[Feature Request] Multi-modal RAG(Retrieval-Augmented Generation)

[FUYICC]

Required prerequisites

• [X] I have searched the Issue Tracker and Discussions that this hasn't already been reported. (+1 or comment there if it has.)
• [X] Consider asking first in a Discussion.

Motivation

The current RAG can only retrieve text-type information, and cannot retrieve and extract information from images, audio, and other information.

Solution

For images:

• [ ] Integrate VLM matching embedding model (CLIP)
• [ ] RAG function with image retrieval

For other modalities: tbc