SQuARE: Sequential Question Answering Reasoning Engine for Enhanced Chain-of-Thought in Large Language Models
Daniel Fleischer, Moshe Berchansky, Gad Markovits, Moshe Wasserblat
2025-02-14
Summary
This paper talks about SQuARE, a new way to make AI language models better at solving complex problems by teaching them to ask and answer their own questions before tackling the main problem.
What's the problem?
Big AI language models are getting really good at understanding language, but they still struggle with complicated reasoning tasks. The current methods, like chain-of-thought prompting, help a bit but don't fully use the AI's potential to think through problems.
What's the solution?
The researchers created SQuARE, which stands for Sequential Question Answering Reasoning Engine. It's like giving the AI a study guide that teaches it to break down big questions into smaller ones. The AI asks itself these smaller questions, answers them, and then uses all that information to solve the main problem. They tested this method on different AI models and found that it works much better than older methods for helping AI reason through tough questions.
Why it matters?
This matters because it could make AI much smarter at solving real-world problems that require careful thinking. By teaching AI to ask good questions and think step-by-step, we might be able to use it for more complex tasks in fields like science, medicine, or education. It's a big step towards making AI that can reason more like humans do when faced with tricky situations.
Abstract
In the rapidly evolving field of Natural Language Processing, Large Language Models (LLMs) are tasked with increasingly complex reasoning challenges. Traditional methods like chain-of-thought prompting have shown promise but often fall short in fully leveraging a model's reasoning capabilities. This paper introduces SQuARE (Sequential Question Answering Reasoning Engine), a novel prompting technique designed to improve reasoning through a self-interrogation paradigm. Building upon CoT frameworks, SQuARE prompts models to generate and resolve multiple auxiliary questions before tackling the main query, promoting a more thorough exploration of various aspects of a topic. Our expansive evaluations, conducted with Llama 3 and GPT-4o models across multiple question-answering datasets, demonstrate that SQuARE significantly surpasses traditional CoT prompts and existing rephrase-and-respond methods. By systematically decomposing queries, SQuARE advances LLM capabilities in reasoning tasks. The code is publicly available at https://github.com/IntelLabs/RAG-FiT/tree/square.