Python & PyTorch for Inference

The PyTorch habits inference work relies on: reading a tensor, keeping it on the right device and format, using inference mode and pinning the environment.