microsoft

/

Phi-3.5-vision-instruct

like 560

Follow

Microsoft 4,868

Hi @haohoo 👋

I'm currently working with your model on multi-image tasks. I've noticed that it's not possible to just give the base image (336x336) to the model. Even with a num_crops=1, I'm still getting pixel_values in the form (4,2,3,336,336) for four images with base image size.
Is it possible to just pass on the base image?