Metraipolis › archive › dc_8a86ee737ead

On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction — arXiv:2610.22543

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.22543
License (as archived): CC0-1.0
Archived: 2026-10-19T08:48:20.479Z
SHA-256 of the archived text: 431e6b9a10a8e0481acfea3eb43f85c94a4941608007e9a2acaed0e6fb9c6a0b
Stable URL of this page: https://metraipolis.com/documento/dc_8a86ee737ead

How to cite

On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction — arXiv:2610.22543 — arXiv (en), https://arxiv.org/abs/2610.22543 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8a86ee737ead (Metraipolis archive dc_8a86ee737ead, sha256 431e6b9a10a8e0481acfea3eb43f85c94a4941608007e9a2acaed0e6fb9c6a0b)

BibTeX

@misc{dc8a86ee737ead,
  title = {On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction — arXiv:2610.22543},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.22543},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8a86ee737ead; sha256 431e6b9a10a8e0481acfea3eb43f85c94a4941608007e9a2acaed0e6fb9c6a0b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.22543. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Diffusion language models (DLMs) promise parallel text generation, but in practice face an unresolved speed-quality trade-off. Exact decoding fills in only one masked position per model pass, while faster multi-token decoding schemes are lossy. Speculative decoding is the standard lossless acceleration for autoregressive models. However, it does not transfer directly to DLMs, which must choose both what to generate and where. We show that this order is the missing piece: the model's own predictions from the previous pass already give the right token at 92-100% of positions, but not which position comes next. Based on this observation, we introduce Speculating Positions Order Trajectory (SPOT), which learns to predict this order. A 0.62M-parameter model reads only the target's output distributions and proposes a tree of candidate continuations, and the target verifies all of them in a single pass by checking whether each equals its own next state. On the verification side, our analysis shows that prior block-diffusion verification approaches either assume $T=0$ or a sequential proposal interface. To support tree-based speculation at any temperature, we propose using coupled-noise match verification for diffusion blocks, following shared-Gumbel-noise coupling approaches. Our comprehensive experiments show that SPOT cuts target passes by up to 6.8x and reaches up to 3.2x the throughput of standard decoding, beating the calibrated and training-free drafting baselines in every task and budget at $T=0$ and $T=1$. In addition, we show that the learned drafter transfers across models, demonstrating successful transfer across six different DLMs. Project page: this https URL

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.22543 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T08:42:40.213Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).