llama-3-8b-base-r-dpo-ultrafeedback-4xh200/train_results.json

{
    "epoch": 0.9989528795811519,
    "total_flos": 0.0,
    "train_loss": 4.32008807264284,
    "train_runtime": 6099.6278,
    "train_samples": 61135,
    "train_samples_per_second": 10.023,
    "train_steps_per_second": 0.078
}