GCoT-Decoding: Unlocking Deep Reasoning Paths for Universal Question Answering

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Luo, Guanran, Qiu, Wentao, Jian, Zhongquan, Wang, Meihong, Wu, Qingqiang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910112303546368
author Luo, Guanran
Qiu, Wentao
Jian, Zhongquan
Wang, Meihong
Wu, Qingqiang
author_facet Luo, Guanran
Qiu, Wentao
Jian, Zhongquan
Wang, Meihong
Wu, Qingqiang
contents Chain-of-Thought reasoning can enhance large language models, but it requires manually designed prompts to guide the model. Recently proposed CoT-decoding enables the model to generate CoT-style reasoning paths without prompts, but it is only applicable to problems with fixed answer sets. To address this limitation, we propose a general decoding strategy GCoT-decoding that extends applicability to a broader range of question-answering tasks. GCoT-decoding employs a two-stage branching method combining Fibonacci sampling and heuristic error backtracking to generate candidate decoding paths. It then splits each path into a reasoning span and an answer span to accurately compute path confidence, and finally aggregates semantically similar paths to identify a consensus answer, replacing traditional majority voting. We conduct extensive experiments on six datasets covering both fixed and free QA tasks. Our method not only maintains strong performance on fixed QA but also achieves significant improvements on free QA, demonstrating its generality.
format Preprint
id arxiv_https___arxiv_org_abs_2604_06794
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GCoT-Decoding: Unlocking Deep Reasoning Paths for Universal Question Answering
Luo, Guanran
Qiu, Wentao
Jian, Zhongquan
Wang, Meihong
Wu, Qingqiang
Computation and Language
Chain-of-Thought reasoning can enhance large language models, but it requires manually designed prompts to guide the model. Recently proposed CoT-decoding enables the model to generate CoT-style reasoning paths without prompts, but it is only applicable to problems with fixed answer sets. To address this limitation, we propose a general decoding strategy GCoT-decoding that extends applicability to a broader range of question-answering tasks. GCoT-decoding employs a two-stage branching method combining Fibonacci sampling and heuristic error backtracking to generate candidate decoding paths. It then splits each path into a reasoning span and an answer span to accurately compute path confidence, and finally aggregates semantically similar paths to identify a consensus answer, replacing traditional majority voting. We conduct extensive experiments on six datasets covering both fixed and free QA tasks. Our method not only maintains strong performance on fixed QA but also achieves significant improvements on free QA, demonstrating its generality.
title GCoT-Decoding: Unlocking Deep Reasoning Paths for Universal Question Answering
topic Computation and Language
url https://arxiv.org/abs/2604.06794