trainer_state.json

{
  "best_metric": 0.9955015182117884,
  "best_model_checkpoint": "./models/portuguese_diff_cse",
  "epoch": 2.0,
  "global_step": 436,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "electra_acc": 0.1145,
      "electra_fix_acc": 0.8984,
      "electra_rep_acc": 0.0633,
      "epoch": 0.0,
      "learning_rate": 6.98394495412844e-06,
      "loss": 0.3321,
      "neg_sim": -0.0092,
      "pos_sim": 0.548,
      "step": 1
    },
    {
      "epoch": 0.57,
      "eval_stsb_spearman": 0.9955015182117884,
      "step": 125
    },
    {
      "epoch": 1.15,
      "eval_stsb_spearman": 0.9947652924624643,
      "step": 250
    },
    {
      "epoch": 1.72,
      "eval_stsb_spearman": 0.9947550447251673,
      "step": 375
    },
    {
      "epoch": 2.0,
      "step": 436,
      "train_runtime": 507.537,
      "train_samples_per_second": 0.859
    }
  ],
  "max_steps": 436,
  "num_train_epochs": 2,
  "total_flos": 6073324447075440,
  "trial_name": null,
  "trial_params": null
}