Skip to main content

Guides

This section provides task-oriented, step-by-step guides for developers, platform engineers, and AI operations teams working with MatrixHub.

đŸ“Ļ Model Repository Operations​

Learn how to manage and distribute actual model files and weight repositories:

  • 👉 Upload & Download Models: Learn how to upload custom fine-tuned weights or download cached weights using standard Hugging Face tools, the Python SDK, and Web UI.
  • 👉 Model Project Settings: Configure visibility parameters, tag locks, and metadata descriptors for your model repositories.

đŸ‘Ĩ Project & Member Governance​

Manage access control scopes and team collaborations:

  • 👉 Project Lifecycle: Step-by-step guide on creating secure logical projects to isolate different research teams or model domains.
  • 👉 Member Roles & RBAC Mapping: Invite members to your workspace projects and assign specific permission roles (Owner, Manager, Reporter).

🔑 User Profiles & CLI Authentication​

Generate secure credentials for command-line tools:

  • 👉 Access Tokens: Learn how to create, configure, and rotate private API tokens to authenticate huggingface-cli or pipeline scripts.

⚙ī¸ Platform Administration​

Enterprise settings and multi-region replication controls for platform engineers:

  • 👉 User & Account Management: Manage global accounts, set usage quotas, and configure corporate single sign-on (SSO/LDAP) configurations.
  • 👉 Registry Management: Configure upstream connection endpoints and credentials to access external registry sources (Hugging Face, ModelScope).
  • 👉 Cross-Region Remote Sync: Configure asynchronous, policy-driven remote replication links between geographical data center hubs.