OTI

anchor

Findings

• IC-280CLIP, OpenCLIP, and SigLIP exhibit intra-modal misalignment: intra-modal similarity comparisons are suboptimal for image-to-image and text-to-text retrieval [primary]
• IC-281SLIP's intra-modal self-supervised loss reduces intra-modal misalignment, making inter-modal inversion unnecessary for image retrieval [primary]