm": 9941.001953125,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 49.9763,
      "step": 152
    },
    {
      "epoch": 0.0610656555577729,
      "grad_norm": 9290.3984375,
      "learning_rate": 1.435357758543015e-05,
      "loss": 49.6699,
      "step": 153
    },
    {
      "epoch": 0.06146477748952305,
      "grad_norm": 13493.3046875,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 48.8268,
      "step": 154
    },
    {
      "epoch": 0.0618638994212732,
      "grad_norm": 12148.2978515625,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 48.4094,
      "step": 155
    },
    {
      "epoch": 0.06226302135302335,
      "grad_norm": 8232.302734375,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 47.4758,
      "step": 156
    },
    {
      "epoch": 0.0626621432847735,
      "grad_norm": 7785.50048828125,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 45.3292,
      "step": 157
    },
    {
      "epoch": 0.06306126521652365,
      "grad_norm": 15800.1875,
      "learning_rate": 1.157994445715706e-05,
      "loss": 51.3398,
      "step": 158
    },
    {
      "epoch": 0.0634603871482738,
      "grad_norm": 14390.69921875,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 46.0982,
      "step": 159
    },
    {
      "epoch": 0.06385950908002394,
      "grad_norm": 11618.111328125,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 46.2737,
      "step": 160
    },
    {
      "epoch": 0.06425863101177409,
      "grad_norm": 11376.5029296875,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 47.809,
      "step": 161
    },
    {
      "epoch": 0.06465775294352424,
      "grad_norm": 23942.490234375,
      "learning_rate": 9.549150281252633e-06,
      "loss": 49.5283,
      "step": 162
    },
    {
      "epoch": 0.0650568748752744,
      "grad_norm": 23093.396484375,
      "learning_rate": 9.068759265665384e-06,
      "loss": 45.9041,
      "step": 163
    },
    {
      "epoch": 0.06545599680702455,
      "grad_norm": 26425.5546875,
      "learning_rate": 8.599558442598998e-06,
      "loss": 49.9965,
      "step": 164
    },
    {
      "epoch": 0.0658551187387747,
      "grad_norm": 39641.6171875,
      "learning_rate": 8.141676086873572e-06,
      "loss": 47.1454,
      "step": 165
    },
    {
      "epoch": 0.06625424067052485,
      "grad_norm": 30851.595703125,
      "learning_rate": 7.695237378953223e-06,
      "loss": 50.2823,
      "step": 166
    },
    {
      "epoch": 0.066653362602275,
      "grad_norm": 33112.140625,
      "learning_rate": 7.260364370723044e-06,
      "loss": 51.5328,
      "step": 167
    },
    {
      "epoch": 0.06705248453402514,
      "grad_norm": 25474.642578125,
      "learning_rate": 6.837175952121306e-06,
      "loss": 50.6706,
      "step": 168
    },
    {
      "epoch": 0.06745160646577529,
      "grad_norm": 18275.091796875,
      "learning_rate": 6.425787818636131e-06,
      "loss": 52.5591,
      "step": 169
    },
    {
      "epoch": 0.06785072839752544,
      "grad_norm": 24002.634765625,
      "learning_rate": 6.026312439675552e-06,
      "loss": 56.9931,
      "step": 170
    },
    {
      "epoch": 0.0682498503292756,
      "grad_norm": 23333.41015625,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 54.5407,
      "step": 171
    },
    {
      "epoch": 0.06864897226102575,
      "grad_norm": 36934.09765625,
      "learning_rate": 5.263533508961827e-06,
      "loss": 48.5398,
      "step": 172
    },
    {
      "epoch": 0.0690480941927759,
      "grad_norm": 22509.69140625,
      "learning_rate": 4.900438493352055e-06,
      "loss": 55.8961,
      "step": 173
    },
    {
      "epoch": 0.06944721612452605,
      "grad_norm": 18083.5859375,
      "learning_rate": 4.549673247541875e-06,
      "loss": 49.2919,
      "step": 174
    },
    {
      "epoch": 0.06984633805627619,
      "grad_norm": 37785.8984375,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 58.4751,
      "step": 175
    },
    {
      "epoch": 0.07024545998802634,
      "grad_norm": 37978.54296875,
      "learning_rate": 3.885512251130763e-06,
      "loss": 57.1662,
      "step": 176
    },
    {
      "epoch": 0.07064458191977649,
      "grad_norm": 42680.03515625,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 53.9309,
      "step": 177
    },
    {
      "epoch": 0.07104370385152664,
      "grad_norm": 37030.38671875,
      "learning_rate": 3.271776770026963e-06,
      "loss": 56.6007,
      "step": 178
    },
    {
      "epoch": 0.07144282578327679,
      "grad_norm": 54973.8671875,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 54.804,
      "step": 179
    },
    {
      "epoch": 0.07184194771502694,
      "grad_norm": 16794.76171875,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 54.2363,
      "step": 180
    },
    {
      "epoch": 0.0722410696467771,
      "grad_norm": 400968.1875,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 73.1373,
      "step": 181
    },
    {
      "epoch": 0.07264019157852725,
      "grad_norm": 24410.708984375,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 58.0236,
      "step": 182
    },
    {
      "epoch": 0.07303931351027738,
      "grad_norm": 32319.859375,
      "learning_rate": 1.962316193157593e-06,
      "loss": 58.926,
      "step": 183
    },
    {
      "epoch": 0.07343843544202754,
      "grad_norm": 71104.375,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 58.6824,
      "step": 184
    },
    {
      "epoch": 0.07383755737377769,
      "grad_norm": 43221.58984375,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 63.3094,
      "step": 185
    },
    {
      "epoch": 0.07423667930552784,
      "grad_norm": 32009.09375,
      "learning_rate": 1.333670137599713e-06,
      "loss": 57.5569,
      "step": 186
    },
    {
      "epoch": 0.07463580123727799,
      "grad_norm": 37873.08984375,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 58.1378,
      "step": 187
    },
    {
      "epoch": 0.07503492316902814,
      "grad_norm": 43808.8828125,
      "learning_rate": 9.810017062595322e-07,
      "loss": 59.5114,
      "step": 188
    },
    {
      "epoch": 0.0754340451007783,
      "grad_norm": 38293.0703125,
      "learning_rate": 8.247462563808817e-07,
      "loss": 65.9775,
      "step": 189
    },
    {
      "epoch": 0.07583316703252843,
      "grad_norm": 71795.5546875,
      "learning_rate": 6.819348298638839e-07,
      "loss": 66.5328,
      "step": 190
    },
    {
      "epoch": 0.07623228896427858,
      "grad_norm": 56646.703125,
      "learning_rate": 5.526064699265753e-07,
      "loss": 57.6959,
      "step": 191
    },
    {
      "epoch": 0.07663141089602873,
      "grad_norm": 47954.93359375,
      "learning_rate": 4.367965336512403e-07,
      "loss": 59.0449,
      "step": 192
    },
    {
      "epoch": 0.07703053282777889,
      "grad_norm": 70545.4765625,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 53.302,
      "step": 193
    },
    {
      "epoch": 0.07742965475952904,
      "grad_norm": 31944.169921875,
      "learning_rate": 2.458548727494292e-07,
      "loss": 54.1979,
      "step": 194
    },
    {
      "epoch": 0.07782877669127919,
      "grad_norm": 90626.34375,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 60.9848,
      "step": 195
    },
    {
      "epoch": 0.07822789862302934,
      "grad_norm": 28368.2890625,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 56.897,
      "step": 196
    },
    {
      "epoch": 0.07862702055477949,
      "grad_norm": 40957.1328125,
      "learning_rate": 6.150154258476315e-08,
      "loss": 70.1486,
      "step": 197
    },
    {
      "epoch": 0.07902614248652963,
      "grad_norm": 46892.70703125,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 65.6722,
      "step": 198
    },
    {
      "epoch": 0.07942526441827978,
      "grad_norm": 100613.1484375,
      "learning_rate": 6.834750376549792e-09,
      "loss": 67.2353,
      "step": 199
    },
    {
      "epoch": 0.07982438635002993,
      "grad_norm": 39987.96875,
      "learning_rate": 0.0,
      "loss": 82.7899,
      "step": 200
    },
    {
      "epoch": 0.07982438635002993,
      "eval_loss": 13.871726989746094,
      "eval_runtime": 14.93,
      "eval_samples_per_second": 282.652,
      "eval_steps_per_second": 70.663,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1887280167911424.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}