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

{
    "epoch": 0.9989528795811519,
    "total_flos": 0.0,
    "train_loss": 1.6700679651096433,
    "train_runtime": 3395.1223,
    "train_samples": 61135,
    "train_samples_per_second": 18.007,
    "train_steps_per_second": 0.14
}