Quick Start¶
Quick Start includes a series of hands-on tutorials that help you quickly experience the features of d.run AI Operating System, from installation, compute onboarding, and model deployment to model training, agents, and operations insight. Two learning paths are provided by role:
- Admin view: deploy the platform and onboard compute;
- User view: deploy models, call models, train models, create agents, and review operations.
Complete the tutorials in order. Each tutorial can be performed independently and helps you quickly get familiar with the core features of d.run AI Operating System.
After the three steps above, the platform is ready to use and compute is onboarded. As an admin, you can also continue with the tutorials in the user view to explore the full capabilities of the platform.
Sign in with the account created by the admin to get started. No local deployment is required; model deployment and invocation are all completed in the console.
Prerequisites¶
Before you start, prepare a browser that can access the console, and confirm the following by view:
- Installation materials are ready: the offline image package, the ISO OS image, the osPackage offline package, and a machine that serves as the seed node. See prerequisites for details.
- Sign in with an account that has the
AdminorFolder Adminrole. - (Creating a work cluster) Several GPU nodes are ready, and the nodes run the same OS and CPU architecture.
- The admin has completed platform deployment, account creation and authorization, and work cluster creation (admin tutorials 1–3).
- Sign in with the account created and authorized by the admin.
- GPU resources are available in the work cluster.
- (AI Lab training) A StorageClass with
ReadWriteManyaccess mode (NFS or JuiceFS recommended) is configured in the cluster.
Download Center Learn about Token Factory Learn about d.run AI Operating System