Fast MoE
===

An easy-to-use but efficient implementation of the Mixture of Experts (MoE) 
model for PyTorch