THOUGHTSCULPT: Reasoning with Intermediate Revision and Search

arXiv:2404.05966v1 Announce Type: new
Abstract: We present THOUGHTSCULPT, a general reasoning and search method for tasks with outputs that can be decomposed into components. THOUGHTSCULPT explores a search tree of potential solutions using Monte Carlo Tree Search (MCTS), building solutions one action at a time and evaluating according to any domain-specific heuristic, which in practice is often simply an LLM evaluator. Critically, our action space includes revision actions: THOUGHTSCULPT may choose to revise part of its previous output rather than …

abstract arxiv building components cs.ai cs.cl domain general intermediate reasoning search solutions tasks tree type