Minimal HTTP inference server in OpenAI API with Pytorch and CUDA - View it on GitHub