{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9999948732934474,
  "global_step": 292584,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9914725343832886e-05,
      "loss": 3.5525,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.982927979657124e-05,
      "loss": 3.4523,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.97438342493096e-05,
      "loss": 3.4223,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.965838870204796e-05,
      "loss": 3.393,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9572943154786325e-05,
      "loss": 3.3943,
      "step": 2500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.948749760752468e-05,
      "loss": 3.3711,
      "step": 3000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9402052060263034e-05,
      "loss": 3.3417,
      "step": 3500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9316606513001396e-05,
      "loss": 3.3459,
      "step": 4000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.923116096573975e-05,
      "loss": 3.3445,
      "step": 4500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.914571541847812e-05,
      "loss": 3.336,
      "step": 5000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9060269871216474e-05,
      "loss": 3.3265,
      "step": 5500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.8974824323954835e-05,
      "loss": 3.3165,
      "step": 6000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.888972055888224e-05,
      "loss": 3.3303,
      "step": 6500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8804275011620595e-05,
      "loss": 3.3164,
      "step": 7000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.871882946435896e-05,
      "loss": 3.3109,
      "step": 7500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.863338391709731e-05,
      "loss": 3.3003,
      "step": 8000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.854793836983567e-05,
      "loss": 3.2919,
      "step": 8500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8462492822574034e-05,
      "loss": 3.2962,
      "step": 9000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8377047275312396e-05,
      "loss": 3.286,
      "step": 9500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.829160172805075e-05,
      "loss": 3.2937,
      "step": 10000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8206156180789105e-05,
      "loss": 3.2646,
      "step": 10500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.812088152462199e-05,
      "loss": 3.2726,
      "step": 11000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.803543597736035e-05,
      "loss": 3.2714,
      "step": 11500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.7949990430098705e-05,
      "loss": 3.2637,
      "step": 12000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.786454488283707e-05,
      "loss": 3.2654,
      "step": 12500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.777927022666995e-05,
      "loss": 3.2713,
      "step": 13000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769382467940831e-05,
      "loss": 3.2532,
      "step": 13500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7608379132146666e-05,
      "loss": 3.2651,
      "step": 14000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752293358488503e-05,
      "loss": 3.2504,
      "step": 14500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.743748803762338e-05,
      "loss": 3.2566,
      "step": 15000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.7352042490361744e-05,
      "loss": 3.2514,
      "step": 15500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.7266596943100105e-05,
      "loss": 3.2398,
      "step": 16000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.718115139583846e-05,
      "loss": 3.2395,
      "step": 16500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.709570584857682e-05,
      "loss": 3.2432,
      "step": 17000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.7010431192409705e-05,
      "loss": 3.2395,
      "step": 17500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.692498564514806e-05,
      "loss": 3.2361,
      "step": 18000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.683971098898094e-05,
      "loss": 3.2326,
      "step": 18500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.675443633281383e-05,
      "loss": 3.2229,
      "step": 19000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.666899078555219e-05,
      "loss": 3.2272,
      "step": 19500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.658354523829054e-05,
      "loss": 3.2294,
      "step": 20000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6498099691028904e-05,
      "loss": 3.224,
      "step": 20500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.6412654143767266e-05,
      "loss": 3.2235,
      "step": 21000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.632720859650562e-05,
      "loss": 3.2235,
      "step": 21500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.624176304924398e-05,
      "loss": 3.2216,
      "step": 22000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.615631750198234e-05,
      "loss": 3.2249,
      "step": 22500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.607104284581522e-05,
      "loss": 3.2184,
      "step": 23000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.598559729855358e-05,
      "loss": 3.2069,
      "step": 23500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5900151751291936e-05,
      "loss": 3.2169,
      "step": 24000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.58147062040303e-05,
      "loss": 3.2062,
      "step": 24500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.572926065676866e-05,
      "loss": 3.2148,
      "step": 25000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5643815109507014e-05,
      "loss": 3.214,
      "step": 25500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.55585404533399e-05,
      "loss": 3.2113,
      "step": 26000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.547309490607826e-05,
      "loss": 3.2076,
      "step": 26500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5387649358816614e-05,
      "loss": 3.1962,
      "step": 27000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5302203811554975e-05,
      "loss": 3.2037,
      "step": 27500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.521675826429334e-05,
      "loss": 3.198,
      "step": 28000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.513131271703169e-05,
      "loss": 3.1867,
      "step": 28500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5046038060864575e-05,
      "loss": 3.188,
      "step": 29000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.4960592513602936e-05,
      "loss": 3.1905,
      "step": 29500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.487514696634129e-05,
      "loss": 3.1942,
      "step": 30000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.478970141907965e-05,
      "loss": 3.1952,
      "step": 30500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4704255871818014e-05,
      "loss": 3.1954,
      "step": 31000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.461881032455637e-05,
      "loss": 3.1869,
      "step": 31500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.453336477729473e-05,
      "loss": 3.1834,
      "step": 32000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4448090121127614e-05,
      "loss": 3.1838,
      "step": 32500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.436264457386597e-05,
      "loss": 3.1838,
      "step": 33000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.427719902660433e-05,
      "loss": 3.1886,
      "step": 33500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4191753479342685e-05,
      "loss": 3.177,
      "step": 34000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.410647882317557e-05,
      "loss": 3.1853,
      "step": 34500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.402103327591392e-05,
      "loss": 3.1732,
      "step": 35000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.393558772865229e-05,
      "loss": 3.1898,
      "step": 35500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3850142181390646e-05,
      "loss": 3.1705,
      "step": 36000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.376469663412901e-05,
      "loss": 3.1822,
      "step": 36500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.367925108686736e-05,
      "loss": 3.1821,
      "step": 37000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.359380553960572e-05,
      "loss": 3.1656,
      "step": 37500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3508359992344085e-05,
      "loss": 3.1706,
      "step": 38000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.342308533617696e-05,
      "loss": 3.1748,
      "step": 38500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3337810680009845e-05,
      "loss": 3.1566,
      "step": 39000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.32523651327482e-05,
      "loss": 3.1627,
      "step": 39500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.316691958548657e-05,
      "loss": 3.1665,
      "step": 40000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.3081644929319445e-05,
      "loss": 3.1663,
      "step": 40500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.2996199382057806e-05,
      "loss": 3.1609,
      "step": 41000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.291075383479616e-05,
      "loss": 3.1545,
      "step": 41500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.282530828753452e-05,
      "loss": 3.1663,
      "step": 42000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.273986274027288e-05,
      "loss": 3.168,
      "step": 42500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2654417193011245e-05,
      "loss": 3.1615,
      "step": 43000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.25689716457496e-05,
      "loss": 3.1474,
      "step": 43500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.2483526098487955e-05,
      "loss": 3.1532,
      "step": 44000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2398080551226316e-05,
      "loss": 3.1674,
      "step": 44500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.231263500396467e-05,
      "loss": 3.1692,
      "step": 45000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.222718945670304e-05,
      "loss": 3.1621,
      "step": 45500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2141743909441394e-05,
      "loss": 3.1624,
      "step": 46000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.205646925327428e-05,
      "loss": 3.1554,
      "step": 46500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.197102370601263e-05,
      "loss": 3.1669,
      "step": 47000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.1885578158750994e-05,
      "loss": 3.1535,
      "step": 47500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.180013261148935e-05,
      "loss": 3.1582,
      "step": 48000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.171468706422771e-05,
      "loss": 3.1469,
      "step": 48500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.162941240806059e-05,
      "loss": 3.15,
      "step": 49000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.154396686079895e-05,
      "loss": 3.1568,
      "step": 49500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1458521313537316e-05,
      "loss": 3.1583,
      "step": 50000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.137324665737019e-05,
      "loss": 3.1547,
      "step": 50500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1287801110108555e-05,
      "loss": 3.1383,
      "step": 51000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.120235556284691e-05,
      "loss": 3.1239,
      "step": 51500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.111691001558527e-05,
      "loss": 3.1581,
      "step": 52000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.1031464468323625e-05,
      "loss": 3.1414,
      "step": 52500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.094601892106199e-05,
      "loss": 3.128,
      "step": 53000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.086057337380035e-05,
      "loss": 3.1435,
      "step": 53500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.07751278265387e-05,
      "loss": 3.1491,
      "step": 54000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0689682279277064e-05,
      "loss": 3.1431,
      "step": 54500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.060423673201542e-05,
      "loss": 3.1397,
      "step": 55000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.051879118475378e-05,
      "loss": 3.1366,
      "step": 55500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.043334563749214e-05,
      "loss": 3.1329,
      "step": 56000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.0347900090230504e-05,
      "loss": 3.1343,
      "step": 56500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.026262543406338e-05,
      "loss": 3.1444,
      "step": 57000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0177350777896264e-05,
      "loss": 3.1396,
      "step": 57500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0091905230634625e-05,
      "loss": 3.1388,
      "step": 58000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.000645968337298e-05,
      "loss": 3.135,
      "step": 58500
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.992101413611134e-05,
      "loss": 3.1425,
      "step": 59000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9835568588849696e-05,
      "loss": 3.1412,
      "step": 59500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.975012304158806e-05,
      "loss": 3.1293,
      "step": 60000
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.966467749432642e-05,
      "loss": 3.1421,
      "step": 60500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.95794028381593e-05,
      "loss": 3.1438,
      "step": 61000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.949395729089766e-05,
      "loss": 3.1413,
      "step": 61500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.940851174363602e-05,
      "loss": 3.1393,
      "step": 62000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9323066196374374e-05,
      "loss": 3.1355,
      "step": 62500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.9237620649112735e-05,
      "loss": 3.1297,
      "step": 63000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.9152175101851097e-05,
      "loss": 3.1345,
      "step": 63500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.906672955458945e-05,
      "loss": 3.1229,
      "step": 64000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.898128400732781e-05,
      "loss": 3.131,
      "step": 64500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8896009351160696e-05,
      "loss": 3.1207,
      "step": 65000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.881056380389905e-05,
      "loss": 3.1154,
      "step": 65500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.872511825663741e-05,
      "loss": 3.129,
      "step": 66000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.863967270937577e-05,
      "loss": 3.1382,
      "step": 66500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.855422716211413e-05,
      "loss": 3.1161,
      "step": 67000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.846895250594701e-05,
      "loss": 3.1299,
      "step": 67500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.838367784977989e-05,
      "loss": 3.1149,
      "step": 68000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.829823230251826e-05,
      "loss": 3.1248,
      "step": 68500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.821278675525661e-05,
      "loss": 3.1218,
      "step": 69000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.812734120799497e-05,
      "loss": 3.1164,
      "step": 69500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.804189566073333e-05,
      "loss": 3.1235,
      "step": 70000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.795645011347169e-05,
      "loss": 3.1156,
      "step": 70500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.787100456621005e-05,
      "loss": 3.1222,
      "step": 71000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7785559018948406e-05,
      "loss": 3.1187,
      "step": 71500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.770011347168677e-05,
      "loss": 3.1255,
      "step": 72000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.761466792442512e-05,
      "loss": 3.1188,
      "step": 72500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7529393268258005e-05,
      "loss": 3.1261,
      "step": 73000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.744394772099637e-05,
      "loss": 3.1207,
      "step": 73500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.735850217373472e-05,
      "loss": 3.1123,
      "step": 74000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.727305662647308e-05,
      "loss": 3.1136,
      "step": 74500
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7187611079211444e-05,
      "loss": 3.1299,
      "step": 75000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.710233642304433e-05,
      "loss": 3.1131,
      "step": 75500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.701689087578268e-05,
      "loss": 3.1225,
      "step": 76000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6931445328521044e-05,
      "loss": 3.1117,
      "step": 76500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.68459997812594e-05,
      "loss": 3.0971,
      "step": 77000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.676072512509228e-05,
      "loss": 3.1096,
      "step": 77500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.667527957783064e-05,
      "loss": 3.11,
      "step": 78000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6589834030569005e-05,
      "loss": 3.0992,
      "step": 78500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.650438848330736e-05,
      "loss": 3.1196,
      "step": 79000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6419113827140244e-05,
      "loss": 3.1152,
      "step": 79500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6333668279878605e-05,
      "loss": 3.1123,
      "step": 80000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.624822273261696e-05,
      "loss": 3.1149,
      "step": 80500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6162777185355314e-05,
      "loss": 3.1118,
      "step": 81000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.6077331638093676e-05,
      "loss": 3.0959,
      "step": 81500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.599188609083204e-05,
      "loss": 3.1101,
      "step": 82000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.59064405435704e-05,
      "loss": 3.0982,
      "step": 82500
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5820994996308753e-05,
      "loss": 3.1139,
      "step": 83000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.573572034014164e-05,
      "loss": 3.1021,
      "step": 83500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.565027479288e-05,
      "loss": 3.1286,
      "step": 84000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.556482924561835e-05,
      "loss": 3.0923,
      "step": 84500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.547938369835671e-05,
      "loss": 3.1152,
      "step": 85000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5393938151095076e-05,
      "loss": 3.1174,
      "step": 85500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.530866349492795e-05,
      "loss": 3.0943,
      "step": 86000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5223217947666314e-05,
      "loss": 3.1051,
      "step": 86500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5137772400404676e-05,
      "loss": 3.0888,
      "step": 87000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.505232685314303e-05,
      "loss": 3.1071,
      "step": 87500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.4966881305881385e-05,
      "loss": 3.1063,
      "step": 88000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.488143575861975e-05,
      "loss": 3.1077,
      "step": 88500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.479599021135811e-05,
      "loss": 3.0985,
      "step": 89000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.471054466409647e-05,
      "loss": 3.0979,
      "step": 89500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.462527000792935e-05,
      "loss": 3.0953,
      "step": 90000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.453982446066771e-05,
      "loss": 3.0865,
      "step": 90500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.445437891340606e-05,
      "loss": 3.0956,
      "step": 91000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4368933366144424e-05,
      "loss": 3.103,
      "step": 91500
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.428348781888278e-05,
      "loss": 3.0955,
      "step": 92000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.419804227162115e-05,
      "loss": 3.095,
      "step": 92500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.41125967243595e-05,
      "loss": 3.0928,
      "step": 93000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.402715117709786e-05,
      "loss": 3.0907,
      "step": 93500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.394187652093074e-05,
      "loss": 3.0863,
      "step": 94000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.38564309736691e-05,
      "loss": 3.0914,
      "step": 94500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.3770985426407456e-05,
      "loss": 3.0989,
      "step": 95000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.3685539879145824e-05,
      "loss": 3.098,
      "step": 95500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.36002652229787e-05,
      "loss": 3.1002,
      "step": 96000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.351481967571706e-05,
      "loss": 3.0808,
      "step": 96500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.342954501954994e-05,
      "loss": 3.092,
      "step": 97000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.33440994722883e-05,
      "loss": 3.0876,
      "step": 97500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.325865392502666e-05,
      "loss": 3.0774,
      "step": 98000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.317320837776502e-05,
      "loss": 3.0681,
      "step": 98500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.308776283050338e-05,
      "loss": 3.0655,
      "step": 99000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.300231728324173e-05,
      "loss": 3.0572,
      "step": 99500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.29168717359801e-05,
      "loss": 3.0677,
      "step": 100000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.283159707981298e-05,
      "loss": 3.0807,
      "step": 100500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.274615153255134e-05,
      "loss": 3.0616,
      "step": 101000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.2660705985289694e-05,
      "loss": 3.0715,
      "step": 101500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2575260438028056e-05,
      "loss": 3.0786,
      "step": 102000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.248981489076641e-05,
      "loss": 3.0762,
      "step": 102500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.2404540234599294e-05,
      "loss": 3.0679,
      "step": 103000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.2319094687337655e-05,
      "loss": 3.0596,
      "step": 103500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.223364914007602e-05,
      "loss": 3.0598,
      "step": 104000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.214820359281437e-05,
      "loss": 3.0626,
      "step": 104500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.206275804555273e-05,
      "loss": 3.0616,
      "step": 105000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.197731249829109e-05,
      "loss": 3.072,
      "step": 105500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.189186695102945e-05,
      "loss": 3.0721,
      "step": 106000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.180642140376781e-05,
      "loss": 3.0707,
      "step": 106500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.172114674760069e-05,
      "loss": 3.0553,
      "step": 107000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.163570120033905e-05,
      "loss": 3.0615,
      "step": 107500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.155025565307741e-05,
      "loss": 3.0536,
      "step": 108000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.1464810105815765e-05,
      "loss": 3.0731,
      "step": 108500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.137953544964865e-05,
      "loss": 3.0594,
      "step": 109000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.129408990238701e-05,
      "loss": 3.0576,
      "step": 109500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.1208644355125365e-05,
      "loss": 3.0633,
      "step": 110000
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.1123198807863726e-05,
      "loss": 3.0699,
      "step": 110500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.103775326060209e-05,
      "loss": 3.0633,
      "step": 111000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.095230771334044e-05,
      "loss": 3.0682,
      "step": 111500
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.0866862166078804e-05,
      "loss": 3.066,
      "step": 112000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.078141661881716e-05,
      "loss": 3.055,
      "step": 112500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.069614196265004e-05,
      "loss": 3.0603,
      "step": 113000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.0610696415388404e-05,
      "loss": 3.0653,
      "step": 113500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.052525086812676e-05,
      "loss": 3.0617,
      "step": 114000
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0439976211959642e-05,
      "loss": 3.065,
      "step": 114500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0354530664698007e-05,
      "loss": 3.0733,
      "step": 115000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0269085117436365e-05,
      "loss": 3.0707,
      "step": 115500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.018363957017472e-05,
      "loss": 3.0643,
      "step": 116000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0098194022913078e-05,
      "loss": 3.0527,
      "step": 116500
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.0012748475651436e-05,
      "loss": 3.0608,
      "step": 117000
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.992747381948432e-05,
      "loss": 3.0666,
      "step": 117500
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9842028272222677e-05,
      "loss": 3.0607,
      "step": 118000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9756582724961042e-05,
      "loss": 3.0624,
      "step": 118500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9671137177699397e-05,
      "loss": 3.0688,
      "step": 119000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9585691630437755e-05,
      "loss": 3.0639,
      "step": 119500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9500246083176113e-05,
      "loss": 3.0542,
      "step": 120000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9414971427008997e-05,
      "loss": 3.0586,
      "step": 120500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9329525879747355e-05,
      "loss": 3.049,
      "step": 121000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9244080332485713e-05,
      "loss": 3.0606,
      "step": 121500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9158634785224078e-05,
      "loss": 3.0571,
      "step": 122000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.9073189237962432e-05,
      "loss": 3.0629,
      "step": 122500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.898774369070079e-05,
      "loss": 3.0634,
      "step": 123000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8902469034533674e-05,
      "loss": 3.0511,
      "step": 123500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8817023487272032e-05,
      "loss": 3.0574,
      "step": 124000
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.873157794001039e-05,
      "loss": 3.061,
      "step": 124500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8646132392748748e-05,
      "loss": 3.0489,
      "step": 125000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.856068684548711e-05,
      "loss": 3.0578,
      "step": 125500
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.847541218931999e-05,
      "loss": 3.063,
      "step": 126000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.838996664205835e-05,
      "loss": 3.0652,
      "step": 126500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.830452109479671e-05,
      "loss": 3.0499,
      "step": 127000
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8219075547535067e-05,
      "loss": 3.0534,
      "step": 127500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8133630000273425e-05,
      "loss": 3.0795,
      "step": 128000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.804818445301179e-05,
      "loss": 3.0544,
      "step": 128500
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.7962738905750145e-05,
      "loss": 3.0582,
      "step": 129000
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7877293358488503e-05,
      "loss": 3.0497,
      "step": 129500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.779184781122686e-05,
      "loss": 3.0507,
      "step": 130000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.770640226396522e-05,
      "loss": 3.0572,
      "step": 130500
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7621127607798103e-05,
      "loss": 3.0776,
      "step": 131000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.753568206053646e-05,
      "loss": 3.0371,
      "step": 131500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7450236513274822e-05,
      "loss": 3.0666,
      "step": 132000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7364961857107703e-05,
      "loss": 3.0543,
      "step": 132500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7279516309846064e-05,
      "loss": 3.0533,
      "step": 133000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7194070762584422e-05,
      "loss": 3.051,
      "step": 133500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.710862521532278e-05,
      "loss": 3.0591,
      "step": 134000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.7023179668061138e-05,
      "loss": 3.037,
      "step": 134500
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6937734120799496e-05,
      "loss": 3.0503,
      "step": 135000
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6852288573537858e-05,
      "loss": 3.0466,
      "step": 135500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6766843026276216e-05,
      "loss": 3.0635,
      "step": 136000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6681397479014574e-05,
      "loss": 3.0433,
      "step": 136500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6595951931752932e-05,
      "loss": 3.0365,
      "step": 137000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6510677275585816e-05,
      "loss": 3.046,
      "step": 137500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6425231728324174e-05,
      "loss": 3.0267,
      "step": 138000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.6339786181062532e-05,
      "loss": 3.0468,
      "step": 138500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6254340633800893e-05,
      "loss": 3.0376,
      "step": 139000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.616889508653925e-05,
      "loss": 3.0292,
      "step": 139500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6083620430372135e-05,
      "loss": 3.0482,
      "step": 140000
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.5998174883110493e-05,
      "loss": 3.0303,
      "step": 140500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.591272933584885e-05,
      "loss": 3.0571,
      "step": 141000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.582728378858721e-05,
      "loss": 3.0465,
      "step": 141500
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5741838241325567e-05,
      "loss": 3.0387,
      "step": 142000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.565639269406393e-05,
      "loss": 3.0624,
      "step": 142500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5571118037896812e-05,
      "loss": 3.0421,
      "step": 143000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.548567249063517e-05,
      "loss": 3.0486,
      "step": 143500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.540022694337353e-05,
      "loss": 3.0479,
      "step": 144000
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5314781396111886e-05,
      "loss": 3.055,
      "step": 144500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.522950673994477e-05,
      "loss": 3.0396,
      "step": 145000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5144061192683128e-05,
      "loss": 3.0608,
      "step": 145500
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.505878653651601e-05,
      "loss": 3.0466,
      "step": 146000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.4973340989254366e-05,
      "loss": 3.0266,
      "step": 146500
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4887895441992728e-05,
      "loss": 3.0399,
      "step": 147000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4802449894731086e-05,
      "loss": 3.044,
      "step": 147500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4717004347469447e-05,
      "loss": 3.038,
      "step": 148000
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4631558800207805e-05,
      "loss": 3.0358,
      "step": 148500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4546113252946163e-05,
      "loss": 3.042,
      "step": 149000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.446066770568452e-05,
      "loss": 3.0303,
      "step": 149500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4375393049517405e-05,
      "loss": 3.0556,
      "step": 150000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4289947502255763e-05,
      "loss": 3.0424,
      "step": 150500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.420450195499412e-05,
      "loss": 3.038,
      "step": 151000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4119056407732483e-05,
      "loss": 3.0476,
      "step": 151500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.403361086047084e-05,
      "loss": 3.048,
      "step": 152000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3948507095398246e-05,
      "loss": 3.0392,
      "step": 152500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3863061548136605e-05,
      "loss": 3.0379,
      "step": 153000
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3777616000874963e-05,
      "loss": 3.046,
      "step": 153500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.369217045361332e-05,
      "loss": 3.0432,
      "step": 154000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3606724906351682e-05,
      "loss": 3.0436,
      "step": 154500
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.352127935909004e-05,
      "loss": 3.0397,
      "step": 155000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3435833811828402e-05,
      "loss": 3.0425,
      "step": 155500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.335038826456676e-05,
      "loss": 3.0454,
      "step": 156000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3264942717305114e-05,
      "loss": 3.0313,
      "step": 156500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3179497170043476e-05,
      "loss": 3.0229,
      "step": 157000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3094051622781834e-05,
      "loss": 3.0417,
      "step": 157500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3008606075520195e-05,
      "loss": 3.0428,
      "step": 158000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2923331419353076e-05,
      "loss": 3.0354,
      "step": 158500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2837885872091437e-05,
      "loss": 3.0351,
      "step": 159000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2752440324829792e-05,
      "loss": 3.0383,
      "step": 159500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2667165668662675e-05,
      "loss": 3.0325,
      "step": 160000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2581720121401033e-05,
      "loss": 3.0345,
      "step": 160500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2496274574139395e-05,
      "loss": 3.0255,
      "step": 161000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2410829026877753e-05,
      "loss": 3.0434,
      "step": 161500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.232538347961611e-05,
      "loss": 3.0221,
      "step": 162000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2239937932354473e-05,
      "loss": 3.0277,
      "step": 162500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2154492385092827e-05,
      "loss": 3.0253,
      "step": 163000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.206904683783119e-05,
      "loss": 3.0294,
      "step": 163500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.198377218166407e-05,
      "loss": 3.0334,
      "step": 164000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.189832663440243e-05,
      "loss": 3.034,
      "step": 164500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.181288108714079e-05,
      "loss": 3.0383,
      "step": 165000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1727435539879146e-05,
      "loss": 3.0337,
      "step": 165500
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1641989992617508e-05,
      "loss": 3.0201,
      "step": 166000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1556544445355863e-05,
      "loss": 3.0331,
      "step": 166500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1471098898094224e-05,
      "loss": 3.0212,
      "step": 167000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1385653350832582e-05,
      "loss": 3.0303,
      "step": 167500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.130020780357094e-05,
      "loss": 3.0309,
      "step": 168000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1214933147403824e-05,
      "loss": 3.0221,
      "step": 168500
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1129487600142185e-05,
      "loss": 3.0169,
      "step": 169000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.104404205288054e-05,
      "loss": 3.0293,
      "step": 169500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.0958596505618898e-05,
      "loss": 3.0363,
      "step": 170000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.087332184945178e-05,
      "loss": 3.0129,
      "step": 170500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.078787630219014e-05,
      "loss": 3.0359,
      "step": 171000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.07024307549285e-05,
      "loss": 3.0324,
      "step": 171500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.061698520766686e-05,
      "loss": 3.0219,
      "step": 172000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0531710551499743e-05,
      "loss": 3.0434,
      "step": 172500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.04462650042381e-05,
      "loss": 3.0436,
      "step": 173000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.036081945697646e-05,
      "loss": 3.0355,
      "step": 173500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0275373909714817e-05,
      "loss": 3.0462,
      "step": 174000
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0189928362453175e-05,
      "loss": 3.0251,
      "step": 174500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0104482815191537e-05,
      "loss": 3.0178,
      "step": 175000
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0019037267929895e-05,
      "loss": 3.0321,
      "step": 175500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.9933591720668253e-05,
      "loss": 3.0464,
      "step": 176000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9848317064501133e-05,
      "loss": 3.0294,
      "step": 176500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9762871517239494e-05,
      "loss": 3.0237,
      "step": 177000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9677425969977852e-05,
      "loss": 3.0314,
      "step": 177500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9591980422716214e-05,
      "loss": 3.0219,
      "step": 178000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9506534875454572e-05,
      "loss": 3.0153,
      "step": 178500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9421431110381978e-05,
      "loss": 3.018,
      "step": 179000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9335985563120336e-05,
      "loss": 3.0142,
      "step": 179500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9250540015858694e-05,
      "loss": 3.0281,
      "step": 180000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9165094468597052e-05,
      "loss": 3.0175,
      "step": 180500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9079648921335413e-05,
      "loss": 3.0225,
      "step": 181000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.899420337407377e-05,
      "loss": 3.026,
      "step": 181500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8908928717906655e-05,
      "loss": 3.0149,
      "step": 182000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8823483170645013e-05,
      "loss": 3.0129,
      "step": 182500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.873803762338337e-05,
      "loss": 3.0328,
      "step": 183000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.865259207612173e-05,
      "loss": 3.032,
      "step": 183500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8567146528860087e-05,
      "loss": 3.0216,
      "step": 184000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.848170098159845e-05,
      "loss": 3.0179,
      "step": 184500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8396255434336807e-05,
      "loss": 3.033,
      "step": 185000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8310809887075165e-05,
      "loss": 3.01,
      "step": 185500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.822553523090805e-05,
      "loss": 3.0294,
      "step": 186000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8140260574740932e-05,
      "loss": 3.0366,
      "step": 186500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8054815027479287e-05,
      "loss": 3.0392,
      "step": 187000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7969369480217648e-05,
      "loss": 3.0288,
      "step": 187500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7883923932956006e-05,
      "loss": 3.0142,
      "step": 188000
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7798478385694364e-05,
      "loss": 3.0021,
      "step": 188500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7713032838432726e-05,
      "loss": 3.0224,
      "step": 189000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7627587291171084e-05,
      "loss": 3.028,
      "step": 189500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7542141743909442e-05,
      "loss": 3.033,
      "step": 190000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7456867087742322e-05,
      "loss": 3.0213,
      "step": 190500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7371421540480684e-05,
      "loss": 3.0241,
      "step": 191000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.728597599321904e-05,
      "loss": 3.0286,
      "step": 191500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7200530445957403e-05,
      "loss": 3.0215,
      "step": 192000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.711508489869576e-05,
      "loss": 3.0168,
      "step": 192500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7029810242528645e-05,
      "loss": 3.0193,
      "step": 193000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6944364695267e-05,
      "loss": 3.0237,
      "step": 193500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6858919148005358e-05,
      "loss": 3.0177,
      "step": 194000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.677347360074372e-05,
      "loss": 3.0329,
      "step": 194500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6688028053482077e-05,
      "loss": 3.0101,
      "step": 195000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.660275339731496e-05,
      "loss": 3.0115,
      "step": 195500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.651730785005332e-05,
      "loss": 3.0242,
      "step": 196000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6431862302791677e-05,
      "loss": 3.002,
      "step": 196500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6346416755530035e-05,
      "loss": 2.9974,
      "step": 197000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6260971208268396e-05,
      "loss": 3.0068,
      "step": 197500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6175696552101277e-05,
      "loss": 3.009,
      "step": 198000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.609042189593416e-05,
      "loss": 2.9938,
      "step": 198500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6004976348672518e-05,
      "loss": 3.0221,
      "step": 199000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.591953080141088e-05,
      "loss": 3.001,
      "step": 199500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5834085254149238e-05,
      "loss": 3.0025,
      "step": 200000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5748639706887592e-05,
      "loss": 2.994,
      "step": 200500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5663194159625954e-05,
      "loss": 3.0117,
      "step": 201000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5577748612364312e-05,
      "loss": 3.0014,
      "step": 201500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5492303065102673e-05,
      "loss": 2.9978,
      "step": 202000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.540685751784103e-05,
      "loss": 2.999,
      "step": 202500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5321582861673915e-05,
      "loss": 3.0163,
      "step": 203000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5236137314412271e-05,
      "loss": 3.014,
      "step": 203500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5150691767150633e-05,
      "loss": 2.9924,
      "step": 204000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.506524621988899e-05,
      "loss": 3.0152,
      "step": 204500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4979800672627347e-05,
      "loss": 2.9974,
      "step": 205000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4894355125365709e-05,
      "loss": 3.0066,
      "step": 205500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4809080469198589e-05,
      "loss": 2.9912,
      "step": 206000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4723805813031471e-05,
      "loss": 3.0037,
      "step": 206500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4638360265769832e-05,
      "loss": 3.0103,
      "step": 207000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.455291471850819e-05,
      "loss": 2.9896,
      "step": 207500
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4467469171246548e-05,
      "loss": 3.0116,
      "step": 208000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4382023623984908e-05,
      "loss": 2.9955,
      "step": 208500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4296578076723266e-05,
      "loss": 2.9942,
      "step": 209000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4211132529461626e-05,
      "loss": 3.0063,
      "step": 209500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4125686982199984e-05,
      "loss": 2.9989,
      "step": 210000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4040241434938342e-05,
      "loss": 2.9947,
      "step": 210500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3954795887676702e-05,
      "loss": 3.0103,
      "step": 211000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3869521231509582e-05,
      "loss": 3.0076,
      "step": 211500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3784075684247944e-05,
      "loss": 3.0042,
      "step": 212000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3698630136986302e-05,
      "loss": 2.995,
      "step": 212500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3613184589724662e-05,
      "loss": 2.99,
      "step": 213000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.352773904246302e-05,
      "loss": 3.0179,
      "step": 213500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3442293495201378e-05,
      "loss": 2.9983,
      "step": 214000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3356847947939737e-05,
      "loss": 3.004,
      "step": 214500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3271573291772621e-05,
      "loss": 2.9958,
      "step": 215000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3186127744510979e-05,
      "loss": 3.0041,
      "step": 215500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3100682197249337e-05,
      "loss": 3.0125,
      "step": 216000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3015236649987697e-05,
      "loss": 3.0022,
      "step": 216500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2929791102726055e-05,
      "loss": 3.0062,
      "step": 217000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2844516446558939e-05,
      "loss": 3.0123,
      "step": 217500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.275924179039182e-05,
      "loss": 2.9988,
      "step": 218000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2673796243130179e-05,
      "loss": 3.0072,
      "step": 218500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2588350695868537e-05,
      "loss": 2.9979,
      "step": 219000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2502905148606896e-05,
      "loss": 2.9923,
      "step": 219500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2417459601345254e-05,
      "loss": 3.0027,
      "step": 220000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2332014054083614e-05,
      "loss": 2.9938,
      "step": 220500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2246568506821974e-05,
      "loss": 2.9908,
      "step": 221000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2161122959560332e-05,
      "loss": 3.0022,
      "step": 221500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.2075848303393214e-05,
      "loss": 2.9952,
      "step": 222000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1990402756131574e-05,
      "loss": 3.0089,
      "step": 222500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1904957208869932e-05,
      "loss": 3.0001,
      "step": 223000
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.181951166160829e-05,
      "loss": 2.9888,
      "step": 223500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1734237005441173e-05,
      "loss": 3.0018,
      "step": 224000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1648791458179533e-05,
      "loss": 2.9915,
      "step": 224500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.156334591091789e-05,
      "loss": 2.9905,
      "step": 225000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.147790036365625e-05,
      "loss": 2.9962,
      "step": 225500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1392625707489133e-05,
      "loss": 2.9813,
      "step": 226000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1307180160227491e-05,
      "loss": 2.9983,
      "step": 226500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1221734612965849e-05,
      "loss": 3.0039,
      "step": 227000
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1136289065704209e-05,
      "loss": 3.0069,
      "step": 227500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1050843518442567e-05,
      "loss": 2.9831,
      "step": 228000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0965568862275449e-05,
      "loss": 2.9884,
      "step": 228500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0880123315013809e-05,
      "loss": 2.9754,
      "step": 229000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0794677767752168e-05,
      "loss": 3.0021,
      "step": 229500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0709232220490526e-05,
      "loss": 2.9892,
      "step": 230000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0623786673228884e-05,
      "loss": 2.9917,
      "step": 230500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0538341125967244e-05,
      "loss": 2.998,
      "step": 231000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0452895578705602e-05,
      "loss": 2.9833,
      "step": 231500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0367450031443962e-05,
      "loss": 2.9971,
      "step": 232000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0282175375276844e-05,
      "loss": 2.9786,
      "step": 232500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0196729828015204e-05,
      "loss": 2.989,
      "step": 233000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0111284280753562e-05,
      "loss": 2.9926,
      "step": 233500
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0025838733491922e-05,
      "loss": 2.9961,
      "step": 234000
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.94039318623028e-06,
      "loss": 2.9916,
      "step": 234500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.854947638968638e-06,
      "loss": 2.9876,
      "step": 235000
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.769843873896043e-06,
      "loss": 2.996,
      "step": 235500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.684398326634403e-06,
      "loss": 2.9994,
      "step": 236000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.598952779372763e-06,
      "loss": 2.9965,
      "step": 236500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.513507232111121e-06,
      "loss": 2.997,
      "step": 237000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.428061684849479e-06,
      "loss": 2.9894,
      "step": 237500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.342616137587839e-06,
      "loss": 3.0015,
      "step": 238000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.257170590326197e-06,
      "loss": 2.9954,
      "step": 238500
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.171725043064557e-06,
      "loss": 2.9944,
      "step": 239000
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.086279495802916e-06,
      "loss": 2.9938,
      "step": 239500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.000833948541273e-06,
      "loss": 3.0076,
      "step": 240000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.915388401279633e-06,
      "loss": 3.0041,
      "step": 240500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.830113745112516e-06,
      "loss": 2.9965,
      "step": 241000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.744668197850874e-06,
      "loss": 3.003,
      "step": 241500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.659222650589232e-06,
      "loss": 2.9935,
      "step": 242000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.573777103327592e-06,
      "loss": 2.9944,
      "step": 242500
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.488502447160474e-06,
      "loss": 3.0002,
      "step": 243000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.403056899898832e-06,
      "loss": 2.976,
      "step": 243500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.317611352637192e-06,
      "loss": 2.9861,
      "step": 244000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.232336696470074e-06,
      "loss": 2.9925,
      "step": 244500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.146891149208433e-06,
      "loss": 2.9808,
      "step": 245000
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.061445601946792e-06,
      "loss": 2.9895,
      "step": 245500
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.976000054685151e-06,
      "loss": 3.0043,
      "step": 246000
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.89055450742351e-06,
      "loss": 2.9998,
      "step": 246500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.805108960161867e-06,
      "loss": 2.9981,
      "step": 247000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.719663412900227e-06,
      "loss": 2.9903,
      "step": 247500
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.634217865638587e-06,
      "loss": 3.0037,
      "step": 248000
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.548772318376945e-06,
      "loss": 2.9895,
      "step": 248500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.463497662209827e-06,
      "loss": 2.9751,
      "step": 249000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.378052114948186e-06,
      "loss": 3.0095,
      "step": 249500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.292606567686546e-06,
      "loss": 3.0019,
      "step": 250000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.207161020424905e-06,
      "loss": 2.9831,
      "step": 250500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.1218863642577865e-06,
      "loss": 2.9905,
      "step": 251000
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.036440816996145e-06,
      "loss": 2.9734,
      "step": 251500
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.950995269734504e-06,
      "loss": 2.9781,
      "step": 252000
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.865549722472862e-06,
      "loss": 2.9798,
      "step": 252500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.780104175211221e-06,
      "loss": 2.9919,
      "step": 253000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.694658627949581e-06,
      "loss": 2.9718,
      "step": 253500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.60921308068794e-06,
      "loss": 2.991,
      "step": 254000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.523767533426299e-06,
      "loss": 2.9908,
      "step": 254500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.438492877259181e-06,
      "loss": 3.0,
      "step": 255000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.35304732999754e-06,
      "loss": 2.9851,
      "step": 255500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.267601782735899e-06,
      "loss": 3.009,
      "step": 256000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.182156235474258e-06,
      "loss": 2.9899,
      "step": 256500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.096710688212616e-06,
      "loss": 2.9888,
      "step": 257000
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.011436032045498e-06,
      "loss": 2.9916,
      "step": 257500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.925990484783857e-06,
      "loss": 2.9975,
      "step": 258000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.840544937522216e-06,
      "loss": 2.9845,
      "step": 258500
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.755270281355098e-06,
      "loss": 2.9899,
      "step": 259000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.669824734093458e-06,
      "loss": 3.0112,
      "step": 259500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.584379186831816e-06,
      "loss": 2.9947,
      "step": 260000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.498933639570175e-06,
      "loss": 2.9966,
      "step": 260500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.413488092308534e-06,
      "loss": 3.0007,
      "step": 261000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.328042545046893e-06,
      "loss": 2.9996,
      "step": 261500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.242596997785252e-06,
      "loss": 2.9927,
      "step": 262000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.1571514505236106e-06,
      "loss": 2.985,
      "step": 262500
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.0718767943564924e-06,
      "loss": 2.9818,
      "step": 263000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.986431247094852e-06,
      "loss": 2.9953,
      "step": 263500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.90098569983321e-06,
      "loss": 2.987,
      "step": 264000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.815540152571569e-06,
      "loss": 2.9929,
      "step": 264500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.730265496404451e-06,
      "loss": 2.9898,
      "step": 265000
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.644819949142811e-06,
      "loss": 2.994,
      "step": 265500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.55937440188117e-06,
      "loss": 2.9936,
      "step": 266000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.473928854619528e-06,
      "loss": 3.0111,
      "step": 266500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.388483307357888e-06,
      "loss": 2.9822,
      "step": 267000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.3032086511907695e-06,
      "loss": 2.9807,
      "step": 267500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.217763103929128e-06,
      "loss": 2.9816,
      "step": 268000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.132317556667487e-06,
      "loss": 2.9889,
      "step": 268500
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.046872009405845e-06,
      "loss": 2.9941,
      "step": 269000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.961426462144205e-06,
      "loss": 2.9986,
      "step": 269500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.876151805977087e-06,
      "loss": 2.9774,
      "step": 270000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.7907062587154456e-06,
      "loss": 2.9729,
      "step": 270500
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.7054316025483283e-06,
      "loss": 2.9833,
      "step": 271000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.619986055286687e-06,
      "loss": 2.9837,
      "step": 271500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.5345405080250457e-06,
      "loss": 2.9881,
      "step": 272000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.449094960763405e-06,
      "loss": 2.9795,
      "step": 272500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.363649413501764e-06,
      "loss": 2.9806,
      "step": 273000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.2782038662401225e-06,
      "loss": 2.9997,
      "step": 273500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.192758318978482e-06,
      "loss": 2.9997,
      "step": 274000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1073127717168403e-06,
      "loss": 2.9724,
      "step": 274500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.0218672244551992e-06,
      "loss": 2.9822,
      "step": 275000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9365925682880815e-06,
      "loss": 2.9985,
      "step": 275500
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.8511470210264405e-06,
      "loss": 2.9753,
      "step": 276000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.7657014737647994e-06,
      "loss": 3.0126,
      "step": 276500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6802559265031583e-06,
      "loss": 2.9844,
      "step": 277000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.594810379241517e-06,
      "loss": 2.9867,
      "step": 277500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.5095357230743995e-06,
      "loss": 2.9821,
      "step": 278000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.424090175812758e-06,
      "loss": 3.0027,
      "step": 278500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.338644628551117e-06,
      "loss": 2.9909,
      "step": 279000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.2535408634785226e-06,
      "loss": 2.9832,
      "step": 279500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1680953162168815e-06,
      "loss": 2.9681,
      "step": 280000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.08264976895524e-06,
      "loss": 2.9832,
      "step": 280500
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.9972042216935994e-06,
      "loss": 2.9858,
      "step": 281000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.9117586744319583e-06,
      "loss": 2.9736,
      "step": 281500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.826313127170317e-06,
      "loss": 2.9777,
      "step": 282000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7408675799086757e-06,
      "loss": 2.9864,
      "step": 282500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6554220326470346e-06,
      "loss": 2.993,
      "step": 283000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.5699764853853938e-06,
      "loss": 2.9912,
      "step": 283500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4845309381237525e-06,
      "loss": 2.9849,
      "step": 284000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.3990853908621114e-06,
      "loss": 2.9938,
      "step": 284500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.3136398436004704e-06,
      "loss": 2.9879,
      "step": 285000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.2283651874333526e-06,
      "loss": 2.9761,
      "step": 285500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1429196401717114e-06,
      "loss": 2.9885,
      "step": 286000
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0576449840045936e-06,
      "loss": 2.9912,
      "step": 286500
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.721994367429523e-07,
      "loss": 2.9883,
      "step": 287000
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.867538894813114e-07,
      "loss": 2.9839,
      "step": 287500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.013083422196702e-07,
      "loss": 3.0088,
      "step": 288000
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.158627949580292e-07,
      "loss": 2.9898,
      "step": 288500
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.304172476963881e-07,
      "loss": 2.9815,
      "step": 289000
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.44971700434747e-07,
      "loss": 2.9874,
      "step": 289500
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.5952615317310585e-07,
      "loss": 2.9896,
      "step": 290000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.7425149700598803e-07,
      "loss": 2.9839,
      "step": 290500
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.8880594974434696e-07,
      "loss": 2.9745,
      "step": 291000
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.033604024827058e-07,
      "loss": 2.9837,
      "step": 291500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.1791485522106473e-07,
      "loss": 2.9712,
      "step": 292000
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.246930795942362e-08,
      "loss": 2.9916,
      "step": 292500
    },
    {
      "epoch": 3.0,
      "step": 292584,
      "total_flos": 1.22320206692352e+18,
      "train_loss": 3.0709511061687094,
      "train_runtime": 441947.7546,
      "train_samples_per_second": 5.296,
      "train_steps_per_second": 0.662
    }
  ],
  "max_steps": 292584,
  "num_train_epochs": 3,
  "total_flos": 1.22320206692352e+18,
  "trial_name": null,
  "trial_params": null
}