DominoTree: Conditional Tree-Structured Drafting with Domino for Speculative Decoding

• URL: http://arxiv.org/abs/2607.08642v2
• Date: Fri, 10 Jul 2026 04:07:45 GMT
• Title: DominoTree: Conditional Tree-Structured Drafting with Domino for Speculative Decoding
• Abstract summary: We introduce DominoTree, a training-free best-first draft tree scored by Domino's conditional, non-factorized correction along each root-to-node path.<n>DominoTree reaches up to 6.6x speedup over greedy decoding and the highest mean accepted length of any evaluated method.
• Score: 6.82231493613379
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Speculative decoding accelerates LLM inference by drafting several tokens and verifying them in parallel. Block-diffusion drafters such as DFlash produce a draft block in one pass but model only per-position marginals, and best-first tree methods such as DDTree expand candidate trees from those marginals. The released Domino drafter adds a GRU-based causal correction that makes each draft token distribution path-dependent, a structure DDTree's factorized formulation cannot represent. We introduce DominoTree, a training-free best-first draft tree scored by Domino's conditional, non-factorized correction along each root-to-node path, made practical by restricting the per-node correction to a candidate top-M set. On Qwen3-4B across eight benchmarks, DominoTree reaches up to 6.6x speedup over autoregressive decoding and the highest mean accepted length of any evaluated method, up to 10.7 tokens per round, at every tested temperature. DominoTree constructs its tree with a GPU-native CUDA-graph builder that is bit-identical to a reference Python implementation, so acceptance is unchanged, while keeping per-round tree construction cheap. With this builder as default, DominoTree improves throughput over the released Domino decoder, the drafter it builds on, at every tested temperature: 9% to 10% overall on Qwen3-4B and up to 22% on Alpaca. It also outperforms DDTree and CaDDTree at every tested temperature, not only under greedy decoding. On Qwen3-8B, DominoTree keeps the highest accepted length at every temperature and gives a 24% throughput gain over DDTree at T=0; at higher temperature its edge over DDTree and CaDDTree narrows to a tie and a small loss, while its aggregate gains over DFlash and Domino persist.

Related papers

• DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees [40.61855455828842]
  Speculative decoding losslessly accelerates autoregressive language models by verifying multiple draft tokens in parallel.<n>Dartree is a training-free speculative decoding method that extends a pretrained AR correction head from chains to trees.
  arXiv  Detail & Related papers  (2026-08-13T17:43:44Z)
• Cost-Aware Diffusion Draft Trees for Speculative Decoding [17.900184934329666]
  We introduce textbfCaDDTree (Cost-aware Diffusion Draft Tree), a method that directly optimize token throughput (expected tokens generated per unit time) by jointly selecting the tree structure and node budget.<n>Experiments on Qwen3-4B and Qwen3-8B across eight benchmarks spanning reasoning, coding, and instruction-following tasks show that caDDTree matches or surpasses DDTree with oracle budget selection on nearly all tasks.
  arXiv  Detail & Related papers  (2026-06-01T07:29:56Z)
• TreeGrad-Ranker: Feature Ranking via $O(L)$-Time Gradients for Decision Trees [73.0940890296463]
  probabilistic values are used to rank features for explaining local predicted values of decision trees.<n>TreeGrad computes the gradients of the multilinear extension of the joint objective in $O(L)$ time for decision trees with $L$ leaves.<n>TreeGrad-Ranker aggregates the gradients while optimizing the joint objective to produce feature rankings.<n>TreeGrad-Shap is a numerically stable algorithm for computing Beta Shapley values with integral parameters.
  arXiv  Detail & Related papers  (2026-02-12T06:17:12Z)
• Fast Inference of Visual Autoregressive Model with Adjacency-Adaptive Dynamical Draft Trees [50.230925890958936]
  We propose an adjacency-adaptive dynamic draft tree that adjusts draft tree depth and width by leveraging adjacent token states and prior acceptance rates.<n>ADT-Tree achieves speedups of 3.13xand 3.05x, respectively, and integrates seamlessly with relaxed sampling methods such as LANTERN.
  arXiv  Detail & Related papers  (2025-12-26T04:45:49Z)
• Designing ReLU Generative Networks to Enumerate Trees with a Given Tree Edit Distance [3.746431245783676]
  We establish the existence and construction of generative networks capable of producing trees similar to a given tree with respect to the tree edit distance.<n>The networks successfully generated all valid trees within the specified tree edit distance.<n>These findings provide a theoretical foundation towards construction of compact generative models and open new directions for exact and valid tree-structured data generation.
  arXiv  Detail & Related papers  (2025-10-12T17:07:49Z)
• OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure [40.9990864658776]
  Speculative decoding employs a "draft and then verify" mechanism to allow multiple tokens to be generated in one step.<n>Existing methods mainly adopt fixed draft structures, which fail to adapt to different situations.<n>We propose OPT-Tree, an algorithm to construct adaptive and scalable draft trees.
  arXiv  Detail & Related papers  (2024-06-25T04:45:53Z)
• Learning a Decision Tree Algorithm with Transformers [75.96920867382859]
  We introduce MetaTree, a transformer-based model trained via meta-learning to directly produce strong decision trees. We fit both greedy decision trees and globally optimized decision trees on a large number of datasets, and train MetaTree to produce only the trees that achieve strong generalization performance.
  arXiv  Detail & Related papers  (2024-02-06T07:40:53Z)
• MAPTree: Beating "Optimal" Decision Trees with Bayesian Decision Trees [2.421336072915701]
  We present a Bayesian approach to decision tree induction via maximum a posteriori inference of a posterior distribution over trees. We propose an AND/OR search algorithm, dubbed MAPTree, which is able to recover the maximum a posteriori tree.
  arXiv  Detail & Related papers  (2023-09-26T23:43:37Z)
• Structure-Unified M-Tree Coding Solver for MathWord Problem [57.825176412485504]
  In previous work, models designed by taking into account the properties of the binary tree structure of mathematical expressions at the output side have achieved better performance. In this paper, we propose the Structure-Unified M-Tree Coding Coding (S-UMCr), which applies a tree with any M branches (M-tree) to unify the output structures. Experimental results on the widely used MAWPS and Math23K datasets have demonstrated that SUMC-r not only outperforms several state-of-the-art models but also performs much better under low-resource conditions.
  arXiv  Detail & Related papers  (2022-10-22T12:20:36Z)
• SoftTreeMax: Policy Gradient with Tree Search [72.9513807133171]
  We introduce SoftTreeMax, the first approach that integrates tree-search into policy gradient. On Atari, SoftTreeMax demonstrates up to 5x better performance in faster run-time compared with distributed PPO.
  arXiv  Detail & Related papers  (2022-09-28T09:55:47Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.