unsloth/Llama-3.2-11B-Vision-Instruct Support for multiple images

[Sosycs]

Hello,
Thank you very much for the great work.

I am trying to use the model with multiple images along with the question and I could not figure the correct formatting. This is my try:

[
{ "role": "user",
  "content": [{"type": "text",  "text": instruction}, {"type": "image", "image": image}, 
{"type": "image", "image": image},
 ]
},
{ "role": "assistant",
  "content": [{"type": "text",  "text": answer} ]
},
]

This is not exactly the format I want to provide for fine-tuning which is:
A reference image then a question on a different image.
Example: image of complete tooth with details, then a question like ( what does this part of tooth called with an image of this part).
So can you please help me with the correct messages?

Thanks in advance