SPOT通过提交顺序预测增强扩散语言模型的推测解码。
On the SPOT: Enhancing Speculative Decoding for Diffusion Language Models via Commit-Order Prediction
SPOT 通过预测 token 生成顺序并行验证候选，将扩散语言模型的遍历次数减少高达 6.8 倍，同时保持精确解码。该 0.62M 参数模型读取目标输出分布，预测顺序并提出候选树，目标模型通过状态等式在单次遍历中验证。SPOT 在 SDAR-8B 上将目标遍历次数减少至标准解码的 1/6.8，吞吐量提升 3.4 倍，并在所有任务和预算下超越校准无训练基线。
Published on Oct 26
Authors:
,
Abstract
SPOT accelerates diffusion language models by predicting token generation order and verifying candidates in parallel, reducing passes up to 6.8x while maintaining exact decoding.
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 over 90% of positions, but not which position comes next. We introduce Speculating Positions Order Trajectory (SPOT), a 0.62M-parameter model that reads only the target's output distributions, predicts this order and proposes a tree of candidate continuations, which the target verifies in a single pass by state equality. Coupled-noise match, which shares Gumbel noise between drafter and target, keeps this verification exact at any temperature. This approach also inspired us to introduce SPOT-Sets for dynamic threshold decoding, where a pass commits several positions: SPOT-Sets drafts whole commit sets together with their tokens. Our comprehensive experiments show that SPOT cuts target passes by up to 6.8times and reaches up to 3.4times the throughput of standard decoding on SDAR-8B, matching or beating the calibrated and training-free drafting baselines in every task and budget at T=0 and T=1. In addition, we show that SPOT's learned drafter transfers across models, demonstrating successful transfer across six different DLMs; with a light calibration, the SPOT-Sets drafter transfers as well. On top of dynamic threshold decoding, SPOT-Sets improves the throughput of SDAR-8B by 59% on average over the threshold decoder alone. Project page: https://eliezer318.github.io/SPOT-dLLM/
View arXiv page View PDF Project page GitHub 7 Add to collection
Get this paper in your agent:
hf papers read 2610.22543
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash
Models citing this paper 1
Other • Updated 1 day ago • 1EliezerS/SPOT-dLLM
Datasets citing this paper 0
No dataset linking this paper
Cite arxiv.org/abs/2610.22543 in a dataset README.md to link it from this page.
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2610.22543 in a Space README.md to link it from this page.
Collections including this paper 1
来源: Hugging Face Daily Papers · huggingface.co