all_results.json

{
    "epoch": 0.9982631930527722,
    "eval_logits/chosen": 0.07677101343870163,
    "eval_logits/rejected": 0.05415331944823265,
    "eval_logps/chosen": -0.46171072125434875,
    "eval_logps/rejected": -0.6754984259605408,
    "eval_loss": 1.1737171411514282,
    "eval_rewards/accuracies": 0.7642276287078857,
    "eval_rewards/chosen": -0.46171072125434875,
    "eval_rewards/margins": 0.21378767490386963,
    "eval_rewards/rejected": -0.6754984259605408,
    "eval_runtime": 423.2412,
    "eval_samples": 1961,
    "eval_samples_per_second": 4.633,
    "eval_steps_per_second": 0.291,
    "total_flos": 0.0,
    "train_loss": 1.2022870587587866,
    "train_runtime": 20947.8363,
    "train_samples": 59876,
    "train_samples_per_second": 2.858,
    "train_steps_per_second": 0.022
}