{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.933304952462041,
  "global_step": 140000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 4.9822619554420325e-05,
      "loss": 7.0399,
      "step": 500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.964523910884065e-05,
      "loss": 6.9038,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.946785866326096e-05,
      "loss": 6.8142,
      "step": 1500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9290478217681286e-05,
      "loss": 6.7261,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.911309777210161e-05,
      "loss": 6.6601,
      "step": 2500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.893571732652193e-05,
      "loss": 6.607,
      "step": 3000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8758336880942247e-05,
      "loss": 6.565,
      "step": 3500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.858095643536257e-05,
      "loss": 6.5083,
      "step": 4000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.840357598978289e-05,
      "loss": 6.4739,
      "step": 4500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8226195544203214e-05,
      "loss": 6.4472,
      "step": 5000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.804881509862353e-05,
      "loss": 6.4133,
      "step": 5500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.787143465304385e-05,
      "loss": 6.3857,
      "step": 6000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7694054207464175e-05,
      "loss": 6.3575,
      "step": 6500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.75166737618845e-05,
      "loss": 6.3337,
      "step": 7000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.733929331630481e-05,
      "loss": 6.3248,
      "step": 7500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.7161912870725136e-05,
      "loss": 6.2974,
      "step": 8000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.698453242514546e-05,
      "loss": 6.2786,
      "step": 8500
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.680715197956578e-05,
      "loss": 6.2569,
      "step": 9000
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.6629771533986097e-05,
      "loss": 6.2477,
      "step": 9500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.645239108840642e-05,
      "loss": 6.2252,
      "step": 10000
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.627501064282674e-05,
      "loss": 6.212,
      "step": 10500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.609763019724706e-05,
      "loss": 6.1802,
      "step": 11000
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.592024975166738e-05,
      "loss": 6.1081,
      "step": 11500
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.5742869306087696e-05,
      "loss": 6.0248,
      "step": 12000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.556548886050802e-05,
      "loss": 5.8882,
      "step": 12500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.538810841492834e-05,
      "loss": 5.7505,
      "step": 13000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.5210727969348656e-05,
      "loss": 5.5657,
      "step": 13500
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.503334752376898e-05,
      "loss": 5.4052,
      "step": 14000
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.48559670781893e-05,
      "loss": 5.2834,
      "step": 14500
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.4678586632609624e-05,
      "loss": 5.1588,
      "step": 15000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.450120618702994e-05,
      "loss": 5.0617,
      "step": 15500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.432382574145026e-05,
      "loss": 4.9254,
      "step": 16000
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.4146445295870585e-05,
      "loss": 4.7733,
      "step": 16500
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.396906485029091e-05,
      "loss": 4.6491,
      "step": 17000
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.379168440471122e-05,
      "loss": 4.5447,
      "step": 17500
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.3614303959131546e-05,
      "loss": 4.4222,
      "step": 18000
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.343692351355187e-05,
      "loss": 4.3276,
      "step": 18500
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.325954306797219e-05,
      "loss": 4.2297,
      "step": 19000
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.3082162622392507e-05,
      "loss": 4.169,
      "step": 19500
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.290478217681283e-05,
      "loss": 4.1032,
      "step": 20000
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.272740173123315e-05,
      "loss": 4.0278,
      "step": 20500
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.2550021285653474e-05,
      "loss": 3.9557,
      "step": 21000
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.237264084007379e-05,
      "loss": 3.8852,
      "step": 21500
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.219526039449411e-05,
      "loss": 3.8525,
      "step": 22000
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.2017879948914435e-05,
      "loss": 3.8033,
      "step": 22500
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.184049950333476e-05,
      "loss": 3.7525,
      "step": 23000
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.166311905775507e-05,
      "loss": 3.7301,
      "step": 23500
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.1485738612175396e-05,
      "loss": 3.6619,
      "step": 24000
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.130835816659572e-05,
      "loss": 3.6457,
      "step": 24500
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.113097772101604e-05,
      "loss": 3.6079,
      "step": 25000
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.0953597275436357e-05,
      "loss": 3.569,
      "step": 25500
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.077621682985668e-05,
      "loss": 3.5467,
      "step": 26000
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.0598836384277e-05,
      "loss": 3.5302,
      "step": 26500
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.0421455938697324e-05,
      "loss": 3.4952,
      "step": 27000
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.024407549311764e-05,
      "loss": 3.4743,
      "step": 27500
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.006669504753796e-05,
      "loss": 3.4558,
      "step": 28000
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.9889314601958285e-05,
      "loss": 3.4161,
      "step": 28500
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.971193415637861e-05,
      "loss": 3.3899,
      "step": 29000
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.953455371079892e-05,
      "loss": 3.3836,
      "step": 29500
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.9357173265219246e-05,
      "loss": 3.3521,
      "step": 30000
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.917979281963957e-05,
      "loss": 3.3418,
      "step": 30500
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.900241237405989e-05,
      "loss": 3.3161,
      "step": 31000
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.8825031928480207e-05,
      "loss": 3.3091,
      "step": 31500
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.864765148290053e-05,
      "loss": 3.2917,
      "step": 32000
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.847027103732085e-05,
      "loss": 3.268,
      "step": 32500
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.829289059174117e-05,
      "loss": 3.2515,
      "step": 33000
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.811551014616149e-05,
      "loss": 3.2415,
      "step": 33500
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.793812970058181e-05,
      "loss": 3.2177,
      "step": 34000
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.776074925500213e-05,
      "loss": 3.2102,
      "step": 34500
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.758336880942245e-05,
      "loss": 3.2186,
      "step": 35000
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.740598836384277e-05,
      "loss": 3.1951,
      "step": 35500
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.722860791826309e-05,
      "loss": 3.1766,
      "step": 36000
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.705122747268341e-05,
      "loss": 3.1603,
      "step": 36500
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.6873847027103734e-05,
      "loss": 3.1269,
      "step": 37000
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.669646658152405e-05,
      "loss": 3.1431,
      "step": 37500
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.651908613594437e-05,
      "loss": 3.112,
      "step": 38000
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.6341705690364695e-05,
      "loss": 3.1211,
      "step": 38500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.616432524478502e-05,
      "loss": 3.0896,
      "step": 39000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.598694479920533e-05,
      "loss": 3.0972,
      "step": 39500
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.5809564353625656e-05,
      "loss": 3.0821,
      "step": 40000
    },
    {
      "epoch": 2.87,
      "learning_rate": 3.563218390804598e-05,
      "loss": 3.0626,
      "step": 40500
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.54548034624663e-05,
      "loss": 3.0586,
      "step": 41000
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.5277423016886617e-05,
      "loss": 3.0464,
      "step": 41500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.510004257130694e-05,
      "loss": 3.0368,
      "step": 42000
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.492266212572726e-05,
      "loss": 3.0282,
      "step": 42500
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.4745281680147584e-05,
      "loss": 2.9903,
      "step": 43000
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.45679012345679e-05,
      "loss": 2.9912,
      "step": 43500
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.439052078898822e-05,
      "loss": 2.9902,
      "step": 44000
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.4213140343408545e-05,
      "loss": 2.995,
      "step": 44500
    },
    {
      "epoch": 3.19,
      "learning_rate": 3.403575989782887e-05,
      "loss": 2.9736,
      "step": 45000
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.385837945224918e-05,
      "loss": 2.9724,
      "step": 45500
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.3680999006669506e-05,
      "loss": 2.9636,
      "step": 46000
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.350361856108983e-05,
      "loss": 2.957,
      "step": 46500
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.332623811551015e-05,
      "loss": 2.9563,
      "step": 47000
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.3148857669930467e-05,
      "loss": 2.9381,
      "step": 47500
    },
    {
      "epoch": 3.41,
      "learning_rate": 3.297147722435079e-05,
      "loss": 2.926,
      "step": 48000
    },
    {
      "epoch": 3.44,
      "learning_rate": 3.279409677877111e-05,
      "loss": 2.933,
      "step": 48500
    },
    {
      "epoch": 3.48,
      "learning_rate": 3.2616716333191434e-05,
      "loss": 2.9144,
      "step": 49000
    },
    {
      "epoch": 3.51,
      "learning_rate": 3.243933588761175e-05,
      "loss": 2.9084,
      "step": 49500
    },
    {
      "epoch": 3.55,
      "learning_rate": 3.226195544203207e-05,
      "loss": 2.9116,
      "step": 50000
    },
    {
      "epoch": 3.58,
      "learning_rate": 3.2084574996452395e-05,
      "loss": 2.8892,
      "step": 50500
    },
    {
      "epoch": 3.62,
      "learning_rate": 3.190719455087272e-05,
      "loss": 2.8828,
      "step": 51000
    },
    {
      "epoch": 3.65,
      "learning_rate": 3.172981410529303e-05,
      "loss": 2.8843,
      "step": 51500
    },
    {
      "epoch": 3.69,
      "learning_rate": 3.1552433659713356e-05,
      "loss": 2.8799,
      "step": 52000
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.137505321413368e-05,
      "loss": 2.8586,
      "step": 52500
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.1197672768554e-05,
      "loss": 2.8601,
      "step": 53000
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.1020292322974317e-05,
      "loss": 2.8523,
      "step": 53500
    },
    {
      "epoch": 3.83,
      "learning_rate": 3.084291187739464e-05,
      "loss": 2.8517,
      "step": 54000
    },
    {
      "epoch": 3.87,
      "learning_rate": 3.066553143181496e-05,
      "loss": 2.8347,
      "step": 54500
    },
    {
      "epoch": 3.9,
      "learning_rate": 3.048815098623528e-05,
      "loss": 2.8427,
      "step": 55000
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.0310770540655597e-05,
      "loss": 2.8471,
      "step": 55500
    },
    {
      "epoch": 3.97,
      "learning_rate": 3.013339009507592e-05,
      "loss": 2.8201,
      "step": 56000
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.995600964949624e-05,
      "loss": 2.8236,
      "step": 56500
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.9778629203916564e-05,
      "loss": 2.8043,
      "step": 57000
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.960124875833688e-05,
      "loss": 2.7957,
      "step": 57500
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.9423868312757202e-05,
      "loss": 2.8042,
      "step": 58000
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.9246487867177525e-05,
      "loss": 2.7836,
      "step": 58500
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.9069107421597847e-05,
      "loss": 2.7875,
      "step": 59000
    },
    {
      "epoch": 4.22,
      "learning_rate": 2.8891726976018163e-05,
      "loss": 2.7737,
      "step": 59500
    },
    {
      "epoch": 4.26,
      "learning_rate": 2.8714346530438486e-05,
      "loss": 2.7783,
      "step": 60000
    },
    {
      "epoch": 4.29,
      "learning_rate": 2.8536966084858808e-05,
      "loss": 2.7689,
      "step": 60500
    },
    {
      "epoch": 4.33,
      "learning_rate": 2.835958563927913e-05,
      "loss": 2.7689,
      "step": 61000
    },
    {
      "epoch": 4.36,
      "learning_rate": 2.8182205193699447e-05,
      "loss": 2.7737,
      "step": 61500
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.800482474811977e-05,
      "loss": 2.7595,
      "step": 62000
    },
    {
      "epoch": 4.43,
      "learning_rate": 2.7827444302540088e-05,
      "loss": 2.7657,
      "step": 62500
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.765006385696041e-05,
      "loss": 2.7664,
      "step": 63000
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.747268341138073e-05,
      "loss": 2.7394,
      "step": 63500
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.729530296580105e-05,
      "loss": 2.7357,
      "step": 64000
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.711792252022137e-05,
      "loss": 2.7346,
      "step": 64500
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.6940542074641694e-05,
      "loss": 2.7351,
      "step": 65000
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.676316162906201e-05,
      "loss": 2.7268,
      "step": 65500
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.6585781183482332e-05,
      "loss": 2.7319,
      "step": 66000
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.6408400737902655e-05,
      "loss": 2.7282,
      "step": 66500
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.6231020292322977e-05,
      "loss": 2.7297,
      "step": 67000
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.6053639846743293e-05,
      "loss": 2.7112,
      "step": 67500
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.5876259401163616e-05,
      "loss": 2.7315,
      "step": 68000
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.5698878955583938e-05,
      "loss": 2.704,
      "step": 68500
    },
    {
      "epoch": 4.9,
      "learning_rate": 2.552149851000426e-05,
      "loss": 2.7036,
      "step": 69000
    },
    {
      "epoch": 4.93,
      "learning_rate": 2.5344118064424577e-05,
      "loss": 2.7103,
      "step": 69500
    },
    {
      "epoch": 4.97,
      "learning_rate": 2.51667376188449e-05,
      "loss": 2.7022,
      "step": 70000
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.498935717326522e-05,
      "loss": 2.6949,
      "step": 70500
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.481197672768554e-05,
      "loss": 2.681,
      "step": 71000
    },
    {
      "epoch": 5.07,
      "learning_rate": 2.4634596282105863e-05,
      "loss": 2.6636,
      "step": 71500
    },
    {
      "epoch": 5.11,
      "learning_rate": 2.4457215836526182e-05,
      "loss": 2.6868,
      "step": 72000
    },
    {
      "epoch": 5.14,
      "learning_rate": 2.4279835390946505e-05,
      "loss": 2.6683,
      "step": 72500
    },
    {
      "epoch": 5.18,
      "learning_rate": 2.4102454945366824e-05,
      "loss": 2.6765,
      "step": 73000
    },
    {
      "epoch": 5.21,
      "learning_rate": 2.3925074499787147e-05,
      "loss": 2.6665,
      "step": 73500
    },
    {
      "epoch": 5.25,
      "learning_rate": 2.3747694054207466e-05,
      "loss": 2.6672,
      "step": 74000
    },
    {
      "epoch": 5.29,
      "learning_rate": 2.3570313608627785e-05,
      "loss": 2.6604,
      "step": 74500
    },
    {
      "epoch": 5.32,
      "learning_rate": 2.3392933163048107e-05,
      "loss": 2.6552,
      "step": 75000
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.3215552717468427e-05,
      "loss": 2.6561,
      "step": 75500
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.3038172271888746e-05,
      "loss": 2.6418,
      "step": 76000
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.2860791826309068e-05,
      "loss": 2.6552,
      "step": 76500
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.2683411380729387e-05,
      "loss": 2.6404,
      "step": 77000
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.250603093514971e-05,
      "loss": 2.6317,
      "step": 77500
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.232865048957003e-05,
      "loss": 2.6343,
      "step": 78000
    },
    {
      "epoch": 5.57,
      "learning_rate": 2.215127004399035e-05,
      "loss": 2.6446,
      "step": 78500
    },
    {
      "epoch": 5.61,
      "learning_rate": 2.197388959841067e-05,
      "loss": 2.6201,
      "step": 79000
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.1796509152830993e-05,
      "loss": 2.6266,
      "step": 79500
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.1619128707251312e-05,
      "loss": 2.6229,
      "step": 80000
    },
    {
      "epoch": 5.71,
      "learning_rate": 2.1441748261671635e-05,
      "loss": 2.6297,
      "step": 80500
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.1264367816091954e-05,
      "loss": 2.6183,
      "step": 81000
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.1086987370512277e-05,
      "loss": 2.6183,
      "step": 81500
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.0909606924932596e-05,
      "loss": 2.6164,
      "step": 82000
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.073222647935292e-05,
      "loss": 2.6069,
      "step": 82500
    },
    {
      "epoch": 5.89,
      "learning_rate": 2.0554846033773237e-05,
      "loss": 2.6219,
      "step": 83000
    },
    {
      "epoch": 5.92,
      "learning_rate": 2.037746558819356e-05,
      "loss": 2.592,
      "step": 83500
    },
    {
      "epoch": 5.96,
      "learning_rate": 2.020008514261388e-05,
      "loss": 2.5975,
      "step": 84000
    },
    {
      "epoch": 6.0,
      "learning_rate": 2.00227046970342e-05,
      "loss": 2.6029,
      "step": 84500
    },
    {
      "epoch": 6.03,
      "learning_rate": 1.984532425145452e-05,
      "loss": 2.5913,
      "step": 85000
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.9667943805874843e-05,
      "loss": 2.5867,
      "step": 85500
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.9490563360295162e-05,
      "loss": 2.5899,
      "step": 86000
    },
    {
      "epoch": 6.14,
      "learning_rate": 1.931318291471548e-05,
      "loss": 2.5923,
      "step": 86500
    },
    {
      "epoch": 6.17,
      "learning_rate": 1.91358024691358e-05,
      "loss": 2.5742,
      "step": 87000
    },
    {
      "epoch": 6.21,
      "learning_rate": 1.8958422023556123e-05,
      "loss": 2.5859,
      "step": 87500
    },
    {
      "epoch": 6.24,
      "learning_rate": 1.8781041577976442e-05,
      "loss": 2.5683,
      "step": 88000
    },
    {
      "epoch": 6.28,
      "learning_rate": 1.8603661132396765e-05,
      "loss": 2.5808,
      "step": 88500
    },
    {
      "epoch": 6.31,
      "learning_rate": 1.8426280686817084e-05,
      "loss": 2.5663,
      "step": 89000
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.8248900241237407e-05,
      "loss": 2.5677,
      "step": 89500
    },
    {
      "epoch": 6.39,
      "learning_rate": 1.8071519795657726e-05,
      "loss": 2.5647,
      "step": 90000
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.7894139350078048e-05,
      "loss": 2.5656,
      "step": 90500
    },
    {
      "epoch": 6.46,
      "learning_rate": 1.7716758904498367e-05,
      "loss": 2.5682,
      "step": 91000
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.753937845891869e-05,
      "loss": 2.5645,
      "step": 91500
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.736199801333901e-05,
      "loss": 2.5586,
      "step": 92000
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.718461756775933e-05,
      "loss": 2.5488,
      "step": 92500
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.700723712217965e-05,
      "loss": 2.5547,
      "step": 93000
    },
    {
      "epoch": 6.63,
      "learning_rate": 1.6829856676599973e-05,
      "loss": 2.5485,
      "step": 93500
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.6652476231020292e-05,
      "loss": 2.5503,
      "step": 94000
    },
    {
      "epoch": 6.7,
      "learning_rate": 1.6475095785440615e-05,
      "loss": 2.547,
      "step": 94500
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.6297715339860934e-05,
      "loss": 2.5468,
      "step": 95000
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.6120334894281257e-05,
      "loss": 2.5416,
      "step": 95500
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.5942954448701576e-05,
      "loss": 2.5562,
      "step": 96000
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.57655740031219e-05,
      "loss": 2.5311,
      "step": 96500
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.5588193557542217e-05,
      "loss": 2.5481,
      "step": 97000
    },
    {
      "epoch": 6.92,
      "learning_rate": 1.5410813111962537e-05,
      "loss": 2.5362,
      "step": 97500
    },
    {
      "epoch": 6.95,
      "learning_rate": 1.5233432666382857e-05,
      "loss": 2.5161,
      "step": 98000
    },
    {
      "epoch": 6.99,
      "learning_rate": 1.505605222080318e-05,
      "loss": 2.5354,
      "step": 98500
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.4878671775223499e-05,
      "loss": 2.5214,
      "step": 99000
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.4701291329643822e-05,
      "loss": 2.5271,
      "step": 99500
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.452391088406414e-05,
      "loss": 2.5024,
      "step": 100000
    },
    {
      "epoch": 7.13,
      "learning_rate": 1.4346530438484462e-05,
      "loss": 2.5096,
      "step": 100500
    },
    {
      "epoch": 7.17,
      "learning_rate": 1.4169149992904782e-05,
      "loss": 2.5094,
      "step": 101000
    },
    {
      "epoch": 7.2,
      "learning_rate": 1.3991769547325103e-05,
      "loss": 2.5151,
      "step": 101500
    },
    {
      "epoch": 7.24,
      "learning_rate": 1.3814389101745422e-05,
      "loss": 2.5141,
      "step": 102000
    },
    {
      "epoch": 7.27,
      "learning_rate": 1.3637008656165745e-05,
      "loss": 2.5102,
      "step": 102500
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.3459628210586064e-05,
      "loss": 2.5091,
      "step": 103000
    },
    {
      "epoch": 7.34,
      "learning_rate": 1.3282247765006387e-05,
      "loss": 2.5025,
      "step": 103500
    },
    {
      "epoch": 7.38,
      "learning_rate": 1.3104867319426706e-05,
      "loss": 2.5025,
      "step": 104000
    },
    {
      "epoch": 7.41,
      "learning_rate": 1.2927486873847028e-05,
      "loss": 2.5051,
      "step": 104500
    },
    {
      "epoch": 7.45,
      "learning_rate": 1.2750106428267347e-05,
      "loss": 2.5084,
      "step": 105000
    },
    {
      "epoch": 7.49,
      "learning_rate": 1.257272598268767e-05,
      "loss": 2.4955,
      "step": 105500
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.2395345537107989e-05,
      "loss": 2.4915,
      "step": 106000
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.221796509152831e-05,
      "loss": 2.5053,
      "step": 106500
    },
    {
      "epoch": 7.59,
      "learning_rate": 1.204058464594863e-05,
      "loss": 2.4989,
      "step": 107000
    },
    {
      "epoch": 7.63,
      "learning_rate": 1.1863204200368952e-05,
      "loss": 2.4957,
      "step": 107500
    },
    {
      "epoch": 7.66,
      "learning_rate": 1.1685823754789272e-05,
      "loss": 2.4909,
      "step": 108000
    },
    {
      "epoch": 7.7,
      "learning_rate": 1.1508443309209593e-05,
      "loss": 2.492,
      "step": 108500
    },
    {
      "epoch": 7.73,
      "learning_rate": 1.1331062863629914e-05,
      "loss": 2.4929,
      "step": 109000
    },
    {
      "epoch": 7.77,
      "learning_rate": 1.1153682418050235e-05,
      "loss": 2.4871,
      "step": 109500
    },
    {
      "epoch": 7.8,
      "learning_rate": 1.0976301972470556e-05,
      "loss": 2.4843,
      "step": 110000
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.0798921526890877e-05,
      "loss": 2.4849,
      "step": 110500
    },
    {
      "epoch": 7.88,
      "learning_rate": 1.0621541081311197e-05,
      "loss": 2.4903,
      "step": 111000
    },
    {
      "epoch": 7.91,
      "learning_rate": 1.0444160635731518e-05,
      "loss": 2.4766,
      "step": 111500
    },
    {
      "epoch": 7.95,
      "learning_rate": 1.0266780190151837e-05,
      "loss": 2.4895,
      "step": 112000
    },
    {
      "epoch": 7.98,
      "learning_rate": 1.0089399744572158e-05,
      "loss": 2.4818,
      "step": 112500
    },
    {
      "epoch": 8.02,
      "learning_rate": 9.912019298992479e-06,
      "loss": 2.4782,
      "step": 113000
    },
    {
      "epoch": 8.05,
      "learning_rate": 9.7346388534128e-06,
      "loss": 2.4671,
      "step": 113500
    },
    {
      "epoch": 8.09,
      "learning_rate": 9.55725840783312e-06,
      "loss": 2.4665,
      "step": 114000
    },
    {
      "epoch": 8.12,
      "learning_rate": 9.379877962253442e-06,
      "loss": 2.4669,
      "step": 114500
    },
    {
      "epoch": 8.16,
      "learning_rate": 9.202497516673762e-06,
      "loss": 2.4552,
      "step": 115000
    },
    {
      "epoch": 8.19,
      "learning_rate": 9.025117071094083e-06,
      "loss": 2.459,
      "step": 115500
    },
    {
      "epoch": 8.23,
      "learning_rate": 8.847736625514404e-06,
      "loss": 2.455,
      "step": 116000
    },
    {
      "epoch": 8.27,
      "learning_rate": 8.670356179934725e-06,
      "loss": 2.472,
      "step": 116500
    },
    {
      "epoch": 8.3,
      "learning_rate": 8.492975734355046e-06,
      "loss": 2.4517,
      "step": 117000
    },
    {
      "epoch": 8.34,
      "learning_rate": 8.315595288775365e-06,
      "loss": 2.458,
      "step": 117500
    },
    {
      "epoch": 8.37,
      "learning_rate": 8.138214843195686e-06,
      "loss": 2.4597,
      "step": 118000
    },
    {
      "epoch": 8.41,
      "learning_rate": 7.960834397616007e-06,
      "loss": 2.4567,
      "step": 118500
    },
    {
      "epoch": 8.44,
      "learning_rate": 7.783453952036327e-06,
      "loss": 2.4561,
      "step": 119000
    },
    {
      "epoch": 8.48,
      "learning_rate": 7.606073506456648e-06,
      "loss": 2.4614,
      "step": 119500
    },
    {
      "epoch": 8.51,
      "learning_rate": 7.428693060876969e-06,
      "loss": 2.4592,
      "step": 120000
    },
    {
      "epoch": 8.55,
      "learning_rate": 7.25131261529729e-06,
      "loss": 2.456,
      "step": 120500
    },
    {
      "epoch": 8.59,
      "learning_rate": 7.073932169717611e-06,
      "loss": 2.4593,
      "step": 121000
    },
    {
      "epoch": 8.62,
      "learning_rate": 6.896551724137932e-06,
      "loss": 2.452,
      "step": 121500
    },
    {
      "epoch": 8.66,
      "learning_rate": 6.719171278558252e-06,
      "loss": 2.4414,
      "step": 122000
    },
    {
      "epoch": 8.69,
      "learning_rate": 6.5417908329785725e-06,
      "loss": 2.4558,
      "step": 122500
    },
    {
      "epoch": 8.73,
      "learning_rate": 6.364410387398893e-06,
      "loss": 2.4417,
      "step": 123000
    },
    {
      "epoch": 8.76,
      "learning_rate": 6.187029941819214e-06,
      "loss": 2.4364,
      "step": 123500
    },
    {
      "epoch": 8.8,
      "learning_rate": 6.009649496239535e-06,
      "loss": 2.4543,
      "step": 124000
    },
    {
      "epoch": 8.83,
      "learning_rate": 5.832269050659856e-06,
      "loss": 2.4419,
      "step": 124500
    },
    {
      "epoch": 8.87,
      "learning_rate": 5.654888605080176e-06,
      "loss": 2.446,
      "step": 125000
    },
    {
      "epoch": 8.9,
      "learning_rate": 5.477508159500497e-06,
      "loss": 2.4395,
      "step": 125500
    },
    {
      "epoch": 8.94,
      "learning_rate": 5.3001277139208175e-06,
      "loss": 2.4466,
      "step": 126000
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.122747268341138e-06,
      "loss": 2.4301,
      "step": 126500
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.945366822761459e-06,
      "loss": 2.4382,
      "step": 127000
    },
    {
      "epoch": 9.05,
      "learning_rate": 4.767986377181779e-06,
      "loss": 2.4406,
      "step": 127500
    },
    {
      "epoch": 9.08,
      "learning_rate": 4.5906059316021e-06,
      "loss": 2.4312,
      "step": 128000
    },
    {
      "epoch": 9.12,
      "learning_rate": 4.413225486022421e-06,
      "loss": 2.4409,
      "step": 128500
    },
    {
      "epoch": 9.15,
      "learning_rate": 4.235845040442742e-06,
      "loss": 2.4256,
      "step": 129000
    },
    {
      "epoch": 9.19,
      "learning_rate": 4.0584645948630625e-06,
      "loss": 2.4301,
      "step": 129500
    },
    {
      "epoch": 9.22,
      "learning_rate": 3.881084149283383e-06,
      "loss": 2.4338,
      "step": 130000
    },
    {
      "epoch": 9.26,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 2.4367,
      "step": 130500
    },
    {
      "epoch": 9.29,
      "learning_rate": 3.5263232581240246e-06,
      "loss": 2.4198,
      "step": 131000
    },
    {
      "epoch": 9.33,
      "learning_rate": 3.348942812544345e-06,
      "loss": 2.4228,
      "step": 131500
    },
    {
      "epoch": 9.37,
      "learning_rate": 3.171562366964666e-06,
      "loss": 2.4257,
      "step": 132000
    },
    {
      "epoch": 9.4,
      "learning_rate": 2.9941819213849867e-06,
      "loss": 2.4286,
      "step": 132500
    },
    {
      "epoch": 9.44,
      "learning_rate": 2.8168014758053075e-06,
      "loss": 2.4153,
      "step": 133000
    },
    {
      "epoch": 9.47,
      "learning_rate": 2.6394210302256283e-06,
      "loss": 2.4235,
      "step": 133500
    },
    {
      "epoch": 9.51,
      "learning_rate": 2.4620405846459487e-06,
      "loss": 2.4243,
      "step": 134000
    },
    {
      "epoch": 9.54,
      "learning_rate": 2.2846601390662696e-06,
      "loss": 2.4235,
      "step": 134500
    },
    {
      "epoch": 9.58,
      "learning_rate": 2.1072796934865904e-06,
      "loss": 2.4297,
      "step": 135000
    },
    {
      "epoch": 9.61,
      "learning_rate": 1.929899247906911e-06,
      "loss": 2.4356,
      "step": 135500
    },
    {
      "epoch": 9.65,
      "learning_rate": 1.7525188023272317e-06,
      "loss": 2.4223,
      "step": 136000
    },
    {
      "epoch": 9.68,
      "learning_rate": 1.5751383567475523e-06,
      "loss": 2.4218,
      "step": 136500
    },
    {
      "epoch": 9.72,
      "learning_rate": 1.397757911167873e-06,
      "loss": 2.4211,
      "step": 137000
    },
    {
      "epoch": 9.76,
      "learning_rate": 1.2203774655881937e-06,
      "loss": 2.4233,
      "step": 137500
    },
    {
      "epoch": 9.79,
      "learning_rate": 1.0429970200085144e-06,
      "loss": 2.4107,
      "step": 138000
    },
    {
      "epoch": 9.83,
      "learning_rate": 8.65616574428835e-07,
      "loss": 2.4176,
      "step": 138500
    },
    {
      "epoch": 9.86,
      "learning_rate": 6.882361288491557e-07,
      "loss": 2.4227,
      "step": 139000
    },
    {
      "epoch": 9.9,
      "learning_rate": 5.108556832694765e-07,
      "loss": 2.4229,
      "step": 139500
    },
    {
      "epoch": 9.93,
      "learning_rate": 3.334752376897971e-07,
      "loss": 2.4089,
      "step": 140000
    }
  ],
  "max_steps": 140940,
  "num_train_epochs": 10,
  "total_flos": 690009058341027840,
  "trial_name": null,
  "trial_params": null
}