[Feature] Colbert(ColPali-like) support

[VoVAllen]

Motivation

https://github.com/illuin-tech/colpali

Recently Colbert+PaliGemma showed huge improvement on pdf file retrieval by using multimodal model instead of OCR+LLM. Would be nice if InternVL can support Colbert-like usage for downstream retrieval tasks.

Related resources

No response

Additional context

No response

[czczup]

Hi, thanks for your advice.