Huawei has announced that its framework for AI app development MindSpore is now open source and available on GiHub and Gitee.
Chief scientist at Huawei MindSpore and an IEEE Fellow, Professor Chen Li explained how MindSpore can scale across devices in a press release, saying:
- Huawei says it is back to 90 per cent capacity
- Google launches machine learning framework for training quantum models
- Amazon trialing AI-based customer support agents
"MindSpore natively adapts to all scenarios across the device, edge, and cloud. We implement AI Algorithms As Code through on-demand collaboration for easier model development, and provide cutting-edge technologies, and co-optimization with Huawei Ascend AI processors to improve runtime efficiency and computing performance. We also support other processors such as GPU and CPU."
MindSpore already has the backing of a number of partners including the University of Edinburgh, Peking University, Imperial College London and the robotics startup Milvus.
The framework is able to run on processors, graphics cards and dedicated neural processing units such as the one in Huawei's own Ascend AI chips. MindSpore also has 20 percent fewer lines of codes than other frameworks when dealing with natural language processing models and the company claims that this leads to an average efficiency boost of 50 percent. Additionally, the framework supports parallel training across hardware as well as dynamic debugging which enables developers to isolate bugs while taking less time to train AI models.
One other interesting thing about MindSpore is the fact that it doesn't process any data on its own but instead, ingests gradient and model information that has already been processed. As a result of this, the framework preserves sensitive data even in “cross-scenario” environments while ensuring that models remain robust.
MindSpore currently requires Python 3.7+ to run but the framework will soon support other languages such as C++, Rust and Julia.
- We've also highlighted the best cloud hosting services