qwen3-8b-finetuned-train/train_results.json

{
    "epoch": 3.0,
    "total_flos": 4.717303430077481e+17,
    "train_loss": 0.15186350211219263,
    "train_runtime": 38487.8681,
    "train_samples_per_second": 1.115,
    "train_steps_per_second": 0.139
}