CSGHub Server is a backend platform for managing Large Language Model (LLM) assets, including models and datasets. It provides REST APIs for user and organization management, auto-tagging, content preview and moderation, and activity tracking. The platform is designed for LLM developers and researchers seeking a centralized, reliable system for asset lifecycle management.

How It Works

The server leverages a modular architecture, supporting integration with various Git servers (like Gitea) and flexible LFS storage solutions, including S3-compatible cloud storage. This allows users to customize their backend infrastructure. It also enables on-demand content moderation through third-party services, enhancing asset safety and compliance.

Quick Start & Requirements

• Installation: Deploy via docker-compose.
• Prerequisites: Docker, 4c CPU/8GB memory, Ubuntu 22 tested. Requires an API token (>= 128 chars).
• Resources: docker-compose.yml available for download.
• Docs: Demo video available, OpenCSG website for experience.

Highlighted Details

• Supports Gitea, with plans for other Git servers.
• Git LFS support for large files and web UI downloads.
• Dataset preview for LFS format data.
• Model/Dataset auto-tagging with custom metadata.
• S3 protocol support (e.g., MinIO) for storage.
• One-click model deployment integration with llm-inference.

Maintenance & Community

• Active development with a roadmap including more Git server support and model format conversion.
• Community channels: Discord, Slack, WeChat group.
• Contribution guidelines and backend developer guides are available.

Licensing & Compatibility

• Licensed under Apache 2.0.
• Permissive license suitable for commercial use and integration with closed-source projects.

Limitations & Caveats

The project is currently tested on Ubuntu 22 and primarily supports Gitea, with broader Git server compatibility planned. Model format conversion is a future feature.