We propose Neural Priming, a technique for adapting large pretrained mod...
Large multimodal datasets have been instrumental in recent breakthroughs...
Does progress on ImageNet transfer to real-world datasets? We investigat...
Compositional representations of the world are a promising step towards
...
Contrastively trained image-text models such as CLIP, ALIGN, and BASIC h...
We investigate the connections between neural networks and simple buildi...