Software & Tools

Open Source Software

The OpenMoss Lab is committed to developing and releasing open-source tools and frameworks for the AI research community.

Major Projects

MOSS

One of the earliest open-source conversational large language models in China. MOSS demonstrates our commitment to making advanced AI technology accessible and transparent.

  • Repository: GitHub
  • Use Cases: Conversational AI, dialogue systems, research platform

FudanNLP

A comprehensive natural language processing toolkit developed at Fudan University.

  • Features: Chinese NLP, tokenization, parsing, named entity recognition
  • Repository: GitHub

FastNLP

A modular and extensible NLP library built for rapid prototyping and experimentation.

  • Features: Modular design, easy experimentation, comprehensive documentation
  • Repository: GitHub

CoLLiE

An efficient framework for fine-tuning large language models.

  • Features: Memory-efficient training, distributed computing support, easy-to-use API
  • Repository: GitHub

Research Tools

Optimizers

  • LOMO: Low-memory optimizer for large model training
  • AdaLomo: Adaptive low-memory optimizer

Multimodal Tools

  • SpeechGPT: Speech-language model
  • SpeechTokenizer: Speech tokenization toolkit
  • AnyGPT: Multimodal generation model

Agent Frameworks

  • UnifiedToolHub: Tool-augmented LLM framework
  • FamilyTool: Collaborative tool-use system
  • R3-RAG: Retrieval-augmented generation system

Evaluation Benchmarks

  • VLA-bench: Vision-language-action reasoning benchmark
  • VehicleWorld: Real-world simulation environment

Coming Soon

Additional tools and frameworks are under development and will be released to the community soon.

For the latest releases and updates, visit our GitHub organization.

Community

We welcome contributions, bug reports, and feature requests from the community. Please see individual project repositories for contribution guidelines.

Citation

If you use our software in your research, please cite the relevant papers. Citation information is available in each project’s repository.