Edge Developer Toolbox Developer Guide
Large Language Model (LLM) Sample Application
This feature is only available to premium customers, and it may not be shown on your page.
Tokens Per Second (TPS) metrics are now available in the Customer Support LLM sample application’s telemetry dashboard. To view these metrics, interact with the Personal Q&A Assistant. The TPS panels in the dashboard will only populate while using this feature.
Navigate to the Develop AI Application section and choose the LLM Customer Support on the Edge sample and select Benchmark on Intel Hardware:
Once launched you will see the sample running and under output choose the third icon (square with the arrow) and select the link that shows up. It will bring up the Assistant in a new tab:
Click the Personal Q&A Assistant tab, enter a prompt, and submit it. Once the Assistant has answered the prompt, you can view the tokens per second that was obtained for the response.
Open the telemetry dashboard as you normally would, and you will see the TPS panels in addition to the system telemetry panels: