Chain-of-Thoughts Prompting with Language Models for Accurate Math Problem-Solving

Large Language Models (LLMs) have gained usage across various domains, especially in education. However, the current state-of-the-art LLMs fail in numerical calculations due to their reliance on the pre-trained dataset that does not focus on mathematical oversight. Prompting is crucial to guide LLMs to yield desired outputs for mathematical problems. This paper explores a new Chain-of-Thoughts (CoT) prompting framework, leveraging Python-based tools like LLM Math, LLM symbolic math, and SerpAPI. We also evaluate the existing works with the CoT prompting framework for math problem-solving. Students can utilize this framework to obtain more precise solutions and comprehensive explanations for their queries.