hybrid-mamba-transformer

Tag

Cards List
#hybrid-mamba-transformer

Nemotron-TwoTower: Diffusion Language Modeling with Pretrained Autoregressive Context

arXiv cs.CL · 2026-06-26 Cached

The paper proposes Nemotron-TwoTower, a diffusion language model that decouples context representation and denoising using a frozen autoregressive tower and a trainable diffusion denoiser, achieving 98.7% of baseline quality with 2.42x throughput.

0 favorites 0 likes
← Back to home

Submit Feedback