finetune/cola/trainer_state.json

{
  "best_metric": 0.7777040477770405,
  "best_model_checkpoint": "final_models/transformer_base_final_2/finetune/cola/checkpoint-400",
  "epoch": 10.0,
  "global_step": 690,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.8,
      "eval_accuracy": 0.6712462902069092,
      "eval_f1": 0.7777040477770405,
      "eval_loss": 0.7399011254310608,
      "eval_mcc": 0.15981907397227785,
      "eval_runtime": 1.4063,
      "eval_samples_per_second": 724.62,
      "eval_steps_per_second": 91.022,
      "step": 400
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.3768115942028985e-05,
      "loss": 0.4934,
      "step": 500
    },
    {
      "epoch": 10.0,
      "step": 690,
      "total_flos": 5372103024537600.0,
      "train_loss": 0.42006420190783517,
      "train_runtime": 267.2668,
      "train_samples_per_second": 305.463,
      "train_steps_per_second": 2.582
    }
  ],
  "max_steps": 690,
  "num_train_epochs": 10,
  "total_flos": 5372103024537600.0,
  "trial_name": null,
  "trial_params": null
}