[2606.17609] The Benchmark Illusion: Pruned LLMs Can Pass Multiple Choice but Fail to Answer

-->

Computer Science > Computation and Language

arXiv:2606.17609 (cs)

[Submitted on 16 Jun 2026]

Title:The Benchmark Illusion: Pruned LLMs Can Pass Multiple Choice but Fail to Answer

Authors:Rui Wen, Lu Sun, Jiayang Liu, Zesheng Xu, Tianshuo Cong, Zheng Li<br>View a PDF of the paper titled The Benchmark Illusion: Pruned LLMs Can Pass Multiple Choice but Fail to Answer, by Rui Wen and 5 other authors

View PDF<br>HTML (experimental)

Abstract:Compressing large language models reduces memory use and inference cost, but it can also create failures that standard benchmarks miss. A pruned model may still perform well on multiple-choice evaluations, yet fail to answer the same question in open generation. We ask what pruning changes: does it erase the correct answer, or does it make the answer harder to produce as the top output?

We study this question with multilingual question answering, tracking the same questions before and after pruning. We find a benchmark illusion. Under high-sparsity pruning, especially Wanda, models often fail in greedy open generation while still selecting the correct answer under multiple-choice scoring. In these recognition-only errors, the answer is usually not gone, but demoted: it often reappears with beam search, sampling, or one in-context example. Overall, multiple-choice benchmarks can overstate the usability of compressed LLMs, creating an evaluation blind spot. Compressed models should be tested on what they can produce, not only on what they can recognize.

Subjects:

Computation and Language (cs.CL)

Cite as:<br>arXiv:2606.17609 [cs.CL]

(or<br>arXiv:2606.17609v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2606.17609

Focus to learn more

arXiv-issued DOI via DataCite (pending registration)

Submission history<br>From: Rui Wen [view email]<br>[v1]<br>Tue, 16 Jun 2026 07:14:52 UTC (354 KB)

Full-text links:<br>Access Paper:

View a PDF of the paper titled The Benchmark Illusion: Pruned LLMs Can Pass Multiple Choice but Fail to Answer, by Rui Wen and 5 other authors<br>View PDF<br>HTML (experimental)<br>TeX Source

view license

Current browse context:

cs.CL

next >

new<br>recent<br>| 2026-06

Change to browse by:

cs

References & Citations

NASA ADS<br>Google Scholar

Semantic Scholar

export BibTeX citation<br>Loading...

BibTeX formatted citation

&times;

loading...

Data provided by:

Bookmark

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Which authors of this paper are endorsers? |<br>Disable MathJax (What is MathJax?)