Strange behaviour of Llama3.2-vision - it behaves like text model

#9
by jirkazcech - opened

I am running ollama on my notebook (windows 11, 8 cores, 32 Gb ram)

Yesterday I have succesfully installed Llam-vision model:
ollama run llama3.2-vision

I was able to connect two pictures and analyze them.....

Once trying to connect third picture I am getting the message:

It looks like you're sharing a link to an image file!

Unfortunately, I'm a text-based AI and don't have the capability to directly view or access external files. But I
can try to help with something related to the image.

I have tried various ways to resolve it:

/clear
ollama stop llama3.2-vision
ollama rm llama3.2-vision and then ollama run llama3.2-vision

I am really confused, what is going on....

Can somebody help me?

Unsloth AI org

I am running ollama on my notebook (windows 11, 8 cores, 32 Gb ram)

Yesterday I have succesfully installed Llam-vision model:
ollama run llama3.2-vision

I was able to connect two pictures and analyze them.....

Once trying to connect third picture I am getting the message:

It looks like you're sharing a link to an image file!

Unfortunately, I'm a text-based AI and don't have the capability to directly view or access external files. But I
can try to help with something related to the image.

I have tried various ways to resolve it:

/clear
ollama stop llama3.2-vision
ollama rm llama3.2-vision and then ollama run llama3.2-vision

I am really confused, what is going on....

Can somebody help me?

Mmm that's weird you didn't use a link to the image file right and it is using this model?

Can you try using this one instead? https://huggingface.co./unsloth/Llama-3.2-11B-Vision-Instruct-unsloth-bnb-4bit

Sign up or log in to comment