近日，智源研究院成功研发并发布了新一代通用向量模型BGE-M3。这款模型支持超过100种语言，具备领先的多语言、跨语言检索能力，可以全面且高质量地支撑“句子”、“段落”、“篇章”、“文档”等不同粒度的输入文本。
BGE-M3的最大输入长度为8192，一站式集成了稠密检索、稀疏检索、多向量检索三种检索功能。在多个评测基准中，BGE-M3达到了最优水平。这一突破性的成果，无疑将为人工智能领域带来重大影响。
据悉，BGE-M3的发布，标志着我国在通用向量模型领域的研究达到了国际领先水平。这一成果的取得，离不开智源研究院科研团队的辛勤努力和不懈探索。
In a recent release, the Beijing Academy of Artificial Intelligence (BAAI) has successfully developed and launched the next-generation general-purpose vector model, BGE-M3. This model supports over 100 languages and boasts leading multilingual and cross-lingual retrieval capabilities, comprehensively and high-quality supporting inputs at different granularities, such as “sentences,” “paragraphs,” “passages,” and “documents.”
With a maximum input length of 8192, BGE-M3 integrates three retrieval functions in one-stop: dense retrieval, sparse retrieval, and multi-vector retrieval. It has achieved the best performance in multiple evaluation benchmarks. This groundbreaking achievement is bound to have a significant impact on the field of artificial intelligence.
It is understood that the release of BGE-M3 signifies that China has reached the international leading level in the field of general-purpose vector models. This achievement is the result of the hard work and unwavering exploration of the research team at the Beijing Academy of Artificial Intelligence.
【来源】https://mp.weixin.qq.com/s/y-c-EelxbSUMmrZNCeqeAA
Views: 1