Google Gemma Announces EmbeddingGemma 2: A Lightweight Multimodal Embedding Model
Summary
Google Gemma announced EmbeddingGemma 2, a lightweight, multimodal embedding model that maps text, code, images, video, and audio into a single embedding space for on-device use. It features a 740M parameter form factor, Matryoshka Representation Learning, an 8K context window, and an Apache 2.0 license, with use cases in local codebase indexing and semantic code search.