Self-Consistency Improves Chain of Thought Reasoning in Language Models
2022/2023 — parallel CoT paths with aggregation (ICLR 2023)
Sample multiple chain-of-thought reasoning paths and select the most consistent answer — parallel path topology.
Wang, X., Wei, J., Schuurmans, D., Le, Q., Chi, E., Narang, S., Chowdhery, A., & Zhou, D. (2023). Self-Consistency Improves Chain of Thought Reasoning in Language Models. International Conference on Learning Representations (ICLR 2023); preprint arXiv:2203.11171.
Why we cite it
Documents parallel linear traces — several CoT paths from one prompt, then aggregate — as a distinct reasoning topology (multiple paths, single merge). Related to multi-agent fan-out on the event log under one correlationId (Thinking by Writing).
Claims we use
- Parallel paths — diverse reasoning paths before answer selection.
- Merge point — final answer is a function of many traces, not one chain alone.
Author (primary)
Xuezhi Wang — full author list in the original paper.
Corpus stance
Context — decoding strategy in single-model prompting; we cite topology shape only.