Target-conditioned diffusion drafters predict candidate tokens in parallel, yet each new block must wait for updated target-model hidden states. This forces drafting and verification to run sequential
D3D is a standalone diffusion drafter that removes target-model hidden-state dependencies to enable asynchronous speculative decoding with high speedup.
Target-conditioned diffusion drafters predict candidate tokens in parallel, yet each new block must wait for updated target-model hidden states. This forces drafting and verification to run sequentially, limiting speedup even though each block is generated in one pass. We introduce D3D, a standalone diffusion drafter that removes this inter-round dependency. D3D combines blockwise denoising with auxiliary autoregressive supervision to achieve long accepted prefixes without target hidden states. This independence lets D3D draft successor blocks during verification in one packed pass, enabling asynchronous speculative decoding. We progressively prune layers according to their effect on…
Summary and abstract excerpt via Hugging Face Papers ↗ · arXiv.
D3D: Decoupled Diffusion Drafting for Asynchronous Speculative Decoding was first seen on Hugging Face · Daily papers on Nov 6, 2026. Since then it has appeared in 32 checks on 3 different days. Its best position was #11 of 30, reached on Nov 6, 21:28 UTC. It is #14 right now, holding steady at #14 over the last 6 checks.
Position on Hugging Face · Daily papers at every check (lower is better).
Best rank per day (UTC).
▲ as reported by Hugging Face.
Archived daily, weekly and monthly lists that include this entry.
Yes. As of Nov 8, 17:28 UTC it is #14, holding steady at #14 over the last 6 checks.
#11 on Hugging Face · Daily papers, on Nov 6, 21:28 UTC. Trendoday checks this list every 120 minutes.
It was first seen on Nov 6, 2026 and has been on the list on 3 days (32 checks) since.