SOURCE-LINKED INTELLIGENCE

Trajectory-Level Speculative Decoding for Diffusion Language Models

arXiv · AI, language, vision and robotics · article · Aug 27, 2026 · UTC

Diffusion-based language models (dLLMs) enable parallel token generation through iterative denoising, but existing decoding strategies collapse to single-token generation under low confidence, severely limiting throughput. Unlike autoregressive models where speculative decoding operates on token sequences in a fixed left-to-right order, dLLMs require speculating over denoising trajectories-sequences of multi-token updates with explicit positions and unmasking orders. We develop a trajectory-level speculative framework that constructs draft denoising trajectories via confidence-stratified tree

Read original source ↗ Open in workspace

recordType

paper

region

Global