Cannot use a local model for sentencetransformers

[cesar-cortez5]

LocalAI version:
docker image version: localai/localai:latest-gpu-nvidia-cuda-12

Environment, CPU architecture, OS, and Version:
Linux gcgpu3 5.10.0-33-cloud-amd64 #1 SMP Debian 5.10.226-1 (2024-10-03) x86_64 GNU/Linux

Describe the bug
I would like to load a sentencetransformer model from a local file/directory, so I can use LocalAI in a machine without internet. To do this I downloaded a huggingfacemodel to a flash drive, and then transfer these files to the machine. Currently, I cannot do this as when setting the model parameter to where the local model is located (which is under /build/models), it fails with an error saying "path not found"

Also, when using an absolute path (eg: /build/models/path-to-model), it also fails when it is parsing the config file.

To Reproduce

Download a model's directory from huggingface, for my testing I was using https://huggingface.co/mixedbread-ai/mxbai-embed-large-v1. I then store this in a models/mxbai-embed-large-v1 directory.

docker-compose.yml file:

image: localai/localai:latest-gpu-nvidia-cuda-12
ports:
    - 8080:8080
environment:
    DEBUG=true
volumes:
    - ./models:/build/models

Setup this config file:
config.yaml

name: text-embedding-ada-002
backend: sentencetransformers
parameters:
    model: ./mxbai-embed-large-v1
embeddings: true

Then docker compose up and call the embedding endpoint

Expected behavior
Correct model path to be used.

Logs
When using path relative to the models path

localai-1  | 9:47PM ERR Server error error="failed to load model with internal loader: could not load model (no success): Unexpected err=ValueError('Path ./mxbai-embed-large-v1'), type(err)=<class 'ValueError'>" ip=172.19.0.1 latenc

When using absolute path:

localai-1  | 9:43PM ERR config is not valid                                                                                                                   

Additional context