mlsanityData to 14 Oct 2026

DC-Leap: Training-Free Acceleration of dLLMs via Draft-Guided Contiguous Leaping Decoding

arXiv 2607.20467 ↗First posted 19 May 2026cs.AI cs.LGcited by 0

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

Abstract (arXiv)

While parallel decoding is central to the efficiency of Diffusion Large Language Models (dLLMs), current strategies are often hindered by overly conservative confidence thresholds. These thresholds, necessitated by the Joint Probability Dependence Error (JPDE), result in redundant denoising iterations and suboptimal inference speeds. To overcome this, we propose DC-Leap, a training-free framework that enables reliable acceleration of dLLMs in the moderate-confidence regime. DC-Leap introduces a Dynamic Contiguous Verification strategy that integrates strictly-ordered causal constraints into the parallel decoding process. By progressively validating token dependencies, this mechanism effectively neutralizes the JPDE, enabling reliable acceleration with comparable performance. Furthermore, DC-Leap incorporates the draft-guided decoding mechanism, where the draft helps extend the context by leaping forward across multiple tokens, providing look-ahead context and retaining the structural benefits of bidirectional attention during inference. Extensive experiments on standard benchmarks demonstrate that DC-Leap achieves substantial speedups, up to 53.19x on MBPP for long-sequence generation, and up to 105.02x when combined with KV-Cache with comparable generation quality. Code is available at https://github.com/ffh-wyls/DC-Leap .

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Diffusion LLM and MoE inference, one of 12 directions in Efficient LLM inference and serving. Builds on 57 of its 66 references that are in the map. First posted on arXiv 19 May 2026.

Builds on

57 of 66 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025Accelerating Diffusion LLM Inference via Local Determinism Propagation8 Oct
  2. 2025dParallel: Learnable Parallel Decoding for dLLMs30 Sep
  3. 2025Learning to Parallel: Accelerating Diffusion Large Language Models via Learnable Parallel Decoding29 Sep
  4. 2025Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding28 May

Other references · 53

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  3. 2025Qwen3 Technical Reportcited by 8,820
  4. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  6. 2017On Calibration of Modern Neural Networkscited by 4,960
  7. 2021Program Synthesis with Large Language Modelscited by 3,637
  8. 2025Qwen3-VL Technical Reportcited by 2,793
  9. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  10. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,653
  11. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 1,599
  12. 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,298
38 more
  1. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  2. 2022Diffusion-LM Improves Controllable Text Generationcited by 1,027
  3. 2022Token Merging: Your ViT But Fastercited by 1,021
  4. 2025Large Language Diffusion Modelscited by 912
  5. 2021DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsificationcited by 864
  6. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Headscited by 843
  7. 2023Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generationcited by 819
  8. 2025Dream 7B: Diffusion Large Language Modelscited by 544
  9. 2022Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizationscited by 363
  10. 2024VisionZip: Longer is Better but Not Necessary in Vision Language Modelscited by 313
  11. 2025LLaDA 1.5: Variance-Reduced Preference Optimization for Large Language Diffusion Modelscited by 285
  12. 2024Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMscited by 273
  13. 2020Prior Guided Feature Enrichment Network for Few-Shot Segmentationcited by 271
  14. 2025Deep Think with Confidencecited by 205
  15. 2022SSD-LM: Semi-autoregressive Simplex-based Diffusion Language Model for Text Generation and Modular Controlcited by 194
  16. 2025dLLM-Cache: Accelerating Diffusion Large Language Models with Adaptive Cachingcited by 169
  17. 2025Accelerated Sampling from Masked Diffusion Models via Entropy Bounded Unmaskingcited by 131
  18. 2021Semi-supervised Semantic Segmentation with Directional Context-aware Consistencycited by 112
  19. 2021ResLT: Residual Learning for Long-tailed Recognitioncited by 96
  20. 2025Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Modelscited by 90
  21. 2021Guided Point Contrastive Learning for Semi-supervised Point Cloud Semantic Segmentationcited by 87
  22. 2023Towards Large-scale 3D Representation Learning with Multi-dataset Point Prompt Trainingcited by 86
  23. 2023Hierarchical Dense Correlation Distillation for Few-Shot Segmentationcited by 84
  24. 2024Explore the Potential of CLIP for Training-Free Open Vocabulary Semantic Segmentationcited by 61
  25. 2020Generalized Few-shot Semantic Segmentationcited by 59
  26. 2024Confidence Matters: Revisiting Intrinsic Self-Correction Capabilities of Large Language Modelscited by 47
  27. 2022Generalized Parametric Contrastive Learningcited by 45
  28. 2025Concerto: Joint 2D-3D Self-Supervised Learning Emerges Spatial Representationscited by 45
  29. 2024OA-CNNs: Omni-Adaptive Sparse CNNs for 3D Semantic Segmentationcited by 44
  30. 2025DPad: Efficient Diffusion Language Models with Suffix Dropoutcited by 39
  31. 2025DeCLIP: Decoupled Learning for Open-Vocabulary Dense Perceptioncited by 33
  32. 2021PFENet++: Boosting Few-shot Semantic Segmentation with the Noise-filtered Context-aware Prior Maskcited by 31
  33. 2024Unified Language-driven Zero-shot Domain Adaptationcited by 30
  34. 2024Scalable Language Model with Generalized Continual Learningcited by 28
  35. 2025Mitigating Object Hallucinations via Sentence-Level Early Interventioncited by 28
  36. 2024GroupContrast: Semantic-aware Self-supervised Representation Learning for 3D Understandingcited by 26
  37. 2023Learning Context-aware Classifier for Semantic Segmentationcited by 23
  38. 2023Boosting Few-shot 3D Point Cloud Segmentation via Query-Guided Enhancementcited by 19

50 most cited of 53 All 53 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026LEAP: Unlocking dLLM Parallelism via Lookahead Early-Convergence Token Detection
  2. 2026Dynamic-dLLM: Dynamic Cache-Budget and Adaptive Parallel Decoding for Training-Free Acceleration of Diffusion LLM
  3. 2026Early-Bird Decoding: Accelerating Diffusion LLMs with Learnable Block Sizes and Parallel Sampling
  4. 2025Free Draft-and-Verification: Toward Lossless Parallel Decoding for Diffusion Large Language Models
  5. 2026Streaming-dLLM: Accelerating Diffusion LLMs via Suffix Pruning and Dynamic Decoding
  6. 2026CForce: Boosting Parallel Decoding for dLLMs via Consistency Forcing
  7. 2026CD4LM: Consistency Distillation and aDaptive Decoding for Diffusion Language Models
  8. 2026Roll Out and Roll Back: Diffusion LLMs are Their Own Efficiency Teachers