返回首页
arXiv AI··论文与技术

Grab a Coffee: Future-Aware Guidance for Discrete Diffusion with Compiled Objectives

中文摘要

这项研究提出“Grab a Coffee”框架,通过编译目标函数优化离散扩散模型的引导过程,高效处理序列级生成任务,有效解决了传统引导计算复杂性指数级增长的问题。

English Summary

"Grab a Coffee" introduces a framework using compiled objectives to guide discrete diffusion models, efficiently addressing the exponential computational complexity of sequence-level generation in parallel token resolution.

原文节选

arXiv:2609.35924v1 Announce Type: new Abstract: Discrete diffusion models generate sequences by iteratively resolving multiple tokens in parallel, offering a flexible alternative to left-to-right generation. However, guiding this process with a sequence-level objective is difficult because the value of one unresolved token depends on the other tokens with which it can form a high-reward sequence. Enumerating all such completions makes the whole guidance computation grow exponentially with the number of unresolved positions. We introduce COFFEE, a plug-and-play framework that avoids this enumeration by separating sequence dependence from the objective. At each diffusion step, a target-free carrier absorbs the marginal token distributions predicted by the denoiser to construct a joint model over the unresolved tokens, while a compiled finite-state model records how their combinations affect the sequence-level preference. Pairing their states allows COFFEE to transfer global preferences to unresolved positions and sample a clean reconstruction without retraining the diffusion model. The same framework supports explicit hard constraints and learned soft objectives. We evaluate COFFEE a…