probing-vits/class-saliency

This Space shows the spatial-class relationship of a CaiT model by plotting its class saliency map w.r.t an input image. For speed, this Space uses the smallest CaiT model (cait_xxs24_224) but you can try it for others too (model collection if here). You'd just need to adjust the code accordingly.

CaiT was proposed in Going deeper with Image Transformers (Touvron et al.).