MXNet

/ˌɛm-ɛks-ˈnɛt/

n. “An open-source deep learning framework designed for efficiency, scalability, and flexible model building.”

MXNet is a machine learning library that supports building and training deep neural networks across multiple CPUs and GPUs. It was originally developed by the Apache Software Foundation and is designed to provide both high performance and flexibility for research and production workloads. MXNet supports imperative (dynamic) and symbolic (static) programming, making it suitable for both experimentation and deployment.