NVIDIA Team Releases Sigma, First Large-Scale Continuous Diffusion Language Model
NVIDIA researchers have introduced Sigma, the first large-scale (3B/8B) continuous diffusion language model trained with a likelihood objective. It matches discrete diffusion and autoregressive baselines on math and coding tasks while maintaining quality with fewer inference steps.
2026-10-05 ~ 2026-10-06 · 3 related posts
- Sigma: first large-scale 3B/8B continuous diffusion LM matches AR models on math and coding — LucaAmb · 2026-10-05
- First large-scale 3B/8B continuous diffusion LMs match pass@1 and beat pass@k vs masked dLMs — ArashVahdat · 2026-10-06
1 near-duplicate retellings: ArashVahdat