all_results.json

{
    "epoch": 0.9880609304240429,
    "eval_logits/chosen": 9.238580703735352,
    "eval_logits/rejected": 10.054126739501953,
    "eval_logps/chosen": -0.3649774193763733,
    "eval_logps/rejected": -0.3733733296394348,
    "eval_loss": 1.383278250694275,
    "eval_rewards/accuracies": 0.5259259343147278,
    "eval_rewards/chosen": -0.7299548387527466,
    "eval_rewards/margins": 0.016791854053735733,
    "eval_rewards/rejected": -0.7467466592788696,
    "eval_runtime": 163.2381,
    "eval_samples": 1080,
    "eval_samples_per_second": 6.616,
    "eval_steps_per_second": 1.654,
    "total_flos": 0.0,
    "train_loss": 1.372072113355001,
    "train_runtime": 7572.4989,
    "train_samples": 9715,
    "train_samples_per_second": 1.283,
    "train_steps_per_second": 0.01
}