19000

A Highly Efficient Distributed Deep Learning System For Automatic Speech Recognition

Wei Zhang, Xiaodong Cui, Ulrich Finkler, George Saon, Abdullah Kayi, Alper Buyuktosunoglu, Brian Kingsbury, David Kung, Michael Picheny
IBM Research
arXiv:1907.05701 [eess.AS], (10 Jul 2019)

@article{zhang2019highly,

   title={A Highly Efficient Distributed Deep Learning System For Automatic Speech Recognition},

   author={Zhang, Wei and Cui, Xiaodong and Finkler, Ulrich and Saon, George and Kayi, Abdullah and Buyuktosunoglu, Alper and Kingsbury, Brian and Kung, David and Picheny, Michael},

   journal={arXiv preprint arXiv:1907.05701},

   year={2019}

}

Download Download (PDF)   View View   Source Source   

302

views

Modern Automatic Speech Recognition (ASR) systems rely on distributed deep learning to for quick training completion. To enable efficient distributed training, it is imperative that the training algorithms can converge with a large mini-batch size. In this work, we discovered that Asynchronous Decentralized Parallel Stochastic Gradient Descent (ADPSGD) can work with much larger batch size than commonly used Synchronous SGD (SSGD) algorithm. On commonly used public SWB-300 and SWB-2000 ASR datasets, ADPSGD can converge with a batch size 3X as large as the one used in SSGD, thus enable training at a much larger scale. Further, we proposed a Hierarchical-ADPSGD (H-ADPSGD) system in which learners on the same computing node construct a super learner via a fast allreduce implementation, and super learners deploy ADPSGD algorithm among themselves. On a 64 Nvidia V100 GPU cluster connected via a 100Gb/s Ethernet network, our system is able to train SWB-2000 to reach a 7.6% WER on the Hub5-2000 Switchboard (SWB) test-set and a 13.2% WER on the Call-home (CH) test-set in 5.2 hours. To the best of our knowledge, this is the fastest ASR training system that attains this level of model accuracy for SWB-2000 task to be ever reported in the literature.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2019 hgpu.org

All rights belong to the respective authors

Contact us: