arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2212.04780cs.LGcs.CV
收藏

Genie:为量化展示数据

Genie: Show Me the Data for Quantization

Samsung Research(三星研究院)

Yongkweon Jeon, Chungman Lee, Ho-young Kim

AI总结:

本文提出零样本量化框架Genie,通过合成适合量化的数据并结合训练后量化算法,在数小时内获得无需真实数据且性能媲美少样本量化的鲁棒量化模型。

摘要:

零样本量化是一种在数据因成本以及隐私相关问题等各种原因而无法获取时,开发轻量级深度神经网络的有前景的方法。通过利用FP32预训练模型中批归一化层的学习参数($μ$和$σ$),零样本量化方案专注于生成合成数据。随后,它们将知识从预训练模型(教师)蒸馏到量化模型(学生),使得量化模型能够利用合成数据集进行优化。然而,迄今为止,零样本量化主要是在量化感知训练方法的背景下讨论的,这些方法需要任务特定的损失和与重新训练一样多的长期优化。因此,我们引入了一种用于零样本量化的训练后量化方案,该方案能在几小时内生成高质量的量化网络。此外,我们提出了一个名为Genie的框架,用于生成适合量化的数据。利用Genie合成的数据,我们可以在没有真实数据集的情况下生成鲁棒的量化模型,其性能可与少样本量化相媲美。我们还提出了一种训练后量化算法来提升量化模型的性能。通过将它们结合起来,我们能够弥合零样本与少样本量化之间的差距,同时与现有方法相比显著提高量化性能。换句话说,我们可以获得一种独特的、最先进的零样本量化方法。代码可在\url{https://github.com/SamsungLabs/Genie}获取。

英文摘要:

Zero-shot quantization is a promising approach for developing lightweight deep neural networks when data is inaccessible owing to various reasons, including cost and issues related to privacy. By exploiting the learned parameters ($μ$ and $σ$) of batch normalization layers in an FP32-pre-trained model, zero-shot quantization schemes focus on generating synthetic data. Subsequently, they distill knowledge from the pre-trained model (teacher) to the quantized model (student) such that the quantized model can be optimized with the synthetic dataset. However, thus far, zero-shot quantization has primarily been discussed in the context of quantization-aware training methods, which require task-specific losses and long-term optimization as much as retraining. We thus introduce a post-training quantization scheme for zero-shot quantization that produces high-quality quantized networks within a few hours. Furthermore, we propose a framework called Genie~that generates data suited for quantization. With the data synthesized by Genie, we can produce robust quantized models without real datasets, which is comparable to few-shot quantization. We also propose a post-training quantization algorithm to enhance the performance of quantized models. By combining them, we can bridge the gap between zero-shot and few-shot quantization while significantly improving the quantization performance compared to that of existing approaches. In other words, we can obtain a unique state-of-the-art zero-shot quantization approach. The code is available at \url{https://github.com/SamsungLabs/Genie}.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑