Tree of Problems：通过组合性提升结构化问题求解能力
Tree of Problems: Improving structured problem solving with compositionality
AI总结：
针对大语言模型复杂推理难题，提出更简化的思维树变体Tree of Problems（ToP），适用于可拆解为相同子任务的复杂任务，实验证明其性能优于ToT、GoT及思维链提示，相关代码已开源。
摘要：
大语言模型（LLM）通过上下文学习在多项任务中展现出卓越性能。对于需要逐步思考的复杂推理任务，思维链（Chain-of-Thought, CoT）提示已取得令人瞩目的成果，尤其是与自洽性策略结合时效果更佳。尽管如此，部分任务对LLM而言仍然极具挑战性。思维树（Tree of Thoughts, ToT）和思维图（Graph of Thoughts, GoT）作为替代方案应运而生，它们将复杂问题拆分为由子问题构成的路径。本文提出Tree of Problems（ToP，问题树）——一种更简化的ToT变体，我们假设它在可拆解为相同子任务的复杂任务上能发挥更优效果。实证结果表明，我们的方法性能优于ToT和GoT，此外在复杂推理任务上的表现也优于CoT。本文所有代码已公开于：https://github.com/ArmelRandy/tree-of-problems。
英文摘要：
Large Language Models (LLMs) have demonstrated remarkable performance across multiple tasks through in-context learning. For complex reasoning tasks that require step-by-step thinking, Chain-of-Thought (CoT) prompting has given impressive results, especially when combined with self-consistency. Nonetheless, some tasks remain particularly difficult for LLMs to solve. Tree of Thoughts (ToT) and Graph of Thoughts (GoT) emerged as alternatives, dividing the complex problem into paths of subproblems. In this paper, we propose Tree of Problems (ToP), a simpler version of ToT, which we hypothesise can work better for complex tasks that can be divided into identical subtasks. Our empirical results show that our approach outperforms ToT and GoT, and in addition performs better than CoT on complex reasoning tasks. All code for this paper is publicly available here: https://github.com/ArmelRandy/tree-of-problems.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
