Show HN: Running EmbeddingGemma2 on the Browser

Category: ai-ml

Tags: browser, embedding, multimodal, semantic-search, transformers-js, webgpu

Score: 5.3/10 (Innovation: 5, Technical: 6, Documentation: 2, Utility: 5)

A browser-based semantic search tool for personal media collections, using EmbeddingGemma 2 via transformers.js with WebGPU/WASM backends. It supports text, image, audio, and video embeddings in a shared vector space, enabling cross-modal search. Interesting for demonstrating on-device multimodal embedding and the trade-offs of client-side model inference.

Target audience: frontend developers, AI enthusiasts, end users seeking personal media search

Repository: https://searchyourcollection.com/examples

View on Hacker News