This repository provides a curated collection of AI application examples built with MindSpore, covering domains such as CV, NLP, GANs, Diffusion Models, LLMs, RAG, and Agents. Each case is organized by domain and model, offering reusable implementations for representative tasks. It serves as a practical reference for developers to explore and extend AI solutions within the MindSpore ecosystem.
- 2025-11-18 [Feature Optimization]：Repository refactored for clearer application navigation; added Issue and Pull Request templates for more standardized contributions.
Before starting this course, you should be familiar with:
- Basic Python programming
- Basic Linux commands
- Using Jupyter Notebook
- Using Docker images
You can take the Prerequisite Test to assess your readiness.
To ensure all example code runs smoothly, set up your environment using one of the following methods.
Confirm your Python version meets the course requirements, then run:
pip install -r requirements.txtPrebuilt Dockerfiles are provided to simplify environment setup.
You can find all course images in the dockerfile directory and pull the one that fits your hardware.
For details, see Using Docker Images in Wiki.
The notebooks are organized by domain. Within each domain, notebooks are further grouped by model to provide a clear and scalable structure.
| No. | Domain | Description |
|---|---|---|
| 1 | CV | Vision models and tasks (classification, detection, segmentation). |
| 2 | NLP | Text processing, sequence modeling, and language understanding tasks. |
| 3 | GAN | GAN models for image synthesis and style transfer. |
| 4 | Audio | Audio classification, speech tasks, and signal processing examples. |
| 5 | Diffusion | Diffusion-based generation models and training workflows. |
| 6 | LLM | Large language models for text generation, reasoning, and instruction tasks. |
| 7 | Multi-Modal | Models combining text, vision, or audio modalities. |
| 8 | OrangePi | Edge-AI applications on OrangePi with MindSpore. |
| 9 | RAG | Retrieval-augmented generation pipelines and examples. |
| 10 | Agent | Agent-style workflows and task-oriented intelligent systems. |
- ChatPDF: an application that allows users to upload PDF files and interact with pdf using a chatbot.
- Emotional-News-Anchor: an AI-powered news reader that uses LLMs to analyze and broadcast news with emotion-aligned narration.
This repository is updated in sync with MindSpore and the MindSpore NLP Suite.
| Branch/Version | Python | MindSpore | MindSpore NLP |
|---|---|---|---|
| dev | >=3.9, <=3.11 | 2.7.0 | 0.5.1 |
See the FAQ in the Wiki for details.
-
Issue: We welcome bug reports, suggestions and feature requests via Issues.
-
Pull Requests： Developers may contribute bug fixes or code enhancements by submitting a Pull Request. Before submitting, please review the Contributing Guidelines. Each PR will be reviewed and merged by Committer @xing-yiren and at least one additional committer. Your contributions help continuously improve the project.
-
Open-Source Project Submissions: If you would like to recommend or self-nominate a qualified open-source project, please first align the repository with the project repository guidelines. Then submit the project via email to contact@public.mindspore.cn with the subject line:
【MindSpore + OrangePi Project Submission】<Project Name>. Include a brief project introduction and a link to the source repository in the email body.
Special thanks to all contributors for improving this project!