Title: Rubrics as Visual-Repair Context for Self-Evolving UI-to-Code Generation</p>\n<p>This paper introduces RubSE, a rubric-guided self-evolution framework that structures visual feedback into typed, targeted rubrics to achieve stable and effective iterative refinement in UI-to-code generation.</p>\n","updatedAt":"2026-08-27T05:11:06.843Z","author":{"_id":"6570977f87a92b76922c9950","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6570977f87a92b76922c9950/AQGto1w6ugBvH2yCV46YU.jpeg","fullname":"Tianyi Xiong","name":"txiong23","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":11,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8083258271217346},"editors":["txiong23"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/6570977f87a92b76922c9950/AQGto1w6ugBvH2yCV46YU.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.24138","authors":[{"_id":"6a8fc6243bd48bb654ea68be","name":"Tianyi Xiong","hidden":false},{"_id":"6a8fc6243bd48bb654ea68bf","name":"Zhengyuan Yang","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c0","name":"Xiaofei Wang","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c1","name":"Chung-Ching Lin","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c2","name":"Ruichun Ma","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c3","name":"Kevin Lin","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c4","name":"Zhendong Wang","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c5","name":"Linjie Li","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c6","name":"Chenxi Liu","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c7","name":"Ruibo Chen","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c8","name":"Ramani Duraiswami","hidden":false},{"_id":"6a8fc6243bd48bb654ea68c9","name":"Heng Huang","hidden":false},{"_id":"6a8fc6243bd48bb654ea68ca","name":"Lijuan Wang","hidden":false}],"publishedAt":"2026-08-25T00:00:00.000Z","submittedOnDailyAt":"2026-08-27T00:00:00.000Z","title":"Rubrics as Visual-Repair Context for Self-Evolving UI-to-Code Generation","submittedOnDailyBy":{"_id":"6570977f87a92b76922c9950","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6570977f87a92b76922c9950/AQGto1w6ugBvH2yCV46YU.jpeg","isPro":false,"fullname":"Tianyi Xiong","user":"txiong23","type":"user","name":"txiong23"},"summary":"Large vision-language models have shown strong progress in UI-to-code generation, yet their test-time self-evolution remains unstable. We first identify a fundamental obstacle, termed visual repair coupling: a local code edit may propagate through layout, style, and component dependencies, correcting one visual mismatch while degrading regions that were previously faithful. To address this issue, we present RubSE, a Rubric-guided Self-Evolution framework that uses rubrics to represent visual feedback as a structured visual-repair context. At each refinement round, RubSE generates typed candidate rubrics, selects one prioritized repair target, and stores previously selected rubrics as history, thereby steering each revision toward a well-scoped visual repair while discouraging repeated or over-broad changes. Evaluations across six VLMs and three UI-to-code benchmarks demonstrate that RubSE substantially outperforms naïve self-evolution in final-round and best-round settings, achieving more stable refinement trajectories and a higher trajectory-level performance ceiling. Further analysis shows that RubSE mitigates trajectory collapse by improving recovery from severe visual regressions, and that stronger rubric generators can transfer effective visual-repair guidance to weaker code improvers.","upvotes":11,"discussionId":"6a8fc6243bd48bb654ea68cb","projectPage":"https://tyxiong23.github.io/rubse","ai_summary":"RubSE improves UI-to-code generation stability by using rubric-guided self-evolution to prevent visual repair coupling and trajectory collapse.","ai_keywords":["visual repair coupling","RubSE","rubric-guided self-evolution","typed candidate rubrics","visual-repair context","trajectory collapse"],"ai_summary_model":"thinkingmachines/Inkling-Small"},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6570977f87a92b76922c9950","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6570977f87a92b76922c9950/AQGto1w6ugBvH2yCV46YU.jpeg","isPro":false,"fullname":"Tianyi Xiong","user":"txiong23","type":"user"},{"_id":"646c1715c7f672003c87730a","avatarUrl":"/avatars/731e608bb542ff21622ac981144d90bc.svg","isPro":false,"fullname":"Shuning Zhang","user":"Shuningz","type":"user"},{"_id":"656741b757c58ae7f82a52ac","avatarUrl":"/avatars/65ddd7147d7745b56290f9d3ef24077a.svg","isPro":false,"fullname":"Shihao Wang","user":"exiawsh","type":"user"},{"_id":"686f806aeb53e7adba46c3de","avatarUrl":"/avatars/db11c84d602cefa72ba409c8292e4191.svg","isPro":true,"fullname":"guoguoc","user":"woshichaoren123","type":"user"},{"_id":"6488a4039d6109a4dd058e44","avatarUrl":"/avatars/2eac1e3a3e8189bbf8f712ef96290401.svg","isPro":false,"fullname":"YANSHUO CHEN","user":"alegendaryfish","type":"user"},{"_id":"6897d62aa51bf50752188f7c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/ctBIt0--kHciWoqJ1qSM8.jpeg","isPro":false,"fullname":"Tianao Li","user":"tianaoli","type":"user"},{"_id":"6623ea65b642e29cdf90a1b4","avatarUrl":"/avatars/e32e90574c1162b2be87ed78604e3e4d.svg","isPro":false,"fullname":"TongZheng","user":"TongZheng1999","type":"user"},{"_id":"68d2bce19ccd51b0b4040d5e","avatarUrl":"/avatars/d8f9fc8e146d3813028f07017f6798b8.svg","isPro":false,"fullname":"Kaishen Wang","user":"tuunan","type":"user"},{"_id":"66184315a219e8a8e42917ed","avatarUrl":"/avatars/90f84a1ebf1b31c92b0348f2086bd73f.svg","isPro":false,"fullname":"Ruibo Chen","user":"RayRuiboChen","type":"user"},{"_id":"651255b8446eba391bc314f9","avatarUrl":"/avatars/1d9e1f95f96b69cfc3daa65a0af1640d.svg","isPro":false,"fullname":"Dong Xu","user":"xudong2023ucas","type":"user"},{"_id":"6623e265609d7a39a1107cc9","avatarUrl":"/avatars/887f1f5b74f6669ce7d560d95b05b530.svg","isPro":false,"fullname":"Dawn","user":"LegendaryDawn","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.24138.md","query":{}}">
Rubrics as Visual-Repair Context for Self-Evolving UI-to-Code Generation
Abstract
RubSE improves UI-to-code generation stability by using rubric-guided self-evolution to prevent visual repair coupling and trajectory collapse.
Large vision-language models have shown strong progress in UI-to-code generation, yet their test-time self-evolution remains unstable. We first identify a fundamental obstacle, termed visual repair coupling: a local code edit may propagate through layout, style, and component dependencies, correcting one visual mismatch while degrading regions that were previously faithful. To address this issue, we present RubSE, a Rubric-guided Self-Evolution framework that uses rubrics to represent visual feedback as a structured visual-repair context. At each refinement round, RubSE generates typed candidate rubrics, selects one prioritized repair target, and stores previously selected rubrics as history, thereby steering each revision toward a well-scoped visual repair while discouraging repeated or over-broad changes. Evaluations across six VLMs and three UI-to-code benchmarks demonstrate that RubSE substantially outperforms naïve self-evolution in final-round and best-round settings, achieving more stable refinement trajectories and a higher trajectory-level performance ceiling. Further analysis shows that RubSE mitigates trajectory collapse by improving recovery from severe visual regressions, and that stronger rubric generators can transfer effective visual-repair guidance to weaker code improvers.
Community
Title: Rubrics as Visual-Repair Context for Self-Evolving UI-to-Code Generation
This paper introduces RubSE, a rubric-guided self-evolution framework that structures visual feedback into typed, targeted rubrics to achieve stable and effective iterative refinement in UI-to-code generation.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2608.24138 in a model README.md to link it from this page.
Cite arxiv.org/abs/2608.24138 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2608.24138 in a Space README.md 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.