trainer_state.json

{
  "best_metric": 0.013692905195057392,
  "best_model_checkpoint": "cat_ds/checkpoint-52",
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 52,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.77,
      "learning_rate": 0.00016153846153846155,
      "loss": 0.4194,
      "step": 10
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.07365512102842331,
      "eval_runtime": 5.5846,
      "eval_samples_per_second": 4.298,
      "eval_steps_per_second": 0.537,
      "step": 13
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001230769230769231,
      "loss": 0.063,
      "step": 20
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.026600712910294533,
      "eval_runtime": 5.1625,
      "eval_samples_per_second": 4.649,
      "eval_steps_per_second": 0.581,
      "step": 26
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.461538461538461e-05,
      "loss": 0.051,
      "step": 30
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.015066702850162983,
      "eval_runtime": 4.9529,
      "eval_samples_per_second": 4.846,
      "eval_steps_per_second": 0.606,
      "step": 39
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.615384615384616e-05,
      "loss": 0.0308,
      "step": 40
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.014,
      "step": 50
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 1.0,
      "eval_loss": 0.013692905195057392,
      "eval_runtime": 4.8268,
      "eval_samples_per_second": 4.972,
      "eval_steps_per_second": 0.622,
      "step": 52
    },
    {
      "epoch": 4.0,
      "step": 52,
      "total_flos": 6.416336740088218e+16,
      "train_loss": 0.11166981543199374,
      "train_runtime": 294.988,
      "train_samples_per_second": 2.807,
      "train_steps_per_second": 0.176
    }
  ],
  "logging_steps": 10,
  "max_steps": 52,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 100,
  "total_flos": 6.416336740088218e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}