Google DeepMind 发布开放轻量多模态嵌入模型 EmbeddingGemma 2
原标题：EmbeddingGemma 2: an open, lightweight multimodal embedding model
AI 摘要
Google DeepMind 发布 EmbeddingGemma 2，一款基于 Gemma 4 架构、采用 Apache 2.0 许可的开放轻量级多模态嵌入模型，参数量 7.4 亿，可在设备端运行。该模型将代码、图像、视频和音频统一到共享嵌入空间，支持 8K 上下文窗口，并通过 MRL 实现向量维度动态截断以节省存储。相比前代，其代码性能在 MTEB Code 上提升 9.92 分，并可与 Gemma 4 配合构建端侧多模态 RAG 流程。
正文节选
EmbeddingGemma 2: an open, lightweight multimodal embedding model We introduced EmbeddingGemma last year to provide a lightweight option for high-quality text embeddings, to help your apps organize, search, and connect information directly on consumer hardware. The developer community’s response blew past our expectations. With more than 20 million downloads, builders have used it to power smarter on-device search tools and privacy-first retrieval augmented generation (RAG) pipelines. Today, we’