UEmbed：统一稀疏与稠密的多模态嵌入模型
原标题：UEmbed: Unified Sparse and Dense Multimodal Embeddings
AI 摘要
Hugging Face 每日论文介绍了 UEmbed，一种仅解码器的多模态嵌入模型，能在一次因果前向传播中同时生成稀疏词级和稠密表示。UEmbed 通过添加可学习特殊标记并将词汇表划分为不相交子集来实现稀疏预测，发布了 2B、4B 和 9B 三种规模，其中 9B 模型在 MMEB-v2 上达到 71.8（稠密）和 71.0（稀疏）的分数，优于现有公开数据训练的多模态嵌入模型，并在 BEIR 基准上保持竞争力。该模型统一了稠密和稀疏嵌入，并将稀疏检索扩展到文本和多模态输入，展示了在效率、效果和智能体应用方面的实用性。
正文节选
UEmbed: Unified Sparse and Dense Multimodal Embeddings Abstract Sparse retrieval underpins modern search systems, from web search to retrieval-augmented generation. Existing work has introduced Learned Sparse Retrieval (LSR) to push beyond exact lexical matching toward richer semantics. Yet LSR has so far remained tied to encoder-style bidirectional architectures, and its extension to multimodal settings still relies heavily on auxiliary cross-modal modules. To address these limitations, we intr