eval_results.json

{
    "epoch": 1.0,
    "eval_logits/chosen": 98.21707153320312,
    "eval_logits/rejected": 98.18864440917969,
    "eval_logps/chosen": -32.851280212402344,
    "eval_logps/rejected": -36.62327575683594,
    "eval_loss": 0.9524344801902771,
    "eval_rewards/accuracies": 0.530315637588501,
    "eval_rewards/chosen": -0.08162158727645874,
    "eval_rewards/margins": 0.04972809553146362,
    "eval_rewards/rejected": -0.13134966790676117,
    "eval_runtime": 103.7326,
    "eval_samples": 343,
    "eval_samples_per_second": 3.307,
    "eval_steps_per_second": 0.415
}