Unable to up deploy a custom trained yolov5 model on robotflow

``
It has been impossible to upload a yolov5 model trained on a custom dataset. My folder

rf = roboflow.Roboflow(api_key="..")
project = rf.workspace().project("...")

version = project.version(2)

version.deploy("yolov5", "yolov5/runs/train/exp2")

I've got this error

It' an issue that arise as resulting of calling the function torch.load on the parameter : os.path.join(model_path, filename). This function attempts to load the yolov5 model using the same argument used as input for the function deploy. Recall the functions deploy inputs :

(method) def deploy(
model_type: str,
model_path: str,
filename: str = "weights/best.pt"
) -> None
Uploads provided weights file to Roboflow.

Args:
model_type (str): The type of the model to be deployed.
model_path (str): File path to the model weights to be uploaded.
filename (str, optional): The name of the weights file. Defaults to "weights/best.pt".

I tried :

torch.load(os.path.join("yolov5/runs/train/exp2", "weights/best.pt"))

but still getting the same error. According to my research, loading former yolo model versions has usually been done using torch.hub.load rather than torch.load. In fact

torch.hub.load('.', 'custom', path='/path/to/yolov5/runs/train/exp5/weights/best.pt', source='local')

works fine.

Start at version.deploy(model_type, model_path, filename) and trace the torch.load call for yolov5 weights. Reproduce the failure with the supplied custom model path and compare it with the working torch.hub.load example. Done means a custom-trained yolov5 weights file can be uploaded successfully through deploy.