Language Modeling with Gated Convolutional Networks

TLDR

Recurrent neural networks dominate language modeling largely because they capture unbounded context. The study introduces a finite‑context, stacked‑convolutional language model with a simplified gating mechanism that outperforms prior work and examines key architectural choices. The model employs stacked convolutions and a novel gating scheme to enable parallel token processing. It achieves state‑of‑the‑art results on WikiText‑103, competitive performance on Google Billion Words, reduces sentence‑scoring latency by an order of magnitude versus recurrent baselines, and is the first non‑recurrent model competitive with strong recurrent models on large‑scale language tasks.

Abstract

The pre-dominant approach to language modeling to date is based on recurrent neural networks. Their success on this task is often linked to their ability to capture unbounded context. In this paper we develop a finite context approach through stacked convolutions, which can be more efficient since they allow parallelization over sequential tokens. We propose a novel simplified gating mechanism that outperforms Oord et al (2016) and investigate the impact of key architectural decisions. The proposed approach achieves state-of-the-art on the WikiText-103 benchmark, even though it features long-term dependencies, as well as competitive results on the Google Billion Words benchmark. Our model reduces the latency to score a sentence by an order of magnitude compared to a recurrent baseline. To our knowledge, this is the first time a non-recurrent approach is competitive with strong recurrent models on these large scale language tasks.