mlsanityData to 7 Oct 2026

DFlash: Block Diffusion for Flash Speculative Decoding

arXiv 2602.06036 ↗First posted 5 Feb 2026cs.CLcited by 120

On the map: Diffusion models → Discrete and text diffusion

also in 1 more
Abstract (arXiv)

Autoregressive large language models (LLMs) deliver strong performance but require inherently sequential decoding, leading to high inference latency and poor GPU utilization. Speculative decoding mitigates this bottleneck by using a fast draft model whose outputs are verified in parallel by the target LLM; however, existing methods still rely on autoregressive drafting, which remains sequential and limits practical speedups. Diffusion LLMs offer a promising alternative by enabling parallel generation, but current diffusion models typically underperform compared with autoregressive models. In this paper, we introduce DFlash, a speculative decoding framework that employs a lightweight block diffusion model for parallel drafting. By generating draft tokens in a single forward pass and conditioning the draft model on context features extracted from the target model, DFlash enables efficient drafting with high-quality outputs and higher acceptance rates. Experiments show that DFlash achieves over 6x lossless acceleration across a range of models and tasks, delivering up to 2.5x higher speedup than the state-of-the-art speculative decoding method EAGLE-3.

Key facts. Cited by 120 papers in the map (data through 7 Oct 2026). Placed in Discrete and text diffusion, one of 14 directions in Diffusion models. Builds on 27 of its 30 references that are in the map. First posted on arXiv 5 Feb 2026.

Builds on

27 of 30 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025SpecDiff-2: Scaling Diffusion Drafter Alignment For Faster Speculative Decoding1 Nov
  2. 2025SDAR: A Synergistic Diffusion-AutoRegression Paradigm for Scalable Sequence Generation7 Oct
  3. 2025DiffuSpec: Unlocking Diffusion Language Models for Speculative Decoding28 Sep
  4. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models12 Mar

Other references · 23

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  3. 2021Evaluating Large Language Models Trained on Codecited by 8,053
  4. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  5. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
  6. 2023Let's Verify Step by Stepcited by 3,598
  7. 2021Program Synthesis with Large Language Modelscited by 3,509
  8. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
  9. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,615
  10. 2023LongBench: A Bilingual, Multitask Benchmark for Long Context Understandingcited by 1,378
  11. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  12. 2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,107
11 more
  1. 2023Enhancing Chat Language Models by Scaling High-quality Instructional Conversationscited by 825
  2. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Headscited by 821
  3. 2024EAGLE: Speculative Sampling Requires Rethinking Feature Uncertaintycited by 564
  4. 2024EAGLE-2: Faster Inference of Language Models with Dynamic Draft Treescited by 362
  5. 2025EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Testcited by 318
  6. 2024Flex Attention: A Programming Model for Generating Optimized Attention Kernelscited by 180
  7. 2025Fast-dLLM v2: Efficient Block-Diffusion LLMcited by 128
  8. 2024LongAlign: A Recipe for Long Context Alignment of Large Language Modelscited by 102
  9. 2024MagicDec: Breaking the Latency-Throughput Tradeoff for Long Context Generation with Speculative Decodingcited by 87
  10. 2025Your LLM Knows the Future: Uncovering Its Multi-Token Prediction Potentialcited by 46
  11. 2026d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillationcited by 34

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 120.

120 papers.

Sort:

By area:

  1. 2026Domino: Decoupling Causal Modeling from Autoregressive Drafting in Speculative Decodingcited by 31
  2. 2026Accelerating Speculative Decoding with Block Diffusion Draft Treescited by 27
  3. 2026D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Draftingcited by 16
  4. 2026DMax: Aggressive Parallel Decoding for dLLMscited by 14
  5. 2026HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Bettercited by 14
  6. 2026Speculative Speculative Decodingcited by 13
  7. 2026JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Draftingcited by 12
  8. 2026Why Diffusion Language Models Struggle with Truly Parallel (Non-Autoregressive) Decoding?cited by 11
  9. 2026Draft-OPD: On-Policy Distillation for Speculative Draft Modelscited by 10
  10. 2026DFlare: Scaling Up Draft Capacity for Block Diffusion Speculative Decodingcited by 9
  11. 2026Breaking Entropy Bounds: Accelerating RL Training via MTP with Rejection Samplingcited by 9
  12. 2026Faster-WAM: Do World Action Models Need Deep Action Modules?cited by 9
38 more
  1. 2026Fast-dVLM: Efficient Block-Diffusion VLM via Direct Conversion from Autoregressive VLMcited by 8
  2. 2026PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decodingcited by 7
  3. 2026TAPS: Target-Aware Prefix Tree Selection for Diffusion-Drafted Speculative Decodingcited by 7
  4. 2026xPress: Parallel Refinement for Diffusion Drafters in Speculative Decodingcited by 7
  5. 2026Draft Less, Retrieve More: Hybrid Tree Construction for Speculative Decodingcited by 5
  6. 2026DARTree: Speculative Diffusion Decoding with Autoregressive Draft Treescited by 5
  7. 2026Verification-Aware Training for Speculative Decodingcited by 5
  8. 2026When Hidden States Drift: Can KV Caches Rescue Long-Range Speculative Decoding?cited by 4
  9. 2026Attention Drift: What Autoregressive Speculative Decoding Models Learncited by 4
  10. 2026TreeFlash: Parallel AR-Approximation for Faster Speculative Decodingcited by 4
  11. 2026D-cut: Adaptive Verification Depth Pruning for Batched Speculative Decodingcited by 4
  12. 2026AngelSpec: Towards Real-World High Performance Inference with Speculative Decodingcited by 4
  13. 2026Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decodingcited by 3
  14. 2026Test-Time Speculationcited by 3
  15. 2026Orthrus: Memory-Efficient Parallel Token Generation via Dual-View Diffusioncited by 3
  16. 2026Cost-Aware Diffusion Draft Trees for Speculative Decodingcited by 3
  17. 2026Teaching Diffusion to Speculate Left-to-Rightcited by 3
  18. 2026Spec-AUF: Accept-Until-Fail Training under Train-Inference Misalignment for Masked Block Drafterscited by 3
  19. 2026Predict, Don't Iterate: Efficient Adaptive-Length Infilling for Diffusion Language Modelscited by 3
  20. 2026A Comparative analysis of Layer-wise Representational Capacity in AR and Diffusion LLMscited by 2
  21. 2026Future Validity is the Missing Statistic: From Impossibility to Phi-Estimation for Grammar-Faithful Speculative Decodingcited by 2
  22. 2026FlexDraft: Flexible Speculative Decoding via Attention Tuning and Bonus-Guided Calibrationcited by 2
  23. 2026SimSD: Simple Speculative Decoding in Diffusion Language Modelscited by 2
  24. 2026BlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decodingcited by 2
  25. 2026Trees from Marginals: Autoregressive drafting with factorized priorscited by 2
  26. 2026DeLS-Spec: Decoupled Long-Short Contexts for Parallel Speculative Draftingcited by 2
  27. 2026DominoTree: Conditional Tree-Structured Drafting with Domino for Speculative Decodingcited by 2
  28. 2026A Practical Investigation of Training-free Relaxed Speculative Decodingcited by 2
  29. 2026PRESTO: Prefix-Aligned Tree Drafting for Diffusion Speculative Decodingcited by 2
  30. 2026From Chains to Trees: Parent-Conditioned Drafting for Semi-Autoregressive Speculative Decodingcited by 2
  31. 2026LiLiCorr: Lightweight Likelihood Correlation of Parallel Drafts for Speculative Decodingcited by 2
  32. 2026TailSieve: Partial-Rollout-Guided Tail Routing for LLM Rolloutscited by 2
  33. 2026ReTrace: Rejected-Trajectory Conditioning for Speculative Decodingcited by 2
  34. 2026DFlow: Enabling Verifier Information Flow in Block Diffusion Speculative Decodingcited by 2
  35. 2026Online Draft Co-Training for Speculative Decoding in Large-Scale, Long-Context RL Post-Trainingcited by 2
  36. 2026LoopSpec: Pipelined Self-Speculative Decoding for Looped Transformerscited by 2
  37. 2026SMART: When is it Actually Worth Expanding a Speculative Tree?cited by 1
  38. 2026Bastion: Budget-Aware Speculative Decoding with Tree-structured Block Diffusion Draftingcited by 1

50 most cited of 120 All 120 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Speculative Diffusion Decoding: Accelerating Language Generation through Diffusion
  2. 2025Fail Fast, Win Big: Rethinking the Drafting Strategy in Speculative Decoding via Diffusion LLMs
  3. 2026AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters
  4. 2025Self Speculative Decoding for Diffusion Large Language Models
  5. 2026DART: Diffusion-Inspired Speculative Decoding for Fast LLM Inference
  6. 2024Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decoding
  7. 2026SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding
  8. 2025Accelerating Diffusion LLMs via Adaptive Parallel Decoding