trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 3,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 43.0,
      "learning_rate": 0.0019982240555854445,
      "loss": 3.7765,
      "step": 1
    },
    {
      "epoch": 0.06,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.6743,
      "eval_covost2-en-de_samples_per_second": 1.762,
      "eval_covost2-en-de_steps_per_second": 0.352,
      "step": 3
    },
    {
      "epoch": 0.12,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.8825,
      "eval_covost2-en-de_samples_per_second": 1.7,
      "eval_covost2-en-de_steps_per_second": 0.34,
      "step": 6
    },
    {
      "epoch": 0.18,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.771,
      "eval_covost2-en-de_samples_per_second": 1.733,
      "eval_covost2-en-de_steps_per_second": 0.347,
      "step": 9
    },
    {
      "epoch": 0.24,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.829,
      "eval_covost2-en-de_samples_per_second": 1.716,
      "eval_covost2-en-de_steps_per_second": 0.343,
      "step": 12
    },
    {
      "epoch": 0.3,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 7.429,
      "eval_covost2-en-de_samples_per_second": 1.346,
      "eval_covost2-en-de_steps_per_second": 0.269,
      "step": 15
    },
    {
      "epoch": 0.36,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.8247,
      "eval_covost2-en-de_samples_per_second": 1.717,
      "eval_covost2-en-de_steps_per_second": 0.343,
      "step": 18
    },
    {
      "epoch": 0.42,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 6.1196,
      "eval_covost2-en-de_samples_per_second": 1.634,
      "eval_covost2-en-de_steps_per_second": 0.327,
      "step": 21
    },
    {
      "epoch": 0.48,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.4491,
      "eval_covost2-en-de_samples_per_second": 1.835,
      "eval_covost2-en-de_steps_per_second": 0.367,
      "step": 24
    },
    {
      "epoch": 0.54,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.6325,
      "eval_covost2-en-de_samples_per_second": 1.775,
      "eval_covost2-en-de_steps_per_second": 0.355,
      "step": 27
    },
    {
      "epoch": 0.6,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 7.1218,
      "eval_covost2-en-de_samples_per_second": 1.404,
      "eval_covost2-en-de_steps_per_second": 0.281,
      "step": 30
    },
    {
      "epoch": 0.66,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.9797,
      "eval_covost2-en-de_samples_per_second": 1.672,
      "eval_covost2-en-de_steps_per_second": 0.334,
      "step": 33
    },
    {
      "epoch": 0.72,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.3852,
      "eval_covost2-en-de_samples_per_second": 1.857,
      "eval_covost2-en-de_steps_per_second": 0.371,
      "step": 36
    },
    {
      "epoch": 0.78,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.7768,
      "eval_covost2-en-de_samples_per_second": 1.731,
      "eval_covost2-en-de_steps_per_second": 0.346,
      "step": 39
    },
    {
      "epoch": 0.84,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.5233,
      "eval_covost2-en-de_samples_per_second": 1.811,
      "eval_covost2-en-de_steps_per_second": 0.362,
      "step": 42
    },
    {
      "epoch": 0.9,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.6144,
      "eval_covost2-en-de_samples_per_second": 1.781,
      "eval_covost2-en-de_steps_per_second": 0.356,
      "step": 45
    },
    {
      "epoch": 0.96,
      "eval_covost2-en-de_model_preparation_time": 0.0137,
      "eval_covost2-en-de_runtime": 5.4255,
      "eval_covost2-en-de_samples_per_second": 1.843,
      "eval_covost2-en-de_steps_per_second": 0.369,
      "step": 48
    }
  ],
  "logging_steps": 100,
  "max_steps": 50,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 13,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 368614469566464.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}