Monte Carlo Tree Search Boosts Reasoning via Iterative Preference Learning

arXiv:2405.00451v1 Announce Type: cross
Abstract: We introduce an approach aimed at enhancing the reasoning capabilities of Large Language Models (LLMs) through an iterative preference learning process inspired by the successful strategy employed by AlphaZero. Our work leverages Monte Carlo Tree Search (MCTS) to iteratively collect preference data, utilizing its look-ahead ability to break down instance-level rewards into more granular step-level signals. To enhance consistency in intermediate steps, we combine outcome validation and stepwise self-evaluation, continually updating the quality assessment of …

abstract alphazero arxiv capabilities cs.ai cs.lg data iterative language language models large language large language models llms look process reasoning search strategy through tree type via work