@inproceedings{xiao-zhong-2026-latent,
title = "Latent Thoughts, Certain Gains: {B}ayesian Optimization for {LLM} Reasoning",
author = "Xiao, Xin and
Zhong, Jiang",
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.1128/",
pages = "25844--25853",
ISBN = "979-8-89176-459-0",
abstract = "Recent advances in Large Language Models (LLMs) have shifted toward latent reasoning, where intermediate processes are represented as continuous vectors rather than text tokens. While this paradigm reduces inference cost, existing approaches often suffer from instability on complex, out-of-distribution tasks where robust reasoning is critical. To address this, we propose Bayesian Latent Thought Optimization (BLTO), a training-free framework that enhances frozen LLMs entirely at test time. BLTO treats latent thought tokens as stochastic variables and formulates reasoning as a Bayesian inference problem. We introduce a variational posterior over latent vectors and optimize it by maximizing a reward-weighted objective that balances reasoning quality with KL regularization toward a standard Gaussian prior. The reward is derived from the model{'}s predictive consistency via output entropy, enabling fully self-supervised optimization without external supervision. BLTO employs reparameterized sampling and reward-weighted updates within a variational inference framework, together with an adaptive controller that dynamically determines inference steps based on reward improvement and uncertainty convergence. Extensive experiments on challenging reasoning benchmarks demonstrate that BLTO consistently improves frozen LLM performance."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="xiao-zhong-2026-latent">
<titleInfo>
<title>Latent Thoughts, Certain Gains: Bayesian Optimization for LLM Reasoning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Xiao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiang</namePart>
<namePart type="family">Zhong</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>Recent advances in Large Language Models (LLMs) have shifted toward latent reasoning, where intermediate processes are represented as continuous vectors rather than text tokens. While this paradigm reduces inference cost, existing approaches often suffer from instability on complex, out-of-distribution tasks where robust reasoning is critical. To address this, we propose Bayesian Latent Thought Optimization (BLTO), a training-free framework that enhances frozen LLMs entirely at test time. BLTO treats latent thought tokens as stochastic variables and formulates reasoning as a Bayesian inference problem. We introduce a variational posterior over latent vectors and optimize it by maximizing a reward-weighted objective that balances reasoning quality with KL regularization toward a standard Gaussian prior. The reward is derived from the model’s predictive consistency via output entropy, enabling fully self-supervised optimization without external supervision. BLTO employs reparameterized sampling and reward-weighted updates within a variational inference framework, together with an adaptive controller that dynamically determines inference steps based on reward improvement and uncertainty convergence. Extensive experiments on challenging reasoning benchmarks demonstrate that BLTO consistently improves frozen LLM performance.</abstract>
<identifier type="citekey">xiao-zhong-2026-latent</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1128/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>25844</start>
<end>25853</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Latent Thoughts, Certain Gains: Bayesian Optimization for LLM Reasoning
%A Xiao, Xin
%A Zhong, Jiang
%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 xiao-zhong-2026-latent
%X Recent advances in Large Language Models (LLMs) have shifted toward latent reasoning, where intermediate processes are represented as continuous vectors rather than text tokens. While this paradigm reduces inference cost, existing approaches often suffer from instability on complex, out-of-distribution tasks where robust reasoning is critical. To address this, we propose Bayesian Latent Thought Optimization (BLTO), a training-free framework that enhances frozen LLMs entirely at test time. BLTO treats latent thought tokens as stochastic variables and formulates reasoning as a Bayesian inference problem. We introduce a variational posterior over latent vectors and optimize it by maximizing a reward-weighted objective that balances reasoning quality with KL regularization toward a standard Gaussian prior. The reward is derived from the model’s predictive consistency via output entropy, enabling fully self-supervised optimization without external supervision. BLTO employs reparameterized sampling and reward-weighted updates within a variational inference framework, together with an adaptive controller that dynamically determines inference steps based on reward improvement and uncertainty convergence. Extensive experiments on challenging reasoning benchmarks demonstrate that BLTO consistently improves frozen LLM performance.
%U https://aclanthology.org/2026.emnlp-main.1128/
%P 25844-25853
Markdown (Informal)
[Latent Thoughts, Certain Gains: Bayesian Optimization for LLM Reasoning](https://aclanthology.org/2026.emnlp-main.1128/) (Xiao & Zhong, EMNLP 2026)
ACL