Model Inference
The following sections describe how to run inference using your trained imitation learning model. You can perform inference either through the Web GUI or the LeRobot CLI. After running inference, you can visualize the results to evaluate how your policy performed.
1. Web UI (Physical AI Manager) – Recommended
You can run inference directly through the Web UI provided by the Physical AI Manager.
Option 1Web UI (Recommended)
2. LeRobot CLI – Deprecated (legacy use only)
This method is no longer officially supported, but remains available for legacy users or automated workflows.
Option 2LeRobot CLI
Visualizing Inference Results
After running inference, you can inspect the result data visually to verify how your policy behaved.