AI peer review tool for CVPR, ICLR, ICML, and NeurIPS papers [D]

Came across this GitHub project called Paper Review Harness. Thought it might be useful for people preparing ML research papers.

It uses 30 different AI reviewer personas to simulate the kind of feedback a paper might receive at conferences like CVPR, ICLR, ICML, and NeurIPS.

What's interesting is that it goes beyond simply asking an LLM to "review my paper."

It provides:

• Independent reviews from different reviewer perspectives
• Section-level scores and detailed criticism
• Identification of weaknesses in novelty, methodology, experiments, and clarity
• Prioritized feedback on what needs improvement
• A revision workflow that tracks whether previous concerns have actually been addressed

It also supports different review panels depending on the conference and the type of research contribution.

The project is open-source and designed to work with AI agent environments.

Obviously, simulated reviews aren't a replacement for actual peer review, and the scores shouldn't be treated as acceptance predictions. But it seems like an interesting way to stress-test a paper before submission.

GitHub: https://github.com/Optimizer077/paper-review-harness

Curious what researchers think about this approach. Could tools like this become a useful part of the paper-writing workflow?