Safety
When Modality Gap Reduction Fails: Prediction-Level Hubness in CLIP
arXiv:2609.01103v1 Announce Type: new Abstract: Reducing the modality gap between image and text representations in CLIP is widely expected to improve cross-modal alignment and downstream performance.
arXiv:2609.01103v1 Announce Type: new Abstract: Reducing the modality gap between image and text representations in CLIP is widely expected to improve cross-modal alignment and downstream performance. However, a smaller average image-text gap does not necessarily lead to consistent accuracy gains. We analyze this mismatch from the perspective of the decision structure in zero-shot classification, i.e. selecting the most similar class-text prototype for an input image. Zero-shot accuracy depends not only on average image--text alignment, but also on class-wise decision margins. Using Linear correction as an analytically tractable case, we show that modality gap correction can alter the relative decision structure among classes and cause predictions to concentrate on a small subset of classes. We refer to this output-space failure mode as prediction-level hubness. Furthermore, experiments across multiple datasets show that accuracy degradation under gap correction is consistently associated with increased prediction concentration, both for Linear correction and for learning-based correction methods. This provides a systematic explanation of why modality gap reduction does not consistently improve CLIP zero-shot accuracy from the perspective of downstream decision structure. Our results suggest that gap correction should be evaluated not only by average alignment, but also by its impact on downstream prediction structure.
Related
- Reading, Not Thinking: Understanding and Bridging the Modality Gap When Text Becomes Pixels in Multimodal LLMs
- GLINT: Sparsely Gated Vision-Language Alignment for Fine-Grained Radiology Representations
- SIMMER: Cross-Modal Food Image--Recipe Retrieval via MLLM-Based Embedding
Source: arXiv cs.CL | 2026-09-02