Qwen

/

Qwen-VL-Chat

Like 384

Follow

Qwen 106k

Hi there,
thanks for providing this great model for free.
I'm evaluating multimodal modals on an own dataset. All the other models use PILLOW to load images and combine it with requests to download images upfront from a given URL. You are wrapping the image download in your tokenizer, which makes it hard to use local images.
Can you hint me on how to do it without having to change your code?
Thanks a lot in advance.
René