MindSpore Transformers Documentation

The goal of the MindSpore Transformers suite is to build a full-process development suite for Large model pre-training, fine-tuning, inference, and deployment. It provides mainstream Transformer-based Large Language Models (LLMs) and Multimodal Models (MMs). It is expected to help users easily realize the full process of large model development.

Note

Starting with r2.0.0, MindSpore Transformers has adopted a dynamic graph (PyNative) implementation as its primary development path, and the documentation is now primarily focused on dynamic graphs. All documentation related to the original static graph (GRAPH_MODE) implementation has been moved to the Static Graph Implementation section and marked as deprecated. For information on capabilities not yet covered by dynamic graphs, such as inference, service-oriented deployment, and quantization, please refer to that section.

This section is the initial release of the dynamic graph documentation framework. Currently, only the "Overview" page is available; installation guides, training guides, and feature pages will be added in future releases. The model support library, installation, contribution guides, and FAQ will remain on their existing pages.

The open-source code repository for MindSpore Transformers is located at AtomGit | MindSpore/mindformers. If you have any suggestions for MindSpore Transformers, please contact us via issue and we will handle them promptly.