NVIDIA AI Releases Nemotron 3 Embed: Open Embedding Collection, 8B Parameter Checkpoint Tops RTEB
NVIDIA
NVIDIA AI unveiled its open embedding collection Nemotron 3 Embed. The 8-billion-parameter model checkpoint achieved first place in the RTEB ranking.
NVIDIA AI has announced the release of Nemotron 3 Embed, an open collection of embeddings designed for search and information retrieval tasks. A key feature is the model checkpoint with 8 billion parameters, which, according to the company, achieved first place in the RTEB (Retrieval Text Embedding Benchmark) benchmark. The collection includes several models of varying sizes, providing developers with flexible options for choosing the appropriate solution. The release is aimed at researchers and engineers working in natural language processing and information retrieval.
Source: Meta AI (GNews) —
original