Self-Consistency Improves Chain of Thought Reasoning in Language Models

Authors: Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, Denny Zhou (2022)

arXiv: 2203.11171

TLDR (English)

Self-Consistency is a key improvement to CoT: instead of greedy decoding a single reasoning chain, sample multiple diverse reasoning paths and take the most frequent answer (majority vote). This simple trick improves accuracy by 10-20 percentage points on multiple reasoning benchmarks.

TLDR（中文）

自洽性（Self-Consistency）是对 CoT 的重要改进：不再贪心解码单条推理链， 而是采样多条不同的推理路径，然后取答案出现最多的那个（多数投票）。 这简单的技巧在多个推理基准上将准确率提升了 10-20 个百分点。