MarkTechPost· Asif Razzaq·· 2 hr agoAI score60
NVIDIA AI Releases Nemotron-Labs-Sigma, a Continuous Diffusion Language Model with 14 checkpoints at 3B/8B
NVIDIA AI Releases Nemotron-Labs-Sigma, a Continuous Diffusion Language Model with 14 checkpoints at 3B/8B
AI summary
NVIDIA released Nemotron-Labs-Sigma, a family of 14 open checkpoints at 3B and 8B scale that denoises text in a learned 16-D continuous embedding space rather than on discrete tokens. Sigma-8B base averages 71.16 on NVIDIA's 6-task suite versus 70.22 for NLD-8B, and the self-speculative 8B checkpoint runs 3.46x faster than AR decoding on average on one GB200 with SGLang. Weak spots include 17.00 on AIME25, a 4,096-token base context (8,192 for Sigma-8B-instruct) and GPU-only inference in bfloat16.
Source: MarkTechPost · marktechpost.com