run-37/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7305213825424721,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-37/checkpoint-428",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 642,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 2.5660171508789062,
      "learning_rate": 0.0005426960660121868,
      "loss": 0.3936,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7065026362038664,
      "eval_f1": 0.3275167785234899,
      "eval_loss": 0.3753478229045868,
      "eval_mcc": 0.2588181173815382,
      "eval_precision": 0.6931818181818182,
      "eval_recall": 0.21441124780316345,
      "eval_runtime": 3.1566,
      "eval_samples_per_second": 540.772,
      "eval_steps_per_second": 17.107,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.099513530731201,
      "learning_rate": 0.00040702204950914005,
      "loss": 0.372,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7305213825424721,
      "eval_f1": 0.3994778067885118,
      "eval_loss": 0.366894006729126,
      "eval_mcc": 0.33967587808797833,
      "eval_precision": 0.7766497461928934,
      "eval_recall": 0.2688927943760984,
      "eval_runtime": 3.769,
      "eval_samples_per_second": 452.906,
      "eval_steps_per_second": 14.327,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2766120433807373,
      "learning_rate": 0.0002713480330060934,
      "loss": 0.3619,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7024018746338606,
      "eval_f1": 0.21604938271604937,
      "eval_loss": 0.3670775592327118,
      "eval_mcc": 0.258294733907354,
      "eval_precision": 0.8860759493670886,
      "eval_recall": 0.12302284710017575,
      "eval_runtime": 3.1357,
      "eval_samples_per_second": 544.372,
      "eval_steps_per_second": 17.221,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 1070,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.6249127121707758,
    "learning_rate": 0.0006783700825152334,
    "num_train_epochs": 5,
    "temperature": 13
  }
}