Self-Consistency Improves Chain of Thought Reasoning in Language Models (arXiv:2203.11171)

arXiv:2203.11171. Self-Consistency Improves Chain of Thought Reasoning in Language Models. By Xuezhi Wang et al. (Google). Submitted 2022-03-21. Improves chain-of-thought reasoning by sampling multiple reasoning paths and taking the majority answer. 2026-10-10: Fetched https://arxiv.org/abs/2203.11171. Title: 'Self-Consistency Improves Chain of Thought Reasoning in Language Models'. Peer-reviewed venue confirmed: ICLR 2023. Abstract confirms self-consistency decoding strategy sampling multiple reasoning paths to find marginal consensus.