昇思MindSpore

MindSpore is Huawei's open-source deep learning framework. Known in Chinese as 昇思MindSpore, this AI framework is built to cover edge, cloud and on-device scenarios, so you can train a model once and deploy it in several of them.

It handles both training and inference, and its main uses are computer vision and natural language processing.

The technical core is automatic differentiation based on source code transformation.

MindSpore also offers automatic distributed parallel training, data processing and a graph execution engine, which matter most when models get large and training has to be spread across many devices.

Parallel processing is supported as a built-in capability rather than something you bolt on yourself.

The audience is data scientists and algorithm engineers who build and tune models, not people looking for a no-code product.

Because MindSpore is open source, you can read the code and work with the wider developer community around it.

If your work involves Huawei hardware or you want a framework geared toward device-to-cloud deployment, this deep learning framework is a reasonable one to evaluate next to the more familiar options.