@inproceedings{f10ca4b0cbb9458cb834d07eeed6485d,
title = "Self-Consistency from Only Two Samples: CoT{\textendash}PoT Ensembling for Efficient LLM Reasoning",
abstract = "Self-consistency (SC) is a popular technique for improving the reasoning accuracy of large language models by aggregating multiple sampled outputs, but it comes at a high computational cost due to extensive sampling. We introduce a hybrid ensembling approach that leverages the complementary strengths of two distinct modes of reasoning: Chain-of-Thought (CoT) and Program-of-Thought (PoT). We describe a general framework for combining these two forms of reasoning in self-consistency, as well as particular strategies for both full sampling and early-stopping. We show that CoT-PoT ensembling not only improves overall accuracy, but also drastically reduces the number of samples required for SC by a factor of 9.3{\texttimes}. In particular, the majority of tasks (78.6\%) can be addressed with only two samples, which has not been possible with any prior SC methods.",
author = "Raman Saparkhan and Majd Hawasly and Parvez, \{M. D.Rizwan\} and Mohammad Raza",
note = "Publisher Copyright: {\textcopyright} 2026 Association for Computational Linguistics.; 64th Annual Meeting of the Association for Computational Linguistics, ACL 2026 ; Conference date: 02-07-2026 Through 07-07-2026",
year = "2026",
language = "English",
series = "Proceedings of the Annual Meeting of the Association for Computational Linguistics",
publisher = "Association for Computational Linguistics (ACL)",
pages = "32804--32839",
editor = "Maria Liakata and Moreira, \{Viviane P.\} and Jiajun Zhang and David Jurgens",
booktitle = "64th Annual Meeting of the Association for Computational Linguistics, Findings of the Association for Computational Linguistics",
address = "United States",
}