trainer_state.json

{
  "best_metric": 2.8538482189178467,
  "best_model_checkpoint": "/scratch/skscla001/speech/results/mms-300m-nyagen-balanced-model/checkpoint-100",
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 181,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5524861878453039,
      "grad_norm": 1.579022765159607,
      "learning_rate": 0.000294,
      "loss": 7.0001,
      "step": 100
    },
    {
      "epoch": 0.5524861878453039,
      "eval_loss": 2.8538482189178467,
      "eval_runtime": 12.0623,
      "eval_samples_per_second": 14.011,
      "eval_steps_per_second": 3.565,
      "eval_wer": 1.0,
      "step": 100
    },
    {
      "epoch": 1.0,
      "step": 181,
      "total_flos": 1.3292347177359955e+17,
      "train_loss": 5.12707603833952,
      "train_runtime": 148.2598,
      "train_samples_per_second": 9.746,
      "train_steps_per_second": 1.221
    }
  ],
  "logging_steps": 100,
  "max_steps": 181,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 400,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.3292347177359955e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}