Explore test-time compute scaling methods for LLMs
thought that reasoning into LLMs without modification
