LaunchAI Models
2 hours ago
NVIDIA releases Nemotron-TwoTower-30B-A3B-Base-BF16 diffusion language model
NVIDIA's diffusion-based language model uses a frozen autoregressive context tower and a diffusion denoiser tower for iterative generation. Built on the Nemotron 3 Nano 30B-A3B backbone, it is available in BF16 on Hugging Face.