all_results.json

{
    "epoch": 0.9998604326587579,
    "eval_logits/chosen": -1.0269631147384644,
    "eval_logits/rejected": -1.061506986618042,
    "eval_logps/chosen": -112.56452178955078,
    "eval_logps/rejected": -272.78466796875,
    "eval_loss": 0.000979769160039723,
    "eval_rewards/accuracies": 0.9996050596237183,
    "eval_rewards/chosen": -3.615877628326416,
    "eval_rewards/margins": 16.76466178894043,
    "eval_rewards/rejected": -20.380538940429688,
    "eval_runtime": 151.8806,
    "eval_samples_per_second": 33.302,
    "eval_steps_per_second": 4.168
}