run-37/checkpoint-1935/trainer_state.json

{
  "best_metric": 0.8978286157425359,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-37/checkpoint-1935",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1935,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.3864041566848755,
      "learning_rate": 0.00016868362667848253,
      "loss": 0.2983,
      "step": 645
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8949205118262893,
      "eval_f1": 0.2322946175637394,
      "eval_loss": 0.2774158716201782,
      "eval_mcc": 0.257425944719354,
      "eval_precision": 0.5985401459854015,
      "eval_recall": 0.14411247803163443,
      "eval_runtime": 9.9039,
      "eval_samples_per_second": 520.804,
      "eval_steps_per_second": 16.357,
      "step": 645
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.49007099866867065,
      "learning_rate": 0.0001499410014919845,
      "loss": 0.2764,
      "step": 1290
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8960837533927879,
      "eval_f1": 0.1518987341772152,
      "eval_loss": 0.27505719661712646,
      "eval_mcc": 0.23128061022537116,
      "eval_precision": 0.7619047619047619,
      "eval_recall": 0.0843585237258348,
      "eval_runtime": 9.9176,
      "eval_samples_per_second": 520.084,
      "eval_steps_per_second": 16.335,
      "step": 1290
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7594336867332458,
      "learning_rate": 0.0001311983763054864,
      "loss": 0.2709,
      "step": 1935
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8978286157425359,
      "eval_f1": 0.1700787401574803,
      "eval_loss": 0.2664003074169159,
      "eval_mcc": 0.2572444823436495,
      "eval_precision": 0.8181818181818182,
      "eval_recall": 0.09490333919156414,
      "eval_runtime": 9.9289,
      "eval_samples_per_second": 519.496,
      "eval_steps_per_second": 16.316,
      "step": 1935
    }
  ],
  "logging_steps": 500,
  "max_steps": 6450,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 4760827597800.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.8497931576459636,
    "learning_rate": 0.0001874262518649806,
    "num_train_epochs": 10,
    "temperature": 17
  }
}