PathFinder: MCTS and LLM Feedback-based Path Selection for Multi-Hop Question Answering
This work addresses multi-hop QA for language model applications, presenting an incremental improvement over existing methods.
The paper tackles the problem of multi-hop question answering by addressing LLM hallucinations and incorrect reasoning paths in training-based approaches, proposing PATHFINDER which uses Monte Carlo Tree Search and LLM feedback to improve training data quality, resulting in performance improvements on public benchmark datasets.
Multi-hop question answering is a challenging task in which language models must reason over multiple steps to reach the correct answer. With the help of Large Language Models and their reasoning capabilities, existing systems are able to think and decompose an input question over multiple steps to analyze, retrieve, and reason. However, training-based approaches for this problem still suffer from LLM hallucinations and incorrect reasoning paths that hinder performance. Hence, we propose PATHFINDER, an approach that: (i) uses Monte Carlo Tree Search to generate training path traces, (ii) improves training data quality by filtering erroneous and lengthy traces using sub-answer recall and LLM-as-a-judge verification, and (iii) reformulates sub-queries to handle failed retrieval cases. By following these steps, we demonstrate that PATHFINDER improves the performance of multi-hop QA over public benchmark datasets.