Back to search
A local, multimodal embedding server and vector database using NVIDIA's Omni-Embed-Nemotron-3B. This project allows you to generate embeddings for Text, Images, Audio, and Video on consumer hardware (tested on RTX 5070 Ti / 12GB VRAM), save them to a local index, and perform multimodal semantic search.
Stars
11
Forks
0
Watchers
11
Open Issues
0
Overall repository health assessment
No package.json found
This might not be a Node.js project
4
commits
docs: Refine installation instructions, update hardware recommendations, detail video subsampling, and consolidate dependencies into `requirements.txt`.
e698f51View on GitHubPrevent disk/RAM explosion on video load & fix stats KeyError
59c580aView on GitHub