Note
Building multilingual RAG applications?
Building multilingual RAG applications?
Building multilingual RAG applications?
Nomic AI’s new Mixture of Experts embedding model is here!
Nomic AI has just released nomic-embed-text-v2-moe. It’s an MoE embedding model with SOTA multilingual retrieval performance:
• Performance: competitive with models 2x in size • Multilingual: Supports over 100 languages • Architecture: Mixture of Experts (8 experts with top-2 routing) • Storage: Matryoshka embeddings with up to 3x reductions in storage cost at minimal performance loss (768-256 dimension) • Fully Open-Source: Weights, code, and training data publicly available • License: Commercially permissible Apache 2.0 license
Check it out on the Hugging Face model page: https://lnkd.in/ev5hVhDq
Start using it immediately through Weaviate’s Hugging Face integration: https://lnkd.in/eAqtG9-n
Commentaires
Réponses imbriquées possibles. Vous commentez en tant qu'invité : choisissez un pseudo, et c'est tout. Modération a posteriori : tout est publié immédiatement, je peux retirer si besoin.