Help regarding loading custom trained models

[Capt4in-Levi]

Hi, I need documentation/examples for loading models using the path of the local directory. It is looking for a URL when passed through --model-id, is there any additional tag to pass as an argument?

[OlivierDehaene]

For example:

# make sure that you have git lfs installed
git lfs install

git clone https://huggingface.co/BAAI/bge-base-en-v1.5

# mount the model directory to /model inside the container and pass it to the `--model-id` CLI arg
docker run --gpus all -p 8080:80 -v $PWD/bge-base-en-v1.5:/model --pull always ghcr.io/huggingface/text-embeddings-inference:0.4.0 --model-id /model