Quintin Pope comments on [Linkpost] New multi-modal Deepmind model fusing Chinchilla with images and videos