SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions

Timofey Garaev; Makar Gorokhovik; Alexey Korotkov; Maksim Savkin; Iurii Vizilter; Andrey Savchenko

SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions

Timofey Garaev, Makar Gorokhovik, Alexey Korotkov, Maksim Savkin, Iurii Vizilter, Andrey Savchenko

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

We introduce SHARP, a compact, span-annotated dataset for hallucination detection in model-generated mathematical solutions. Unlike existing resources that rely on coarse step-level labels, SHARP provides fine-grained, span-level annotations generated through an automated pipeline. Our methodology synthesizes realistic reasoning traces with precise error markup, enabling the production of high-fidelity training data with minimal computational overhead. Despite being 10–30 tim smaller than existing open-source datasets, a 1.5B PRM fine-tuned on SHARP achieves state-of-the-art performance on ProcessBench.

Anthology ID:: 2026.findings-emnlp.500
Volume:: Findings of the Association for Computational Linguistics: EMNLP 2026
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: Findings
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 10470–10487
Language:
URL:: https://aclanthology.org/2026.findings-emnlp.500/
DOI:
Bibkey:
Cite (ACL):: Timofey Garaev, Makar Gorokhovik, Alexey Korotkov, Maksim Savkin, Iurii Vizilter, and Andrey Savchenko. 2026. SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 10470–10487, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions (Garaev et al., Findings 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.findings-emnlp.500.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{garaev-etal-2026-sharp,
    title = "{SHARP}: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions",
    author = "Garaev, Timofey  and
      Gorokhovik, Makar  and
      Korotkov, Alexey  and
      Savkin, Maksim  and
      Vizilter, Iurii  and
      Savchenko, Andrey",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-emnlp.500/",
    pages = "10470--10487",
    ISBN = "979-8-89176-462-0",
    abstract = "We introduce SHARP, a compact, span-annotated dataset for hallucination detection in model-generated mathematical solutions. Unlike existing resources that rely on coarse step-level labels, SHARP provides fine-grained, span-level annotations generated through an automated pipeline. Our methodology synthesizes realistic reasoning traces with precise error markup, enabling the production of high-fidelity training data with minimal computational overhead. Despite being 10{--}30 tim smaller than existing open-source datasets, a 1.5B PRM fine-tuned on SHARP achieves state-of-the-art performance on ProcessBench."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="garaev-etal-2026-sharp">
    <titleInfo>
        <title>SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Timofey</namePart>
        <namePart type="family">Garaev</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Makar</namePart>
        <namePart type="family">Gorokhovik</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Alexey</namePart>
        <namePart type="family">Korotkov</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Maksim</namePart>
        <namePart type="family">Savkin</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Iurii</namePart>
        <namePart type="family">Vizilter</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Andrey</namePart>
        <namePart type="family">Savchenko</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>We introduce SHARP, a compact, span-annotated dataset for hallucination detection in model-generated mathematical solutions. Unlike existing resources that rely on coarse step-level labels, SHARP provides fine-grained, span-level annotations generated through an automated pipeline. Our methodology synthesizes realistic reasoning traces with precise error markup, enabling the production of high-fidelity training data with minimal computational overhead. Despite being 10–30 tim smaller than existing open-source datasets, a 1.5B PRM fine-tuned on SHARP achieves state-of-the-art performance on ProcessBench.</abstract>
    <identifier type="citekey">garaev-etal-2026-sharp</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.500/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>10470</start>
            <end>10487</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions
%A Garaev, Timofey
%A Gorokhovik, Makar
%A Korotkov, Alexey
%A Savkin, Maksim
%A Vizilter, Iurii
%A Savchenko, Andrey
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F garaev-etal-2026-sharp
%X We introduce SHARP, a compact, span-annotated dataset for hallucination detection in model-generated mathematical solutions. Unlike existing resources that rely on coarse step-level labels, SHARP provides fine-grained, span-level annotations generated through an automated pipeline. Our methodology synthesizes realistic reasoning traces with precise error markup, enabling the production of high-fidelity training data with minimal computational overhead. Despite being 10–30 tim smaller than existing open-source datasets, a 1.5B PRM fine-tuned on SHARP achieves state-of-the-art performance on ProcessBench.
%U https://aclanthology.org/2026.findings-emnlp.500/
%P 10470-10487

Download as File

Markdown (Informal)

[SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions](https://aclanthology.org/2026.findings-emnlp.500/) (Garaev et al., Findings 2026)

SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions (Garaev et al., Findings 2026)

ACL

Timofey Garaev, Makar Gorokhovik, Alexey Korotkov, Maksim Savkin, Iurii Vizilter, and Andrey Savchenko. 2026. SHARP: Span-level Hallucination Annotation for Reasoning Paths in Generated Mathematical Solutions. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 10470–10487, Budapest, Hungary. Association for Computational Linguistics.