eval_results.json

{
    "epoch": 1.971563981042654,
    "eval_logits/chosen": 111.36813354492188,
    "eval_logits/rejected": 106.11373901367188,
    "eval_logps/chosen": -332.9066467285156,
    "eval_logps/rejected": -364.4402160644531,
    "eval_loss": 0.2980845868587494,
    "eval_rewards/accuracies": 0.6979166865348816,
    "eval_rewards/chosen": 1.5380783081054688,
    "eval_rewards/margins": 1.6565409898757935,
    "eval_rewards/rejected": -0.11846256256103516,
    "eval_runtime": 119.8229,
    "eval_samples": 750,
    "eval_samples_per_second": 6.259,
    "eval_steps_per_second": 0.2
}