run-23/checkpoint-642/trainer_state.json

{
  "best_metric": 0.7311072056239016,
  "best_model_checkpoint": "tiny-bert-sst2-distilled/run-23/checkpoint-642",
  "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": 8.174213409423828,
      "learning_rate": 0.00021193737357349756,
      "loss": 0.6053,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7041593438781488,
      "eval_f1": 0.4107351225204201,
      "eval_loss": 0.5531450510025024,
      "eval_mcc": 0.2654659366009494,
      "eval_precision": 0.6111111111111112,
      "eval_recall": 0.3093145869947276,
      "eval_runtime": 3.1397,
      "eval_samples_per_second": 543.682,
      "eval_steps_per_second": 17.199,
      "step": 214
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.549020290374756,
      "learning_rate": 0.00018544520187681036,
      "loss": 0.5455,
      "step": 428
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7229056824838899,
      "eval_f1": 0.4569460390355913,
      "eval_loss": 0.5337331295013428,
      "eval_mcc": 0.32023248430720136,
      "eval_precision": 0.6589403973509934,
      "eval_recall": 0.34973637961335674,
      "eval_runtime": 3.1777,
      "eval_samples_per_second": 537.185,
      "eval_steps_per_second": 16.994,
      "step": 428
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.2418973445892334,
      "learning_rate": 0.00015895303018012317,
      "loss": 0.514,
      "step": 642
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7311072056239016,
      "eval_f1": 0.43122676579925656,
      "eval_loss": 0.5188717246055603,
      "eval_mcc": 0.3396283739174578,
      "eval_precision": 0.7310924369747899,
      "eval_recall": 0.30579964850615116,
      "eval_runtime": 3.1525,
      "eval_samples_per_second": 541.47,
      "eval_steps_per_second": 17.129,
      "step": 642
    }
  ],
  "logging_steps": 500,
  "max_steps": 1926,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "total_flos": 1574326993320.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.9926400370862213,
    "learning_rate": 0.00023842954527018476,
    "num_train_epochs": 9,
    "temperature": 23
  }
}