Trajectory-Level Speculative Decoding for Diffusion Language Models

arXiv:2608.27514v1 Announce Type: new Abstract: 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 exploration and verifies them through blockwise parallel evaluation with bidirectional attention masking. Our method further introduces inter-block speculation, exploiting diffusion models' bidirectional structure to perform cross-block lookahead. We formally characterize when this approach is exact and identify trajectory drift as the fundamental cost of increased parallelism. Building on Fast-dLLM's dual-cache infrastructure, our framework reduces denoising iterations by 30-40% and increases tokens-per-step from 2.6 to 4.3, achieving 7-14x speedup over vanilla dLLMs and 1.3x over Fast-dLLM with less than 1% accuracy change across reasoning and code benchmarks.

Related

• Factorization-Error-Free Discrete Diffusion Language Model via Speculative Decoding
• Don't Commit Alone: Joint Token Commitment in Diffusion Large Language Models
• Read the Trace, Steer the Path: Trajectory-Aware Reinforcement Learning for Diffusion Language Models
• Dependency-Aware Revocable Decoding for Efficient Diffusion Large Language Model Inference

Source: arXiv cs.CL | 2026-08-31