HuggingFace Daily Papers·· 2 天前
T²Quant：面向扩散语言模型的任务与时间感知混合精度量化
Different Tasks, Different Timesteps: Temporal Task-Aware Quantization for Diffusion Language Models
AI 导读
T²Quant 是一种面向扩散语言模型（DLM）的任务与时间感知混合精度量化框架，利用无标签敏感度信号，在固定比特预算下按去噪步分配激活精度。研究显示量化敏感度同时依赖任务和去噪时间步，该方法在 3 个 DLM 和 6 个 benchmark 上稳定提升现有量化方法平均准确率，同等预算下最高提升 6.7 个百分点。
正文 · 原文
Abstract:Diffusion Language Models (DLMs) introduce unique quantization challenges due to their iterative denoising process. In this work, we reveal that quantization sensitivity is both taskand time-dependent, with different downstream tasks exhibiting distinct precision requirements across denoising steps. We further show that these patterns can be predicted without groundtruth labels or task-specific evaluation metrics. Building on these findings, we introduce T 2Quant, a task and time-aware mixedprecision quantization framework that uses a label-free sensitivity signal to optimally allocate activation precision across denoising steps under a fixed bit-width budget. Extensive experiments across three DLMs and six benchmarks demonstrate that T 2Quant consistently improves the average accuracy of existing quantization methods, achieving gains of up to 6.7 percentage points under the same budget.
| Comments: | preprint |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.31198 [cs.AI] |
| (or arXiv:2610.31198v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.31198 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Eliron Rahimi [view email]
[v1]
Thu, 22 Oct 2026 12:45:17 UTC (194 KB)
来源：HuggingFace Daily Papers · arxiv.org