trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "episode": 2,
  "epoch": 6.55694708543702e-05,
  "eval_steps": 500,
  "global_step": 1,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "episode": 2,
      "epoch": 6.55694708543702e-05,
      "eps": 0,
      "loss/policy_avg": 0.21204827725887299,
      "loss/value_avg": 33.20888137817383,
      "lr": 3e-06,
      "objective/entropy": 70.37702941894531,
      "objective/kl": 427.1629638671875,
      "objective/non_score_reward": -21.3581485748291,
      "objective/rlhf_reward": -17.6393985748291,
      "objective/scores": 3.71875,
      "policy/approxkl_avg": 0.31922289729118347,
      "policy/clipfrac_avg": 0.1320754736661911,
      "policy/entropy_avg": 1.321390151977539,
      "step": 1,
      "val/clipfrac_avg": 0.0,
      "val/num_eos_tokens": 1,
      "val/ratio": 1.5207229852676392,
      "val/ratio_var": 0.5327765941619873
    }
  ],
  "logging_steps": 500,
  "max_steps": 1,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3.27847354271851e-05,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0,
  "train_batch_size": null,
  "trial_name": null,
  "trial_params": null
}