Papers2

#compositionality

Half-Truths Break Similarity-Based Retrieval

Bora Kargi, Arnas Uselis et al.Feb 27arXiv

Similarity-based image–text models like CLIP can be fooled by “half-truths,” where adding one plausible but wrong detail makes a caption look more similar to an image instead of less similar.

#half-truth vulnerability#similarity-based retrieval#CLIP

Show, Don't Tell: Morphing Latent Reasoning into Image Generation

Intermediate

Harold Haodong Chen, Xinxiang Yin et al.Feb 2arXiv

LatentMorph teaches an image-making AI to quietly think in its head while it draws, instead of stopping to write out its thoughts in words.

#latent reasoning#text-to-image generation#autoregressive models