add image2text generation

[leostormrage]

add image2text generation

[OllieBroadhurst]

Have you tried Vision Encoder Decoder?

[qgallouedec]

Hey @leostormrage, sorry for the very slow reply, we're catching up on the backlog. Image-to-text generation is well covered in transformers now, with captioning models like BLIP and GIT plus the image-text-to-text pipeline.