Two Roads Through Diffusion Space

Two tutorial posts on Hacker News today trace the two distinct technical lineages of diffusion language models — discrete masked and continuous embedding-space — both of which have now reached production deployments with 2–10× throughput gains over autoregressive models. The quality gap with AR is measurably narrowing.

Read more →

The Integral Shortcut Through Diffusion Space

Sander Dieleman's post on flow maps frames diffusion model distillation as learning to compute the integral of the velocity field directly, rather than stepping along tangent directions. The reformulation unifies 20+ recent papers under three consistency constraints and explains why single-step sampling is achievable without sacrificing bijectivity.

Read more →