Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling
Abstract
Lingvo is a Tensorflow framework offering a complete solution for collaborative deep learning research, with a particular focus towards sequence-to-sequence models. Lingvo models are composed of modular building blocks that are flexible and easily extensible, and experiment configurations are centralized and highly customizable. Distributed training and quantized inference are supported directly within the framework, and it contains existing implementations of a large number of utilities, helper functions, and the newest research ideas. Lingvo has been used in collaboration by dozens of researchers in more than 20 papers over the last two years. This document outlines the underlying design of Lingvo and serves as an introduction to the various pieces of the framework, while also offering examples of advanced features that showcase the capabilities of the framework.
Cite
@article{arxiv.1902.08295,
title = {Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling},
author = {Jonathan Shen and Patrick Nguyen and Yonghui Wu and Zhifeng Chen and Mia X. Chen and Ye Jia and Anjuli Kannan and Tara Sainath and Yuan Cao and Chung-Cheng Chiu and Yanzhang He and Jan Chorowski and Smit Hinsu and Stella Laurenzo and James Qin and Orhan Firat and Wolfgang Macherey and Suyog Gupta and Ankur Bapna and Shuyuan Zhang and Ruoming Pang and Ron J. Weiss and Rohit Prabhavalkar and Qiao Liang and Benoit Jacob and Bowen Liang and HyoukJoong Lee and Ciprian Chelba and Sébastien Jean and Bo Li and Melvin Johnson and Rohan Anil and Rajat Tibrewal and Xiaobing Liu and Akiko Eriguchi and Navdeep Jaitly and Naveen Ari and Colin Cherry and Parisa Haghani and Otavio Good and Youlong Cheng and Raziel Alvarez and Isaac Caswell and Wei-Ning Hsu and Zongheng Yang and Kuan-Chieh Wang and Ekaterina Gonina and Katrin Tomanek and Ben Vanik and Zelin Wu and Llion Jones and Mike Schuster and Yanping Huang and Dehao Chen and Kazuki Irie and George Foster and John Richardson and Klaus Macherey and Antoine Bruguier and Heiga Zen and Colin Raffel and Shankar Kumar and Kanishka Rao and David Rybach and Matthew Murray and Vijayaditya Peddinti and Maxim Krikun and Michiel A. U. Bacchiani and Thomas B. Jablin and Rob Suderman and Ian Williams and Benjamin Lee and Deepti Bhatia and Justin Carlson and Semih Yavuz and Yu Zhang and Ian McGraw and Max Galkin and Qi Ge and Golan Pundak and Chad Whipkey and Todd Wang and Uri Alon and Dmitry Lepikhin and Ye Tian and Sara Sabour and William Chan and Shubham Toshniwal and Baohua Liao and Michael Nirschl and Pat Rondon},
journal= {arXiv preprint arXiv:1902.08295},
year = {2019}
}