trainer_state.json

{
  "best_metric": 0.4050879180431366,
  "best_model_checkpoint": "/content/deberta-v3-small/checkpoint-535",
  "epoch": 5.0,
  "global_step": 2675,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 0.4050879180431366,
      "eval_matthews_correlation": 0.6333205721749096,
      "eval_runtime": 2.031,
      "eval_samples_per_second": 513.528,
      "eval_steps_per_second": 64.499,
      "step": 535
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8785046728971963e-05,
      "loss": 0.3371,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4455100893974304,
      "eval_matthews_correlation": 0.6530734183134956,
      "eval_runtime": 2.0334,
      "eval_samples_per_second": 512.931,
      "eval_steps_per_second": 64.424,
      "step": 1070
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.5754692554473877,
      "eval_matthews_correlation": 0.6499333084534608,
      "eval_runtime": 2.0345,
      "eval_samples_per_second": 512.647,
      "eval_steps_per_second": 64.388,
      "step": 1605
    },
    {
      "epoch": 3.74,
      "learning_rate": 7.570093457943924e-06,
      "loss": 0.1305,
      "step": 2000
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.7188061475753784,
      "eval_matthews_correlation": 0.6553451139053764,
      "eval_runtime": 2.0152,
      "eval_samples_per_second": 517.554,
      "eval_steps_per_second": 65.004,
      "step": 2140
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.8047384023666382,
      "eval_matthews_correlation": 0.6700089525749314,
      "eval_runtime": 2.0026,
      "eval_samples_per_second": 520.818,
      "eval_steps_per_second": 65.414,
      "step": 2675
    },
    {
      "epoch": 5.0,
      "step": 2675,
      "total_flos": 1415961343234560.0,
      "train_loss": 0.19247490570923992,
      "train_runtime": 168.1314,
      "train_samples_per_second": 254.295,
      "train_steps_per_second": 15.91
    }
  ],
  "max_steps": 2675,
  "num_train_epochs": 5,
  "total_flos": 1415961343234560.0,
  "trial_name": null,
  "trial_params": null
}