LLM Digest

Story

arxiv_cs_ai · Nov 9, 2026 · paper

Source brief

DForkTree: Accelerating Speculative Decoding with Fork-Aware Block Diffusion Draft Trees

arxiv.orgNov 9, 2026
original source linked

In brief

Speculative decoding speeds up language models by letting a lightweight drafter propose several future tokens that the target model verifies in parallel. Block diffusion drafters such as DFlash draft a whole block in one forward pass, and DDTree builds a draft tree from their per

Read the original at arxiv.org →Open in live feed Read that day’s brief

DForkTree: Accelerating Speculative Decoding with Fork-Aware Block Diffusion Draft Trees

Earlier in this thread 4 items

DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees

NgramRL: Accelerating RL Post-Training for Structured Generation via Rollout-Aware N-gram Speculative Decoding

Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds

Draft Less, Retrieve More: Hybrid Tree Construction for Speculative Decoding