Running 'jinaai/jina-clip-v2' model

[roberto-corno-ntt]

Hi I'm trying to run the afore mentioned model in LocalAI.
I've created the proper yaml config file but once downloaded, the python backend is complaining about missing packages (timm and torchvision), so I created a docker file to install dependencies I need.
My rather simple dockerfile is

`
FROM localai/localai:latest-aio-gpu-nvidia-cuda-12

USER root
RUN apt-get update && apt-get install -y python3-pip
RUN (. /build/backend/python/transformers/venv/bin/activate && pip install torchvision timm)
`

but I still encounter the error :

10:18AM DBG GRPC(jinaai/jina-clip-v2-127.0.0.1:39503): stderr Encountered exception while importing timm: No module named 'timm' 10:18AM DBG GRPC(jinaai/jina-clip-v2-127.0.0.1:39503): stderr Encountered exception while importing torchvision: No module named 'torchvision' 10:18AM DBG GRPC(jinaai/jina-clip-v2-127.0.0.1:39503): stderr Error: This modeling file requires the following packages that were not found in your environment: timm, torchvision. Runpip install timm torchvision10:18AM ERR Server error error="failed to load model with internal loader: could not load model (no success): Unexpected err=ImportError('This modeling file requires the following packages that were not found in your environment: timm, torchvision. Runpip install timm torchvision'), type(err)=<class 'ImportError'>" latency=6.227169438s method=POST status=500 url=/v1/embeddings

I tried to install the dependencies at root level as well without any luck
What am I doing wrong ?

[roberto-corno-ntt]

Fixed by myself, reporting here the solution for anyone that might be interested.
Digging into the scripts I discovered that python venv is managed by uv so installing dependencies shall just have a slightly different form

`
FROM localai/localai:latest-aio-gpu-nvidia-cuda-12

USER root
RUN apt-get update && apt-get install -y python3-pip
RUN (. /build/backend/python/transformers/venv/bin/activate && uv pip install torchvision==0.19.1 timm)