Repolode
Submit
mindspore-ai

mindspore-ai/mindspore

MindSpore is an open source deep learning training and inference framework for mobile, edge and cloud scenarios, with native support for Huawei Ascend AI processors.

Open-source alternative to:PyTorchTensorFlow

View on GitHubWebsite
Stars
4,719
Forks
746
Open issues
253
Last commit
11 days ago

What is it?

MindSpore is a deep learning framework for training and running neural networks across mobile, edge and cloud environments. It is aimed at data scientists and algorithm engineers who want a Python-based development experience alongside efficient execution, and it is designed with hardware and software co-optimization in mind.

The framework uses source-transformation automatic differentiation, which compiles the program into a static graph and supports control flow and higher-order functions. Its automatic parallel feature aims to combine data, model and hybrid parallelism and select a low-cost splitting strategy for distributed training without manual partitioning.

Installation is available via pip or from source for CPU, NVIDIA GPU (CUDA 10.1) and Huawei Ascend hardware on Ubuntu, CentOS, EulerOS and Windows (CPU), and Docker images are published for CPU and GPU backends.

Key features

  • 01Automatic differentiation based on source transformation, supporting control flow and higher-order functions.
  • 02Automatic parallelism that combines data, model and hybrid parallel training.
  • 03Static graph compilation for performance optimization, with a Pythonic programming interface.
  • 04Native support for Huawei Ascend AI processors alongside CPU and CUDA GPU backends.
  • 05Installation via pip, source compilation, or Docker images.

Use cases

  • Training deep learning models on Huawei Ascend hardware.
  • Running distributed training that needs automatic model or data parallelism.
  • Deploying trained neural networks to mobile, edge or cloud targets.
  • Developing AI models in Python with a static-graph compilation path.

Quick start

  1. 1

    Install MindSpore CPU build with pip (wheel from the README)

    pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/1.2.0-rc1/MindSpore/cpu/ubuntu_x86/mindspore-1.2.0rc1-cp37-cp37m-linux_x86_64.whl
  2. 2

    Run the verification Mul example from the README in Python

Things to know

  • The README's pip example installs an old 1.2.0rc1 wheel for Python 3.7 on Ubuntu x86, so newer setups need the current install guides.
  • Native support and most documented builds target Huawei Ascend hardware, while the GPU build is listed only for CUDA 10.1.

Topics

#deep-learning#machine-learning#neural-network#distributed-training#automatic-differentiation#ascend#edge-ai#python

FAQ

What is mindspore?
MindSpore is an open source deep learning training and inference framework for mobile, edge and cloud scenarios, with native support for Huawei Ascend AI processors. MindSpore is a deep learning framework for training and running neural networks across mobile, edge and cloud environments. It is aimed at data scientists and algorithm engineers who want a Python-based development experience alongside efficient execution, and it is designed with hardware and software co-optimization in mind.
Is mindspore free to use? What license does it have?
mindspore is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is mindspore still maintained?
Actively maintained. The last commit to mindspore was on Oct 28, 2026.
What can I use mindspore for?
Training deep learning models on Huawei Ascend hardware. Running distributed training that needs automatic model or data parallelism. Deploying trained neural networks to mobile, edge or cloud targets. Developing AI models in Python with a static-graph compilation path.
How do I install mindspore?
Install MindSpore CPU build with pip (wheel from the README): pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/1.2.0-rc1/MindSpore/cpu/ubuntu_x86/mindspore-1.2.0rc1-cp37-cp37m-linux_x86_64.whl Run the verification Mul example from the README in Python
Where can I get mindspore?
The source code and installation instructions are on GitHub: https://github.com/mindspore-ai/mindspore

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects