/SKILL

Research Paper Review

Perform a rigorous, structured review of an AI/ML research paper.

Review Structure

Always organize the review under these sections:

1. Summary (2–3 sentences)

State the paper's core claim, the approach taken, and the key result. No jargon — write so that a senior researcher in an adjacent field can follow.

2. Contributions

List the paper's stated and actual contributions. Note if the framing overstates novelty.

3. Methodology

• Is the method described with enough detail to reproduce?
• Are design choices justified or arbitrary?
• Are there ablations that isolate each component's effect?

4. Experiments

• Are baselines appropriate and up-to-date?
• Is the evaluation dataset standard, or cherry-picked?
• Are error bars / statistical significance reported?
• Is the compute budget disclosed?

5. Limitations

Identify limitations the authors acknowledge, then add any they missed. Be specific.

6. Related Work

Note whether the paper engages honestly with prior work, including concurrent work.

7. Verdict

Rate the paper: Accept / Weak Accept / Borderline / Weak Reject / Reject Give a one-paragraph justification.

Tone

• Be constructive, not dismissive.
• Distinguish between fatal flaws and minor weaknesses.
• Praise genuine novelty — do not penalize incremental but solid work.

Output Format

Produce the review as markdown with the sections above. Summarize the verdict at the top in a blockquote before the full review body.