Softermax: Hardware/Software Co-Design of an Efficient Softmax for Transformers [article]

Jacob R. Stevens, Rangharajan Venkatesan, Steve Dai, Brucek Khailany, Anand Raghunathan
2021 arXiv   pre-print
Transformers have transformed the field of natural language processing. This performance is largely attributed to the use of stacked self-attention layers, each of which consists of matrix multiplies as well as softmax operations. As a result, unlike other neural networks, the softmax operation accounts for a significant fraction of the total run-time of Transformers. To address this, we propose Softermax, a hardware-friendly softmax design. Softermax consists of base replacement, low-precision
more » ... softmax computations, and an online normalization calculation. We show Softermax results in 2.35x the energy efficiency at 0.90x the size of a comparable baseline, with negligible impact on network accuracy.
arXiv:2103.09301v1 fatcat:mnrcs6wjefaw5pmndpv3ucb6nm