mlsanityData to 14 Oct 2026

AsyncLane: Decoupling Refinement from Advancement in Diffusion Language Model Decoding

arXiv 2606.08411 ↗First posted 7 Jun 2026cs.CLcited by 0

On the map: Diffusion models → Discrete and text diffusion

Abstract (arXiv)

Block-wise semi-autoregressive decoding is the standard inference paradigm for diffusion large language models (DLMs), but it imposes a strict dependency between blocks: the next block cannot begin until the current block is fully decoded or its denoising budget is exhausted. We observe that once a block exposes a reliable delimiter boundary or stable semantic prefix, continuation generation need not wait for every residual token to be resolved. We propose AsyncLane, a training-free decoding scheduler that decouples refinement from advancement. AsyncLane forks a generate lane at observed delimiter boundaries into a refine lane and a continuation generate lane: the prefix remains editable, while the continuation advances before prefix refinement finishes. The resulting lane tree records decoding dependencies and output order, while execution proceeds over the active lane set. To make this asynchronous schedule efficient under bidirectional attention, AsyncLane combines shared-prefix lane batching, lookahead draft reuse, cascading termination, and compact cache refresh with refresh-logit reuse, preventing model-call cost from scaling directly with the number of lanes. AsyncLane is a drop-in replacement for block-wise DLM samplers and requires no retraining. Experiments on mathematical reasoning and code generation show that AsyncLane consistently improves throughput while maintaining competitive quality. Across LLaDA and Dream backbones, AsyncLane achieves the highest TPS in all evaluated benchmark-length settings; relative to the fastest competing baseline, it reaches peak speedups of 2.95x on LLaDA and 3.04x on Dream, with especially large gains under longer generation budgets.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Discrete and text diffusion, one of 14 directions in Diffusion models. Builds on 18 of its 18 references that are in the map. First posted on arXiv 7 Jun 2026.

Builds on

18 of 18 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025Dream 7B: Diffusion Large Language Models21 Aug
  2. 2025FlashDLM: Accelerating Diffusion Language Model Inference via Efficient KV Caching and Guided Diffusion27 May
  3. 2025dKV-Cache: The Cache for Diffusion Language Models21 May
  4. 2025Large Language Diffusion Models14 Feb
  5. 2024Simple and Effective Masked Diffusion Language Models11 Jun
  6. 2023Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution25 Oct

Other references · 12

  1. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  3. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  4. 2021Program Synthesis with Large Language Modelscited by 3,637
  5. 2022Diffusion-LM Improves Controllable Text Generationcited by 1,027
  6. 2025Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decodingcited by 420
  7. 2025Accelerating Diffusion LLMs via Adaptive Parallel Decodingcited by 112
  8. 2026d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillationcited by 37
  9. 2025Attention Is All You Need for KV Cache in Diffusion LLMscited by 32
  10. 2025AdaBlock-dLLM: Semantic-Aware Diffusion LLM Inference via Adaptive Block Sizecited by 31
  11. 2025Learning to Parallel: Accelerating Diffusion Large Language Models via Learnable Parallel Decodingcited by 27
  12. 2025d2Cache: Accelerating Diffusion-Based LLMs via Dual Adaptive Cachingcited by 25

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Roll Out and Roll Back: Diffusion LLMs are Their Own Efficiency Teachers
  2. 2026CAI-DLLM: Convergence Aware Inference for Diffusion Language Models
  3. 2026CD4LM: Consistency Distillation and aDaptive Decoding for Diffusion Language Models
  4. 2026S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculation
  5. 2026Speculative Refinement: A Hybrid Autoregressive Diffusion Decoding Strategy and Its Behavior Across Benchmarks
  6. 2026Window-Diffusion: Accelerating Diffusion Language Model Inference with Windowed Token Pruning and Caching
  7. 2026SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language Models
  8. 2026BlockBatch: Multi-Scale Consensus Decoding for Efficient Diffusion Language Model Inference