← 返回论文库

Tree of Thoughts: Deliberate Problem Solving with Large Language Models

Yao, Yu, Zhao, et al. (Princeton) · 2023
L3.1 · LLM Agent Patterns & FrameworksNeurIPS 2023#reasoning#tree-search
CORE IDEA
树形 reasoning：每个 thought 是 state，LLM 同时评估多个 thought 选最有希望的扩展，BFS/DFS/beam 可选。
CONCRETE EXAMPLE
Game of 24 上 4% → 74% 解题率（GPT-4 base）。
L-ANCHOR · 为什么在这一层重要
tree-structured reasoning
arXiv:2305.10601 ↗源码 ↗

相关论文

AI Scientist v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search
L12025
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
L3.12022
Self-Consistency Improves Chain of Thought Reasoning in Language Models
L3.12022
Graph of Thoughts: Solving Elaborate Problems with Large Language Models
L3.12023