Llama 3.2 11B Currently Only Supports Single Image

[Jack-Khuu]

🐛 Describe the bug

Currently, Llama 3.2 11B only supports a single optional image prompt in torchchat. The base torchtune model backing Llama3.2 11B should* be capable of supporting multiturn with:

• Multiple Simultaneous Images
• Replacing the previous image

This Issue acts as a tracker for the development of these 2 extensions to Llama 3.2 11B functionality

E.g. Via OpenAI API/Browser you can currently provide text prompts similar to LLama3.1 8B, but you are unable to replace the image once one is provided

*Should being the operative word as it may require additional changes to the torchtune repo

Versions

NA

[srikary12]

Would like to work on this. Would love some guidance on the approach.