🚀 Excited to share ActReview!<br>We ask: Can LLM-generated peer reviews not only identify weaknesses, but also help authors revise their papers?<br>Our key insight is that reviewer comments reveal what is wrong, while author rebuttals often reveal how the concern can be addressed. We use this signal to build ActReview-40K, aligning reviewer weaknesses with rebuttal responses and localized paper evidence.<br>We train ActReview with multi-task SFT followed by GRPO using weakness-specific rubric rewards, and introduce ActReview-Bench, a human-curated benchmark of 1,000 examples.<br>ActReview produces feedback that is more actionable and grounded than prior specialized review-generation models.</p>\n","updatedAt":"2026-09-11T15:21:12.408Z","author":{"_id":"69227307168e7fa5935fc7a6","avatarUrl":"/avatars/20f959a3314ce098fcd224ae999e2a8f.svg","fullname":"YilingMa","name":"YilingMa","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9322383403778076},"editors":["YilingMa"],"editorAvatarUrls":["/avatars/20f959a3314ce098fcd224ae999e2a8f.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2609.09076","authors":[{"_id":"6aa1453bff4bf7311191aaf0","name":"Yiling Ma","hidden":false},{"_id":"6aa1453bff4bf7311191aaf1","name":"Yilun Zhao","hidden":false},{"_id":"6aa1453bff4bf7311191aaf2","name":"Sihong Wu","hidden":false},{"_id":"6aa1453bff4bf7311191aaf3","name":"Ziyu Chen","hidden":false},{"_id":"6aa1453bff4bf7311191aaf4","name":"Manasi Patwardhan","hidden":false},{"_id":"6aa1453bff4bf7311191aaf5","name":"Arman Cohan","hidden":false}],"publishedAt":"2026-09-08T00:00:00.000Z","submittedOnDailyAt":"2026-09-11T00:00:00.000Z","title":"ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation","submittedOnDailyBy":{"_id":"69227307168e7fa5935fc7a6","avatarUrl":"/avatars/20f959a3314ce098fcd224ae999e2a8f.svg","isPro":false,"fullname":"YilingMa","user":"YilingMa","type":"user","name":"YilingMa"},"summary":"As LLMs are increasingly used for pre-submission self-review, there is growing demand for feedback that not only identifies weaknesses but also guides authors toward concrete revisions. We study this as Actionable Peer-review Generation and decompose it into two subtasks: diagnostic claim generation and revision suggestion generation. We introduce ActReview, a rebuttal-guided post-training framework that connects paper-specific diagnoses to concrete, grounded revision plans. Our central insight is that author rebuttals reveal plausible actions for addressing reviewer concerns and can therefore provide latent supervision for revision-oriented feedback. From real review-rebuttal threads on OpenReview, we construct ActReview-40K by aligning reviewer weaknesses with author responses and grounding the resulting feedback in localized paper evidence. We post-train Qwen3-8B-Base with multi-task supervised fine-tuning followed by GRPO using candidate-aware, weakness-specific rubric rewards. We also introduce ActReview-Bench, a human-curated benchmark of 1,000 instances for evaluating diagnostic quality and revision usefulness. Experiments show that ActReview outperforms prior specialized review-generation models on actionability and grounding while remaining competitive with strong prompt-based LLMs. Human evaluation confirms improved revision usefulness while revealing a remaining gap in technical accuracy, and additional analyses support generalization to held-out papers and robustness across independent judges.","upvotes":0,"discussionId":"6aa1453cff4bf7311191aaf6","githubRepo":"https://github.com/Yiling-Ma/ActReview","githubRepoAddedBy":"user","ai_summary":"ActReview is a rebuttal-guided post-training framework that generates diagnostic claims and concrete revision suggestions for peer review by leveraging author responses as latent supervision.","ai_keywords":["Actionable Peer-review Generation","diagnostic claim generation","revision suggestion generation","ActReview","rebuttal-guided post-training","multi-task supervised fine-tuning","GRPO","candidate-aware rubric rewards","ActReview-40K","ActReview-Bench"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0,"organization":{"_id":"6532df27d690f3012efde84c","name":"yale-nlp","fullname":"Yale NLP Lab","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"organization":{"_id":"6532df27d690f3012efde84c","name":"yale-nlp","fullname":"Yale NLP Lab","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/65204db5b0e0d57453cb1809/9OAeiZ-BrN2g1h1yd6-1W.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2609/2609.09076.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2609.09076 in a model README.md to link it from this page.

Cite arxiv.org/abs/2609.09076 in a dataset README.md to link it from this page.

Cite arxiv.org/abs/2609.09076 in a Space README.md to link it from this page.

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