Visual Document Retrieval
ColPali
Safetensors
English
vidore
vidore-experimental

Is this ready to use now? Will it be benchmarked?

#1
by jjovalle99 - opened

I am asking because to make it work I have to use:

pip install git+https://github.com/illuin-tech/colpali@colqwen2_5
pip install git+https://github.com/huggingface/transformers
ILLUIN Vidore org

Hey for now Qwen2.5-VL is not released yet in the latest PyPi version fo transformers, we are waiting for transformers 4.49 to be released before merging the PR. So your solution is the only way for now to use the model.

It is already benchmarked here https://huggingface.co./spaces/vidore/vidore-leaderboard (you may need to refresh), along with slightly better version 0.2.

QuentinJG changed discussion status to closed

@QuentinJG Hello! Was the 0.2 version removed?

ILLUIN Vidore org

Hello @jjovalle99 , I put it back it was a mistake on our end, thanks for pointing it out.

Great, thanks for the quick response. By the way, with the new release from transformers I think the colqwen2_5 branch can be merged now, cant it? I will be more than happy to help if something is needed.

ILLUIN Vidore org

i'll do this by the end of today

Sign up or log in to comment