Think&Cite: Improving Attributed Text Generation with Self-Guided Tree Search and Progress Reward Modeling
This addresses the issue of factual inaccuracies in LLM-generated text for users relying on reliable information, representing a novel method rather than an incremental improvement.
The paper tackles the problem of hallucination in large language models by proposing Think&Cite, a framework for attributed text generation that formulates it as a multi-step reasoning problem with search, resulting in significant outperformance over baseline approaches on three datasets.
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.