Post-training Quantization with Multiple Points: Mixed Precision without Mixed Precision

Authors

  • Xingchao Liu University of Texas at Austin
  • Mao Ye University of Texas at Austin
  • Dengyong Zhou Google Brain
  • Qiang Liu University of Texas at Austin

Keywords:

Learning on the Edge & Model Compression

Abstract

We consider the post-training quantization problem, which discretizes the weights of pre-trained deep neural networks without re-training the model. We propose multipoint quantization, a quantization method that approximates a full-precision weight vector using a linear combination of multiple vectors of low-bit numbers; this is in contrast to typical quantization methods that approximate each weight using a single low precision number. Computationally, we construct the multipoint quantization with an efficient greedy selection procedure, and adaptively decides the number of low precision points on each quantized weight vector based on the error of its output. This allows us to achieve higher precision levels for important weights that greatly influence the outputs, yielding an ``effect of mixed precision'' but without physical mixed precision implementations (which requires specialized hardware accelerators). Empirically, our method can be implemented by common operands, bringing almost no memory and computation overhead. We show that our method outperforms a range of state-of-the-art methods on ImageNet classification and it can be generalized to more challenging tasks like PASCAL VOC object detection.

Downloads

Published

2021-05-18

How to Cite

Liu, X., Ye, M., Zhou, D., & Liu, Q. (2021). Post-training Quantization with Multiple Points: Mixed Precision without Mixed Precision. Proceedings of the AAAI Conference on Artificial Intelligence, 35(10), 8697-8705. Retrieved from https://ojs.aaai.org/index.php/AAAI/article/view/17054

Issue

Section

AAAI Technical Track on Machine Learning III