{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 124482,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.979916775116081e-05,
      "loss": 2.8339,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9598335502321625e-05,
      "loss": 2.1894,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9397503253482433e-05,
      "loss": 1.9553,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.919667100464325e-05,
      "loss": 1.8148,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.899583875580405e-05,
      "loss": 1.7143,
      "step": 2500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8795006506964865e-05,
      "loss": 1.6377,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.859417425812567e-05,
      "loss": 1.5807,
      "step": 3500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.839334200928649e-05,
      "loss": 1.5233,
      "step": 4000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8192509760447296e-05,
      "loss": 1.4835,
      "step": 4500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.799167751160811e-05,
      "loss": 1.4557,
      "step": 5000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.779084526276892e-05,
      "loss": 1.4174,
      "step": 5500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.759001301392973e-05,
      "loss": 1.4232,
      "step": 6000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.7389180765090535e-05,
      "loss": 1.3836,
      "step": 6500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.718834851625134e-05,
      "loss": 1.3605,
      "step": 7000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.698751626741216e-05,
      "loss": 1.3425,
      "step": 7500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6786684018572966e-05,
      "loss": 1.313,
      "step": 8000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.658585176973378e-05,
      "loss": 1.3125,
      "step": 8500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.638501952089459e-05,
      "loss": 1.3025,
      "step": 9000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6184187272055404e-05,
      "loss": 1.2743,
      "step": 9500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.598335502321621e-05,
      "loss": 1.2873,
      "step": 10000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.578252277437702e-05,
      "loss": 1.2505,
      "step": 10500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.558169052553783e-05,
      "loss": 1.2546,
      "step": 11000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.538085827669864e-05,
      "loss": 1.2394,
      "step": 11500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.518002602785945e-05,
      "loss": 1.2235,
      "step": 12000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.497919377902026e-05,
      "loss": 1.2016,
      "step": 12500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4778361530181074e-05,
      "loss": 1.1981,
      "step": 13000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.457752928134188e-05,
      "loss": 1.2103,
      "step": 13500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.43766970325027e-05,
      "loss": 1.1915,
      "step": 14000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.41758647836635e-05,
      "loss": 1.1745,
      "step": 14500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.3975032534824313e-05,
      "loss": 1.1911,
      "step": 15000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.377420028598512e-05,
      "loss": 1.16,
      "step": 15500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3573368037145936e-05,
      "loss": 1.163,
      "step": 16000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3372535788306745e-05,
      "loss": 1.1559,
      "step": 16500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.317170353946756e-05,
      "loss": 1.1553,
      "step": 17000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.297087129062837e-05,
      "loss": 1.1327,
      "step": 17500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.2770039041789176e-05,
      "loss": 1.1294,
      "step": 18000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.256920679294999e-05,
      "loss": 1.1315,
      "step": 18500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.236837454411079e-05,
      "loss": 1.1124,
      "step": 19000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.216754229527161e-05,
      "loss": 1.1222,
      "step": 19500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.1966710046432415e-05,
      "loss": 1.1068,
      "step": 20000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.176587779759323e-05,
      "loss": 1.1042,
      "step": 20500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.156504554875404e-05,
      "loss": 1.0997,
      "step": 21000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.136421329991485e-05,
      "loss": 1.0787,
      "step": 21500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.116338105107566e-05,
      "loss": 1.1018,
      "step": 22000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.096254880223647e-05,
      "loss": 1.0729,
      "step": 22500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.076171655339728e-05,
      "loss": 1.0816,
      "step": 23000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.056088430455809e-05,
      "loss": 1.0697,
      "step": 23500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.03600520557189e-05,
      "loss": 1.0598,
      "step": 24000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.015921980687971e-05,
      "loss": 1.0738,
      "step": 24500
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.995838755804052e-05,
      "loss": 1.0632,
      "step": 25000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.975755530920133e-05,
      "loss": 1.0478,
      "step": 25500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9556723060362146e-05,
      "loss": 1.061,
      "step": 26000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9355890811522954e-05,
      "loss": 1.0457,
      "step": 26500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.915505856268376e-05,
      "loss": 1.0467,
      "step": 27000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.895422631384457e-05,
      "loss": 1.0261,
      "step": 27500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8753394065005385e-05,
      "loss": 1.043,
      "step": 28000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.855256181616619e-05,
      "loss": 1.0238,
      "step": 28500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.835172956732701e-05,
      "loss": 1.0345,
      "step": 29000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8150897318487816e-05,
      "loss": 1.0246,
      "step": 29500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7950065069648624e-05,
      "loss": 1.0292,
      "step": 30000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.774923282080944e-05,
      "loss": 1.0156,
      "step": 30500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.754840057197025e-05,
      "loss": 1.0081,
      "step": 31000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7347568323131056e-05,
      "loss": 1.0134,
      "step": 31500
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7146736074291864e-05,
      "loss": 0.9982,
      "step": 32000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.694590382545268e-05,
      "loss": 1.0128,
      "step": 32500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.674507157661349e-05,
      "loss": 0.9917,
      "step": 33000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.65442393277743e-05,
      "loss": 0.9924,
      "step": 33500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.634340707893511e-05,
      "loss": 1.0014,
      "step": 34000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.614257483009592e-05,
      "loss": 0.9885,
      "step": 34500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.594174258125673e-05,
      "loss": 0.9911,
      "step": 35000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.574091033241754e-05,
      "loss": 0.9903,
      "step": 35500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.554007808357835e-05,
      "loss": 0.9797,
      "step": 36000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.533924583473916e-05,
      "loss": 0.982,
      "step": 36500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.513841358589997e-05,
      "loss": 0.9639,
      "step": 37000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.493758133706078e-05,
      "loss": 0.9804,
      "step": 37500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4736749088221595e-05,
      "loss": 0.9665,
      "step": 38000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.45359168393824e-05,
      "loss": 0.9761,
      "step": 38500
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.433508459054322e-05,
      "loss": 0.9674,
      "step": 39000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.413425234170402e-05,
      "loss": 0.9838,
      "step": 39500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.3933420092864834e-05,
      "loss": 0.9663,
      "step": 40000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.373258784402564e-05,
      "loss": 0.9562,
      "step": 40500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.353175559518646e-05,
      "loss": 0.9513,
      "step": 41000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3330923346347265e-05,
      "loss": 0.9456,
      "step": 41500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.313009109750807e-05,
      "loss": 0.8644,
      "step": 42000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.292925884866889e-05,
      "loss": 0.881,
      "step": 42500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.2728426599829696e-05,
      "loss": 0.8684,
      "step": 43000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2527594350990504e-05,
      "loss": 0.8653,
      "step": 43500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.232676210215131e-05,
      "loss": 0.871,
      "step": 44000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.212592985331213e-05,
      "loss": 0.8633,
      "step": 44500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.1925097604472935e-05,
      "loss": 0.8571,
      "step": 45000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.172426535563375e-05,
      "loss": 0.8561,
      "step": 45500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.152343310679456e-05,
      "loss": 0.8585,
      "step": 46000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.1322600857955367e-05,
      "loss": 0.8568,
      "step": 46500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.112176860911618e-05,
      "loss": 0.8657,
      "step": 47000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.092093636027699e-05,
      "loss": 0.8605,
      "step": 47500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.07201041114378e-05,
      "loss": 0.8585,
      "step": 48000
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0519271862598606e-05,
      "loss": 0.8618,
      "step": 48500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.031843961375942e-05,
      "loss": 0.8481,
      "step": 49000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0117607364920232e-05,
      "loss": 0.8529,
      "step": 49500
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.991677511608104e-05,
      "loss": 0.8512,
      "step": 50000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9715942867241852e-05,
      "loss": 0.8576,
      "step": 50500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9515110618402663e-05,
      "loss": 0.8572,
      "step": 51000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9314278369563475e-05,
      "loss": 0.8491,
      "step": 51500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.911344612072428e-05,
      "loss": 0.8444,
      "step": 52000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.891261387188509e-05,
      "loss": 0.846,
      "step": 52500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8711781623045902e-05,
      "loss": 0.8557,
      "step": 53000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8510949374206714e-05,
      "loss": 0.8339,
      "step": 53500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8310117125367525e-05,
      "loss": 0.848,
      "step": 54000
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8109284876528337e-05,
      "loss": 0.8371,
      "step": 54500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7908452627689145e-05,
      "loss": 0.8448,
      "step": 55000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7707620378849957e-05,
      "loss": 0.8414,
      "step": 55500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7506788130010765e-05,
      "loss": 0.8436,
      "step": 56000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7305955881171573e-05,
      "loss": 0.8606,
      "step": 56500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7105123632332384e-05,
      "loss": 0.8425,
      "step": 57000
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6904291383493196e-05,
      "loss": 0.8325,
      "step": 57500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6703459134654007e-05,
      "loss": 0.8371,
      "step": 58000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.650262688581482e-05,
      "loss": 0.846,
      "step": 58500
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.630179463697563e-05,
      "loss": 0.8228,
      "step": 59000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6100962388136442e-05,
      "loss": 0.8408,
      "step": 59500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5900130139297253e-05,
      "loss": 0.8447,
      "step": 60000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5699297890458058e-05,
      "loss": 0.828,
      "step": 60500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.549846564161887e-05,
      "loss": 0.8183,
      "step": 61000
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.529763339277968e-05,
      "loss": 0.8116,
      "step": 61500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.509680114394049e-05,
      "loss": 0.8251,
      "step": 62000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.48959688951013e-05,
      "loss": 0.8293,
      "step": 62500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4695136646262112e-05,
      "loss": 0.8161,
      "step": 63000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.449430439742292e-05,
      "loss": 0.8185,
      "step": 63500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.429347214858373e-05,
      "loss": 0.7976,
      "step": 64000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4092639899744543e-05,
      "loss": 0.8214,
      "step": 64500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3891807650905355e-05,
      "loss": 0.8223,
      "step": 65000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3690975402066163e-05,
      "loss": 0.8155,
      "step": 65500
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3490143153226974e-05,
      "loss": 0.8108,
      "step": 66000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3289310904387786e-05,
      "loss": 0.8111,
      "step": 66500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3088478655548594e-05,
      "loss": 0.8221,
      "step": 67000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2887646406709405e-05,
      "loss": 0.8319,
      "step": 67500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2686814157870213e-05,
      "loss": 0.8093,
      "step": 68000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2485981909031025e-05,
      "loss": 0.7943,
      "step": 68500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2285149660191836e-05,
      "loss": 0.7968,
      "step": 69000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2084317411352645e-05,
      "loss": 0.8062,
      "step": 69500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1883485162513456e-05,
      "loss": 0.8092,
      "step": 70000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1682652913674268e-05,
      "loss": 0.8016,
      "step": 70500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.148182066483508e-05,
      "loss": 0.7967,
      "step": 71000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1280988415995887e-05,
      "loss": 0.8161,
      "step": 71500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.10801561671567e-05,
      "loss": 0.821,
      "step": 72000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.087932391831751e-05,
      "loss": 0.8022,
      "step": 72500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0678491669478318e-05,
      "loss": 0.8005,
      "step": 73000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.047765942063913e-05,
      "loss": 0.7895,
      "step": 73500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0276827171799938e-05,
      "loss": 0.8075,
      "step": 74000
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.007599492296075e-05,
      "loss": 0.802,
      "step": 74500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.987516267412156e-05,
      "loss": 0.7947,
      "step": 75000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9674330425282372e-05,
      "loss": 0.7987,
      "step": 75500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.947349817644318e-05,
      "loss": 0.7948,
      "step": 76000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9272665927603992e-05,
      "loss": 0.7977,
      "step": 76500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9071833678764803e-05,
      "loss": 0.7879,
      "step": 77000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8871001429925615e-05,
      "loss": 0.7855,
      "step": 77500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8670169181086423e-05,
      "loss": 0.7893,
      "step": 78000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8469336932247235e-05,
      "loss": 0.7824,
      "step": 78500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8268504683408043e-05,
      "loss": 0.7855,
      "step": 79000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8067672434568854e-05,
      "loss": 0.7967,
      "step": 79500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7866840185729662e-05,
      "loss": 0.7882,
      "step": 80000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7666007936890474e-05,
      "loss": 0.7847,
      "step": 80500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7465175688051285e-05,
      "loss": 0.7802,
      "step": 81000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7264343439212097e-05,
      "loss": 0.7755,
      "step": 81500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7063511190372905e-05,
      "loss": 0.7905,
      "step": 82000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6862678941533716e-05,
      "loss": 0.7872,
      "step": 82500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6661846692694528e-05,
      "loss": 0.7773,
      "step": 83000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.646101444385534e-05,
      "loss": 0.6754,
      "step": 83500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6260182195016147e-05,
      "loss": 0.685,
      "step": 84000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.605934994617696e-05,
      "loss": 0.6836,
      "step": 84500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5858517697337767e-05,
      "loss": 0.6946,
      "step": 85000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.565768544849858e-05,
      "loss": 0.6975,
      "step": 85500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5456853199659387e-05,
      "loss": 0.6767,
      "step": 86000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5256020950820198e-05,
      "loss": 0.6802,
      "step": 86500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.505518870198101e-05,
      "loss": 0.6896,
      "step": 87000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4854356453141821e-05,
      "loss": 0.6859,
      "step": 87500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4653524204302633e-05,
      "loss": 0.6826,
      "step": 88000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.445269195546344e-05,
      "loss": 0.6811,
      "step": 88500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4251859706624252e-05,
      "loss": 0.6925,
      "step": 89000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4051027457785062e-05,
      "loss": 0.6835,
      "step": 89500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3850195208945874e-05,
      "loss": 0.6758,
      "step": 90000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3649362960106682e-05,
      "loss": 0.6719,
      "step": 90500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3448530711267493e-05,
      "loss": 0.6731,
      "step": 91000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3247698462428305e-05,
      "loss": 0.6848,
      "step": 91500
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3046866213589115e-05,
      "loss": 0.6572,
      "step": 92000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2846033964749923e-05,
      "loss": 0.6778,
      "step": 92500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2645201715910734e-05,
      "loss": 0.6749,
      "step": 93000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2444369467071546e-05,
      "loss": 0.6797,
      "step": 93500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2243537218232355e-05,
      "loss": 0.6752,
      "step": 94000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.2042704969393167e-05,
      "loss": 0.6799,
      "step": 94500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1841872720553977e-05,
      "loss": 0.6758,
      "step": 95000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1641040471714787e-05,
      "loss": 0.6838,
      "step": 95500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1440208222875596e-05,
      "loss": 0.6816,
      "step": 96000
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1239375974036408e-05,
      "loss": 0.6668,
      "step": 96500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1038543725197218e-05,
      "loss": 0.6774,
      "step": 97000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0837711476358029e-05,
      "loss": 0.6754,
      "step": 97500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0636879227518839e-05,
      "loss": 0.6514,
      "step": 98000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0436046978679649e-05,
      "loss": 0.6726,
      "step": 98500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0235214729840459e-05,
      "loss": 0.6624,
      "step": 99000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.003438248100127e-05,
      "loss": 0.6662,
      "step": 99500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.83355023216208e-06,
      "loss": 0.6722,
      "step": 100000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.632717983322891e-06,
      "loss": 0.6621,
      "step": 100500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.431885734483701e-06,
      "loss": 0.6782,
      "step": 101000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.231053485644511e-06,
      "loss": 0.6717,
      "step": 101500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.03022123680532e-06,
      "loss": 0.6753,
      "step": 102000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.829388987966132e-06,
      "loss": 0.6744,
      "step": 102500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.628556739126944e-06,
      "loss": 0.6663,
      "step": 103000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.427724490287754e-06,
      "loss": 0.6726,
      "step": 103500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.226892241448563e-06,
      "loss": 0.6772,
      "step": 104000
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.026059992609373e-06,
      "loss": 0.6579,
      "step": 104500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.825227743770185e-06,
      "loss": 0.6568,
      "step": 105000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.624395494930994e-06,
      "loss": 0.6586,
      "step": 105500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.423563246091805e-06,
      "loss": 0.6653,
      "step": 106000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.222730997252615e-06,
      "loss": 0.6542,
      "step": 106500
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.021898748413426e-06,
      "loss": 0.6693,
      "step": 107000
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.821066499574235e-06,
      "loss": 0.6662,
      "step": 107500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.620234250735047e-06,
      "loss": 0.6628,
      "step": 108000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.419402001895857e-06,
      "loss": 0.6595,
      "step": 108500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.2185697530566664e-06,
      "loss": 0.6769,
      "step": 109000
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.017737504217477e-06,
      "loss": 0.665,
      "step": 109500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.8169052553782886e-06,
      "loss": 0.6536,
      "step": 110000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.616073006539098e-06,
      "loss": 0.6633,
      "step": 110500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.415240757699909e-06,
      "loss": 0.6693,
      "step": 111000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.21440850886072e-06,
      "loss": 0.6738,
      "step": 111500
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.0135762600215295e-06,
      "loss": 0.6541,
      "step": 112000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.81274401118234e-06,
      "loss": 0.6637,
      "step": 112500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.611911762343151e-06,
      "loss": 0.6702,
      "step": 113000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.411079513503961e-06,
      "loss": 0.6665,
      "step": 113500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.210247264664771e-06,
      "loss": 0.6541,
      "step": 114000
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.009415015825582e-06,
      "loss": 0.648,
      "step": 114500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.8085827669863917e-06,
      "loss": 0.6468,
      "step": 115000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.6077505181472023e-06,
      "loss": 0.6418,
      "step": 115500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.4069182693080126e-06,
      "loss": 0.651,
      "step": 116000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.2060860204688228e-06,
      "loss": 0.6553,
      "step": 116500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.0052537716296334e-06,
      "loss": 0.652,
      "step": 117000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.8044215227904437e-06,
      "loss": 0.6528,
      "step": 117500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.603589273951254e-06,
      "loss": 0.6454,
      "step": 118000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.4027570251120645e-06,
      "loss": 0.6543,
      "step": 118500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.2019247762728748e-06,
      "loss": 0.665,
      "step": 119000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.001092527433685e-06,
      "loss": 0.6613,
      "step": 119500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8002602785944954e-06,
      "loss": 0.6469,
      "step": 120000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.5994280297553059e-06,
      "loss": 0.6404,
      "step": 120500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.3985957809161165e-06,
      "loss": 0.6435,
      "step": 121000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1977635320769267e-06,
      "loss": 0.6495,
      "step": 121500
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.969312832377372e-07,
      "loss": 0.6544,
      "step": 122000
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.960990343985477e-07,
      "loss": 0.6562,
      "step": 122500
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.952667855593579e-07,
      "loss": 0.6433,
      "step": 123000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.9443453672016844e-07,
      "loss": 0.6494,
      "step": 123500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.9360228788097877e-07,
      "loss": 0.6397,
      "step": 124000
    },
    {
      "epoch": 3.0,
      "step": 124482,
      "total_flos": 1.2516123971946086e+17,
      "train_loss": 0.8944096128791305,
      "train_runtime": 35169.3032,
      "train_samples_per_second": 35.394,
      "train_steps_per_second": 3.54
    }
  ],
  "max_steps": 124482,
  "num_train_epochs": 3,
  "total_flos": 1.2516123971946086e+17,
  "trial_name": null,
  "trial_params": null
}