Hugging Face Daily Papers · · 3 min read

FineVerify: Scaling Test-Time Compute with Fine-Grained Self-Verification for Agentic Search

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

Code and data are available at: <a href=\"https://github.com/XuZhao0/fineverify\" rel=\"nofollow\">https://github.com/XuZhao0/fineverify</a></p>\n","updatedAt":"2026-06-02T07:52:47.791Z","author":{"_id":"64ca18318d2d187c24df20ec","avatarUrl":"/avatars/cada297547bf4c84934c6196d2ee6abd.svg","fullname":"James X. Zhao","name":"JamesXZ","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.5939863920211792},"editors":["JamesXZ"],"editorAvatarUrls":["/avatars/cada297547bf4c84934c6196d2ee6abd.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.00660","authors":[{"_id":"6a1e8b91808ddbc3c7d43f73","name":"James Xu Zhao","hidden":false},{"_id":"6a1e8b91808ddbc3c7d43f74","name":"Hui Chen","hidden":false},{"_id":"6a1e8b91808ddbc3c7d43f75","name":"Bryan Hooi","hidden":false},{"_id":"6a1e8b91808ddbc3c7d43f76","name":"See-Kiong Ng","hidden":false}],"publishedAt":"2026-05-30T00:00:00.000Z","submittedOnDailyAt":"2026-06-02T00:00:00.000Z","title":"FineVerify: Scaling Test-Time Compute with Fine-Grained Self-Verification for Agentic Search","submittedOnDailyBy":{"_id":"64ca18318d2d187c24df20ec","avatarUrl":"/avatars/cada297547bf4c84934c6196d2ee6abd.svg","isPro":false,"fullname":"James X. Zhao","user":"JamesXZ","type":"user","name":"JamesXZ"},"summary":"Agentic search requires language model agents to explore many sources and answer complex information-seeking questions. Scaling test-time compute is a promising way to improve these agents, but current approaches can fail, because correct answers are often sparse and score-based selection depends on model calibration. We propose FineVerify, a fine-grained self-verification framework that decomposes each question into checkable sub-questions, verifies sampled candidates against each sub-question, and selects the candidate with the highest aggregated score. This per-check structure turns selection into simpler local judgments and produces scores under the same explicit criteria. Across four agentic search benchmarks and two models, FineVerify consistently outperforms standard scaling baselines. With only four sampled trajectories, it improves GPT-5-mini by 8.2 accuracy points and Gemini-3-flash by 5.6% on average. With 12 samples, FineVerify enables GPT-5-mini to surpass frontier GPT-5 on BrowseComp-Plus. Beyond accuracy, FineVerify produces interpretable verification traces that help audit benchmark errors, suggesting broader applications for inspecting agentic search systems. Code and data are available at https://github.com/XuZhao0/fineverify","upvotes":2,"discussionId":"6a1e8b91808ddbc3c7d43f77","githubRepo":"https://github.com/XuZhao0/fineverify","githubRepoAddedBy":"user","ai_summary":"FineVerify is a self-verification framework for agentic search that improves accuracy through decomposed sub-question checking and trajectory selection.","ai_keywords":["agentic search","language model agents","test-time compute","score-based selection","fine-grained self-verification","checkable sub-questions","sampled candidates","aggregated score","trajectory selection","BrowseComp-Plus"],"githubStars":2},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64ca18318d2d187c24df20ec","avatarUrl":"/avatars/cada297547bf4c84934c6196d2ee6abd.svg","isPro":false,"fullname":"James X. Zhao","user":"JamesXZ","type":"user"},{"_id":"61166c4328c98bfd5b92e7c5","avatarUrl":"/avatars/f4bb0f0cc2c5b84428c28bddaa479b61.svg","isPro":false,"fullname":"Hui Chen","user":"chchenhui","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0}">
Papers
arxiv:2606.00660

FineVerify: Scaling Test-Time Compute with Fine-Grained Self-Verification for Agentic Search

Published on May 30
· Submitted by
James X. Zhao
on Jun 2
Authors:
,
,
,

Abstract

FineVerify is a self-verification framework for agentic search that improves accuracy through decomposed sub-question checking and trajectory selection.

AI-generated summary

Agentic search requires language model agents to explore many sources and answer complex information-seeking questions. Scaling test-time compute is a promising way to improve these agents, but current approaches can fail, because correct answers are often sparse and score-based selection depends on model calibration. We propose FineVerify, a fine-grained self-verification framework that decomposes each question into checkable sub-questions, verifies sampled candidates against each sub-question, and selects the candidate with the highest aggregated score. This per-check structure turns selection into simpler local judgments and produces scores under the same explicit criteria. Across four agentic search benchmarks and two models, FineVerify consistently outperforms standard scaling baselines. With only four sampled trajectories, it improves GPT-5-mini by 8.2 accuracy points and Gemini-3-flash by 5.6% on average. With 12 samples, FineVerify enables GPT-5-mini to surpass frontier GPT-5 on BrowseComp-Plus. Beyond accuracy, FineVerify produces interpretable verification traces that help audit benchmark errors, suggesting broader applications for inspecting agentic search systems. Code and data are available at https://github.com/XuZhao0/fineverify

Community

Paper submitter about 2 hours ago

Code and data are available at: https://github.com/XuZhao0/fineverify

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images

· Sign up or log in to comment

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2606.00660 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2606.00660 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2606.00660 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

Discussion (0)

Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.

Sign in →

No comments yet. Sign in and be the first to say something.

More from Hugging Face Daily Papers