NVIDIA Releases Nemotron-Labs-Sigma-Distill-8B-NFE16 Model

NVIDIA has launched the Nemotron-Labs-Sigma-distill-8B-nfe16, an 8 billion-parameter diffusion language model.

NVIDIA has introduced the Nemotron-Labs-Sigma-distill-8B-nfe16, an 8 billion-parameter continuous diffusion language model designed for fast few-step text generation. This model, distilled using Phased Distribution Matching Distillation from a multiscale SUSReg-regularized 8B teacher, features a maximum context length of 4,096 tokens and generates text in 32-token blocks. It is available for both commercial and non-commercial use under the NVIDIA Nemotron Open Model License. The model runs offline on the user’s hardware, specifically requiring a CUDA-capable NVIDIA GPU, as CPU inference is not supported. NVIDIA claims that by leveraging their GPU-accelerated systems, the model achieves faster training and inference times compared to CPU-only solutions. The model is particularly suited for latency-sensitive generation research and deployment-oriented experimentation, according to the model card.