Inspired by the principles of speed reading, we introduce Skim-RNN, a recurrent neural network (RNN) that dynamically decides to update only a small fraction of the hidden state for relatively unimportant input tokens. Skim-RNN gives computational advantage over an RNN that always updates the entire hidden state. Skim-RNN uses the same input and output interfaces as a standard RNN and can be easily used instead of RNNs in existing models. In our experiments, we show that Skim-RNN can achieve significantly reduced computational cost without losing accuracy compared to standard RNNs across five different natural language tasks. In addition, we demonstrate that the trade-off between accuracy and speed of Skim-RNN can be dynamically controlled during inference time in a stable manner. Our analysis also shows that Skim-RNN running on a single CPU offers lower latency compared to standard RNNs on GPUs.
@article{arxiv.1711.02085,
title = {Neural Speed Reading via Skim-RNN},
author = {Minjoon Seo and Sewon Min and Ali Farhadi and Hannaneh Hajishirzi},
journal= {arXiv preprint arXiv:1711.02085},
year = {2018}
}