all_results.json

{
    "epoch": 2.97196261682243,
    "eval_logits/chosen": -2.328775644302368,
    "eval_logits/rejected": -2.302034378051758,
    "eval_logps/chosen": -264.0635681152344,
    "eval_logps/rejected": -198.01995849609375,
    "eval_loss": 0.5211486220359802,
    "eval_rewards/accuracies": 0.8229166865348816,
    "eval_rewards/chosen": 1.5578466653823853,
    "eval_rewards/margins": 2.5716841220855713,
    "eval_rewards/rejected": -1.0138373374938965,
    "eval_runtime": 99.9476,
    "eval_samples_per_second": 15.208,
    "eval_steps_per_second": 0.24,
    "total_flos": 1874604243025920.0,
    "train_loss": 0.33642754952112836,
    "train_runtime": 6089.7617,
    "train_samples_per_second": 6.737,
    "train_steps_per_second": 0.026
}