qwen3-1.7b-dabstep-reasoning-30-fixed-reasoning-sharegpt-sft/train_results.json

{
    "epoch": 5.0,
    "total_flos": 213154529280.0,
    "train_loss": 0.7237587083663259,
    "train_runtime": 212.7239,
    "train_samples_per_second": 0.658,
    "train_steps_per_second": 0.329
}