Token Pooling-Aware Training Cuts Multi-Vector Embedding Storage by 4x Without Quality Loss
Hugging Face demonstrates a training technique that preserves retrieval quality while dramatically reducing the vector storage footprint of ColBERT-style multi-vector embedding models.
Box integrates Nvidia Nemotron embeddings to accelerate agentic document retrieval
Box Content Retrieval, powered by Nvidia's Nemotron embedding models, promises 4x faster searches with 20% fewer errors for enterprise agents.
NVIDIA Nemotron-3-Embed-8B Tops RTEB Leaderboard, Reshaping Production RAG Deployment
NVIDIA releases three open-weights embedding models, with the 8B variant claiming the #1 spot on RTEB's multilingual leaderboard and targeting enterprise agentic retrieval workloads.
Web Data Infrastructure Emerges as Critical AI Bottleneck
Real-time web data retrieval is becoming essential for enterprise AI systems to reduce hallucinations and ground models in current information, driving investment in specialized infrastructure.
Hugging Face Releases Six Open-Weights Ettin Reranker Models, From 17M to 1B Parameters
New CrossEncoder rerankers built on ModernBERT achieve state-of-the-art performance at multiple scales with published training recipes.
IBM Granite Embedding Multilingual R2: 97M and 311M Parameter Models Top MTEB Multilingual Retrieval Charts
IBM releases two Apache 2.0 multilingual embedding models built on ModernBERT, with 32K-token context and coverage for 200+ languages.