{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 42985,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 1.976736070722345e-05,
      "loss": 0.6539,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.95347214144469e-05,
      "loss": 0.503,
      "step": 1000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.930208212167035e-05,
      "loss": 0.4624,
      "step": 1500
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9069442828893803e-05,
      "loss": 0.4344,
      "step": 2000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.8836803536117253e-05,
      "loss": 0.4245,
      "step": 2500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.8604164243340703e-05,
      "loss": 0.4102,
      "step": 3000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.8371524950564152e-05,
      "loss": 0.4059,
      "step": 3500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.8138885657787602e-05,
      "loss": 0.3898,
      "step": 4000
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.790624636501105e-05,
      "loss": 0.3791,
      "step": 4500
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.76736070722345e-05,
      "loss": 0.3763,
      "step": 5000
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.744096777945795e-05,
      "loss": 0.3773,
      "step": 5500
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7208328486681404e-05,
      "loss": 0.3731,
      "step": 6000
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.6975689193904853e-05,
      "loss": 0.36,
      "step": 6500
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.67430499011283e-05,
      "loss": 0.358,
      "step": 7000
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6510410608351752e-05,
      "loss": 0.356,
      "step": 7500
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6277771315575202e-05,
      "loss": 0.3576,
      "step": 8000
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.604513202279865e-05,
      "loss": 0.3501,
      "step": 8500
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.58124927300221e-05,
      "loss": 0.3034,
      "step": 9000
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.557985343724555e-05,
      "loss": 0.2965,
      "step": 9500
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.5347214144469004e-05,
      "loss": 0.2923,
      "step": 10000
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.5114574851692452e-05,
      "loss": 0.3006,
      "step": 10500
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.4881935558915903e-05,
      "loss": 0.2882,
      "step": 11000
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.4649296266139352e-05,
      "loss": 0.2911,
      "step": 11500
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.4416656973362802e-05,
      "loss": 0.2882,
      "step": 12000
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.4184017680586253e-05,
      "loss": 0.2886,
      "step": 12500
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.3951378387809701e-05,
      "loss": 0.2867,
      "step": 13000
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.371873909503315e-05,
      "loss": 0.2905,
      "step": 13500
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.3486099802256602e-05,
      "loss": 0.2897,
      "step": 14000
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.3253460509480052e-05,
      "loss": 0.2852,
      "step": 14500
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.3020821216703503e-05,
      "loss": 0.2893,
      "step": 15000
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.2788181923926953e-05,
      "loss": 0.2905,
      "step": 15500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.2555542631150402e-05,
      "loss": 0.288,
      "step": 16000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.2322903338373853e-05,
      "loss": 0.287,
      "step": 16500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.2090264045597301e-05,
      "loss": 0.285,
      "step": 17000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.1857624752820753e-05,
      "loss": 0.2427,
      "step": 17500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.1624985460044202e-05,
      "loss": 0.2328,
      "step": 18000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.1392346167267652e-05,
      "loss": 0.2206,
      "step": 18500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.1159706874491103e-05,
      "loss": 0.2262,
      "step": 19000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.0927067581714553e-05,
      "loss": 0.2358,
      "step": 19500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.0694428288938004e-05,
      "loss": 0.2221,
      "step": 20000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0461788996161452e-05,
      "loss": 0.2233,
      "step": 20500
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.0229149703384902e-05,
      "loss": 0.2266,
      "step": 21000
    },
    {
      "epoch": 2.5,
      "learning_rate": 9.996510410608353e-06,
      "loss": 0.2241,
      "step": 21500
    },
    {
      "epoch": 2.56,
      "learning_rate": 9.763871117831802e-06,
      "loss": 0.2305,
      "step": 22000
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.531231825055252e-06,
      "loss": 0.2242,
      "step": 22500
    },
    {
      "epoch": 2.68,
      "learning_rate": 9.298592532278703e-06,
      "loss": 0.2323,
      "step": 23000
    },
    {
      "epoch": 2.73,
      "learning_rate": 9.065953239502153e-06,
      "loss": 0.227,
      "step": 23500
    },
    {
      "epoch": 2.79,
      "learning_rate": 8.833313946725602e-06,
      "loss": 0.2313,
      "step": 24000
    },
    {
      "epoch": 2.85,
      "learning_rate": 8.600674653949052e-06,
      "loss": 0.2318,
      "step": 24500
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.368035361172503e-06,
      "loss": 0.2313,
      "step": 25000
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.135396068395953e-06,
      "loss": 0.2292,
      "step": 25500
    },
    {
      "epoch": 3.02,
      "learning_rate": 7.902756775619403e-06,
      "loss": 0.2063,
      "step": 26000
    },
    {
      "epoch": 3.08,
      "learning_rate": 7.670117482842852e-06,
      "loss": 0.1837,
      "step": 26500
    },
    {
      "epoch": 3.14,
      "learning_rate": 7.437478190066303e-06,
      "loss": 0.1811,
      "step": 27000
    },
    {
      "epoch": 3.2,
      "learning_rate": 7.204838897289753e-06,
      "loss": 0.1777,
      "step": 27500
    },
    {
      "epoch": 3.26,
      "learning_rate": 6.972199604513203e-06,
      "loss": 0.1847,
      "step": 28000
    },
    {
      "epoch": 3.32,
      "learning_rate": 6.739560311736652e-06,
      "loss": 0.1849,
      "step": 28500
    },
    {
      "epoch": 3.37,
      "learning_rate": 6.506921018960103e-06,
      "loss": 0.1811,
      "step": 29000
    },
    {
      "epoch": 3.43,
      "learning_rate": 6.274281726183553e-06,
      "loss": 0.1833,
      "step": 29500
    },
    {
      "epoch": 3.49,
      "learning_rate": 6.0416424334070035e-06,
      "loss": 0.1816,
      "step": 30000
    },
    {
      "epoch": 3.55,
      "learning_rate": 5.809003140630452e-06,
      "loss": 0.1754,
      "step": 30500
    },
    {
      "epoch": 3.61,
      "learning_rate": 5.576363847853903e-06,
      "loss": 0.1808,
      "step": 31000
    },
    {
      "epoch": 3.66,
      "learning_rate": 5.343724555077353e-06,
      "loss": 0.184,
      "step": 31500
    },
    {
      "epoch": 3.72,
      "learning_rate": 5.111085262300804e-06,
      "loss": 0.1812,
      "step": 32000
    },
    {
      "epoch": 3.78,
      "learning_rate": 4.878445969524253e-06,
      "loss": 0.1835,
      "step": 32500
    },
    {
      "epoch": 3.84,
      "learning_rate": 4.645806676747703e-06,
      "loss": 0.1797,
      "step": 33000
    },
    {
      "epoch": 3.9,
      "learning_rate": 4.413167383971153e-06,
      "loss": 0.1816,
      "step": 33500
    },
    {
      "epoch": 3.95,
      "learning_rate": 4.180528091194604e-06,
      "loss": 0.1766,
      "step": 34000
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.947888798418053e-06,
      "loss": 0.1744,
      "step": 34500
    },
    {
      "epoch": 4.07,
      "learning_rate": 3.7152495056415033e-06,
      "loss": 0.1536,
      "step": 35000
    },
    {
      "epoch": 4.13,
      "learning_rate": 3.482610212864953e-06,
      "loss": 0.1549,
      "step": 35500
    },
    {
      "epoch": 4.19,
      "learning_rate": 3.2499709200884033e-06,
      "loss": 0.1483,
      "step": 36000
    },
    {
      "epoch": 4.25,
      "learning_rate": 3.017331627311853e-06,
      "loss": 0.1529,
      "step": 36500
    },
    {
      "epoch": 4.3,
      "learning_rate": 2.7846923345353034e-06,
      "loss": 0.1519,
      "step": 37000
    },
    {
      "epoch": 4.36,
      "learning_rate": 2.552053041758753e-06,
      "loss": 0.1518,
      "step": 37500
    },
    {
      "epoch": 4.42,
      "learning_rate": 2.3194137489822034e-06,
      "loss": 0.1474,
      "step": 38000
    },
    {
      "epoch": 4.48,
      "learning_rate": 2.0867744562056534e-06,
      "loss": 0.1523,
      "step": 38500
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.8541351634291032e-06,
      "loss": 0.1491,
      "step": 39000
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.6214958706525532e-06,
      "loss": 0.1474,
      "step": 39500
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.3888565778760033e-06,
      "loss": 0.1469,
      "step": 40000
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.1562172850994535e-06,
      "loss": 0.1447,
      "step": 40500
    },
    {
      "epoch": 4.77,
      "learning_rate": 9.235779923229034e-07,
      "loss": 0.1515,
      "step": 41000
    },
    {
      "epoch": 4.83,
      "learning_rate": 6.909386995463534e-07,
      "loss": 0.1473,
      "step": 41500
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.5829940676980343e-07,
      "loss": 0.151,
      "step": 42000
    },
    {
      "epoch": 4.94,
      "learning_rate": 2.256601139932535e-07,
      "loss": 0.15,
      "step": 42500
    },
    {
      "epoch": 5.0,
      "step": 42985,
      "total_flos": 4.625905344087859e+17,
      "train_runtime": 34216.2349,
      "train_samples_per_second": 1.256
    }
  ],
  "max_steps": 42985,
  "num_train_epochs": 5,
  "total_flos": 4.625905344087859e+17,
  "trial_name": null,
  "trial_params": null
}