Neurips : Pushing anonymous repo after rebuttal [D]

A NeurIPS author faces a nuanced ethical and procedural challenge. After the submission deadline, they notice a misleading label in their paper. The numerical metrics are accurate, but the label misrepresents what those metrics mean. The author wants to clarify this by updating the anonymous repository's README with corrected labels, while explicitly noting that numbers are unchanged. However, NeurIPS double-blind review policies strictly forbid post-deadline modifications that could alter the paper's interpretation. Even though it's not a new result, a label change could be seen as a revision. The author questions whether this correction qualifies as a permissible clarification or an impermissible correction, especially given the anonymity requirements.

Alternative approaches include posting a comment on OpenReview (if allowed) or addressing it during the rebuttal phase. The safest route is likely the rebuttal, where authors can clarify under controlled conditions. Updating the repository README carries risk because it is part of the submission and might be checked for consistency. However, a minimal, disclaimed clarification might be acceptable if it does not change results or add new experiments. Ultimately, the author should consult NeurIPS guidelines or ask the program chairs. This case highlights the tension between scientific accuracy and procedural integrity in peer review. For future submissions, authors may consider double-checking all labels before the deadline or using a detailed README that anticipates such corrections.