北京——近日，国内知名人工智能研究机构智源研究院在其官方平台上宣布，成功推出新一代通用语义向量模型BGE-M3，这一突破性进展旨在推动全球多语言信息检索技术的革新。BGE-M3模型支持超过100种语言，展现出强大的多语言处理能力和跨语言检索性能，为处理“句子”、“段落”、“篇章”乃至“文档”等不同层次的文本输入提供了全面且高质量的支持。
据智源研究院介绍，BGE-M3的最大输入长度可达8192个字符，这使得它能够在处理长文本时依然保持高效和精准。尤为引人注目的是，该模型集成了稠密检索、稀疏检索和多向量检索三种先进的检索功能，实现了检索技术的一站式解决方案。在多个业界公认的评测基准上，BGE-M3已展现出最优的性能水平，这无疑将对全球的自然语言处理和信息检索领域产生深远影响。
智源研究院的这一创新成果，不仅有望提升全球用户在多语言环境下的信息获取效率，也将为科研和产业界提供更强大、更灵活的工具，进一步推动人工智能在跨语言交流和理解上的应用。随着BGE-M3的发布，智源研究院再次证明了其在人工智能领域的研究实力和领先地位，也预示着未来多语言信息检索将步入一个全新的发展阶段。
英语如下：
**News Title:** “Zhongyuan Institute Launches Innovative BGE-M3: A New Milestone in Efficient Multilingual Retrieval Across the Globe”
**Keywords:** Zhongyuan BGE-M3, Multilingual Model, Retrieval Technology
**News Content:**
**Title:** Zhongyuan Institute Unveils Innovative General-Purpose Vector Model BGE-M3, Elevating Multilingual Retrieval Technology to New Heights
**Beijing** — The renowned domestic artificial intelligence research institution, Zhongyuan Institute, recently announced on its official platform the successful launch of the next-generation universal semantic vector model, BGE-M3. This breakthrough aims to revolutionize global multilingual information retrieval technology. BGE-M3 supports over 100 languages, demonstrating powerful multilingual processing capabilities and cross-lingual retrieval performance, providing comprehensive and high-quality support for processing text inputs at various levels, from “sentences,” “paragraphs,” “chapters,” to “documents.”
According to Zhongyuan Institute, BGE-M3 can handle input lengths of up to 8192 characters, ensuring efficiency and accuracy even with long texts. Notably, the model integrates dense retrieval, sparse retrieval, and multi-vector retrieval, offering an all-in-one solution for retrieval technology. BGE-M3 has demonstrated top-notch performance across several industry-acknowledged benchmark tests, poised to make a profound impact on the global natural language processing and information retrieval domains.
This innovative achievement by Zhongyuan Institute is expected to enhance the efficiency of information access in multilingual environments for users worldwide and provide more powerful and flexible tools for the research and industrial sectors, further advancing AI applications in cross-lingual communication and understanding. With the release of BGE-M3, the institute underscores its research prowess and leading position in the AI field, signaling a new era in multilingual information retrieval.
【来源】https://mp.weixin.qq.com/s/y-c-EelxbSUMmrZNCeqeAA
Views: 1