Improving Language Modeling using Densely Connected Recurrent Neural Networks

2017/07/19 by Godin, Fréderic, Dambre, Joni, De Neve, Wesley · 1 citation
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.1707.06130

Abstract

In this paper, we introduce the novel concept of densely connected layers into recurrent neural networks. We evaluate our proposed architecture on the Penn Treebank language modeling task. We show that we can obtain similar perplexity scores with six times fewer parameters compared to a standard stacked 2-layer LSTM model trained with dropout (Zaremba et al. 2014). In contrast with the current usage of skip connections, we show that densely connecting only a few stacked layers with skip connections already yields significant perplexity reductions.

Cited by

• Residual Matrix Transformers: Scaling the Size of the Residual Stream

Related

• Exploring the Limits of Language Modeling
• Tying Word Vectors and Word Classifiers: A Loss Framework for Language Modeling
• Gradual Learning of Recurrent Neural Networks
• Direct Output Connection for a High-Rank Language Model
• Language Modeling with Gated Convolutional Networks