Saved in:
Bibliographic Details
Main Authors: Xu, Hongfei, Song, Yang, Liu, Qiuhui, van Genabith, Josef, Xiong, Deyi
Format: Preprint
Published: 2020
Subjects:
Online Access:https://arxiv.org/abs/2007.06257
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909160005697536
author Xu, Hongfei
Song, Yang
Liu, Qiuhui
van Genabith, Josef
Xiong, Deyi
author_facet Xu, Hongfei
Song, Yang
Liu, Qiuhui
van Genabith, Josef
Xiong, Deyi
contents Stacking non-linear layers allows deep neural networks to model complicated functions, and including residual connections in Transformer layers is beneficial for convergence and performance. However, residual connections may make the model "forget" distant layers and fail to fuse information from previous layers effectively. Selectively managing the representation aggregation of Transformer layers may lead to better performance. In this paper, we present a Transformer with depth-wise LSTMs connecting cascading Transformer layers and sub-layers. We show that layer normalization and feed-forward computation within a Transformer layer can be absorbed into depth-wise LSTMs connecting pure Transformer attention layers. Our experiments with the 6-layer Transformer show significant BLEU improvements in both WMT 14 English-German / French tasks and the OPUS-100 many-to-many multilingual NMT task, and our deep Transformer experiments demonstrate the effectiveness of depth-wise LSTM on the convergence and performance of deep Transformers.
format Preprint
id arxiv_https___arxiv_org_abs_2007_06257
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle Rewiring the Transformer with Depth-Wise LSTMs
Xu, Hongfei
Song, Yang
Liu, Qiuhui
van Genabith, Josef
Xiong, Deyi
Computation and Language
Stacking non-linear layers allows deep neural networks to model complicated functions, and including residual connections in Transformer layers is beneficial for convergence and performance. However, residual connections may make the model "forget" distant layers and fail to fuse information from previous layers effectively. Selectively managing the representation aggregation of Transformer layers may lead to better performance. In this paper, we present a Transformer with depth-wise LSTMs connecting cascading Transformer layers and sub-layers. We show that layer normalization and feed-forward computation within a Transformer layer can be absorbed into depth-wise LSTMs connecting pure Transformer attention layers. Our experiments with the 6-layer Transformer show significant BLEU improvements in both WMT 14 English-German / French tasks and the OPUS-100 many-to-many multilingual NMT task, and our deep Transformer experiments demonstrate the effectiveness of depth-wise LSTM on the convergence and performance of deep Transformers.
title Rewiring the Transformer with Depth-Wise LSTMs
topic Computation and Language
url https://arxiv.org/abs/2007.06257