← 返回论文库

Plan-and-Solve Prompting

Wang, et al. (Salesforce) · 2023
L3.2 · LLM Agent Patterns & FrameworksICLR 2023#planning
CORE IDEA
先做完整 plan 再 step-by-step 执行，和 ReAct 互补——适合 predictable subgoal。
L-ANCHOR · 为什么在这一层重要
plan-first 范式
arXiv:2305.04091 ↗

相关论文

ReAct: Synergizing Reasoning and Acting in Language Models
L3.22022
ReWOO: Decoupling Reasoning from Observations for Efficient Augmented Language Models
L3.22023