ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation

Hugging Face Daily Papers Papers

Summary

ActReview is a rebuttal-guided post-training framework that generates diagnostic claims and revision suggestions for peer reviews by leveraging author responses as supervision, along with a human-curated benchmark for evaluation.

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.
Original Article
View Cached Full Text

Cached at: 09/11/26, 06:19 PM

Paper page - ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation

Source: https://huggingface.co/papers/2609.09076

Abstract

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.

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 asActionable Peer-review Generationand decompose it into two subtasks:diagnostic claim generationandrevision suggestion generation. We introduceActReview, arebuttal-guided post-trainingframework 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 constructActReview-40Kby aligning reviewer weaknesses with author responses and grounding the resulting feedback in localized paper evidence. We post-train Qwen3-8B-Base withmulti-task supervised fine-tuningfollowed byGRPOusing candidate-aware, weakness-specific rubric rewards. We also introduceActReview-Bench, a human-curated benchmark of 1,000 instances for evaluating diagnostic quality and revision usefulness. Experiments show thatActReviewoutperforms 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.

View arXiv pageView PDFGitHub0Add to collection

Get this paper in your agent:

hf papers read 2609\.09076

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

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

Datasets citing this paper0

No dataset linking this paper

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

Spaces citing this paper0

No Space linking this paper

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

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Author-in-the-Loop Response Generation and Evaluation: Integrating Author Expertise and Intent in Responses to Peer Review

arXiv cs.CL

This paper introduces the Re3Align dataset, REspGen framework, and REspEval evaluation suite for author-in-the-loop response generation in peer review, integrating author expertise and intent signals. The work addresses gaps in NLP formulation of scientific rebuttal writing with comprehensive datasets, controllable generation frameworks, and multi-dimensional evaluation metrics.