OpenCSGs/csghub-server

csghub-server is the backend server for CSGHub which helps user to manage datasets, modes, and also run Model Inference, Finetune and Application Spaces.

What it solves

CSGHub Server provides a centralized management platform for large model assets, allowing users to organize, search, and distribute models and datasets. It addresses the complexity of managing large-scale AI assets across different storage systems and Git servers.

How it works

The server operates as a REST API-driven backend that integrates with Git servers (like Gitea or GitLab) and LFS storage systems (local or S3-compatible cloud storage). It manages user and organization permissions, handles the content moderation of text and images, and tracks activity data such as downloads and likes.

Who it’s for

Developers and organizations that need a reliable, self-hosted asset management system for LLM models and datasets, particularly those requiring integration with Git LFS and S3 storage.

Highlights

  • own a REST API for managing models, datasets, and users
  • Support for Git LFS for handling large files and online downloads
  • Online preview for dataset files, including .parquet files
  • Automatic tagging of model and dataset labels
  • Extensible architecture supporting various Git servers and S3-compatible storage
  • Integration with OpenCSG llm-inference for one-click model deployment