Plug-and-Play Overview

This section provides guidance on configuring and replacing default components of the Ingestion and Retrieval pipelines. These instructions make integrating VSS with unique hardware topologies and existing systems easier. It covers modifications to key elements such as:

• Configure the VLM: Swapping the VLM used by VSS, including fine-tuned models and 3rd-party endpoints (i.e. GPT4o).

• Configure the NIMs: Swapping the NIMs used by VSS, including different available models and sizes.

• Using External Endpoints: Utilize external endpoints (cloud, local, 3rd party) and disable the sub charts.

• Configure Storage Directories: Connect VSS to existing databases for utilizing video streams and other assets.

Note

The above instructions are specific to Docker Compose deployment only.