FSM: A Finite State Machine Based Zero-Shot Prompting Paradigm for Multi-Hop Question Answering
This addresses MHQA challenges like hallucination and error propagation for users of large language models, representing a novel method for a known bottleneck rather than a foundational advance.
The paper tackles multi-hop question answering (MHQA) by proposing a Finite State Machine (FSM) prompting method that iteratively decomposes questions into sub-questions and self-corrects, achieving improved accuracy on challenging datasets like Musique and reducing hallucination.
Large Language Models (LLMs) with chain-of-thought (COT) prompting have demonstrated impressive abilities on simple nature language inference tasks. However, they tend to perform poorly on Multi-hop Question Answering (MHQA) tasks due to several challenges, including hallucination, error propagation and limited context length. We propose a prompting method, Finite State Machine (FSM) to enhance the reasoning capabilities of LLM for complex tasks in addition to improved effectiveness and trustworthiness. Different from COT methods, FSM addresses MHQA by iteratively decomposing a question into multi-turn sub-questions, and self-correcting in time, improving the accuracy of answers in each step. Specifically, FSM addresses one sub-question at a time and decides on the next step based on its current result and state, in an automaton-like format. Experiments on benchmarks show the effectiveness of our method. Although our method performs on par with the baseline on relatively simpler datasets, it excels on challenging datasets like Musique. Moreover, this approach mitigates the hallucination phenomenon, wherein the correct final answer can be recovered despite errors in intermediate reasoning. Furthermore, our method improves LLMs' ability to follow specified output format requirements, significantly reducing the difficulty of answer interpretation and the need for reformatting.