← Back to context Comment by energy123 1 year ago Yeah, once it gets converted into tokens how does "zooming in" somehow increase information content? 1 comment energy123 Reply nutrientharvest 1 year ago It's cropping the original image then tokenizing it again with less downsampling, not cropping its internal representation.
nutrientharvest 1 year ago It's cropping the original image then tokenizing it again with less downsampling, not cropping its internal representation.
It's cropping the original image then tokenizing it again with less downsampling, not cropping its internal representation.