Aug 2026· 2 citations· ⚡ 1 influential· 32 references
Computer Science
TL;DR
TreeSpark is introduced, which reads a parent-conditioned distribution from the drafter's existing Markov head at negligible cost, calibrates it into an edge-acceptance estimate, and lets path survival govern everything else: best-first expansion, per-round stopping, and a load-adaptive serving policy.
Abstract
Speculative decoding accelerates language-model inference by letting a cheap drafter propose tokens that the target model verifies in parallel. Recent block drafters make drafting nearly free: a single backbone pass emits an entire block of draft tokens. Draft trees promise a further gain -- several alternative continuations verified in one target forward -- but existing constructions rank candidates by per-position marginals that ignore which parent a candidate extends, so on semi-autoregressive drafters wider trees mostly add mis-ranked nodes; and a tree of fixed size ignores how much speculation each decoding round, and each serving load, can support. We introduce TreeSpark, which reads a parent-conditioned distribution from the drafter's existing Markov head at negligible cost, calibrates it into an edge-acceptance estimate, and lets path survival govern everything else: best-first expansion, per-round stopping, and a load-adaptive serving policy. Sampling siblings without replacement, with matching residuals in recursive rejection, keeps decoding lossless at any temperature. Adaptive trees improve on matched fixed budgets at every temperature; against a tuned chain on the same drafter, TreeSpark accepts 15-25% more draft tokens per round and decodes 8-14% faster in single-request wall-clock, and under rising load it gracefully shrinks the tree back to the chain. Code and artifacts: https://github.com/PopSoda2002/TreeSpark
DARTree is introduced, a training-free speculative decoding method that extends a pretrained AR correction head from chains to trees, and achieves the highest average acceptance length and speedup in all four model--temperature configurations.
A theoretical framework for training and evaluating parallel and semi-autoregressive drafters by representing speculative decoding as a Markov reward process, which yields the Expected Decoding Rounds (EDR) objective, which weights local rejection costs by state occupancies and exactly equals the expected number of dec...
This work introduces LongSpark, a block-diffusion drafter that achieves state-of-the-art end-to-end efficiency across multiple model scales and realistic serving conditions by extracting fixed-size, multiscale views from the target's verification pass, thereby eliminating the need for a growing persistent state.
Hao-Yuan He, Peng-Fei Liu, Si-Shi Shen et al.· 0 citations
ResiSpec, a framework that strategically reforms the proposal distribution during verification to anchor the residual target mass within the draft model's high-confidence regions, prevents candidate obsolescence and achieves up to 1.92$\times$ speedup over state-of-the-art multi-candidate methods.
Zhi-Kai Chen, Jun Tao, Wei-Hao Mao et al.· 1 citation
ReTrace is introduced, a rejected-trajectory conditioning method that conditions each draft block on the rejected suffix from the previous round rather than generating it from fresh mask placeholders alone, indicating that the draft model can retain useful semantic and structural information despite local token-level e...
Luxi Lin, Zhan-Peng Zeng, Shuang Peng et al.· 2 citations
D-Loop is proposed, which introduces \emph{intra-block causal conditioning} within the original diffusion drafter without additional model components, and can beat DFlash and DSpark on Qwen3-4B and Qwen3-8B with obvious gains.
Ke-Cheng Chen, Yu-Yang He, Cheng Gong et al.· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduNov 19, 2026
An MIT and Dana-Farber Institute-led team used large language models to examine nearly 4,000 studies and hundreds of code repositories, revealing gaps in the information needed to scrutinize and reuse published research.
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.