{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 100,
  "global_step": 63,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 2.08540016412735,
      "epoch": 0.963855421686747,
      "grad_norm": 1.4502723217010498,
      "learning_rate": 1.9000000000000002e-06,
      "logits/chosen": 1.9373448673414857,
      "logits/rejected": 1.4560297831585156,
      "logps/chosen": -43.992462682723996,
      "logps/rejected": -51.50791831016541,
      "loss": 0.6897519111633301,
      "mean_token_accuracy": 0.2624862600117922,
      "num_tokens": 37803.0,
      "rewards/accuracies": 0.7875,
      "rewards/chosen": 0.0014421886393392925,
      "rewards/margins": 0.0068238718846259875,
      "rewards/rejected": -0.005381683184623398,
      "step": 20
    },
    {
      "entropy": 2.1433570188812063,
      "epoch": 1.9156626506024095,
      "grad_norm": 1.7149038314819336,
      "learning_rate": 3.900000000000001e-06,
      "logits/chosen": 2.0310179220420284,
      "logits/rejected": 1.5091439161978566,
      "logps/chosen": -43.38176954245265,
      "logps/rejected": -52.56174864950059,
      "loss": 0.628478717803955,
      "mean_token_accuracy": 0.2560643789511693,
      "num_tokens": 74911.0,
      "rewards/accuracies": 0.9968354430379747,
      "rewards/chosen": 0.02222161173903018,
      "rewards/margins": 0.13752426909683627,
      "rewards/rejected": -0.11530265806219246,
      "step": 40
    },
    {
      "entropy": 2.2126575768748418,
      "epoch": 2.8674698795180724,
      "grad_norm": 1.3426592350006104,
      "learning_rate": 5.9e-06,
      "logits/chosen": 2.04987203167158,
      "logits/rejected": 1.484420240733255,
      "logps/chosen": -43.2671080963521,
      "logps/rejected": -56.65543109555788,
      "loss": 0.4599471569061279,
      "mean_token_accuracy": 0.2588409010933924,
      "num_tokens": 112046.0,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.032298280657092227,
      "rewards/margins": 0.5631697264653218,
      "rewards/rejected": -0.5308714414699168,
      "step": 60
    },
    {
      "epoch": 3.0,
      "eval_entropy": 2.2325300216674804,
      "eval_logits/chosen": 1.9693634026292444,
      "eval_logits/rejected": 1.4018482016627885,
      "eval_logps/chosen": -43.20057983398438,
      "eval_logps/rejected": -61.643408966064456,
      "eval_loss": 0.324849396944046,
      "eval_mean_token_accuracy": 0.2726498693227768,
      "eval_num_tokens": 117000.0,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": -0.06716743037104607,
      "eval_rewards/margins": 0.9790484547615051,
      "eval_rewards/rejected": -1.0462158918380737,
      "eval_runtime": 1.2337,
      "eval_samples_per_second": 14.59,
      "eval_steps_per_second": 4.053,
      "step": 63
    }
  ],
  "logging_steps": 20,
  "max_steps": 63,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 3003135151976448.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}