Fast Generation for Convolutional Autoregressive Models

Prajit Ramachandran, Tom Le Paine, Pooya Khorrami, Mohammad Babaeizadeh, Shiyu Chang, Yang Zhang, Mark A. Hasegawa-Johnson, Roy H. Campbell, Thomas S. Huang

Abstract: Convolutional autoregressive models have recently demonstrated state-of-the-art performance on a number of generation tasks. While fast, parallel training methods have been crucial for their success, generation is typically implemented in a naive fashion where redundant computations are unnecessarily repeated. This results in slow generation, making such models infeasible for production environments. In this work, we describe a method to speed up generation in convolutional autoregressive models. The key idea is to cache hidden states to avoid redundant computation. We apply our fast generation method to the Wavenet and PixelCNN++ models and achieve up to 21x and 183x speedups respectively.

TL;DR: We significantly speedup the generation in autoregressive models like Wavenet and PixelCNN up to 183 times.

Keywords: Deep learning, Applications

Conflicts: illinois.edu, ibm.com