trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.41833943128585815,
  "best_model_checkpoint": "tiny_bert_rand_10_v1_qqp/checkpoint-2844",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 9954,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.6271923780441284,
      "learning_rate": 4.9e-05,
      "loss": 0.4957,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7827850605985655,
      "eval_combined_score": 0.7276225765472314,
      "eval_f1": 0.6724600924958973,
      "eval_loss": 0.4516511857509613,
      "eval_runtime": 12.0702,
      "eval_samples_per_second": 3349.569,
      "eval_steps_per_second": 13.09,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.40657639503479,
      "learning_rate": 4.8e-05,
      "loss": 0.4048,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8025476131585456,
      "eval_combined_score": 0.7654761676856587,
      "eval_f1": 0.7284047222127717,
      "eval_loss": 0.41833943128585815,
      "eval_runtime": 11.8057,
      "eval_samples_per_second": 3424.62,
      "eval_steps_per_second": 13.383,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.13236141204834,
      "learning_rate": 4.7e-05,
      "loss": 0.3455,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8124412564927035,
      "eval_combined_score": 0.7738034871853382,
      "eval_f1": 0.7351657178779729,
      "eval_loss": 0.4275418519973755,
      "eval_runtime": 11.7509,
      "eval_samples_per_second": 3440.578,
      "eval_steps_per_second": 13.446,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.7595160007476807,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2989,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8182785060598565,
      "eval_combined_score": 0.7767561040876241,
      "eval_f1": 0.7352337021153915,
      "eval_loss": 0.4458482563495636,
      "eval_runtime": 11.8253,
      "eval_samples_per_second": 3418.932,
      "eval_steps_per_second": 13.361,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.8723201751708984,
      "learning_rate": 4.5e-05,
      "loss": 0.2624,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8240168191936681,
      "eval_combined_score": 0.7922474265973389,
      "eval_f1": 0.7604780340010099,
      "eval_loss": 0.4298754334449768,
      "eval_runtime": 11.8409,
      "eval_samples_per_second": 3414.438,
      "eval_steps_per_second": 13.344,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.1317880153656006,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2308,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.823695275785308,
      "eval_combined_score": 0.791588493608675,
      "eval_f1": 0.7594817114320421,
      "eval_loss": 0.4603184163570404,
      "eval_runtime": 11.8708,
      "eval_samples_per_second": 3405.843,
      "eval_steps_per_second": 13.31,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.599036455154419,
      "learning_rate": 4.3e-05,
      "loss": 0.2067,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.824956715310413,
      "eval_combined_score": 0.79490850019433,
      "eval_f1": 0.764860285078247,
      "eval_loss": 0.46773019433021545,
      "eval_runtime": 11.8891,
      "eval_samples_per_second": 3400.585,
      "eval_steps_per_second": 13.289,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 6.678940801217126e+16,
      "train_loss": 0.32066750828223894,
      "train_runtime": 1423.9001,
      "train_samples_per_second": 12776.388,
      "train_steps_per_second": 49.933
    }
  ],
  "logging_steps": 1,
  "max_steps": 71100,
  "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": 6.678940801217126e+16,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}