90_variance/mistral-dpo-lr-5.0e-7-beta-0.01/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 0.642561550393249,
    "train_runtime": 7450.8709,
    "train_samples": 16889,
    "train_samples_per_second": 2.267,
    "train_steps_per_second": 0.035
}