This repo deals with several Triton python client examples for novice users as a reference. Most of the examples are simply focused on demonstrating some basics and important functionalitiy of Triton(i.e.,performance might not be severely considered). On a side note, you can also find some Trtion python client examples from the official Triton git repo.
As a prerequisite step for the demonstaration, I recommend cloning this repo to your <local_full_path> first to be free from unexpected errors. Then, launch the Triton container from NGC using the docker command such as:
docker run --gpus all --network host -v <local_full_path>:/workspace \
nvcr.io/nvidia/tritonserver:22.03-py3 <local_full_path> is a local directory where you have cloned this repository. This path is mounted on the Triton container as a docker volume. In addition, host mode networking is configured(see --network host) for simplicity. I guess the launching command would be able to be varied according to your environment(e.g.,even though Triton v22.03 is used in this repo, you could pick a more advanced version).