NV-Embed-v2 от NVIDIA — это абсолютный чемпион рейтинга MTEB, установивший новый стандарт в задачах по встраиванию текста. Модель идеально подходит для создания продвинутых RAG-систем, обеспечивая ИИ-приложениям феноменальную точность поиска информации и понимания контекста.
We present NV-Embed-v2, a generalist embedding model that ranks No. 1 on the Massive Text Embedding Benchmark (MTEB benchmark)(as of Aug 30, 2024) with a score of 72.31 across 56 text embedding tasks. It also holds the No. 1 in the retrieval sub-category (a score of 62.65 across 15 tasks) in the leaderboard, which is essential to the development of RAG technology. NV-Embed-v2 presents several new designs, including having the LLM attend to latent vectors for better pooled embedding output, and demonstrating a two-staged instruction tuning method to enhance the accuracy of both retrieval and non-retrieval tasks. Additionally, NV-Embed-v2 incorporates a novel hard-negative mining methods that take into account the positive relevance score for better false negatives removal.