It will definitely work better with larger datasets, we have plans to improve quality.