eval_results.json

{
    "epoch": 0.9987943737441393,
    "eval_logits/chosen": -2.8429527282714844,
    "eval_logits/rejected": -2.840003728866577,
    "eval_logps/chosen": -0.8574438691139221,
    "eval_logps/rejected": -0.9005469679832458,
    "eval_loss": 1.5836583375930786,
    "eval_rewards/accuracies": 0.5598404407501221,
    "eval_rewards/chosen": -0.8574438691139221,
    "eval_rewards/margins": 0.04310305044054985,
    "eval_rewards/rejected": -0.9005469679832458,
    "eval_runtime": 436.1865,
    "eval_samples": 2994,
    "eval_samples_per_second": 6.864,
    "eval_steps_per_second": 0.431
}