Hugging Face Daily Papers · August 31, 2026 · 3 min read

J-Zero: Unified Challenger--Solver--Judge Co-Evolution from Zero Data

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

Like Read original ↗

<a href=\"https://cdn-uploads.huggingface.co/production/uploads/64ad53921db46d71a5c576e5/LBp_HtTtTd21i3nUFQitN.jpeg\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/64ad53921db46d71a5c576e5/LBp_HtTtTd21i3nUFQitN.jpeg\" alt=\"teaser\"></a><br><strong>Frozen Judge set the upper bound of self-evolving LLMs. By co-evolving Judge with the current frontier of self-evolution, a highly performant and sustained self-evolution can be realized.</strong></p>\n","updatedAt":"2026-08-31T03:24:14.968Z","author":{"_id":"64ad53921db46d71a5c576e5","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64ad53921db46d71a5c576e5/mAPg4l6kZ6x1dvoxdPRm3.png","fullname":"ChuGyouk","name":"ChuGyouk","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":34,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7987746000289917},"editors":["ChuGyouk"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/64ad53921db46d71a5c576e5/mAPg4l6kZ6x1dvoxdPRm3.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.26582","authors":[{"_id":"6a94f284073195fee5157243","name":"Gyouk Chu","hidden":false},{"_id":"6a94f284073195fee5157244","name":"Myeongho Jeon","hidden":false},{"_id":"6a94f284073195fee5157245","name":"Eunho Yang","hidden":false}],"publishedAt":"2026-08-27T00:00:00.000Z","submittedOnDailyAt":"2026-08-31T00:00:00.000Z","title":"J-Zero: Unified Challenger--Solver--Judge Co-Evolution from Zero Data","submittedOnDailyBy":{"_id":"64ad53921db46d71a5c576e5","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64ad53921db46d71a5c576e5/mAPg4l6kZ6x1dvoxdPRm3.png","isPro":true,"fullname":"ChuGyouk","user":"ChuGyouk","type":"user","name":"ChuGyouk"},"summary":"Self-evolving language models have recently emerged as a promising path toward superintelligence, with the advantage of reducing the cost of human supervision. While considerable progress has been made in verifiable domains, self-evolution in unverifiable domains remains substantially less explored. We propose Judge co-adaptation from Zero data (J-Zero), a unified Challenger--Solver--Judge co-evolution framework that supports self-improvement across both domains. The Challenger and Solver co-evolve through an adversarial interaction: the Challenger generates increasingly difficult tasks, while the Solver learns to produce higher-quality responses to them. In parallel, the Judge co-adapts using preference pairs whose ordering is known in advance from how each response was produced, i.e., the Solver's answer over the Challenger's, and its decomposed-and-recombined answer over its one-shot answer, rather than from the Judge's own scores. J-Zero outperforms the baselines by an average of 4.2 points on verifiable and 8.0 points on unverifiable domains, and continues to improve through at least ten iterations, whereas the baselines degrade after two.","upvotes":27,"discussionId":"6a94f284073195fee5157246","projectPage":"https://gyoukchu.github.io/projects/j_zero/","githubRepo":"https://github.com/GyoukChu/J-Zero","githubRepoAddedBy":"user","ai_summary":"J-Zero enables self-improving language models across verifiable and unverifiable domains through adversarial co-evolution of a task generator, solver, and judge using predefined preference pairs.","ai_keywords":["self-evolving language models","Challenger-Solver-Judge co-evolution","adversarial interaction","preference pairs","J-Zero"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":1,"organization":{"_id":"6475760c33192631bad2bb38","name":"kaist-ai","fullname":"KAIST AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64ad53921db46d71a5c576e5","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64ad53921db46d71a5c576e5/mAPg4l6kZ6x1dvoxdPRm3.png","isPro":true,"fullname":"ChuGyouk","user":"ChuGyouk","type":"user"},{"_id":"62845957b410bd779033759c","avatarUrl":"/avatars/4feef73c06f2f7de6abf7a4789ac13f9.svg","isPro":false,"fullname":"Doohyuk Jang","user":"jadohu","type":"user"},{"_id":"6690a286181e2af45c742dd8","avatarUrl":"/avatars/511d0f86386e3b29a17b445d855b3aef.svg","isPro":false,"fullname":"Sohee Kim","user":"joyhee","type":"user"},{"_id":"657a64c91ccc3c2a5ea5cde4","avatarUrl":"/avatars/aee59f3650cc71424591340ec9f862b2.svg","isPro":false,"fullname":"Hangyeol Jung","user":"Hangyeol","type":"user"},{"_id":"66303ce3e1c93377db71efd5","avatarUrl":"/avatars/3d444dfb9799c9324c98cba893f4a10f.svg","isPro":false,"fullname":"Yoon Sik Park","user":"nooynoos","type":"user"},{"_id":"663073e4b108b8d71d4b6f32","avatarUrl":"/avatars/f66de2d81205397a29baf21169157bdd.svg","isPro":false,"fullname":"Inki Park","user":"seoharuss","type":"user"},{"_id":"69bb4d6a3823efb0602370c1","avatarUrl":"/avatars/4327f9f696f960cbffdb214e61f3505b.svg","isPro":false,"fullname":"Yang","user":"KiungYang","type":"user"},{"_id":"65bbe7e2c084467aca4d0994","avatarUrl":"/avatars/a92ce1bcf144699adfda447423593967.svg","isPro":false,"fullname":"Jang","user":"Hyeongwon","type":"user"},{"_id":"61b15ce1a5dd7dc7024406dc","avatarUrl":"/avatars/682ce5ee7d2fec7180dc8e1144cd12ab.svg","isPro":false,"fullname":"Yoonjeon Kim","user":"yjyjyj98","type":"user"},{"_id":"62a4d58e81a4b10e93064ad6","avatarUrl":"/avatars/744d5cbc1745a26b816a458260aba050.svg","isPro":false,"fullname":"hangyulyoon","user":"hangyulmd","type":"user"},{"_id":"63ac5701c21e60a3e9b58aa7","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ac5701c21e60a3e9b58aa7/g6EX7diOpuA94R2ab-rZC.png","isPro":true,"fullname":"Dipankar Sarkar","user":"dipankarsarkar","type":"user"},{"_id":"6667a3090e1819c8025b0a30","avatarUrl":"/avatars/4748a2875ec4d7f0cf3d80d903870beb.svg","isPro":false,"fullname":"Jinbae Im","user":"sjc00215","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":2,"organization":{"_id":"6475760c33192631bad2bb38","name":"kaist-ai","fullname":"KAIST AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.26582.md","query":{}}">

Papers

arxiv:2608.26582

J-Zero: Unified Challenger--Solver--Judge Co-Evolution from Zero Data

Published on Aug 27

· Submitted by

ChuGyouk on Aug 31

#2 Paper of the day

KAIST AI

Upvote

Authors:

Abstract

J-Zero enables self-improving language models across verifiable and unverifiable domains through adversarial co-evolution of a task generator, solver, and judge using predefined preference pairs.

Generated by thinkingmachines/Inkling-Small

Self-evolving language models have recently emerged as a promising path toward superintelligence, with the advantage of reducing the cost of human supervision. While considerable progress has been made in verifiable domains, self-evolution in unverifiable domains remains substantially less explored. We propose Judge co-adaptation from Zero data (J-Zero), a unified Challenger--Solver--Judge co-evolution framework that supports self-improvement across both domains. The Challenger and Solver co-evolve through an adversarial interaction: the Challenger generates increasingly difficult tasks, while the Solver learns to produce higher-quality responses to them. In parallel, the Judge co-adapts using preference pairs whose ordering is known in advance from how each response was produced, i.e., the Solver's answer over the Challenger's, and its decomposed-and-recombined answer over its one-shot answer, rather than from the Judge's own scores. J-Zero outperforms the baselines by an average of 4.2 points on verifiable and 8.0 points on unverifiable domains, and continues to improve through at least ten iterations, whereas the baselines degrade after two.

View arXiv page View PDF Project page GitHub 1 Add to collection

Community

ChuGyouk

Paper submitter about 5 hours ago

Frozen Judge set the upper bound of self-evolving LLMs. By co-evolving Judge with the current frontier of self-evolution, a highly performant and sustained self-evolution can be realized.