AgentSkillsCN

Setup Gpu Workbench

搭建 GPU 工作台

SKILL.md

Setup GPU Workbench

Create a GPU-enabled data science workbench with common ML frameworks pre-installed.

Usage

typescript
import { setupGpuWorkbench } from './skills/setup-gpu-workbench';

await setupGpuWorkbench({
  projectName: "ml-experiments",
  workbenchName: "training-notebook",
  framework: "pytorch",
  gpuCount: 1,
  storageSize: "100Gi"
});

Parameters

ParameterRequiredDescription
projectNameYesTarget OpenShift AI project
workbenchNameYesName for the workbench
frameworkNoML framework: pytorch (default), tensorflow
gpuCountNoNumber of GPUs (default: 1)
containerSizeNoContainer size: Medium (default), Large, X-Large
storageSizeNoPersistent storage size (default: 50Gi)

What This Skill Does

  1. Ensures the project exists
  2. Creates a workbench with GPU support
  3. Configures appropriate container resources
  4. Attaches persistent storage for notebooks and data
  5. Enables OAuth authentication

Framework Options

  • pytorch: CUDA-enabled PyTorch with Jupyter
  • tensorflow: CUDA-enabled TensorFlow with Jupyter