all_results.json

{
    "epoch": 2.9959514170040484,
    "eval_logits/chosen": -1.5362521409988403,
    "eval_logits/rejected": -1.577704906463623,
    "eval_logps/chosen": -191.7516632080078,
    "eval_logps/rejected": -226.06272888183594,
    "eval_loss": 0.31698280572891235,
    "eval_rewards/accuracies": 0.8463855385780334,
    "eval_rewards/chosen": -0.7307406663894653,
    "eval_rewards/margins": 2.3931150436401367,
    "eval_rewards/rejected": -3.1238558292388916,
    "eval_runtime": 23.3432,
    "eval_samples_per_second": 14.137,
    "eval_steps_per_second": 3.556,
    "total_flos": 4.5615607240812134e+17,
    "train_loss": 0.26195224279218965,
    "train_runtime": 3105.2921,
    "train_samples_per_second": 2.862,
    "train_steps_per_second": 0.357
}