Improving the Training and Evaluation Efficiency of Recurrent Neural Network Language Models

Abstract

Recurrent neural network language models (RNNLMs) are becoming increasingly popular for speech recognition. Previously, we have shown that RNNLMs with a full (non-classed) output layer (F-RNNLMs) can be trained efficiently using a GPU giving a large reduction in training time over conventional class-based models (C-RNNLMs) on a standard CPU. However, since test-time RNNLM evaluation is often performed entirely on a CPU, standard F-RNNLMs are inefficient since the entire output layer needs to be calculated for normalisation. In this paper, it is demonstrated that C-RNNLMs can be efficiently trained on a GPU, using our spliced sentence bunch technique which allows good CPU test-time performance $(42\times$ speedup over F-RNNLM). Furthermore, the performance of different classing approaches is investigated. We also examine the use of variance regularisation of the softmax denominator for F-RNNLMs and show that it allows F-RNNLMs to be efficiently used in test $(56\times$ speedup on a CPU). Finally the use of two GPUs for F-RNNLM training using pipelining is described and shown to give a reduction in training time over a single GPU by a factor of $16\times$.