@inproceedings{li-etal-2026-incorporating,
title = "Incorporating Explicit and Implicit Chain-of-Thought as Hybrid Reasoning Paradigm with Guided Latent Space Modeling",
author = "Li, Guangyu and
Ge, Zhekai and
Chen, Zhuo and
Zhang, Zhen and
Wang, Xinyu and
Jiang, Yong and
Tu, Kewei",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1416/",
pages = "32329--32347",
ISBN = "979-8-89176-459-0",
abstract = "Chain-of-Thought (CoT) has proven effective for eliciting reasoning abilities in large language models, but standard autoregressive decoders treat reasoning and answer tokens uniformly as next-token prediction targets, despite their fundamentally different roles. This homogeneous modeling introduces conflicting inductive biases and significant inference overhead. Latent CoT methods attempt to improve efficiency by removing explicit reasoning tokens, yet often degrade performance, indicating that explicit reasoning supervision remains critical for preserving pretrained reasoning capabilities. To address these challenges, we propose $Pre^2$ , a phase-aware reasoning framework that decouples thinking and answering while retaining explicit CoT supervision. Our approach introduces a Guided Latent Space and predict independent pre-thinking and pre-answering latent states to guide the model to generate reasoning tokens and answering tokens. We integrates them with explicit token representations through a hybrid latent{--}token decoding paradigm. This design increases computation along both width and depth dimensions with minimal parameter overhead, enabling stronger reasoning without sacrificing inference efficiency. Training is conducted exclusively on one in-domain dataset, with direct zero-shot evaluation on six out-of-domain benchmarks, and experiments demonstrate consistent and significant improvements over strong baselines and robust generalization."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-etal-2026-incorporating">
<titleInfo>
<title>Incorporating Explicit and Implicit Chain-of-Thought as Hybrid Reasoning Paradigm with Guided Latent Space Modeling</title>
</titleInfo>
<name type="personal">
<namePart type="given">Guangyu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhekai</namePart>
<namePart type="family">Ge</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhuo</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yong</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kewei</namePart>
<namePart type="family">Tu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Chain-of-Thought (CoT) has proven effective for eliciting reasoning abilities in large language models, but standard autoregressive decoders treat reasoning and answer tokens uniformly as next-token prediction targets, despite their fundamentally different roles. This homogeneous modeling introduces conflicting inductive biases and significant inference overhead. Latent CoT methods attempt to improve efficiency by removing explicit reasoning tokens, yet often degrade performance, indicating that explicit reasoning supervision remains critical for preserving pretrained reasoning capabilities. To address these challenges, we propose Pre² , a phase-aware reasoning framework that decouples thinking and answering while retaining explicit CoT supervision. Our approach introduces a Guided Latent Space and predict independent pre-thinking and pre-answering latent states to guide the model to generate reasoning tokens and answering tokens. We integrates them with explicit token representations through a hybrid latent–token decoding paradigm. This design increases computation along both width and depth dimensions with minimal parameter overhead, enabling stronger reasoning without sacrificing inference efficiency. Training is conducted exclusively on one in-domain dataset, with direct zero-shot evaluation on six out-of-domain benchmarks, and experiments demonstrate consistent and significant improvements over strong baselines and robust generalization.</abstract>
<identifier type="citekey">li-etal-2026-incorporating</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1416/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>32329</start>
<end>32347</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Incorporating Explicit and Implicit Chain-of-Thought as Hybrid Reasoning Paradigm with Guided Latent Space Modeling
%A Li, Guangyu
%A Ge, Zhekai
%A Chen, Zhuo
%A Zhang, Zhen
%A Wang, Xinyu
%A Jiang, Yong
%A Tu, Kewei
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F li-etal-2026-incorporating
%X Chain-of-Thought (CoT) has proven effective for eliciting reasoning abilities in large language models, but standard autoregressive decoders treat reasoning and answer tokens uniformly as next-token prediction targets, despite their fundamentally different roles. This homogeneous modeling introduces conflicting inductive biases and significant inference overhead. Latent CoT methods attempt to improve efficiency by removing explicit reasoning tokens, yet often degrade performance, indicating that explicit reasoning supervision remains critical for preserving pretrained reasoning capabilities. To address these challenges, we propose Pre² , a phase-aware reasoning framework that decouples thinking and answering while retaining explicit CoT supervision. Our approach introduces a Guided Latent Space and predict independent pre-thinking and pre-answering latent states to guide the model to generate reasoning tokens and answering tokens. We integrates them with explicit token representations through a hybrid latent–token decoding paradigm. This design increases computation along both width and depth dimensions with minimal parameter overhead, enabling stronger reasoning without sacrificing inference efficiency. Training is conducted exclusively on one in-domain dataset, with direct zero-shot evaluation on six out-of-domain benchmarks, and experiments demonstrate consistent and significant improvements over strong baselines and robust generalization.
%U https://aclanthology.org/2026.emnlp-main.1416/
%P 32329-32347
Markdown (Informal)
[Incorporating Explicit and Implicit Chain-of-Thought as Hybrid Reasoning Paradigm with Guided Latent Space Modeling](https://aclanthology.org/2026.emnlp-main.1416/) (Li et al., EMNLP 2026)
ACL