mlsanityData to 7 Oct 2026

Accelerating Diffusion LLMs via Adaptive Parallel Decoding

arXiv 2506.00413 ↗First posted 31 May 2025cs.CL cs.AI cs.LGcited by 108

On the map: Efficient LLM inference and serving → Diffusion LLM and MoE inference

Abstract (arXiv)

The generation speed of LLMs are bottlenecked by autoregressive decoding, where tokens are predicted sequentially one by one. Alternatively, diffusion large language models (dLLMs) theoretically allow for parallel token generation, but in practice struggle to achieve the speed of autoregressive models without significantly sacrificing quality. We therefore introduce adaptive parallel decoding (APD), a novel method that dynamically adjusts the number of tokens sampled in parallel. We achieve this by defining a multiplicative mixture between the dLLM marginal probabilities and the joint probability of sequences under a small auxiliary autoregressive model. This inverts the standard setup of speculative decoding, where the goal is to sample from a large autoregressive verifier by drafting from a smaller model. We further optimize APD by enabling KV caching and limiting the size of the masked input. Altogether, our method puts forward three tunable parameters to flexibly tradeoff throughput and quality. We show that APD provides markedly higher throughput with minimal quality degradations on downstream benchmarks.

Key facts. Cited by 108 papers in the map (data through 7 Oct 2026). Placed in Diffusion LLM and MoE inference, one of 12 directions in Efficient LLM inference and serving. Builds on 28 of its 40 references that are in the map. First posted on arXiv 31 May 2025.

Builds on

28 of 40 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Reviving Any-Subset Autoregressive Models with Principled Parallel Sampling and Speculative Decoding29 Apr

Other references · 27

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  4. 2020Scaling Laws for Neural Language Modelscited by 6,411
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2024Qwen2.5 Technical Reportcited by 4,713
  7. 2016Categorical Reparameterization with Gumbel-Softmaxcited by 2,708
  8. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,928
  9. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,615
  10. 2023Accelerating Large Language Model Decoding with Speculative Samplingcited by 975
  11. 2025Large Language Diffusion Modelscited by 872
  12. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Headscited by 821
15 more
  1. 2024Simple and Effective Masked Diffusion Language Modelscited by 800
  2. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Modelscited by 385
  3. 2024Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Datacited by 371
  4. 2024Break the Sequential Dependency of LLM Inference Using Lookahead Decodingcited by 295
  5. 2022The case for 4-bit precision: k-bit Inference Scaling Lawscited by 247
  6. 2025Train for the Worst, Plan for the Best: Understanding Token Ordering in Masked Diffusionscited by 208
  7. 2024A Survey on Efficient Inference for Large Language Modelscited by 197
  8. 2025Remasking Discrete Diffusion Models with Inference-Time Scalingcited by 181
  9. 2024Speculative Diffusion Decoding: Accelerating Language Generation through Diffusioncited by 57
  10. 2024Discrete Copula Diffusioncited by 53
  11. 2018Semi-Autoregressive Neural Machine Translationcited by 48
  12. 2024Faster Cascades via Speculative Decodingcited by 38
  13. 2024Parallel Sampling via Countingcited by 7
  14. 2025Enabling Autoregressive Models to Fill In Masked Tokenscited by 7
  15. 2024DynaMo: Accelerating Language Model Inference with Dynamic Multi-Token Samplingcited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 37; 2026: 71.

108 papers.

Sort:

By area:

  1. 2025Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcingcited by 111
  2. 2025A Survey on Diffusion Language Modelscited by 75
  3. 2025Diffusion Language Models Know the Answer Before Decodingcited by 68
  4. 2025dParallel: Learnable Parallel Decoding for dLLMscited by 59
  5. 2025ParallelBench: Understanding the Trade-offs of Parallel Decoding in Diffusion LLMscited by 42
  6. 2025TiDAR: Think in Diffusion, Talk in Autoregressioncited by 35
  7. 2025Beyond Fixed: Training-Free Variable-Length Denoising for Diffusion Large Language Modelscited by 32
  8. 2025AdaBlock-dLLM: Semantic-Aware Diffusion LLM Inference via Adaptive Block Sizecited by 30
  9. 2025Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speedcited by 30
  10. 2025Learning Unmasking Policies for Diffusion Language Modelscited by 29
  11. 2025A Survey on Parallel Text Generation: From Parallel Decoding to Diffusion Language Modelscited by 27
  12. 2025Learning to Parallel: Accelerating Diffusion Large Language Models via Learnable Parallel Decodingcited by 26
38 more
  1. 2025LaDiR: Latent Diffusion Enhances LLMs for Text Reasoningcited by 25
  2. 2025Set Block Decoding is a Language Model Inference Acceleratorcited by 22
  3. 2025Structuring The Future: Diffusion LLM Speculative Decoding via Calibrated Draft Graphscited by 21
  4. 2025Empirical Analysis of Decoding Biases in Masked Diffusion Modelscited by 19
  5. 2025Free Draft-and-Verification: Toward Lossless Parallel Decoding for Diffusion Large Language Modelscited by 19
  6. 2026Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decodingcited by 19
  7. 2025LoPA: Scaling dLLM Inference via Lookahead Parallel Decodingcited by 17
  8. 2025Inpainting-Guided Policy Optimization for Diffusion Large Language Modelscited by 16
  9. 2025ReFusion: A Diffusion Large Language Model with Parallel Autoregressive Decodingcited by 16
  10. 2025Unveiling the Potential of Diffusion Large Language Model in Controllable Generationcited by 14
  11. 2025A Survey on Cache Methods in Diffusion Models: Toward Efficient Multi-Modal Generationcited by 14
  12. 2026DMax: Aggressive Parallel Decoding for dLLMscited by 14
  13. 2026DAPD: Dependency-Aware Parallel Decoding via Attention for Diffusion LLMscited by 13
  14. 2025DiffuGuard: How Intrinsic Safety is Lost and Found in Diffusion Large Language Modelscited by 11
  15. 2025Corrective Diffusion Language Modelscited by 11
  16. 2026Why Diffusion Language Models Struggle with Truly Parallel (Non-Autoregressive) Decoding?cited by 11
  17. 2026Breaking the Factorization Barrier in Diffusion Language Modelscited by 11
  18. 2025dUltra: Ultra-Fast Diffusion Language Models via Reinforcement Learningcited by 10
  19. 2025Blockwise SFT for Diffusion Language Models: Reconciling Bidirectional Attention and Autoregressive Decodingcited by 9
  20. 2025Enabling Approximate Joint Sampling in Diffusion LMscited by 9
  21. 2025A2D: Any-Order, Any-Step Safety Alignment for Diffusion Language Modelscited by 9
  22. 2025Latent Refinement Decoding: Enhancing Diffusion-Based Language Models by Refining Belief Statescited by 9
  23. 2025Planned Diffusioncited by 8
  24. 2026CD4LM: Consistency Distillation and aDaptive Decoding for Diffusion Language Modelscited by 8
  25. 2026Dependency-Guided Parallel Decoding in Discrete Diffusion Language Modelscited by 8
  26. 2025Masks Can Be Distracting: On Context Comprehension in Diffusion Language Modelscited by 7
  27. 2026The Bitter Lesson of Diffusion Language Models for Agentic Workflows: A Comprehensive Reality Checkcited by 6
  28. 2026Stop the Flip-Flop: Context-Preserving Verification for Fast Revocable Diffusion Decodingcited by 6
  29. 2025Finish First, Perfect Later: Test-Time Token-Level Cross-Validation for Diffusion Large Language Modelscited by 5
  30. 2025Guided Star-Shaped Masked Diffusioncited by 5
  31. 2025A Survey on Collaborating Small and Large Language Models for Performance, Cost-effectiveness, Cloud-edge Privacy, and Trustworthinesscited by 5
  32. 2025Taming the Memory Footprint Crisis: System Design for Production Diffusion LLM Servingcited by 5
  33. 2026IDLM: Inverse-distilled Diffusion Language Modelscited by 5
  34. 2025DiffuMamba: High-Throughput Diffusion LMs with Mamba Backbonecited by 4
  35. 2026Rejection Mixing: Fast Semantic Propagation of Mask Tokens for Efficient DLLM Inferencecited by 4
  36. 2026Early Decisions Matter: Proximity Bias and Initial Trajectory Shaping in Non-Autoregressive Diffusion Language Modelscited by 4
  37. 2026Seeing the End at Step Zero: Accelerating Diffusion MLLMs via MLP Sparsity-Aware Truncationcited by 4
  38. 2026Safer by Diffusion, Broken by Context: Diffusion LLM's Safety Blessing and Its Failure Modecited by 3

50 most cited of 108 All 108 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding
  2. 2026SimSD: Simple Speculative Decoding in Diffusion Language Models
  3. 2025Accelerating Diffusion LLM Inference via Local Determinism Propagation
  4. 2026Unlocking Lossless Speedups in LLMs via Discrete Diffusion
  5. 2025Self Speculative Decoding for Diffusion Large Language Models
  6. 2025Fail Fast, Win Big: Rethinking the Drafting Strategy in Speculative Decoding via Diffusion LLMs
  7. 2026PSD: Pushing the Pareto Frontier of Diffusion LLMs via Parallel Speculative Decoding
  8. 2026Dynamic-dLLM: Dynamic Cache-Budget and Adaptive Parallel Decoding for Training-Free Acceleration of Diffusion LLM