arXiv cs.CLPaper
When Modality Gap Reduction Fails: Prediction-Level Hubness in CLIP
A genuine failure mode in multimodal alignment work: gap reduction can concentrate predictions onto few classes even as average alignment improves. If you're using CLIP and applying gap-reduction techniques, this is a reminder to evaluate class-wise margins and decision concentration, not just average metrics. The problem is real but the paper doesn't propose a fix, leaving the takeaway as caution rather than action.