Skip to content

Getting Started with CMF GUI¤

The CMF GUI provides an intuitive, browser-based interface for exploring ML pipeline metadata, visualizing lineage relationships, and synchronizing metadata between multiple CMF servers. Built with React and D3.js, it offers interactive dashboards for artifacts, executions, and pipeline lineage.

Overview¤

The CMF GUI consists of several main sections:

  • Artifacts: Browse and search datasets, models, and metrics
  • Executions: View pipeline runs and execution history
  • Lineage: Visualize data flow and dependencies
  • Metahub: Synchronize metadata between CMF servers
  • TensorBoard: View ML training metrics

Quick Start¤

Accessing the CMF GUI¤

  1. Ensure the CMF Server is running
  2. Open your browser and navigate to the server URL (default: http://your-server-ip:80)
  3. The GUI will display the available pipelines in the sidebar

Artifacts View¤

The Artifacts page displays all datasets, models, and metrics tracked by CMF. You can browse, search, and explore artifact metadata, versions, and lineage.

Key Features¤

  • Artifact Listing: View all artifacts with their types (Dataset, Model, Metrics)
  • Search & Filter: Find specific artifacts by name, type, or properties
  • Artifact Details: Examine metadata, custom properties, and version information

CMF Artifacts Page

Executions View¤

The Executions page displays all pipeline runs and execution history. You can view execution details, parameters, and associated artifacts for each run.

Key Features¤

  • Execution History: View all past executions with timestamps
  • Execution Details: See parameters, properties, and metadata for each run
  • Filtering & Search: Find specific executions by name, type, or properties

CMF Executions Page

Lineage Visualization¤

The Lineage page offers interactive visualizations of data flow and dependencies in your ML pipelines. It provides following different visualization modes:

Visualization Types¤

  1. Artifact Lineage: Hierarchical view of artifact dependencies
  2. Execution Lineage: Hierarchical view of execution flow
  3. Artifact-Execution Lineage: Combined view showing both artifacts and executions

CMF ArtifactExecution Page

Metahub¤

The Metahub feature enables synchronization of metadata between two CMF servers, allowing distributed teams to collaborate and share ML pipeline metadata.

TensorBoard Integration¤

CMF integrates with TensorBoard to visualize training metrics, model graphs, and other ML-specific visualizations alongside CMF metadata.

Prerequisites¤

Before using the CMF GUI, ensure you have:

  1. CMF Server Running: Follow the installation guide
  2. Metadata Pushed: Use cmf metadata push to send metadata to server
  3. Browser Compatibility: Modern browser (Chrome, Firefox, Safari, Edge)

Getting Help¤