NeurIPS 2026 review results spark a thread on noisy scores and rebuttals
Afraid_Difference697 · reddit · 2026-07-22
A NeurIPS review-results discussion thread where authors and attendees compare notes on reviews, rebuttals, and backup conference plans.
The thread highlights a recurring point from NeurIPS consistency experiments: review outcomes are noisy, and many papers accepted in one committee would be rejected by another. The practical advice is to focus rebuttal effort on the reviews that point out real issues, not the ones that merely score poorly or skimmed the work.
Useful discussion prompts in the thread include:
- Review comments that uncovered missing experiments or baselines
- What to contest versus concede in a rebuttal
- Patterns in this cycle such as reproducibility requests, compute comparisons, and ablation depth
- Alternatives if the paper does not land at NeurIPS, such as ICLR, AISTATS, or workshops
More from Research
- Deep Learning Course Adds Mechanistic Interpretability Lecture This Winter — CSProfKGD · 2026-10-26
- INCLUDE-2.0 benchmark evaluates regional knowledge across 91 languages and 86 countries — ABosselut · 2026-10-26
- CoRL paper: flow-matching VLAs can run 1-step Euler inference with no retraining — kaixhin · 2026-10-26
- CoVer at EMNLP 2026: Training Code LLMs to Generate and Self-Verify Simultaneously — algo_diver · 2026-10-26
- Robotic dexterity should be judged on speed, force control and tool use, not copying humans — DJiafei · 2026-10-26
- Turing-complete malware can run in PyTorch's backward pass via a zero-valued loss — retr0jirachi · 2026-10-26