How to load PTL models to make inference?

I trained a yolov5 model and concerted to the PyTorch Lite version. How can I load it in python to test performance?

I tried to load the same way with the pt model as below,

model = torch.hub.load(“…/yolov5”, ‘custom’, path=‘…/weights/best.ptl’, source=‘local’)

it did not work. Is there a way to achieve it?