{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 113705,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.991275669495625e-05,
      "loss": 0.5297,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.982480981487182e-05,
      "loss": 0.4328,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.973686293478739e-05,
      "loss": 0.4064,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.964891605470296e-05,
      "loss": 0.3912,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9560969174618532e-05,
      "loss": 0.3802,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9473022294534103e-05,
      "loss": 0.367,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9385075414449674e-05,
      "loss": 0.3647,
      "step": 3500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9297128534365245e-05,
      "loss": 0.3546,
      "step": 4000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9209181654280816e-05,
      "loss": 0.3546,
      "step": 4500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9121234774196386e-05,
      "loss": 0.3439,
      "step": 5000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9033287894111957e-05,
      "loss": 0.3493,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.8945341014027528e-05,
      "loss": 0.3488,
      "step": 6000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.88573941339431e-05,
      "loss": 0.352,
      "step": 6500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.8769623147618842e-05,
      "loss": 0.3304,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.868167626753441e-05,
      "loss": 0.3298,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.859372938744998e-05,
      "loss": 0.3359,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.850578250736555e-05,
      "loss": 0.3216,
      "step": 8500
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.8417835627281126e-05,
      "loss": 0.3235,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.8329888747196697e-05,
      "loss": 0.3257,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.8241941867112264e-05,
      "loss": 0.3223,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.8153994987027835e-05,
      "loss": 0.3188,
      "step": 10500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.8066224000703575e-05,
      "loss": 0.309,
      "step": 11000
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.797827712061915e-05,
      "loss": 0.3062,
      "step": 11500
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.789033024053472e-05,
      "loss": 0.3238,
      "step": 12000
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.780238336045029e-05,
      "loss": 0.3096,
      "step": 12500
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.771461237412603e-05,
      "loss": 0.311,
      "step": 13000
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.76266654940416e-05,
      "loss": 0.315,
      "step": 13500
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.753871861395717e-05,
      "loss": 0.3201,
      "step": 14000
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.7450771733872743e-05,
      "loss": 0.3127,
      "step": 14500
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.7362824853788314e-05,
      "loss": 0.3108,
      "step": 15000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.7274877973703885e-05,
      "loss": 0.2981,
      "step": 15500
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7186931093619456e-05,
      "loss": 0.3037,
      "step": 16000
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7098984213535027e-05,
      "loss": 0.3054,
      "step": 16500
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7011213227210767e-05,
      "loss": 0.2986,
      "step": 17000
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.6923266347126338e-05,
      "loss": 0.3025,
      "step": 17500
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.6835495360802078e-05,
      "loss": 0.2992,
      "step": 18000
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.674754848071765e-05,
      "loss": 0.2985,
      "step": 18500
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.665960160063322e-05,
      "loss": 0.2913,
      "step": 19000
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.657165472054879e-05,
      "loss": 0.3051,
      "step": 19500
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.648370784046436e-05,
      "loss": 0.2994,
      "step": 20000
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6395760960379932e-05,
      "loss": 0.3035,
      "step": 20500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6307814080295503e-05,
      "loss": 0.282,
      "step": 21000
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6219867200211074e-05,
      "loss": 0.2888,
      "step": 21500
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6132096213886814e-05,
      "loss": 0.2987,
      "step": 22000
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.6044149333802385e-05,
      "loss": 0.2988,
      "step": 22500
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5956378347478124e-05,
      "loss": 0.2587,
      "step": 23000
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5868431467393695e-05,
      "loss": 0.2308,
      "step": 23500
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5780484587309266e-05,
      "loss": 0.2312,
      "step": 24000
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.5692537707224837e-05,
      "loss": 0.2184,
      "step": 24500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.5604766720900577e-05,
      "loss": 0.2283,
      "step": 25000
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.5516819840816148e-05,
      "loss": 0.2321,
      "step": 25500
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.542887296073172e-05,
      "loss": 0.239,
      "step": 26000
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.534092608064729e-05,
      "loss": 0.23,
      "step": 26500
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.525297920056286e-05,
      "loss": 0.2285,
      "step": 27000
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.51652082142386e-05,
      "loss": 0.2449,
      "step": 27500
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.5077261334154171e-05,
      "loss": 0.2317,
      "step": 28000
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.4989314454069744e-05,
      "loss": 0.2414,
      "step": 28500
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.4901367573985315e-05,
      "loss": 0.2335,
      "step": 29000
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.4813420693900884e-05,
      "loss": 0.242,
      "step": 29500
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.4725649707576625e-05,
      "loss": 0.2483,
      "step": 30000
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.4637702827492195e-05,
      "loss": 0.2381,
      "step": 30500
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.4549755947407767e-05,
      "loss": 0.2286,
      "step": 31000
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.4461809067323338e-05,
      "loss": 0.2384,
      "step": 31500
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.4374038080999078e-05,
      "loss": 0.2253,
      "step": 32000
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.4286091200914649e-05,
      "loss": 0.2328,
      "step": 32500
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.4198144320830221e-05,
      "loss": 0.2265,
      "step": 33000
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.411019744074579e-05,
      "loss": 0.2384,
      "step": 33500
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.4022250560661361e-05,
      "loss": 0.2254,
      "step": 34000
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.3934303680576932e-05,
      "loss": 0.2315,
      "step": 34500
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.3846356800492505e-05,
      "loss": 0.2366,
      "step": 35000
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.3758409920408076e-05,
      "loss": 0.2285,
      "step": 35500
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.3670638934083816e-05,
      "loss": 0.2252,
      "step": 36000
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.3582867947759556e-05,
      "loss": 0.2431,
      "step": 36500
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.3495096961435294e-05,
      "loss": 0.2295,
      "step": 37000
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.3407150081350866e-05,
      "loss": 0.2295,
      "step": 37500
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.3319203201266435e-05,
      "loss": 0.2407,
      "step": 38000
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.3231256321182006e-05,
      "loss": 0.2218,
      "step": 38500
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.3143309441097579e-05,
      "loss": 0.231,
      "step": 39000
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.305536256101315e-05,
      "loss": 0.2309,
      "step": 39500
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.296741568092872e-05,
      "loss": 0.2318,
      "step": 40000
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.287946880084429e-05,
      "loss": 0.2171,
      "step": 40500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.2791521920759862e-05,
      "loss": 0.2291,
      "step": 41000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.2703575040675433e-05,
      "loss": 0.2299,
      "step": 41500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.2615804054351173e-05,
      "loss": 0.2195,
      "step": 42000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.2527857174266744e-05,
      "loss": 0.2291,
      "step": 42500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.2439910294182317e-05,
      "loss": 0.2256,
      "step": 43000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.2351963414097886e-05,
      "loss": 0.234,
      "step": 43500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.2264016534013457e-05,
      "loss": 0.2352,
      "step": 44000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.2176069653929028e-05,
      "loss": 0.2208,
      "step": 44500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.2088298667604767e-05,
      "loss": 0.2266,
      "step": 45000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.2000351787520338e-05,
      "loss": 0.222,
      "step": 45500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.1912404907435911e-05,
      "loss": 0.1571,
      "step": 46000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.182445802735148e-05,
      "loss": 0.1712,
      "step": 46500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.1736687041027222e-05,
      "loss": 0.1624,
      "step": 47000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.164874016094279e-05,
      "loss": 0.1595,
      "step": 47500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.156096917461853e-05,
      "loss": 0.1598,
      "step": 48000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.1473022294534102e-05,
      "loss": 0.1787,
      "step": 48500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1385075414449674e-05,
      "loss": 0.1604,
      "step": 49000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.1297304428125412e-05,
      "loss": 0.1747,
      "step": 49500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.1209357548040985e-05,
      "loss": 0.1773,
      "step": 50000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.1121410667956556e-05,
      "loss": 0.1856,
      "step": 50500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.1033463787872125e-05,
      "loss": 0.1759,
      "step": 51000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.0945516907787696e-05,
      "loss": 0.1692,
      "step": 51500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.0857570027703268e-05,
      "loss": 0.1826,
      "step": 52000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.076962314761884e-05,
      "loss": 0.1737,
      "step": 52500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.068167626753441e-05,
      "loss": 0.1838,
      "step": 53000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.059390528121015e-05,
      "loss": 0.1844,
      "step": 53500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.050595840112572e-05,
      "loss": 0.1771,
      "step": 54000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0418011521041292e-05,
      "loss": 0.1776,
      "step": 54500
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.0330064640956863e-05,
      "loss": 0.17,
      "step": 55000
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.0242117760872434e-05,
      "loss": 0.181,
      "step": 55500
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.0154346774548173e-05,
      "loss": 0.1723,
      "step": 56000
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.0066399894463746e-05,
      "loss": 0.1796,
      "step": 56500
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.978453014379315e-06,
      "loss": 0.1745,
      "step": 57000
    },
    {
      "epoch": 2.53,
      "learning_rate": 9.890682028055055e-06,
      "loss": 0.1783,
      "step": 57500
    },
    {
      "epoch": 2.55,
      "learning_rate": 9.802735147970626e-06,
      "loss": 0.1773,
      "step": 58000
    },
    {
      "epoch": 2.57,
      "learning_rate": 9.714788267886197e-06,
      "loss": 0.1864,
      "step": 58500
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.62684138780177e-06,
      "loss": 0.1777,
      "step": 59000
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.53889450771734e-06,
      "loss": 0.1829,
      "step": 59500
    },
    {
      "epoch": 2.64,
      "learning_rate": 9.450947627632911e-06,
      "loss": 0.1873,
      "step": 60000
    },
    {
      "epoch": 2.66,
      "learning_rate": 9.363000747548482e-06,
      "loss": 0.1744,
      "step": 60500
    },
    {
      "epoch": 2.68,
      "learning_rate": 9.275053867464053e-06,
      "loss": 0.1715,
      "step": 61000
    },
    {
      "epoch": 2.7,
      "learning_rate": 9.187282881139791e-06,
      "loss": 0.1787,
      "step": 61500
    },
    {
      "epoch": 2.73,
      "learning_rate": 9.099336001055364e-06,
      "loss": 0.1864,
      "step": 62000
    },
    {
      "epoch": 2.75,
      "learning_rate": 9.011389120970935e-06,
      "loss": 0.1737,
      "step": 62500
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.923442240886505e-06,
      "loss": 0.1884,
      "step": 63000
    },
    {
      "epoch": 2.79,
      "learning_rate": 8.835671254562245e-06,
      "loss": 0.1656,
      "step": 63500
    },
    {
      "epoch": 2.81,
      "learning_rate": 8.747724374477816e-06,
      "loss": 0.1787,
      "step": 64000
    },
    {
      "epoch": 2.84,
      "learning_rate": 8.659777494393387e-06,
      "loss": 0.1726,
      "step": 64500
    },
    {
      "epoch": 2.86,
      "learning_rate": 8.571830614308958e-06,
      "loss": 0.1733,
      "step": 65000
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.484059627984698e-06,
      "loss": 0.1864,
      "step": 65500
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.396112747900269e-06,
      "loss": 0.1839,
      "step": 66000
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.30816586781584e-06,
      "loss": 0.1759,
      "step": 66500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.22021898773141e-06,
      "loss": 0.1853,
      "step": 67000
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.132272107646983e-06,
      "loss": 0.1911,
      "step": 67500
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.044325227562552e-06,
      "loss": 0.18,
      "step": 68000
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.956554241238292e-06,
      "loss": 0.1336,
      "step": 68500
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.868607361153865e-06,
      "loss": 0.0996,
      "step": 69000
    },
    {
      "epoch": 3.06,
      "learning_rate": 7.780836374829603e-06,
      "loss": 0.1093,
      "step": 69500
    },
    {
      "epoch": 3.08,
      "learning_rate": 7.692889494745175e-06,
      "loss": 0.1032,
      "step": 70000
    },
    {
      "epoch": 3.1,
      "learning_rate": 7.6049426146607454e-06,
      "loss": 0.1089,
      "step": 70500
    },
    {
      "epoch": 3.12,
      "learning_rate": 7.516995734576316e-06,
      "loss": 0.1156,
      "step": 71000
    },
    {
      "epoch": 3.14,
      "learning_rate": 7.429048854491887e-06,
      "loss": 0.1174,
      "step": 71500
    },
    {
      "epoch": 3.17,
      "learning_rate": 7.341101974407459e-06,
      "loss": 0.1101,
      "step": 72000
    },
    {
      "epoch": 3.19,
      "learning_rate": 7.25315509432303e-06,
      "loss": 0.1144,
      "step": 72500
    },
    {
      "epoch": 3.21,
      "learning_rate": 7.165208214238601e-06,
      "loss": 0.118,
      "step": 73000
    },
    {
      "epoch": 3.23,
      "learning_rate": 7.0772613341541716e-06,
      "loss": 0.1298,
      "step": 73500
    },
    {
      "epoch": 3.25,
      "learning_rate": 6.9896662415900804e-06,
      "loss": 0.1059,
      "step": 74000
    },
    {
      "epoch": 3.28,
      "learning_rate": 6.901719361505651e-06,
      "loss": 0.1187,
      "step": 74500
    },
    {
      "epoch": 3.3,
      "learning_rate": 6.813772481421222e-06,
      "loss": 0.1087,
      "step": 75000
    },
    {
      "epoch": 3.32,
      "learning_rate": 6.725825601336793e-06,
      "loss": 0.1129,
      "step": 75500
    },
    {
      "epoch": 3.34,
      "learning_rate": 6.637878721252364e-06,
      "loss": 0.1133,
      "step": 76000
    },
    {
      "epoch": 3.36,
      "learning_rate": 6.549931841167935e-06,
      "loss": 0.1134,
      "step": 76500
    },
    {
      "epoch": 3.39,
      "learning_rate": 6.4619849610835065e-06,
      "loss": 0.1203,
      "step": 77000
    },
    {
      "epoch": 3.41,
      "learning_rate": 6.374038080999077e-06,
      "loss": 0.1099,
      "step": 77500
    },
    {
      "epoch": 3.43,
      "learning_rate": 6.286091200914648e-06,
      "loss": 0.1264,
      "step": 78000
    },
    {
      "epoch": 3.45,
      "learning_rate": 6.198144320830219e-06,
      "loss": 0.1103,
      "step": 78500
    },
    {
      "epoch": 3.47,
      "learning_rate": 6.110373334505959e-06,
      "loss": 0.1137,
      "step": 79000
    },
    {
      "epoch": 3.5,
      "learning_rate": 6.022602348181699e-06,
      "loss": 0.1279,
      "step": 79500
    },
    {
      "epoch": 3.52,
      "learning_rate": 5.93465546809727e-06,
      "loss": 0.1172,
      "step": 80000
    },
    {
      "epoch": 3.54,
      "learning_rate": 5.846708588012841e-06,
      "loss": 0.1175,
      "step": 80500
    },
    {
      "epoch": 3.56,
      "learning_rate": 5.7587617079284116e-06,
      "loss": 0.121,
      "step": 81000
    },
    {
      "epoch": 3.58,
      "learning_rate": 5.6708148278439824e-06,
      "loss": 0.1168,
      "step": 81500
    },
    {
      "epoch": 3.61,
      "learning_rate": 5.582867947759554e-06,
      "loss": 0.1138,
      "step": 82000
    },
    {
      "epoch": 3.63,
      "learning_rate": 5.494921067675124e-06,
      "loss": 0.1115,
      "step": 82500
    },
    {
      "epoch": 3.65,
      "learning_rate": 5.406974187590696e-06,
      "loss": 0.1093,
      "step": 83000
    },
    {
      "epoch": 3.67,
      "learning_rate": 5.319203201266435e-06,
      "loss": 0.1159,
      "step": 83500
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.231256321182007e-06,
      "loss": 0.1125,
      "step": 84000
    },
    {
      "epoch": 3.72,
      "learning_rate": 5.143309441097577e-06,
      "loss": 0.1212,
      "step": 84500
    },
    {
      "epoch": 3.74,
      "learning_rate": 5.0553625610131484e-06,
      "loss": 0.1175,
      "step": 85000
    },
    {
      "epoch": 3.76,
      "learning_rate": 4.967415680928719e-06,
      "loss": 0.1032,
      "step": 85500
    },
    {
      "epoch": 3.78,
      "learning_rate": 4.87946880084429e-06,
      "loss": 0.1173,
      "step": 86000
    },
    {
      "epoch": 3.8,
      "learning_rate": 4.791521920759861e-06,
      "loss": 0.1166,
      "step": 86500
    },
    {
      "epoch": 3.83,
      "learning_rate": 4.703750934435602e-06,
      "loss": 0.1132,
      "step": 87000
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.615979948111341e-06,
      "loss": 0.1111,
      "step": 87500
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.5282089617870816e-06,
      "loss": 0.115,
      "step": 88000
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.4402620817026524e-06,
      "loss": 0.1182,
      "step": 88500
    },
    {
      "epoch": 3.91,
      "learning_rate": 4.352315201618223e-06,
      "loss": 0.1116,
      "step": 89000
    },
    {
      "epoch": 3.94,
      "learning_rate": 4.264368321533794e-06,
      "loss": 0.1162,
      "step": 89500
    },
    {
      "epoch": 3.96,
      "learning_rate": 4.176421441449365e-06,
      "loss": 0.1256,
      "step": 90000
    },
    {
      "epoch": 3.98,
      "learning_rate": 4.088474561364936e-06,
      "loss": 0.1113,
      "step": 90500
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.000527681280507e-06,
      "loss": 0.1062,
      "step": 91000
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.912580801196078e-06,
      "loss": 0.0466,
      "step": 91500
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.8246339211116486e-06,
      "loss": 0.06,
      "step": 92000
    },
    {
      "epoch": 4.07,
      "learning_rate": 3.73668704102722e-06,
      "loss": 0.0575,
      "step": 92500
    },
    {
      "epoch": 4.09,
      "learning_rate": 3.6489160547029593e-06,
      "loss": 0.0568,
      "step": 93000
    },
    {
      "epoch": 4.11,
      "learning_rate": 3.5609691746185306e-06,
      "loss": 0.0526,
      "step": 93500
    },
    {
      "epoch": 4.13,
      "learning_rate": 3.473022294534102e-06,
      "loss": 0.0572,
      "step": 94000
    },
    {
      "epoch": 4.16,
      "learning_rate": 3.3852513082098414e-06,
      "loss": 0.0598,
      "step": 94500
    },
    {
      "epoch": 4.18,
      "learning_rate": 3.2973044281254123e-06,
      "loss": 0.0526,
      "step": 95000
    },
    {
      "epoch": 4.2,
      "learning_rate": 3.209357548040983e-06,
      "loss": 0.0637,
      "step": 95500
    },
    {
      "epoch": 4.22,
      "learning_rate": 3.121410667956554e-06,
      "loss": 0.0584,
      "step": 96000
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.0334637878721257e-06,
      "loss": 0.0472,
      "step": 96500
    },
    {
      "epoch": 4.27,
      "learning_rate": 2.9455169077876966e-06,
      "loss": 0.0584,
      "step": 97000
    },
    {
      "epoch": 4.29,
      "learning_rate": 2.8575700277032675e-06,
      "loss": 0.0517,
      "step": 97500
    },
    {
      "epoch": 4.31,
      "learning_rate": 2.7696231476188384e-06,
      "loss": 0.0582,
      "step": 98000
    },
    {
      "epoch": 4.33,
      "learning_rate": 2.6816762675344093e-06,
      "loss": 0.0634,
      "step": 98500
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.59372938744998e-06,
      "loss": 0.0508,
      "step": 99000
    },
    {
      "epoch": 4.38,
      "learning_rate": 2.505782507365551e-06,
      "loss": 0.0587,
      "step": 99500
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.4178356272811223e-06,
      "loss": 0.0516,
      "step": 100000
    },
    {
      "epoch": 4.42,
      "learning_rate": 2.329888747196693e-06,
      "loss": 0.046,
      "step": 100500
    },
    {
      "epoch": 4.44,
      "learning_rate": 2.242117760872433e-06,
      "loss": 0.0569,
      "step": 101000
    },
    {
      "epoch": 4.46,
      "learning_rate": 2.154346774548173e-06,
      "loss": 0.0514,
      "step": 101500
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.066399894463744e-06,
      "loss": 0.0509,
      "step": 102000
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.978453014379315e-06,
      "loss": 0.0484,
      "step": 102500
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.890506134294886e-06,
      "loss": 0.0561,
      "step": 103000
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.8027351479706259e-06,
      "loss": 0.0508,
      "step": 103500
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.7147882678861968e-06,
      "loss": 0.0548,
      "step": 104000
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.6268413878017679e-06,
      "loss": 0.0497,
      "step": 104500
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.538894507717339e-06,
      "loss": 0.0552,
      "step": 105000
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.4509476276329098e-06,
      "loss": 0.0528,
      "step": 105500
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.3631766413086497e-06,
      "loss": 0.0525,
      "step": 106000
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.2752297612242206e-06,
      "loss": 0.0543,
      "step": 106500
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.1872828811397917e-06,
      "loss": 0.0402,
      "step": 107000
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.0993360010553625e-06,
      "loss": 0.0563,
      "step": 107500
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.0115650147311024e-06,
      "loss": 0.0473,
      "step": 108000
    },
    {
      "epoch": 4.77,
      "learning_rate": 9.236181346466735e-07,
      "loss": 0.0491,
      "step": 108500
    },
    {
      "epoch": 4.79,
      "learning_rate": 8.356712545622445e-07,
      "loss": 0.0544,
      "step": 109000
    },
    {
      "epoch": 4.82,
      "learning_rate": 7.477243744778155e-07,
      "loss": 0.0429,
      "step": 109500
    },
    {
      "epoch": 4.84,
      "learning_rate": 6.597774943933865e-07,
      "loss": 0.0533,
      "step": 110000
    },
    {
      "epoch": 4.86,
      "learning_rate": 5.720065080691263e-07,
      "loss": 0.0429,
      "step": 110500
    },
    {
      "epoch": 4.88,
      "learning_rate": 4.840596279846973e-07,
      "loss": 0.0581,
      "step": 111000
    },
    {
      "epoch": 4.9,
      "learning_rate": 3.9611274790026826e-07,
      "loss": 0.0465,
      "step": 111500
    },
    {
      "epoch": 4.93,
      "learning_rate": 3.0816586781583924e-07,
      "loss": 0.0461,
      "step": 112000
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.2021898773141022e-07,
      "loss": 0.0468,
      "step": 112500
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.324480014071501e-07,
      "loss": 0.0358,
      "step": 113000
    },
    {
      "epoch": 4.99,
      "learning_rate": 4.450112132272108e-08,
      "loss": 0.0516,
      "step": 113500
    },
    {
      "epoch": 5.0,
      "step": 113705,
      "total_flos": 4.2384364792971264e+17,
      "train_loss": 0.18087886704874775,
      "train_runtime": 10203.8706,
      "train_samples_per_second": 178.288,
      "train_steps_per_second": 11.143
    }
  ],
  "max_steps": 113705,
  "num_train_epochs": 5,
  "total_flos": 4.2384364792971264e+17,
  "trial_name": null,
  "trial_params": null
}