EmbeddingGemma 2 unifies text, images, audio, and video in shared space
Original titleEmbeddingGemma 2: Unified and Efficient Multimodal Representations
- Outlet
- Hugging Face
- Published
- Importance
- 3/5
- Trending
- 13
Summary
Google presents EmbeddingGemma 2, a multimodal embedding model that natively represents text, images, audio, and video in a unified space. It establishes a new Pareto frontier in embedding benchmarks and supports resource-constrained environments.
Original excerpt
We present EmbeddingGemma 2, an open multimodal embedding model which natively represents text, images, audio, video, and combinations thereof in a unified space. Our approach introduces improvements such as simplified per-layer embeddings and a multi-stage curriculum with modality and objective progression which ensure strong quality across unimodal, cross-modal, and mixed-modal evaluations. Notably, in comprehensive embedding benchmarks across all modalities, EmbeddingGemma 2 establishes a new Pareto frontier, outperforming both specialist models and models two to twelve times larger. Furthermore, our design offers flexibility for resource-constrained environments, supporting selective modality encoder loading, adjustable vision token budgets, truncatable embedding dimensions, and quantized weights. We present ablations analyzing key decisions and release EmbeddingGemma 2 to promote further research.