SOURCE-LINKED INTELLIGENCE

DFlow: Enabling Verifier Information Flow in Block Diffusion Speculative Decoding

arXiv · AI, language, vision and robotics · article · Sep 6, 2026 · UTC

Block diffusion speculative decoding improves LLM inference efficiency by proposing a block of future tokens in parallel and verifying them with a single forward pass through the target model. However, existing methods retain only the accepted prefix and discard the rejected suffix, preventing the computation spent on these positions from benefiting subsequent drafting rounds and forcing the drafter to repeatedly reconstruct representations for future tokens from scratch. We observe that rejection only determines whether a proposed token can be committed, while the verifier representations at

Read original source ↗ Open in workspace

recordType

paper

region

Global