News for hackers. Stuff that matters.
Re:Fast-DLLM: Training-Free Acceleration of Diffusion LLM by yorwba on Friday October 24, 2025 @04:39AM (#45692340)
In particular, part of the paper is about dynamically adjusting the number of tokens generated in parallel while maintaining roughly the same output quality as one-token-at-a-time decoding. The other part is about the KV caching strategy they use to speed up parallel decoding further.

There are no comments on this comment yet.