trainer_state.json

{
  "best_metric": 0.6070845127105713,
  "best_model_checkpoint": "./xls-r-300m-bemba-fullset/checkpoint-500",
  "epoch": 2.0,
  "global_step": 632,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.32,
      "learning_rate": 1.9599999999999995e-05,
      "loss": 11.9405,
      "step": 100
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.96e-05,
      "loss": 3.6915,
      "step": 200
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.959999999999999e-05,
      "loss": 2.984,
      "step": 300
    },
    {
      "epoch": 1.27,
      "learning_rate": 7.959999999999998e-05,
      "loss": 2.7834,
      "step": 400
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.96e-05,
      "loss": 1.6564,
      "step": 500
    },
    {
      "epoch": 1.58,
      "eval_loss": 0.6070845127105713,
      "eval_runtime": 282.4761,
      "eval_samples_per_second": 5.962,
      "eval_steps_per_second": 0.747,
      "eval_wer": 0.9916864608076009,
      "step": 500
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001196,
      "loss": 0.8314,
      "step": 600
    },
    {
      "epoch": 2.0,
      "step": 632,
      "total_flos": 2.9733883621978947e+18,
      "train_loss": 3.814326035825512,
      "train_runtime": 2899.9163,
      "train_samples_per_second": 3.485,
      "train_steps_per_second": 0.218
    }
  ],
  "max_steps": 632,
  "num_train_epochs": 2,
  "total_flos": 2.9733883621978947e+18,
  "trial_name": null,
  "trial_params": null
}