mlsanityData to 7 Oct 2026

Accelerating Speculative Decoding with Block Diffusion Draft Trees

arXiv 2604.12989 ↗First posted 14 Apr 2026cs.CLcited by 27

On the map: Efficient LLM inference and serving → Speculative decoding

Abstract (arXiv)

Speculative decoding accelerates autoregressive language models by using a lightweight drafter to propose multiple future tokens, which the target model then verifies in parallel. DFlash shows that a block diffusion drafter can generate an entire draft block in a single forward pass and achieve state-of-the-art speculative decoding performance, outperforming strong autoregressive drafters such as EAGLE-3. Vanilla DFlash, however, still verifies only a single drafted trajectory per round, potentially limiting its acceptance length. We introduce DDTree (Diffusion Draft Tree), a method that constructs a draft tree directly from the per-position distributions of a block diffusion drafter. Under a fixed node budget, DDTree uses a simple best-first heap algorithm to select the continuations that are most likely to match the target model according to a surrogate defined by the draft model's output. The resulting tree is verified efficiently in a single target model forward pass using an ancestor-only attention mask. Because DDTree builds on DFlash, a leading draft model for speculative decoding, these gains place DDTree among the leading approaches to speculative decoding.

Key facts. Cited by 27 papers in the map (data through 7 Oct 2026). Placed in Speculative decoding, one of 12 directions in Efficient LLM inference and serving. Builds on 23 of its 27 references that are in the map. First posted on arXiv 14 Apr 2026.

Builds on

23 of 27 references.

Closest ancestors · 14

Same direction, newest first

  1. 2026DFlash: Block Diffusion for Flash Speculative Decoding5 Feb
  2. 2026DART: Diffusion-Inspired Speculative Decoding for Fast LLM Inference27 Jan
  3. 2025PARD: Accelerating LLM Inference with Low-Cost PARallel Draft Model Adaptation23 Apr
  4. 2025EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test3 Mar
  5. 2024DySpec: Faster Speculative Decoding with Dynamic Token Tree Structure15 Oct
  6. 2024OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure25 Jun
  7. 2024EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees24 Jun
  8. 2024Recursive Speculative Decoding: Accelerating LLM Inference via Sampling Without Replacement21 Feb
  9. 2024EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty26 Jan
  10. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads19 Jan
  11. 2023Accelerating LLM Inference with Staged Speculative Decoding8 Aug
  12. 2023SpecInfer: Accelerating Generative Large Language Model Serving with Tree-based Speculative Inference and Verification16 May
2 more
  1. 2023Accelerating Large Language Model Decoding with Speculative Sampling2 Feb
  2. 2022Fast Inference from Transformers via Speculative Decoding30 Nov

Other references · 9

  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. 2023Let's Verify Step by Stepcited by 3,598
  5. 2021Program Synthesis with Large Language Modelscited by 3,509
  6. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,193
  7. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,551
  8. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
  9. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Modelscited by 385

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 27.

27 papers.

Sort:

By area:

  1. 2026JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Draftingcited by 12
  2. 2026xPress: Parallel Refinement for Diffusion Drafters in Speculative Decodingcited by 7
  3. 2026DARTree: Speculative Diffusion Decoding with Autoregressive Draft Treescited by 5
  4. 2026TreeFlash: Parallel AR-Approximation for Faster Speculative Decodingcited by 4
  5. 2026Teaching Diffusion to Speculate Left-to-Rightcited by 3
  6. 2026Trees from Marginals: Autoregressive drafting with factorized priorscited by 2
  7. 2026DeLS-Spec: Decoupled Long-Short Contexts for Parallel Speculative Draftingcited by 2
  8. 2026DominoTree: Conditional Tree-Structured Drafting with Domino for Speculative Decodingcited by 2
  9. 2026From Chains to Trees: Parent-Conditioned Drafting for Semi-Autoregressive Speculative Decodingcited by 2
  10. 2026LiLiCorr: Lightweight Likelihood Correlation of Parallel Drafts for Speculative Decodingcited by 2
  11. 2026Bastion: Budget-Aware Speculative Decoding with Tree-structured Block Diffusion Draftingcited by 1
  12. 2026D^2SD: Accelerating Speculative Decoding with Dual Diffusion Draft Modelscited by 1
15 more
  1. 2026Margins, Not Windows: Training-Free Per-Step Lossy Speculative Decodingcited by 1
  2. 2026Mean-Field Parallel Decoding for Discrete Diffusion Language Modelscited by 0
  3. 2026Fully Rotation-Equivariant Spectral-Spatial Learning for Multispectral Object Detectioncited by 0
  4. 2026LibraSpec: Dynamic Diffusion-Based Speculative Decoding via Marginal-Gain-Driven Optimizationcited by 0
  5. 2026Is Multimodal Speculative Decoding Ready for Diffusion-Based Parallel Drafting? A Survey and Empirical Diagnosiscited by 0
  6. 2026Ceiling-Clipped Acceptance Histograms Indicate Stranded Speed-up in Block-Diffusion Speculative Decodingcited by 0
  7. 2026Vision Is Not Overhead: One-Pass Block Drafting for Lossless Speculative Decoding in Vision-Language Modelscited by 0
  8. 2026Carryover Drafting: Recycling Rejected States for Speculative Decodingcited by 0
  9. 2026TreeSpark: Calibrated, Load-Adaptive Draft Trees for Semi-Autoregressive Speculative Decodingcited by 0
  10. 2026Tsubame: Tree Replay for Diffusion-Based Speculative Decodingcited by 0
  11. 2026Draft in Parallel, Condition Through Depth: Adjacent Causal Injection for Speculative Decodingcited by 0
  12. 2026LongSpark: Efficient speculative decoding with a fixed-cost parallel draftercited by 0
  13. 2026UBTree: Parallel Tree Drafting via Unigram and Bigram Models for Speculative Decodingcited by 0
  14. 2026DLoop: Looped Speculative Decodingcited by 0
  15. 2026Where Draft Trees Lose Target Mass: Exit-Guided Speculative Decodingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026PRESTO: Prefix-Aligned Tree Drafting for Diffusion Speculative Decoding
  2. 2026Cost-Aware Diffusion Draft Trees for Speculative Decoding
  3. 2026GRAFT: Adaptive DLM-Based Draft Tree Construction with Target-Distilled Edge Scoring
  4. 2026TAPS: Target-Aware Prefix Tree Selection for Diffusion-Drafted Speculative Decoding
  5. 2025DiffuSpec: Unlocking Diffusion Language Models for Speculative Decoding
  6. 2026D-Loop: Looped Diffusion Drafting for Speculative Decoding
  7. 2026AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters
  8. 2026Unlocking Parallelism in Autoregressive Language Models via Speculative Decoding with Progressive Tree Drafting