checkpoint-8000/trainer_state.json

{
  "best_metric": 27.8203,
  "best_model_checkpoint": "./ko-en_mbartLarge_exp20p_batch64_linear/checkpoint-8000",
  "epoch": 3.712727694628147,
  "eval_steps": 4000,
  "global_step": 8000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.23,
      "learning_rate": 1.25e-05,
      "loss": 1.5426,
      "step": 500
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.5e-05,
      "loss": 1.343,
      "step": 1000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 1.2768,
      "step": 1500
    },
    {
      "epoch": 0.93,
      "learning_rate": 5e-05,
      "loss": 1.242,
      "step": 2000
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.970294676806084e-05,
      "loss": 1.1208,
      "step": 2500
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.940589353612168e-05,
      "loss": 1.0718,
      "step": 3000
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.910884030418251e-05,
      "loss": 1.061,
      "step": 3500
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.881178707224335e-05,
      "loss": 1.0585,
      "step": 4000
    },
    {
      "epoch": 1.86,
      "eval_bleu": 27.494,
      "eval_gen_len": 18.9136,
      "eval_loss": 1.1434648036956787,
      "eval_runtime": 1314.8151,
      "eval_samples_per_second": 13.11,
      "eval_steps_per_second": 1.639,
      "step": 4000
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.851473384030418e-05,
      "loss": 0.9558,
      "step": 4500
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.8217680608365025e-05,
      "loss": 0.8186,
      "step": 5000
    },
    {
      "epoch": 2.55,
      "learning_rate": 4.792062737642586e-05,
      "loss": 0.8398,
      "step": 5500
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.7623574144486695e-05,
      "loss": 0.8454,
      "step": 6000
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.732652091254753e-05,
      "loss": 0.8416,
      "step": 6500
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.702946768060837e-05,
      "loss": 0.6354,
      "step": 7000
    },
    {
      "epoch": 3.48,
      "learning_rate": 4.67324144486692e-05,
      "loss": 0.6545,
      "step": 7500
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.643536121673004e-05,
      "loss": 0.6719,
      "step": 8000
    },
    {
      "epoch": 3.71,
      "eval_bleu": 27.8203,
      "eval_gen_len": 18.6815,
      "eval_loss": 1.204200267791748,
      "eval_runtime": 1295.2995,
      "eval_samples_per_second": 13.307,
      "eval_steps_per_second": 1.664,
      "step": 8000
    }
  ],
  "logging_steps": 500,
  "max_steps": 86160,
  "num_train_epochs": 40,
  "save_steps": 4000,
  "total_flos": 1.109570441183232e+18,
  "trial_name": null,
  "trial_params": null
}