Summary
New understanding on the geometry of raw CLIP embedding, the modality gap, false negatives and the relation to image and text conformity.
Meir Yossef Levi, Guy Gilboa
Proceedings of the 42nd International Conference on Machine Learning (ICML), 2025
Contrastive Language-Image Pre-Training (CLIP) is highly instrumental in machine learning applications within a large variety of domains.
We investigate the geometry of this embedding, which is still not well understood, and show that text and image reside on linearly separable ellipsoid shells, not centered at the origin. We explain the benefits of having this structure, allowing to better embed instances according to their uncertainty during contrastive training.
Frequent concepts in the dataset yield more false negatives, inducing greater uncertainty.
A new notion of conformity is introduced, which measures the average cosine similarity of an instance to any other instance within a representative data set. We prove this measure can be accurately estimated by simply computing the cosine similarity to the modality mean vector. Furthermore, we find that CLIP’s modality gap optimizes the matching of the conformity distributions of image and text.
