trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 20,
  "global_step": 48,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.021052631578947368,
      "grad_norm": 156.80313847716513,
      "learning_rate": 0.0,
      "logits/chosen": -3.19140625,
      "logits/rejected": -3.1171875,
      "logps/chosen": -284.0,
      "logps/rejected": -347.0,
      "loss": 0.6914,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 140.79071517526302,
      "learning_rate": 3.8023508512198257e-07,
      "logits/chosen": NaN,
      "logits/rejected": NaN,
      "logps/chosen": -294.0526428222656,
      "logps/rejected": -305.3552551269531,
      "loss": 0.6032,
      "rewards/accuracies": 0.375,
      "rewards/chosen": -0.007361412048339844,
      "rewards/margins": 0.3393391966819763,
      "rewards/rejected": -0.34671854972839355,
      "step": 20
    },
    {
      "epoch": 0.42105263157894735,
      "eval_logits/chosen": NaN,
      "eval_logits/rejected": NaN,
      "eval_logps/chosen": -293.04254150390625,
      "eval_logps/rejected": -301.6383056640625,
      "eval_loss": 0.5011141896247864,
      "eval_rewards/accuracies": 0.4976728856563568,
      "eval_rewards/chosen": -0.0284495260566473,
      "eval_rewards/margins": 0.8294028639793396,
      "eval_rewards/rejected": -0.8580036759376526,
      "eval_runtime": 292.0408,
      "eval_samples_per_second": 15.368,
      "eval_steps_per_second": 0.243,
      "step": 20
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 108.6076169236696,
      "learning_rate": 5.212587789268649e-08,
      "logits/chosen": NaN,
      "logits/rejected": NaN,
      "logps/chosen": -298.98748779296875,
      "logps/rejected": -308.0874938964844,
      "loss": 0.4221,
      "rewards/accuracies": 0.5316406488418579,
      "rewards/chosen": 0.30016860365867615,
      "rewards/margins": 1.107550024986267,
      "rewards/rejected": -0.8070526123046875,
      "step": 40
    },
    {
      "epoch": 0.8421052631578947,
      "eval_logits/chosen": NaN,
      "eval_logits/rejected": NaN,
      "eval_logps/chosen": -294.68084716796875,
      "eval_logps/rejected": -303.2127685546875,
      "eval_loss": 0.43298768997192383,
      "eval_rewards/accuracies": 0.5176196694374084,
      "eval_rewards/chosen": 0.4743392765522003,
      "eval_rewards/margins": 1.0208220481872559,
      "eval_rewards/rejected": -0.5463438630104065,
      "eval_runtime": 291.3347,
      "eval_samples_per_second": 15.405,
      "eval_steps_per_second": 0.244,
      "step": 40
    },
    {
      "epoch": 1.0,
      "step": 48,
      "total_flos": 0.0,
      "train_loss": 0.48351796468098956,
      "train_runtime": 4731.3673,
      "train_samples_per_second": 3.793,
      "train_steps_per_second": 0.01
    }
  ],
  "logging_steps": 20,
  "max_steps": 48,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 10,
  "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": 16,
  "trial_name": null,
  "trial_params": null
}