SDAR-4B-Math-R1-CoT-SFT/all_results.json

{
    "effective_tokens_per_sec": 16342.776477265266,
    "epoch": 3.9988184324537217,
    "total_flos": 4.3565600076592054e+19,
    "train_loss": 1.895302517482642,
    "train_runtime": 61070.1581,
    "train_samples_per_second": 2.661,
    "train_steps_per_second": 0.042
}