last-checkpoint/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.050955414012738856,
  "eval_steps": 500,
  "global_step": 5,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01019108280254777,
      "grad_norm": 1.9706840515136719,
      "learning_rate": 8e-05,
      "log_odds_chosen": 0.043639570474624634,
      "log_odds_ratio": -0.6778613328933716,
      "logits/chosen": -0.5009905099868774,
      "logits/rejected": -0.5356172919273376,
      "logps/chosen": -2.954733371734619,
      "logps/rejected": -2.996232509613037,
      "loss": 3.6057,
      "nll_loss": 3.5378708839416504,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.2954733371734619,
      "rewards/margins": 0.00414988212287426,
      "rewards/rejected": -0.2996232211589813,
      "step": 1
    },
    {
      "epoch": 0.02038216560509554,
      "grad_norm": 1.9629229307174683,
      "learning_rate": 6e-05,
      "log_odds_chosen": 0.10935741662979126,
      "log_odds_ratio": -0.6454683542251587,
      "logits/chosen": -0.5241304636001587,
      "logits/rejected": -0.5285190939903259,
      "logps/chosen": -2.8890933990478516,
      "logps/rejected": -2.9919257164001465,
      "loss": 3.4851,
      "nll_loss": 3.420555830001831,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.28890931606292725,
      "rewards/margins": 0.010283238254487514,
      "rewards/rejected": -0.2991925776004791,
      "step": 2
    },
    {
      "epoch": 0.030573248407643312,
      "grad_norm": 2.020477533340454,
      "learning_rate": 4e-05,
      "log_odds_chosen": 0.069330133497715,
      "log_odds_ratio": -0.6641735434532166,
      "logits/chosen": -0.4971535801887512,
      "logits/rejected": -0.5353123545646667,
      "logps/chosen": -2.731393575668335,
      "logps/rejected": -2.795962333679199,
      "loss": 3.3144,
      "nll_loss": 3.2479400634765625,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.2731393277645111,
      "rewards/margins": 0.006456875707954168,
      "rewards/rejected": -0.279596209526062,
      "step": 3
    },
    {
      "epoch": 0.04076433121019108,
      "grad_norm": 2.3341853618621826,
      "learning_rate": 2e-05,
      "log_odds_chosen": 0.12229391932487488,
      "log_odds_ratio": -0.6381683945655823,
      "logits/chosen": -0.4815675914287567,
      "logits/rejected": -0.5111523866653442,
      "logps/chosen": -2.8219356536865234,
      "logps/rejected": -2.9366464614868164,
      "loss": 3.3456,
      "nll_loss": 3.281827211380005,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.28219354152679443,
      "rewards/margins": 0.011471111327409744,
      "rewards/rejected": -0.2936646640300751,
      "step": 4
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 2.1892218589782715,
      "learning_rate": 0.0,
      "log_odds_chosen": 0.06512074917554855,
      "log_odds_ratio": -0.6671069860458374,
      "logits/chosen": -0.5229994654655457,
      "logits/rejected": -0.5436381697654724,
      "logps/chosen": -2.7552542686462402,
      "logps/rejected": -2.816582679748535,
      "loss": 3.2584,
      "nll_loss": 3.191640615463257,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.27552542090415955,
      "rewards/margins": 0.006132860202342272,
      "rewards/rejected": -0.2816582918167114,
      "step": 5
    }
  ],
  "logging_steps": 1,
  "max_steps": 5,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}