Interesting viewpoint! I look forward to learning more about the limitations of LLMs for numerical reasoning!