Low-precision training reduces computational cost and produces efficient models. Recent research in developing new low-precision training algorithms often relies on simulation to empirically evaluate the statistical effects of quantization while avoiding the substantial overhead of building specific hardware. To support this empirical research, we introduce QPyTorch, a low-precision arithmetic simulation framework. Built natively in PyTorch, QPyTorch provides a convenient interface that minimizes the efforts needed to reliably convert existing codes to study low-precision training. QPyTorch is general, and supports a variety of combinations of precisions, number formats, and rounding options. Additionally, it leverages an efficient fused-kernel approach to reduce simulator overhead, which enables simulation of large-scale, realistic problems. QPyTorch is publicly available at https://github.com/Tiiiger/QPyTorch.
@article{arxiv.1910.04540,
title = {QPyTorch: A Low-Precision Arithmetic Simulation Framework},
author = {Tianyi Zhang and Zhiqiu Lin and Guandao Yang and Christopher De Sa},
journal= {arXiv preprint arXiv:1910.04540},
year = {2019}
}
Comments
NeurIPS 2019 EMC^2 Workshop on Energy Efficient Machine Learning and Cognitive Computing