Fast-dLLM++ 提出 Fréchet Profile Decoding，加速扩散大语言模型推理
Fast-dLLM++: Fr\'echet Profile Decoding for Faster Diffusion LLM Inference
AI 导读
Fast-dLLM++ 是一项免训练的解码扩展，提出 Fréchet profile decoding，从完整排序置信度分布而非单一最坏情况置信度中选择并行提交集合，是对 Fast-dLLM 同质高置信假设的异质置信度推广。
正文 · 原文
Abstract:Diffusion large language models promise parallel token generation, yet inference remains bottlenecked by deciding which masked tokens can be safely committed together. Fast-dLLM addressed this with KV caching and confidence-guided parallel decoding, but its decoding theory uses a homogeneous high-confidence assumption that effectively reduces each candidate set to its weakest selected token. We argue that this leaves speed on the table because real decoding steps exhibit heterogeneous confidence profiles. We propose Fast-dLLM++, a training-free extension that introduces Fréchet profile decoding: selecting parallel commit sets from the full sorted confidence profile rather than a single worst-case confidence. The resulting rule is a heterogeneous-confidence generalization of Fast-dLLM's factor selector and it recovers the previous rule exactly in the equal-confidence case and adds a provable heterogeneity bonus when the selected tokens have uneven confidences. Fast-dLLM++ leaves the model, diffusion process, and cache implementation entirely unchanged, making it a drop-in replacement for existing Fast-dLLM decoding. Experiments on GSM8K, MATH, HumanEval, and MBPP with the LLaDA-8B model show that the theoretical improvement translates directly into empirical gains: profile-aware selection improves the accuracy-throughput frontier by exploiting safe parallelism that weakest-token rules miss, achieving up to 37% higher throughput at comparable accuracy. Our code release is at this https URL.
| Comments: | Project Page: this https URL An earlier version was accepted at ICML 2026 SPIGM Workshop. This updated version includes substantially expanded experimental results and additional contributions, and has been accepted to the NeurIPS 2026 main track |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2606.02955 [cs.CL] |
| (or arXiv:2606.02955v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2606.02955 arXiv-issued DOI via DataCite |
Submission history
From: Yasong Dai [view email]
[v1]
Mon, 1 Jun 2026 23:18:59 UTC (150 KB)
[v2]
Mon, 15 Jun 2026 16:47:38 UTC (777 KB)
[v3]
Wed, 11 Nov 2026 17:40:04 UTC (900 KB)
来源：arXiv cs.AI · arxiv.org