trainer_state.json

{
  "best_metric": 0.0008013004553504288,
  "best_model_checkpoint": "data/further_pretraining/bert-large-uncased/checkpoint-4310",
  "epoch": 2.0,
  "global_step": 4310,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.23,
      "learning_rate": 1.9535962877030165e-05,
      "loss": 0.0601,
      "step": 500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9071925754060325e-05,
      "loss": 0.0044,
      "step": 1000
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.860788863109049e-05,
      "loss": 0.0018,
      "step": 1500
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.8143851508120652e-05,
      "loss": 0.0012,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.004205523990094662,
      "eval_runtime": 396.7144,
      "eval_samples_per_second": 173.775,
      "eval_steps_per_second": 5.432,
      "step": 2155
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.7679814385150812e-05,
      "loss": 0.0028,
      "step": 2500
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.7215777262180976e-05,
      "loss": 0.0008,
      "step": 3000
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.675174013921114e-05,
      "loss": 0.0018,
      "step": 3500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.62877030162413e-05,
      "loss": 0.0028,
      "step": 4000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.0008013004553504288,
      "eval_runtime": 393.1202,
      "eval_samples_per_second": 175.364,
      "eval_steps_per_second": 5.482,
      "step": 4310
    }
  ],
  "max_steps": 21550,
  "num_train_epochs": 10,
  "total_flos": 3.2126439900312576e+16,
  "trial_name": null,
  "trial_params": null
}