llama-3-8b-base-cpo-ultrafeedback-4xH200-batch-128-rerun/train_results.json

{
    "epoch": 0.9989528795811519,
    "total_flos": 0.0,
    "train_loss": 17.025300079921507,
    "train_runtime": 4689.5603,
    "train_samples": 61135,
    "train_samples_per_second": 13.036,
    "train_steps_per_second": 0.102
}