Hi! I was wondering how many images did IDEFICS see when training on the interleaved image-text data. I wasn't able to find information on this. Is there a limit on the max number of images we can supply at inference time?
· Sign up or log in to comment