checkpoint-4000/trainer_state.json

{
  "best_global_step": 4000,
  "best_metric": 2.2951489639952594,
  "best_model_checkpoint": "./SALAMA_NEW3/checkpoint-4000",
  "epoch": 2.396644697423607,
  "eval_steps": 2000,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.1983223487118035,
      "grad_norm": 1.8670843839645386,
      "learning_rate": 7.572862694300519e-06,
      "loss": 0.1037,
      "step": 2000
    },
    {
      "epoch": 1.1983223487118035,
      "eval_loss": 0.05342990532517433,
      "eval_runtime": 6060.4714,
      "eval_samples_per_second": 2.203,
      "eval_steps_per_second": 0.275,
      "eval_wer": 4.51019657525849,
      "step": 2000
    },
    {
      "epoch": 2.396644697423607,
      "grad_norm": 1.0199387073516846,
      "learning_rate": 4.334520725388601e-06,
      "loss": 0.0334,
      "step": 4000
    },
    {
      "epoch": 2.396644697423607,
      "eval_loss": 0.03055572137236595,
      "eval_runtime": 6051.1495,
      "eval_samples_per_second": 2.206,
      "eval_steps_per_second": 0.276,
      "eval_wer": 2.2951489639952594,
      "step": 4000
    }
  ],
  "logging_steps": 2000,
  "max_steps": 6676,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.693431387455488e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}