Implements an efficient softmax approximation as described in the paper "Efficient softmax approximation for GPUs" (http://arxiv.org/abs/1609.04309)

Decoupled Weight Decay Regularization (ICLR 2019)

Convolutional Neural Network Language Models