all_results.json

{
    "epoch": 0.9997172745264349,
    "eval_logits/chosen": -18.20407485961914,
    "eval_logits/rejected": -18.113718032836914,
    "eval_logps/chosen": -1.7059684991836548,
    "eval_logps/rejected": -2.194978952407837,
    "eval_loss": 3.0558276176452637,
    "eval_rewards/accuracies": 0.7583892345428467,
    "eval_rewards/chosen": -17.05968475341797,
    "eval_rewards/margins": 4.890104293823242,
    "eval_rewards/rejected": -21.949787139892578,
    "eval_runtime": 2126.294,
    "eval_samples_per_second": 1.401,
    "eval_steps_per_second": 0.35,
    "total_flos": 227674672136192.0,
    "train_loss": 0.0,
    "train_runtime": 1.6273,
    "train_samples_per_second": 34774.982,
    "train_steps_per_second": 271.612
}