DIVISION: Memory Efficient Training via Dual Activation Precision

Guanchu Wang, Zirui Liu, Zhimeng Jiang, Ninghao Liu, Na Zou, Xia Hu

Keywords: DNN training, activation compressed training, memory efficient training, frequency domain

TL;DR: A simple and transparent framework to reduce the memory cost of DNN training.

Abstract: Activation compressed training (ACT) has been shown to be a promising way to reduce the memory cost of training deep neural networks (DNNs). However, existing work of ACT relies on searching for optimal bit-width during DNN training to reduce the quantization noise, which makes the procedure complicated and less transparent. To this end, we propose a simple and effective method to compress DNN training. Our method is motivated by an instructive observation: DNN backward propagation mainly utilizes the low-frequency component (LFC) of the activation maps, while the majority of memory is for caching the high-frequency component (HFC) during the training. This indicates the HFC of activation maps is highly redundant and compressible during DNN training, which inspires our proposed Dual Activation Precision (DIVISION). During the training, DIVISION preserves the high-precision copy of LFC and compresses the HFC into a light-weight copy with low numerical precision. This can significantly reduce the memory cost without negatively affecting the precision of backward propagation such that DIVISION maintains competitive model accuracy. Experimental results show DIVISION achieves over 10× compression of activation maps, and significantly higher training throughput than state-of-the-art ACT methods, without loss of model accuracy. The code is available at https://anonymous.4open.science/r/division-5CC0/

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Deep Learning and representational learning