trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.42537251114845276,
  "best_model_checkpoint": "tiny_bert_rand_100_v2_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": 1.579450249671936,
      "learning_rate": 4.9e-05,
      "loss": 0.4966,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7801632451150136,
      "eval_combined_score": 0.7236895303005051,
      "eval_f1": 0.6672158154859967,
      "eval_loss": 0.46046456694602966,
      "eval_runtime": 11.6395,
      "eval_samples_per_second": 3473.517,
      "eval_steps_per_second": 13.574,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.6449062824249268,
      "learning_rate": 4.8e-05,
      "loss": 0.4107,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.800123670541677,
      "eval_combined_score": 0.7559589785355769,
      "eval_f1": 0.7117942865294768,
      "eval_loss": 0.42537251114845276,
      "eval_runtime": 11.5865,
      "eval_samples_per_second": 3489.417,
      "eval_steps_per_second": 13.637,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.029707431793213,
      "learning_rate": 4.7e-05,
      "loss": 0.3516,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8092258224091021,
      "eval_combined_score": 0.7663524157397391,
      "eval_f1": 0.7234790090703761,
      "eval_loss": 0.4325940012931824,
      "eval_runtime": 11.5587,
      "eval_samples_per_second": 3497.813,
      "eval_steps_per_second": 13.669,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.960402250289917,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.3052,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8183527083848627,
      "eval_combined_score": 0.7791111799220998,
      "eval_f1": 0.7398696514593369,
      "eval_loss": 0.42600205540657043,
      "eval_runtime": 11.8713,
      "eval_samples_per_second": 3405.698,
      "eval_steps_per_second": 13.309,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.9639410972595215,
      "learning_rate": 4.5e-05,
      "loss": 0.2689,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8201830324016819,
      "eval_combined_score": 0.7896953847752978,
      "eval_f1": 0.7592077371489137,
      "eval_loss": 0.4374306797981262,
      "eval_runtime": 11.5582,
      "eval_samples_per_second": 3497.946,
      "eval_steps_per_second": 13.67,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.333555221557617,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.2372,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8209497897600792,
      "eval_combined_score": 0.7902456970723664,
      "eval_f1": 0.7595416043846537,
      "eval_loss": 0.4387308657169342,
      "eval_runtime": 11.5223,
      "eval_samples_per_second": 3508.833,
      "eval_steps_per_second": 13.712,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 4.353480815887451,
      "learning_rate": 4.3e-05,
      "loss": 0.2121,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8232995300519417,
      "eval_combined_score": 0.790606051030715,
      "eval_f1": 0.7579125720094884,
      "eval_loss": 0.4771482050418854,
      "eval_runtime": 11.5195,
      "eval_samples_per_second": 3509.689,
      "eval_steps_per_second": 13.716,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 6.678940801217126e+16,
      "train_loss": 0.3260430904524092,
      "train_runtime": 1415.5151,
      "train_samples_per_second": 12852.071,
      "train_steps_per_second": 50.229
    }
  ],
  "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
}