2 papers
cs.CV2026
Birds of a Feather Flock Together: Background-Invariant Representations via Linear Structure in VLMs
Youssef Zaazou, Mark Thomas
Vision-language models (VLMs), such as CLIP and SigLIP 2, are widely used for image classification, yet their vision encoders remain vulnerable to systematic biases that undermine…
hep-th2025
A Master Superspace Action for 3D S-Duality
Adi Armoni, Ricardo Stuardo, Mark Thomas
We formulate a `master' partition function in three-dimensional superspace that realises, upon integrating out complementary superfields, both the electric Maxwell-…