trainer_state.json

{
  "best_global_step": 6,
  "best_metric": 0.7122578620910645,
  "best_model_checkpoint": "tiny_bert_km_100_v2_wnli/checkpoint-6",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 21,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.1270878314971924,
      "learning_rate": 4.9e-05,
      "loss": 0.7076,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.352112676056338,
      "eval_loss": 0.7217209935188293,
      "eval_runtime": 0.0303,
      "eval_samples_per_second": 2346.787,
      "eval_steps_per_second": 33.053,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7225415706634521,
      "learning_rate": 4.8e-05,
      "loss": 0.6992,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.4225352112676056,
      "eval_loss": 0.7122578620910645,
      "eval_runtime": 0.0305,
      "eval_samples_per_second": 2325.801,
      "eval_steps_per_second": 32.758,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7624399065971375,
      "learning_rate": 4.7e-05,
      "loss": 0.6946,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.39436619718309857,
      "eval_loss": 0.7191900014877319,
      "eval_runtime": 0.0859,
      "eval_samples_per_second": 826.329,
      "eval_steps_per_second": 11.638,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.9452669620513916,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.691,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.30985915492957744,
      "eval_loss": 0.7306336760520935,
      "eval_runtime": 0.0417,
      "eval_samples_per_second": 1704.484,
      "eval_steps_per_second": 24.007,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.9200876951217651,
      "learning_rate": 4.5e-05,
      "loss": 0.6924,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.29577464788732394,
      "eval_loss": 0.7442232966423035,
      "eval_runtime": 0.0306,
      "eval_samples_per_second": 2316.7,
      "eval_steps_per_second": 32.63,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.9126797914505005,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6922,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.29577464788732394,
      "eval_loss": 0.758472740650177,
      "eval_runtime": 0.032,
      "eval_samples_per_second": 2215.741,
      "eval_steps_per_second": 31.208,
      "step": 18
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.8488839864730835,
      "learning_rate": 4.3e-05,
      "loss": 0.6857,
      "step": 21
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.28169014084507044,
      "eval_loss": 0.7635892033576965,
      "eval_runtime": 0.0473,
      "eval_samples_per_second": 1501.009,
      "eval_steps_per_second": 21.141,
      "step": 21
    },
    {
      "epoch": 7.0,
      "step": 21,
      "total_flos": 116563804707840.0,
      "train_loss": 0.694670075461978,
      "train_runtime": 7.4862,
      "train_samples_per_second": 4241.146,
      "train_steps_per_second": 20.037
    }
  ],
  "logging_steps": 1,
  "max_steps": 150,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 5
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 116563804707840.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}