trainer_state.json

{
  "best_metric": 44.25531914893617,
  "best_model_checkpoint": "OUTCOMESAI/whisper-small-medical/checkpoint-50",
  "epoch": 50.0,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 12.5,
      "grad_norm": 34.30704125300902,
      "learning_rate": 3.828878651016684e-07,
      "loss": 9.9956,
      "step": 25
    },
    {
      "epoch": 25.0,
      "grad_norm": 15.950854138638713,
      "learning_rate": 4.865337556154919e-07,
      "loss": 7.4337,
      "step": 50
    },
    {
      "epoch": 25.0,
      "eval_loss": 0.6201171875,
      "eval_runtime": 59.7245,
      "eval_samples_per_second": 0.318,
      "eval_steps_per_second": 0.033,
      "eval_wer": 44.25531914893617,
      "step": 50
    },
    {
      "epoch": 37.5,
      "grad_norm": 7.118798058764047,
      "learning_rate": 3.1e-07,
      "loss": 6.3944,
      "step": 75
    },
    {
      "epoch": 50.0,
      "grad_norm": 13.037787941820651,
      "learning_rate": 6e-08,
      "loss": 5.7447,
      "step": 100
    },
    {
      "epoch": 50.0,
      "eval_loss": 0.611328125,
      "eval_runtime": 57.9793,
      "eval_samples_per_second": 0.328,
      "eval_steps_per_second": 0.034,
      "eval_wer": 51.23404255319149,
      "step": 100
    },
    {
      "epoch": 50.0,
      "step": 100,
      "total_flos": 6405677165576192.0,
      "train_loss": 7.392109375,
      "train_runtime": 632.0978,
      "train_samples_per_second": 5.063,
      "train_steps_per_second": 0.158
    }
  ],
  "logging_steps": 25,
  "max_steps": 100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6405677165576192.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}