trainer_state.json

{
  "best_global_step": 2844,
  "best_metric": 0.40068772435188293,
  "best_model_checkpoint": "distilbert_rand_20_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": 1.9076238870620728,
      "learning_rate": 4.9e-05,
      "loss": 0.4777,
      "step": 1422
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7824635171902052,
      "eval_combined_score": 0.7155636405109737,
      "eval_f1": 0.6486637638317421,
      "eval_loss": 0.45324674248695374,
      "eval_runtime": 16.7108,
      "eval_samples_per_second": 2419.396,
      "eval_steps_per_second": 9.455,
      "step": 1422
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.6517744064331055,
      "learning_rate": 4.8e-05,
      "loss": 0.3735,
      "step": 2844
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8172149394014345,
      "eval_combined_score": 0.7825928731520507,
      "eval_f1": 0.7479708069026669,
      "eval_loss": 0.40068772435188293,
      "eval_runtime": 16.4671,
      "eval_samples_per_second": 2455.197,
      "eval_steps_per_second": 9.595,
      "step": 2844
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.003544569015503,
      "learning_rate": 4.7e-05,
      "loss": 0.2989,
      "step": 4266
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8229037843185754,
      "eval_combined_score": 0.7924992074171293,
      "eval_f1": 0.7620946305156832,
      "eval_loss": 0.40749505162239075,
      "eval_runtime": 16.2756,
      "eval_samples_per_second": 2484.085,
      "eval_steps_per_second": 9.708,
      "step": 4266
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.771472454071045,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.2401,
      "step": 5688
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8265149641355429,
      "eval_combined_score": 0.7846263200959405,
      "eval_f1": 0.742737676056338,
      "eval_loss": 0.4642251431941986,
      "eval_runtime": 16.2505,
      "eval_samples_per_second": 2487.926,
      "eval_steps_per_second": 9.723,
      "step": 5688
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.239269495010376,
      "learning_rate": 4.5e-05,
      "loss": 0.193,
      "step": 7110
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8348008904279001,
      "eval_combined_score": 0.8018348155638133,
      "eval_f1": 0.7688687406997267,
      "eval_loss": 0.4590378403663635,
      "eval_runtime": 16.3149,
      "eval_samples_per_second": 2478.103,
      "eval_steps_per_second": 9.684,
      "step": 7110
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.03383731842041,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.1577,
      "step": 8532
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8348256245362354,
      "eval_combined_score": 0.8010307976288706,
      "eval_f1": 0.7672359707215057,
      "eval_loss": 0.48832327127456665,
      "eval_runtime": 16.2537,
      "eval_samples_per_second": 2487.431,
      "eval_steps_per_second": 9.721,
      "step": 8532
    },
    {
      "epoch": 7.0,
      "grad_norm": 4.526302814483643,
      "learning_rate": 4.3e-05,
      "loss": 0.131,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8355181795696265,
      "eval_combined_score": 0.8080593207316955,
      "eval_f1": 0.7806004618937644,
      "eval_loss": 0.5554528832435608,
      "eval_runtime": 16.2607,
      "eval_samples_per_second": 2486.364,
      "eval_steps_per_second": 9.717,
      "step": 9954
    },
    {
      "epoch": 7.0,
      "step": 9954,
      "total_flos": 1.686920659598684e+17,
      "train_loss": 0.2674089492696678,
      "train_runtime": 2423.4789,
      "train_samples_per_second": 7506.688,
      "train_steps_per_second": 29.338
    }
  ],
  "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": 1.686920659598684e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}