Think&Cite: Improving Attributed Text Generation with Self-Guided Tree Search and Progress Reward Modeling

Despite their outstanding capabilities, large language models (LLMs) are prone to hallucination and producing factually incorrect information. This challenge has spurred efforts in attributed text generation, which prompts LLMs to generate content with supporting evidence. In this paper, we propose a novel framework, called Think&Cite, and formulate attributed text generation as a multi-step reasoning problem integrated with search. Specifically, we propose Self-Guided Monte Carlo Tree Search (SG-MCTS), which capitalizes on the self-reflection capability of LLMs to reason about the intermediate states of MCTS for guiding the tree expansion process. To provide reliable and comprehensive feedback, we introduce Progress Reward Modeling to measure the progress of tree search from the root to the current state from two aspects, i.e., generation and attribution progress. We conduct extensive experiments on three datasets and the results show that our approach significantly outperforms baseline approaches. 1 Which is the most rainy place on earth? Question Lloró, Colombia reported an average annual rainfall of 12,717 mm between 1952 and 1989 [1]. However, the official record is held by Mawsynram, India with an average annual rainfall of 11,872 mm [2], Cherrapunji holds the record for most rain in a calendar month for July 1861 and most rain in a year from August 1860 to July 1861 [3].