Figure 5From: RNN language model with word clustering and class-based output layer Detailed comparisons of perplexity and speed for different RNNLMs on Penn Treeback Corpus. Back to article page