trainer_state.json

{
  "best_metric": 0.9405013472869778,
  "best_model_checkpoint": "model_saves/xlnet-large_spell_5k_2_p3/checkpoint-804",
  "epoch": 6.0,
  "global_step": 1608,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9404566240678004,
      "eval_loss": 0.43533819913864136,
      "eval_runtime": 4.8033,
      "eval_samples_per_second": 912.282,
      "eval_steps_per_second": 7.287,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9400093918760273,
      "eval_loss": 0.44132065773010254,
      "eval_runtime": 4.7938,
      "eval_samples_per_second": 914.089,
      "eval_steps_per_second": 7.301,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9405013472869778,
      "eval_loss": 0.4725654423236847,
      "eval_runtime": 4.8004,
      "eval_samples_per_second": 912.849,
      "eval_steps_per_second": 7.291,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.3275,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9397354621585662,
      "eval_loss": 0.5153043270111084,
      "eval_runtime": 4.8338,
      "eval_samples_per_second": 906.524,
      "eval_steps_per_second": 7.241,
      "step": 1072
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9390534330661121,
      "eval_loss": 0.5466041564941406,
      "eval_runtime": 4.7698,
      "eval_samples_per_second": 918.69,
      "eval_steps_per_second": 7.338,
      "step": 1340
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9384832120216013,
      "eval_loss": 0.5921659469604492,
      "eval_runtime": 4.746,
      "eval_samples_per_second": 923.305,
      "eval_steps_per_second": 7.375,
      "step": 1608
    },
    {
      "epoch": 6.0,
      "step": 1608,
      "total_flos": 2.551532092391424e+16,
      "train_loss": 0.2767679157541759,
      "train_runtime": 670.3814,
      "train_samples_per_second": 767.563,
      "train_steps_per_second": 5.997
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 2.551532092391424e+16,
  "trial_name": null,
  "trial_params": null
}