LLM Digest

Story

arxiv_cs_lg · Aug 13, 2026 · paper

Source brief

DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees

arxiv.orgAug 13, 2026
original source linked

In brief

Speculative decoding losslessly accelerates autoregressive language models by verifying multiple draft tokens in parallel. Diffusion-based drafters further reduce proposal latency by predicting an entire token block i...

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

DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees

Earlier in this thread 4 items

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

EAGLE-3 Speculative Decoding on AMD Instinct GPUs: Training and Serving with vLLM and AMD Quark

Achieve state-of-the-art inference latencies with speculative decoding

Boost Inference Performance up to 15x on NVIDIA Blackwell Using DFlash Speculative Decoding - NVIDIA Developer