FP8, NVFP4 (Blackwell), and MLX 4-bit builds — validated on RTEB-open within 0.008 nDCG@10 of NVIDIA's published scores.
-
shadowrock-io/Nemotron-3-Embed-8B-Community-FP8
Sentence Similarity • 8B • Updated -
shadowrock-io/Nemotron-3-Embed-8B-Community-NVFP4
Sentence Similarity • 5B • Updated -
shadowrock-io/Nemotron-3-Embed-8B-Community-MLX-4bit
Sentence Similarity • 1B • Updated -
nvidia/Nemotron-3-Embed-8B-BF16
Sentence Similarity • 8B • Updated • 129k • 84