trainer_state.json

{
  "best_metric": 0.13481491804122925,
  "best_model_checkpoint": "./xls-r-1b-bem-msv/checkpoint-2000",
  "epoch": 4.994797086368366,
  "eval_steps": 500,
  "global_step": 2400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4162330905306972,
      "grad_norm": 11.17747974395752,
      "learning_rate": 1.97e-05,
      "loss": 4.9143,
      "step": 200
    },
    {
      "epoch": 0.8324661810613944,
      "grad_norm": 4.283135414123535,
      "learning_rate": 3.960000000000001e-05,
      "loss": 2.4467,
      "step": 400
    },
    {
      "epoch": 1.0405827263267429,
      "eval_loss": 0.24771228432655334,
      "eval_runtime": 86.9901,
      "eval_samples_per_second": 12.07,
      "eval_steps_per_second": 1.517,
      "eval_wer": 0.8371428571428572,
      "step": 500
    },
    {
      "epoch": 1.2486992715920915,
      "grad_norm": 1.9659106731414795,
      "learning_rate": 4.747368421052632e-05,
      "loss": 0.583,
      "step": 600
    },
    {
      "epoch": 1.6649323621227887,
      "grad_norm": 1.3077179193496704,
      "learning_rate": 4.221052631578948e-05,
      "loss": 0.4601,
      "step": 800
    },
    {
      "epoch": 2.0811654526534857,
      "grad_norm": 1.9944076538085938,
      "learning_rate": 3.6947368421052635e-05,
      "loss": 0.4021,
      "step": 1000
    },
    {
      "epoch": 2.0811654526534857,
      "eval_loss": 0.1869359165430069,
      "eval_runtime": 86.2672,
      "eval_samples_per_second": 12.171,
      "eval_steps_per_second": 1.53,
      "eval_wer": 0.7790476190476191,
      "step": 1000
    },
    {
      "epoch": 2.497398543184183,
      "grad_norm": 2.027698516845703,
      "learning_rate": 3.168421052631579e-05,
      "loss": 0.3278,
      "step": 1200
    },
    {
      "epoch": 2.91363163371488,
      "grad_norm": 1.2876348495483398,
      "learning_rate": 2.6421052631578945e-05,
      "loss": 0.3192,
      "step": 1400
    },
    {
      "epoch": 3.121748178980229,
      "eval_loss": 0.14687703549861908,
      "eval_runtime": 87.0632,
      "eval_samples_per_second": 12.06,
      "eval_steps_per_second": 1.516,
      "eval_wer": 0.6580952380952381,
      "step": 1500
    },
    {
      "epoch": 3.3298647242455774,
      "grad_norm": 0.6025579571723938,
      "learning_rate": 2.1157894736842106e-05,
      "loss": 0.2686,
      "step": 1600
    },
    {
      "epoch": 3.7460978147762747,
      "grad_norm": 0.5916275978088379,
      "learning_rate": 1.5894736842105266e-05,
      "loss": 0.2524,
      "step": 1800
    },
    {
      "epoch": 4.1623309053069715,
      "grad_norm": 1.3612473011016846,
      "learning_rate": 1.0631578947368421e-05,
      "loss": 0.2252,
      "step": 2000
    },
    {
      "epoch": 4.1623309053069715,
      "eval_loss": 0.13481491804122925,
      "eval_runtime": 86.2138,
      "eval_samples_per_second": 12.179,
      "eval_steps_per_second": 1.531,
      "eval_wer": 0.6038095238095238,
      "step": 2000
    },
    {
      "epoch": 4.578563995837669,
      "grad_norm": 1.8633441925048828,
      "learning_rate": 5.36842105263158e-06,
      "loss": 0.1991,
      "step": 2200
    },
    {
      "epoch": 4.994797086368366,
      "grad_norm": 7.069295883178711,
      "learning_rate": 1.3157894736842107e-07,
      "loss": 0.1907,
      "step": 2400
    },
    {
      "epoch": 4.994797086368366,
      "step": 2400,
      "total_flos": 2.1780564758105436e+19,
      "train_loss": 0.8824304628372193,
      "train_runtime": 6499.422,
      "train_samples_per_second": 5.911,
      "train_steps_per_second": 0.369
    }
  ],
  "logging_steps": 200,
  "max_steps": 2400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.1780564758105436e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}