ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation
ActReview post-trains Qwen3-8B-Base on 40K rebuttal-derived instances with rubric rewards to generate actionable, grounded peer-review feedback, plus a 1,000-instance benchmark.
The framework builds ActReview-40K from real OpenReview review-rebuttal threads, aligning reviewer weaknesses with author responses and grounding feedback in localized paper evidence. Qwen3-8B-Base is post-trained with multi-task supervised fine-tuning followed by GRPO using candidate-aware, weakness-specific rubric rewards. Experiments show improved actionability and grounding over prior specialized review-generation models, supported by ActReview-Bench, a human-curated 1,000-instance evaluation set. Human evaluation confirms better revision usefulness while noting a remaining gap in technical accuracy.
- ActReview-40K derived from real OpenReview rebuttal threads
- SFT plus GRPO with rubric-based rewards applied to Qwen3-8B-Base
- ActReview-Bench provides 1,000 human-curated evaluation instances
- Outperforms prior specialized review generators on actionability and grounding
- Products
- ActReviewOpenReview
- AI models
- Qwen3-8B-Base
From the original article
1 min read at arxiv.orgAs 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.
By Yiling Ma · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.