HuggingFace's platform is a resource for developers and researchers working in natural language processing (NLP) and machine learning, offering insights into NLP models, tools, and datasets. Through articles, tutorials, and open-source projects, HuggingFace offers insights into state-of-the-art NLP techniques, transformer architectures, and transfer learning methods. Developers can learn about using pre-trained models, fine-tuning strategies, and deploying NLP applications with HuggingFace's libraries and APIs.

Hugging Face

NVIDIA released Nemotron ColEmbed V2, a family of late-interaction multimodal embedding models in 3B, 4B, and 8B sizes that achieve state-of-the-art performance on ViDoRe V1, V2, and V3 benchmarks for visual document retrieval. The 8B model ranks #1 on ViDoRe V3, using a ColBERT-style multi-vector architecture that enables fine-grained token-level interactions between queries and documents containing text, tables, charts, and images. Built on Qwen3-VL and Llama architectures with bi-directional attention, these models are trained using contrastive learning with hard negative mining and enhanced synthetic multilingual data, targeting research applications where retrieval accuracy is prioritized over storage efficiency.

Nemotron ColEmbed V2: Raising the Bar for Multimodal Retrieval with ViDoRe V3’s Top Model