Apache MXNet is an open-source deep learning framework that is lightweight, memory-efficient, and highly performant. It is known for its ability to work on CPUs and GPUs.

MXNet is also portable to smart devices via its cross-compilation on ARM. Its NumPy-like interface integrates with Gluon 2.0 and supports parallelism through BytePS, Horovod, and ps-lite.

Project Background

• Framework: Apache MXNet
• Author: Carlos Guestrin
• Stable Release: March 2021
• Type: Open-source library for deep learning
• License: Apache License 2.0
• Contains: D2L.ai, GluonCV, GluonNLP, GluonTS, 8 language bindings, distributed training, etc.
• Language: C++, Python, R, Java, Julia, JavaScript, Scala, Go, Perl
• GitHub: incubator-mxnet has 19.7k stars and 864 contributors
• Runs On: Windows, macOS, Linux
• Twitter: MXNet
• Stackflow: MXNet

Applications

• Simplifying common expressions
• Design deep neural networks
• Faster computation
• Accelerate numerical computation
• Face recognition
• Machine translation
• Optical character recognition
• Long short-term memory networks
• Convolutional neural networks (CNNs)
• Exploration and experimentation
• Scale to multiple GPUs
• Augmented reality and scene identification
• Translate voice conversations into text