{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.99998809481291,
  "global_step": 125994,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.980157785291363e-05,
      "loss": 1.2538,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9603155705827266e-05,
      "loss": 1.2611,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9404733558740894e-05,
      "loss": 1.2527,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.920631141165453e-05,
      "loss": 1.2434,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.900788926456816e-05,
      "loss": 1.238,
      "step": 2500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8809467117481786e-05,
      "loss": 1.2442,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.861104497039542e-05,
      "loss": 1.2649,
      "step": 3500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.841262282330905e-05,
      "loss": 1.226,
      "step": 4000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.821420067622268e-05,
      "loss": 1.2163,
      "step": 4500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8015778529136314e-05,
      "loss": 1.204,
      "step": 5000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.781735638204994e-05,
      "loss": 1.2133,
      "step": 5500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.761893423496357e-05,
      "loss": 1.2059,
      "step": 6000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.74205120878772e-05,
      "loss": 1.1878,
      "step": 6500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7222089940790834e-05,
      "loss": 1.1928,
      "step": 7000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.702366779370446e-05,
      "loss": 1.1991,
      "step": 7500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.682524564661809e-05,
      "loss": 1.1784,
      "step": 8000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6626823499531726e-05,
      "loss": 1.1839,
      "step": 8500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6428401352445355e-05,
      "loss": 1.1888,
      "step": 9000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.622997920535899e-05,
      "loss": 1.1972,
      "step": 9500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.603155705827262e-05,
      "loss": 1.1959,
      "step": 10000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5833134911186247e-05,
      "loss": 1.1742,
      "step": 10500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.563471276409988e-05,
      "loss": 1.1769,
      "step": 11000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.543629061701351e-05,
      "loss": 1.1729,
      "step": 11500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.523786846992714e-05,
      "loss": 1.174,
      "step": 12000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5039446322840774e-05,
      "loss": 1.1693,
      "step": 12500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.48410241757544e-05,
      "loss": 1.1711,
      "step": 13000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.464260202866804e-05,
      "loss": 1.1554,
      "step": 13500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4444179881581666e-05,
      "loss": 1.1556,
      "step": 14000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4245757734495294e-05,
      "loss": 1.1582,
      "step": 14500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.404733558740893e-05,
      "loss": 1.1474,
      "step": 15000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.384891344032256e-05,
      "loss": 1.1529,
      "step": 15500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.365049129323619e-05,
      "loss": 1.1511,
      "step": 16000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.345206914614982e-05,
      "loss": 1.1406,
      "step": 16500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.325364699906345e-05,
      "loss": 1.135,
      "step": 17000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.3055224851977085e-05,
      "loss": 1.1401,
      "step": 17500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.285680270489071e-05,
      "loss": 1.1348,
      "step": 18000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.265838055780434e-05,
      "loss": 1.1276,
      "step": 18500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.245995841071797e-05,
      "loss": 1.1316,
      "step": 19000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.22615362636316e-05,
      "loss": 1.1272,
      "step": 19500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.2063114116545234e-05,
      "loss": 1.1244,
      "step": 20000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.186469196945886e-05,
      "loss": 1.119,
      "step": 20500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.16662698223725e-05,
      "loss": 1.125,
      "step": 21000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1467847675286126e-05,
      "loss": 1.1251,
      "step": 21500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1269425528199754e-05,
      "loss": 1.1179,
      "step": 22000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.107100338111339e-05,
      "loss": 1.1202,
      "step": 22500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.087258123402702e-05,
      "loss": 1.1117,
      "step": 23000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0674159086940646e-05,
      "loss": 1.1153,
      "step": 23500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.047573693985428e-05,
      "loss": 1.1111,
      "step": 24000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.027731479276791e-05,
      "loss": 1.1011,
      "step": 24500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.0078892645681545e-05,
      "loss": 1.1047,
      "step": 25000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9880470498595174e-05,
      "loss": 1.111,
      "step": 25500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.96820483515088e-05,
      "loss": 1.0921,
      "step": 26000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.948362620442244e-05,
      "loss": 1.0949,
      "step": 26500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9285204057336066e-05,
      "loss": 1.0997,
      "step": 27000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.90867819102497e-05,
      "loss": 1.1009,
      "step": 27500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.888835976316333e-05,
      "loss": 1.0911,
      "step": 28000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.868993761607696e-05,
      "loss": 1.0958,
      "step": 28500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.849151546899059e-05,
      "loss": 1.0815,
      "step": 29000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.829309332190422e-05,
      "loss": 1.0868,
      "step": 29500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.809467117481785e-05,
      "loss": 1.086,
      "step": 30000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.789624902773148e-05,
      "loss": 1.0915,
      "step": 30500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.7697826880645106e-05,
      "loss": 1.0896,
      "step": 31000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.749940473355874e-05,
      "loss": 1.0855,
      "step": 31500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.730098258647237e-05,
      "loss": 1.0786,
      "step": 32000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7102560439386005e-05,
      "loss": 1.089,
      "step": 32500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.6904138292299634e-05,
      "loss": 1.0829,
      "step": 33000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.670571614521326e-05,
      "loss": 1.0813,
      "step": 33500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.65072939981269e-05,
      "loss": 1.0759,
      "step": 34000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6308871851040526e-05,
      "loss": 1.08,
      "step": 34500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.611044970395416e-05,
      "loss": 1.0769,
      "step": 35000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.591202755686779e-05,
      "loss": 1.0731,
      "step": 35500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.571360540978142e-05,
      "loss": 1.0658,
      "step": 36000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.551518326269505e-05,
      "loss": 1.0749,
      "step": 36500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.531676111560868e-05,
      "loss": 1.0663,
      "step": 37000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.511833896852231e-05,
      "loss": 1.0715,
      "step": 37500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.4919916821435945e-05,
      "loss": 1.067,
      "step": 38000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.472149467434957e-05,
      "loss": 1.0594,
      "step": 38500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.452307252726321e-05,
      "loss": 1.0684,
      "step": 39000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.432465038017684e-05,
      "loss": 1.0654,
      "step": 39500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.4126228233090465e-05,
      "loss": 1.0589,
      "step": 40000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.39278060860041e-05,
      "loss": 1.0634,
      "step": 40500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.372938393891773e-05,
      "loss": 1.0585,
      "step": 41000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3530961791831364e-05,
      "loss": 1.0431,
      "step": 41500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3332539644744986e-05,
      "loss": 1.0542,
      "step": 42000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.313411749765862e-05,
      "loss": 1.0401,
      "step": 42500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.293569535057225e-05,
      "loss": 1.0388,
      "step": 43000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.273727320348588e-05,
      "loss": 1.0434,
      "step": 43500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.253885105639951e-05,
      "loss": 1.0441,
      "step": 44000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.234042890931314e-05,
      "loss": 1.0401,
      "step": 44500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.214200676222677e-05,
      "loss": 1.0292,
      "step": 45000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.1943584615140405e-05,
      "loss": 1.0325,
      "step": 45500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.1745162468054033e-05,
      "loss": 1.0311,
      "step": 46000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.154674032096767e-05,
      "loss": 1.0307,
      "step": 46500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.13483181738813e-05,
      "loss": 1.0292,
      "step": 47000
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.1149896026794925e-05,
      "loss": 1.0431,
      "step": 47500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.095147387970856e-05,
      "loss": 1.0265,
      "step": 48000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.075305173262219e-05,
      "loss": 1.0283,
      "step": 48500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0554629585535824e-05,
      "loss": 1.0306,
      "step": 49000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0356207438449453e-05,
      "loss": 1.028,
      "step": 49500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0157785291363085e-05,
      "loss": 1.0218,
      "step": 50000
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.9959363144276713e-05,
      "loss": 1.0168,
      "step": 50500
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9760940997190345e-05,
      "loss": 1.0265,
      "step": 51000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9562518850103977e-05,
      "loss": 1.016,
      "step": 51500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.936409670301761e-05,
      "loss": 1.0194,
      "step": 52000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9165674555931237e-05,
      "loss": 1.0261,
      "step": 52500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.896725240884487e-05,
      "loss": 1.0163,
      "step": 53000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.87688302617585e-05,
      "loss": 1.0125,
      "step": 53500
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8570408114672125e-05,
      "loss": 1.006,
      "step": 54000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8371985967585757e-05,
      "loss": 1.011,
      "step": 54500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.817356382049939e-05,
      "loss": 1.0113,
      "step": 55000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.7975141673413017e-05,
      "loss": 1.0066,
      "step": 55500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.777671952632665e-05,
      "loss": 1.0087,
      "step": 56000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.757829737924028e-05,
      "loss": 1.0049,
      "step": 56500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7379875232153913e-05,
      "loss": 1.0081,
      "step": 57000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7181453085067545e-05,
      "loss": 1.0083,
      "step": 57500
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6983030937981173e-05,
      "loss": 1.0137,
      "step": 58000
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6784608790894805e-05,
      "loss": 1.0041,
      "step": 58500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6586186643808437e-05,
      "loss": 1.0065,
      "step": 59000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.638776449672207e-05,
      "loss": 1.0062,
      "step": 59500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6189342349635697e-05,
      "loss": 0.9971,
      "step": 60000
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.599092020254933e-05,
      "loss": 1.002,
      "step": 60500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.579249805546296e-05,
      "loss": 1.0017,
      "step": 61000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5594075908376592e-05,
      "loss": 0.9941,
      "step": 61500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.539565376129022e-05,
      "loss": 0.9984,
      "step": 62000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5197231614203852e-05,
      "loss": 0.9963,
      "step": 62500
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.499880946711748e-05,
      "loss": 0.9876,
      "step": 63000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4800387320031113e-05,
      "loss": 0.9897,
      "step": 63500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4601965172944745e-05,
      "loss": 0.9983,
      "step": 64000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4403543025858376e-05,
      "loss": 0.9866,
      "step": 64500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4205120878772005e-05,
      "loss": 0.9843,
      "step": 65000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4006698731685637e-05,
      "loss": 0.9876,
      "step": 65500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.380827658459927e-05,
      "loss": 0.9904,
      "step": 66000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.36098544375129e-05,
      "loss": 0.9966,
      "step": 66500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.341143229042653e-05,
      "loss": 0.9859,
      "step": 67000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.321301014334016e-05,
      "loss": 0.9835,
      "step": 67500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3014587996253792e-05,
      "loss": 0.9767,
      "step": 68000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2816165849167424e-05,
      "loss": 0.9786,
      "step": 68500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2617743702081052e-05,
      "loss": 0.9818,
      "step": 69000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.241932155499468e-05,
      "loss": 0.979,
      "step": 69500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2220899407908313e-05,
      "loss": 0.9835,
      "step": 70000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.2022477260821944e-05,
      "loss": 0.973,
      "step": 70500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1824055113735576e-05,
      "loss": 0.9704,
      "step": 71000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1625632966649208e-05,
      "loss": 0.9771,
      "step": 71500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1427210819562836e-05,
      "loss": 0.9715,
      "step": 72000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1228788672476468e-05,
      "loss": 0.9668,
      "step": 72500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.10303665253901e-05,
      "loss": 0.9703,
      "step": 73000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0831944378303732e-05,
      "loss": 0.9692,
      "step": 73500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.063352223121736e-05,
      "loss": 0.9597,
      "step": 74000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0435100084130992e-05,
      "loss": 0.9684,
      "step": 74500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.023667793704462e-05,
      "loss": 0.973,
      "step": 75000
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0038255789958252e-05,
      "loss": 0.9669,
      "step": 75500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9839833642871884e-05,
      "loss": 0.9669,
      "step": 76000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9641411495785513e-05,
      "loss": 0.9616,
      "step": 76500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9442989348699144e-05,
      "loss": 0.9658,
      "step": 77000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9244567201612776e-05,
      "loss": 0.9723,
      "step": 77500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9046145054526408e-05,
      "loss": 0.9571,
      "step": 78000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8847722907440036e-05,
      "loss": 0.9661,
      "step": 78500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8649300760353668e-05,
      "loss": 0.9607,
      "step": 79000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.84508786132673e-05,
      "loss": 0.9548,
      "step": 79500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8252456466180932e-05,
      "loss": 0.946,
      "step": 80000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8054034319094564e-05,
      "loss": 0.9494,
      "step": 80500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7855612172008192e-05,
      "loss": 0.9506,
      "step": 81000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.765719002492182e-05,
      "loss": 0.959,
      "step": 81500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7458767877835452e-05,
      "loss": 0.9529,
      "step": 82000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7260345730749084e-05,
      "loss": 0.9526,
      "step": 82500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7061923583662716e-05,
      "loss": 0.9492,
      "step": 83000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6863501436576344e-05,
      "loss": 0.9468,
      "step": 83500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6665079289489976e-05,
      "loss": 0.9403,
      "step": 84000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6466657142403608e-05,
      "loss": 0.9363,
      "step": 84500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.626823499531724e-05,
      "loss": 0.9223,
      "step": 85000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6069812848230868e-05,
      "loss": 0.9316,
      "step": 85500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.58713907011445e-05,
      "loss": 0.9442,
      "step": 86000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.567296855405813e-05,
      "loss": 0.9328,
      "step": 86500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.547454640697176e-05,
      "loss": 0.9301,
      "step": 87000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5276124259885392e-05,
      "loss": 0.9304,
      "step": 87500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5077702112799022e-05,
      "loss": 0.9238,
      "step": 88000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4879279965712654e-05,
      "loss": 0.9303,
      "step": 88500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4680857818626284e-05,
      "loss": 0.925,
      "step": 89000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4482435671539916e-05,
      "loss": 0.9213,
      "step": 89500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4284013524453546e-05,
      "loss": 0.9199,
      "step": 90000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4085591377367178e-05,
      "loss": 0.9268,
      "step": 90500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3887169230280808e-05,
      "loss": 0.9315,
      "step": 91000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.368874708319444e-05,
      "loss": 0.9232,
      "step": 91500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.349032493610807e-05,
      "loss": 0.9242,
      "step": 92000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3291902789021701e-05,
      "loss": 0.9136,
      "step": 92500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.309348064193533e-05,
      "loss": 0.9232,
      "step": 93000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.289505849484896e-05,
      "loss": 0.9234,
      "step": 93500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2696636347762592e-05,
      "loss": 0.926,
      "step": 94000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2498214200676222e-05,
      "loss": 0.9153,
      "step": 94500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2299792053589854e-05,
      "loss": 0.9161,
      "step": 95000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2101369906503485e-05,
      "loss": 0.9154,
      "step": 95500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1902947759417116e-05,
      "loss": 0.9187,
      "step": 96000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1704525612330747e-05,
      "loss": 0.9239,
      "step": 96500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1506103465244377e-05,
      "loss": 0.9083,
      "step": 97000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1307681318158008e-05,
      "loss": 0.9173,
      "step": 97500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1109259171071638e-05,
      "loss": 0.9103,
      "step": 98000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.091083702398527e-05,
      "loss": 0.916,
      "step": 98500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0712414876898901e-05,
      "loss": 0.9058,
      "step": 99000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0513992729812531e-05,
      "loss": 0.9128,
      "step": 99500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0315570582726163e-05,
      "loss": 0.9121,
      "step": 100000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0117148435639792e-05,
      "loss": 0.9096,
      "step": 100500
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.918726288553423e-06,
      "loss": 0.9066,
      "step": 101000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.720304141467054e-06,
      "loss": 0.9061,
      "step": 101500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.521881994380685e-06,
      "loss": 0.8937,
      "step": 102000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.323459847294317e-06,
      "loss": 0.8996,
      "step": 102500
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.125037700207947e-06,
      "loss": 0.8985,
      "step": 103000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.926615553121577e-06,
      "loss": 0.8975,
      "step": 103500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.728193406035207e-06,
      "loss": 0.8996,
      "step": 104000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.52977125894884e-06,
      "loss": 0.9021,
      "step": 104500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.33134911186247e-06,
      "loss": 0.8999,
      "step": 105000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.132926964776101e-06,
      "loss": 0.897,
      "step": 105500
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.934504817689731e-06,
      "loss": 0.8979,
      "step": 106000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.736082670603361e-06,
      "loss": 0.9032,
      "step": 106500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.537660523516993e-06,
      "loss": 0.8913,
      "step": 107000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.339238376430624e-06,
      "loss": 0.8933,
      "step": 107500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.140816229344255e-06,
      "loss": 0.8973,
      "step": 108000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.942394082257886e-06,
      "loss": 0.9012,
      "step": 108500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.743971935171517e-06,
      "loss": 0.8874,
      "step": 109000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.545549788085146e-06,
      "loss": 0.8912,
      "step": 109500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.347127640998777e-06,
      "loss": 0.8938,
      "step": 110000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.148705493912408e-06,
      "loss": 0.8922,
      "step": 110500
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.95028334682604e-06,
      "loss": 0.8873,
      "step": 111000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.751861199739671e-06,
      "loss": 0.8932,
      "step": 111500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.553439052653301e-06,
      "loss": 0.8923,
      "step": 112000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.355016905566932e-06,
      "loss": 0.8884,
      "step": 112500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.156594758480563e-06,
      "loss": 0.8895,
      "step": 113000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.958172611394193e-06,
      "loss": 0.8834,
      "step": 113500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.759750464307824e-06,
      "loss": 0.8862,
      "step": 114000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.561328317221456e-06,
      "loss": 0.8871,
      "step": 114500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.362906170135086e-06,
      "loss": 0.8982,
      "step": 115000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.164484023048717e-06,
      "loss": 0.8886,
      "step": 115500
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.966061875962348e-06,
      "loss": 0.8821,
      "step": 116000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.767639728875978e-06,
      "loss": 0.8851,
      "step": 116500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.5692175817896094e-06,
      "loss": 0.8842,
      "step": 117000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.3707954347032404e-06,
      "loss": 0.8871,
      "step": 117500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1723732876168705e-06,
      "loss": 0.881,
      "step": 118000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.973951140530502e-06,
      "loss": 0.8775,
      "step": 118500
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7755289934441324e-06,
      "loss": 0.8803,
      "step": 119000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.5771068463577633e-06,
      "loss": 0.8751,
      "step": 119500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.378684699271394e-06,
      "loss": 0.8817,
      "step": 120000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.180262552185025e-06,
      "loss": 0.883,
      "step": 120500
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.9818404050986558e-06,
      "loss": 0.8762,
      "step": 121000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7834182580122865e-06,
      "loss": 0.8811,
      "step": 121500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.584996110925917e-06,
      "loss": 0.8698,
      "step": 122000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.386573963839548e-06,
      "loss": 0.8762,
      "step": 122500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1881518167531788e-06,
      "loss": 0.8801,
      "step": 123000
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.897296696668095e-07,
      "loss": 0.8732,
      "step": 123500
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.913075225804404e-07,
      "loss": 0.8753,
      "step": 124000
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.928853754940711e-07,
      "loss": 0.8718,
      "step": 124500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.94463228407702e-07,
      "loss": 0.8754,
      "step": 125000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.9604108132133275e-07,
      "loss": 0.8802,
      "step": 125500
    },
    {
      "epoch": 3.0,
      "step": 125994,
      "total_flos": 1.1751527342197965e+18,
      "train_loss": 1.0093599338199357,
      "train_runtime": 128740.5509,
      "train_samples_per_second": 9.787,
      "train_steps_per_second": 0.979
    }
  ],
  "max_steps": 125994,
  "num_train_epochs": 3,
  "total_flos": 1.1751527342197965e+18,
  "trial_name": null,
  "trial_params": null
}