checkpoint-1500/trainer_state.json

{
  "best_metric": 1.531933307647705,
  "best_model_checkpoint": "./tst-translation-output/checkpoint-1500",
  "epoch": 1.1219147344801794,
  "eval_steps": 1500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.37,
      "learning_rate": 5e-05,
      "loss": 1.7074,
      "step": 500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.952812382030955e-05,
      "loss": 1.6178,
      "step": 1000
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.9056247640619104e-05,
      "loss": 1.4354,
      "step": 1500
    },
    {
      "epoch": 1.12,
      "eval_bleu": 11.2991,
      "eval_gen_len": 16.268,
      "eval_loss": 1.531933307647705,
      "eval_runtime": 168.1897,
      "eval_samples_per_second": 15.887,
      "eval_steps_per_second": 0.993,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 53480,
  "num_train_epochs": 40,
  "save_steps": 1500,
  "total_flos": 5.2011114430464e+16,
  "trial_name": null,
  "trial_params": null
}