AI Resonance · 阅读最新日报 · AI 入门推荐

2026-11-05 · America/Los_Angeles · 论文 · #11

D3D: Decoupled Diffusion Drafting for Asynchronous Speculative Decoding

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…

D3D: Decoupled Diffusion Drafting for Asynchronous Speculative Decoding

热度 45.9 / 100;排名与评分保留该期记录。