A new study questions the longstanding view that the visual system is divided into two pathways, one for object-recognition ... actually helps the model learn a better representation, instead ...

a neuroscientist at the University of Cambridge who studies visual object recognition and was not involved in the study. "This is consistent with the similarity of the brain representations of ...

Available under a permissive MIT license, the model ... representation. It comes in two sizes — 232M and 771M parameters — and already excels at tasks such as captioning, object detection ...