Trees from Marginals: Autoregressive drafting with factorized priors

arXiv cs.LG Papers

Summary

This paper introduces Weaver, a lightweight autoregressive adapter that constructs proposal trees from factorized drafters' top-K marginals, achieving a 4.37x speedup over autoregressive decoding and outperforming the DFlash baseline by 24.7%.

arXiv:2607.06763v1 Announce Type: new Abstract: Speculative decoding greatly increases the interactivity of autoregressive language models by trading off computation for extra tokens generated in a single forward pass. Factorized draft models are especially efficient because they predict future-token marginals in parallel, but their independence assumption causes acceptance rates to degrade sharply as the speculative budget grows. We analyze this limitation and introduce Weaver, a lightweight autoregressive adapter that constructs proposal trees from the top-K marginals of a factorized drafter. Weaver restores conditional dependencies between proposed tokens while avoiding a full-vocabulary projection. To support fast verification for models with Gated Delta Net layers, we derive a rollback-free tree-verification algorithm and implement optimized CUDA kernels in SGLang. By combining these model and systems contributions we achieve a 4.37-fold speedup over autoregressive decoding, and outperform a highly optimized DFlash baseline by 24.7%.
Original Article
View Cached Full Text

Cached at: 07/09/26, 07:43 AM

# Trees from Marginals: Autoregressive drafting with factorized priors
Source: [https://arxiv.org/abs/2607.06763](https://arxiv.org/abs/2607.06763)
[View PDF](https://arxiv.org/pdf/2607.06763)

> Abstract:Speculative decoding greatly increases the interactivity of autoregressive language models by trading off computation for extra tokens generated in a single forward pass\. Factorized draft models are especially efficient because they predict future\-token marginals in parallel, but their independence assumption causes acceptance rates to degrade sharply as the speculative budget grows\. We analyze this limitation and introduce Weaver, a lightweight autoregressive adapter that constructs proposal trees from the top\-K marginals of a factorized drafter\. Weaver restores conditional dependencies between proposed tokens while avoiding a full\-vocabulary projection\. To support fast verification for models with Gated Delta Net layers, we derive a rollback\-free tree\-verification algorithm and implement optimized CUDA kernels in SGLang\. By combining these model and systems contributions we achieve a 4\.37\-fold speedup over autoregressive decoding, and outperform a highly optimized DFlash baseline by 24\.7%\.

## Submission history

From: Roman Knyazhitskiy \[[view email](https://arxiv.org/show-email/ece9bcdf/2607.06763)\] **\[v1\]**Tue, 7 Jul 2026 19:48:36 UTC \(956 KB\)

Similar Articles