asr/whisper-small-chinese/checkpoint/trainer_state.json

{
  "best_metric": 82.67898383371825,
  "best_model_checkpoint": "whisper-small-chinese/checkpoint-1000",
  "epoch": 192.30769230769232,
  "eval_steps": 1000,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 19.23076923076923,
      "grad_norm": 1.2309635877609253,
      "learning_rate": 1.7395e-05,
      "loss": 0.2754,
      "step": 500
    },
    {
      "epoch": 38.46153846153846,
      "grad_norm": 0.0204775370657444,
      "learning_rate": 1.556722222222222e-05,
      "loss": 0.006,
      "step": 1000
    },
    {
      "epoch": 38.46153846153846,
      "eval_loss": 0.9139542579650879,
      "eval_runtime": 11.2242,
      "eval_samples_per_second": 8.107,
      "eval_steps_per_second": 0.356,
      "eval_wer": 82.67898383371825,
      "step": 1000
    },
    {
      "epoch": 57.69230769230769,
      "grad_norm": 0.007038629613816738,
      "learning_rate": 1.3622777777777778e-05,
      "loss": 0.0024,
      "step": 1500
    },
    {
      "epoch": 76.92307692307692,
      "grad_norm": 0.004221443086862564,
      "learning_rate": 1.1678333333333333e-05,
      "loss": 0.0003,
      "step": 2000
    },
    {
      "epoch": 76.92307692307692,
      "eval_loss": 0.9765084385871887,
      "eval_runtime": 11.0976,
      "eval_samples_per_second": 8.2,
      "eval_steps_per_second": 0.36,
      "eval_wer": 84.52655889145497,
      "step": 2000
    },
    {
      "epoch": 96.15384615384616,
      "grad_norm": 0.0029376207385212183,
      "learning_rate": 9.733888888888888e-06,
      "loss": 0.0002,
      "step": 2500
    },
    {
      "epoch": 115.38461538461539,
      "grad_norm": 0.0019433604320511222,
      "learning_rate": 7.789444444444444e-06,
      "loss": 0.0001,
      "step": 3000
    },
    {
      "epoch": 115.38461538461539,
      "eval_loss": 1.022544264793396,
      "eval_runtime": 11.1249,
      "eval_samples_per_second": 8.18,
      "eval_steps_per_second": 0.36,
      "eval_wer": 84.06466512702079,
      "step": 3000
    },
    {
      "epoch": 134.6153846153846,
      "grad_norm": 0.001670741941779852,
      "learning_rate": 5.845e-06,
      "loss": 0.0001,
      "step": 3500
    },
    {
      "epoch": 153.84615384615384,
      "grad_norm": 0.0014259717427194118,
      "learning_rate": 3.9005555555555556e-06,
      "loss": 0.0001,
      "step": 4000
    },
    {
      "epoch": 153.84615384615384,
      "eval_loss": 1.0503652095794678,
      "eval_runtime": 11.0802,
      "eval_samples_per_second": 8.213,
      "eval_steps_per_second": 0.361,
      "eval_wer": 83.37182448036951,
      "step": 4000
    },
    {
      "epoch": 173.07692307692307,
      "grad_norm": 0.0013439098838716745,
      "learning_rate": 1.956111111111111e-06,
      "loss": 0.0001,
      "step": 4500
    },
    {
      "epoch": 192.30769230769232,
      "grad_norm": 0.0017614001408219337,
      "learning_rate": 1.1666666666666665e-08,
      "loss": 0.0001,
      "step": 5000
    },
    {
      "epoch": 192.30769230769232,
      "eval_loss": 1.0611838102340698,
      "eval_runtime": 11.1867,
      "eval_samples_per_second": 8.135,
      "eval_steps_per_second": 0.358,
      "eval_wer": 83.37182448036951,
      "step": 5000
    }
  ],
  "logging_steps": 500,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 193,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.876182050504704e+19,
  "train_batch_size": 48,
  "trial_name": null,
  "trial_params": null
}