trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.6917364597320557,
  "best_model_checkpoint": "tiny_bert_rand_50_v2_wnli/checkpoint-3",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 18,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.8235816359519958,
      "learning_rate": 4.9e-05,
      "loss": 0.6957,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5492957746478874,
      "eval_loss": 0.6917364597320557,
      "eval_runtime": 0.0315,
      "eval_samples_per_second": 2253.978,
      "eval_steps_per_second": 31.746,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8038091063499451,
      "learning_rate": 4.8e-05,
      "loss": 0.7026,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5492957746478874,
      "eval_loss": 0.694267213344574,
      "eval_runtime": 0.0366,
      "eval_samples_per_second": 1940.277,
      "eval_steps_per_second": 27.328,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5372172594070435,
      "learning_rate": 4.7e-05,
      "loss": 0.6984,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.4225352112676056,
      "eval_loss": 0.7109375,
      "eval_runtime": 0.032,
      "eval_samples_per_second": 2221.559,
      "eval_steps_per_second": 31.29,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.4333108067512512,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6952,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3380281690140845,
      "eval_loss": 0.7083516120910645,
      "eval_runtime": 0.0307,
      "eval_samples_per_second": 2311.144,
      "eval_steps_per_second": 32.551,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.7413873672485352,
      "learning_rate": 4.5e-05,
      "loss": 0.6948,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.5070422535211268,
      "eval_loss": 0.7027947902679443,
      "eval_runtime": 0.0309,
      "eval_samples_per_second": 2299.402,
      "eval_steps_per_second": 32.386,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.591996967792511,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6959,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.5211267605633803,
      "eval_loss": 0.7051607966423035,
      "eval_runtime": 0.0349,
      "eval_samples_per_second": 2033.29,
      "eval_steps_per_second": 28.638,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 99911832606720.0,
      "train_loss": 0.6970992353227403,
      "train_runtime": 6.8842,
      "train_samples_per_second": 4612.036,
      "train_steps_per_second": 21.789
    }
  ],
  "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": 99911832606720.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}