trainer_state.json

{
  "best_metric": 0.2531541883945465,
  "best_model_checkpoint": "geotrend-mdistill-bert-all-finetuned-toxicity-classification/checkpoint-1126",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1689,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.89,
      "learning_rate": 1.4079336885731204e-05,
      "loss": 0.4604,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.28258055448532104,
      "eval_runtime": 2.7146,
      "eval_samples_per_second": 184.188,
      "eval_steps_per_second": 23.208,
      "step": 563
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.158673771462404e-06,
      "loss": 0.2476,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.2531541883945465,
      "eval_runtime": 2.7258,
      "eval_samples_per_second": 183.43,
      "eval_steps_per_second": 23.112,
      "step": 1126
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.238010657193606e-06,
      "loss": 0.1753,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.34129810333251953,
      "eval_runtime": 2.695,
      "eval_samples_per_second": 185.527,
      "eval_steps_per_second": 23.376,
      "step": 1689
    }
  ],
  "logging_steps": 500,
  "max_steps": 1689,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1788309881856000.0,
  "trial_name": null,
  "trial_params": null
}