{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 21960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 2.9316939890710385e-05,
      "loss": 2.3439,
      "step": 500
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.8633879781420765e-05,
      "loss": 1.6579,
      "step": 1000
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.795081967213115e-05,
      "loss": 1.5036,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.7267759562841533e-05,
      "loss": 1.3947,
      "step": 2000
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.6584699453551913e-05,
      "loss": 1.3372,
      "step": 2500
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.5901639344262294e-05,
      "loss": 1.295,
      "step": 3000
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.5218579234972678e-05,
      "loss": 1.2527,
      "step": 3500
    },
    {
      "epoch": 0.36,
      "learning_rate": 2.453551912568306e-05,
      "loss": 1.2545,
      "step": 4000
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.3852459016393442e-05,
      "loss": 1.1842,
      "step": 4500
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.3169398907103826e-05,
      "loss": 1.168,
      "step": 5000
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.248633879781421e-05,
      "loss": 1.1353,
      "step": 5500
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.180327868852459e-05,
      "loss": 1.1211,
      "step": 6000
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.1120218579234974e-05,
      "loss": 1.1117,
      "step": 6500
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.0437158469945358e-05,
      "loss": 1.0815,
      "step": 7000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.975409836065574e-05,
      "loss": 1.0361,
      "step": 7500
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.907103825136612e-05,
      "loss": 1.0426,
      "step": 8000
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.8387978142076503e-05,
      "loss": 1.053,
      "step": 8500
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7704918032786887e-05,
      "loss": 1.0328,
      "step": 9000
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.7021857923497267e-05,
      "loss": 1.0315,
      "step": 9500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.633879781420765e-05,
      "loss": 0.9997,
      "step": 10000
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.5655737704918035e-05,
      "loss": 1.0027,
      "step": 10500
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.4972677595628415e-05,
      "loss": 0.9692,
      "step": 11000
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.4289617486338798e-05,
      "loss": 0.7429,
      "step": 11500
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.3606557377049181e-05,
      "loss": 0.7197,
      "step": 12000
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2923497267759564e-05,
      "loss": 0.7079,
      "step": 12500
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.2240437158469946e-05,
      "loss": 0.7309,
      "step": 13000
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.1557377049180328e-05,
      "loss": 0.7045,
      "step": 13500
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.087431693989071e-05,
      "loss": 0.6833,
      "step": 14000
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.0191256830601094e-05,
      "loss": 0.7179,
      "step": 14500
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.508196721311476e-06,
      "loss": 0.7007,
      "step": 15000
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.825136612021857e-06,
      "loss": 0.6918,
      "step": 15500
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.14207650273224e-06,
      "loss": 0.6913,
      "step": 16000
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.459016393442623e-06,
      "loss": 0.6691,
      "step": 16500
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.775956284153006e-06,
      "loss": 0.6671,
      "step": 17000
    },
    {
      "epoch": 1.59,
      "learning_rate": 6.092896174863388e-06,
      "loss": 0.6931,
      "step": 17500
    },
    {
      "epoch": 1.64,
      "learning_rate": 5.409836065573771e-06,
      "loss": 0.6655,
      "step": 18000
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.726775956284153e-06,
      "loss": 0.6734,
      "step": 18500
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.043715846994535e-06,
      "loss": 0.6529,
      "step": 19000
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.3606557377049183e-06,
      "loss": 0.6425,
      "step": 19500
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.6775956284153005e-06,
      "loss": 0.6805,
      "step": 20000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.994535519125683e-06,
      "loss": 0.6551,
      "step": 20500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.3114754098360657e-06,
      "loss": 0.6549,
      "step": 21000
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.284153005464482e-07,
      "loss": 0.6653,
      "step": 21500
    },
    {
      "epoch": 2.0,
      "step": 21960,
      "total_flos": 5.164033933049242e+16,
      "train_loss": 0.9565104889305128,
      "train_runtime": 5798.4703,
      "train_samples_per_second": 45.444,
      "train_steps_per_second": 3.787
    }
  ],
  "max_steps": 21960,
  "num_train_epochs": 2,
  "total_flos": 5.164033933049242e+16,
  "trial_name": null,
  "trial_params": null
}