Bayesian Control for Coding Agents

Theodore Papamarkou; Vladislav Smirnov; Viktor Mazanov; Artem Vazhentsev; Preslav Nakov; Timothy Baldwin; Artem Shelmanov

Bayesian Control for Coding Agents

Theodore Papamarkou, Vladislav Smirnov, Viktor Mazanov, Artem Vazhentsev, Preslav Nakov, Timothy Baldwin, Artem Shelmanov

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

Modern coding agents pair LLM generators with various tools, including cheap diagnostics and expensive verifiers. The tool-use decisions are typically governed by orchestrators that often use fixed rules and ignore uncertainty. We formulate orchestration as cost-sensitive sequential hypothesis testing: a Bayesian controller maintains a belief over candidate correctness and dynamically decides whether to gather more evidence, refine the candidate, verify it, or stop. Across six generators and nine coding benchmarks, Bayesian control proves to be most valuable when verification is costly and critics are informative but imperfect. Beyond control, the belief state yields an interpretable correctness score that outperforms token-probability and raw tool-success baselines for uncertainty quantification.

Anthology ID:: 2026.emnlp-main.1548
Volume:: Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: EMNLP
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 35299–35341
Language:
URL:: https://aclanthology.org/2026.emnlp-main.1548/
DOI:
Bibkey:
Cite (ACL):: Theodore Papamarkou, Vladislav Smirnov, Viktor Mazanov, Artem Vazhentsev, Preslav Nakov, Timothy Baldwin, and Artem Shelmanov. 2026. Bayesian Control for Coding Agents. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 35299–35341, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: Bayesian Control for Coding Agents (Papamarkou et al., EMNLP 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.emnlp-main.1548.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{papamarkou-etal-2026-bayesian,
    title = "{B}ayesian Control for Coding Agents",
    author = "Papamarkou, Theodore  and
      Smirnov, Vladislav  and
      Mazanov, Viktor  and
      Vazhentsev, Artem  and
      Nakov, Preslav  and
      Baldwin, Timothy  and
      Shelmanov, Artem",
    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.1548/",
    pages = "35299--35341",
    ISBN = "979-8-89176-459-0",
    abstract = "Modern coding agents pair LLM generators with various tools, including cheap diagnostics and expensive verifiers. The tool-use decisions are typically governed by orchestrators that often use fixed rules and ignore uncertainty. We formulate orchestration as cost-sensitive sequential hypothesis testing: a Bayesian controller maintains a belief over candidate correctness and dynamically decides whether to gather more evidence, refine the candidate, verify it, or stop. Across six generators and nine coding benchmarks, Bayesian control proves to be most valuable when verification is costly and critics are informative but imperfect. Beyond control, the belief state yields an interpretable correctness score that outperforms token-probability and raw tool-success baselines for uncertainty quantification."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="papamarkou-etal-2026-bayesian">
    <titleInfo>
        <title>Bayesian Control for Coding Agents</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Theodore</namePart>
        <namePart type="family">Papamarkou</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Vladislav</namePart>
        <namePart type="family">Smirnov</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Viktor</namePart>
        <namePart type="family">Mazanov</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Artem</namePart>
        <namePart type="family">Vazhentsev</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Preslav</namePart>
        <namePart type="family">Nakov</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Timothy</namePart>
        <namePart type="family">Baldwin</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Artem</namePart>
        <namePart type="family">Shelmanov</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>Modern coding agents pair LLM generators with various tools, including cheap diagnostics and expensive verifiers. The tool-use decisions are typically governed by orchestrators that often use fixed rules and ignore uncertainty. We formulate orchestration as cost-sensitive sequential hypothesis testing: a Bayesian controller maintains a belief over candidate correctness and dynamically decides whether to gather more evidence, refine the candidate, verify it, or stop. Across six generators and nine coding benchmarks, Bayesian control proves to be most valuable when verification is costly and critics are informative but imperfect. Beyond control, the belief state yields an interpretable correctness score that outperforms token-probability and raw tool-success baselines for uncertainty quantification.</abstract>
    <identifier type="citekey">papamarkou-etal-2026-bayesian</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.1548/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>35299</start>
            <end>35341</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T Bayesian Control for Coding Agents
%A Papamarkou, Theodore
%A Smirnov, Vladislav
%A Mazanov, Viktor
%A Vazhentsev, Artem
%A Nakov, Preslav
%A Baldwin, Timothy
%A Shelmanov, Artem
%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 papamarkou-etal-2026-bayesian
%X Modern coding agents pair LLM generators with various tools, including cheap diagnostics and expensive verifiers. The tool-use decisions are typically governed by orchestrators that often use fixed rules and ignore uncertainty. We formulate orchestration as cost-sensitive sequential hypothesis testing: a Bayesian controller maintains a belief over candidate correctness and dynamically decides whether to gather more evidence, refine the candidate, verify it, or stop. Across six generators and nine coding benchmarks, Bayesian control proves to be most valuable when verification is costly and critics are informative but imperfect. Beyond control, the belief state yields an interpretable correctness score that outperforms token-probability and raw tool-success baselines for uncertainty quantification.
%U https://aclanthology.org/2026.emnlp-main.1548/
%P 35299-35341

Download as File

Markdown (Informal)

[Bayesian Control for Coding Agents](https://aclanthology.org/2026.emnlp-main.1548/) (Papamarkou et al., EMNLP 2026)

Bayesian Control for Coding Agents (Papamarkou et al., EMNLP 2026)

ACL

Theodore Papamarkou, Vladislav Smirnov, Viktor Mazanov, Artem Vazhentsev, Preslav Nakov, Timothy Baldwin, and Artem Shelmanov. 2026. Bayesian Control for Coding Agents. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 35299–35341, Budapest, Hungary. Association for Computational Linguistics.