mocha
                                
                                
                                
                                    mocha copied to clipboard
                            
                            
                            
                        Example implementation of Monotonic Chunkwise Attention.
Monotonic Chunkwise Attention (MoChA)
This repository contains a notebook which shows how to efficiently compute the Monotonic Chunkwise Attention (MoChA) distribution using TensorFlow. MoChA is described in "Monotonic Chunkwise Attention" by Chung-Cheng Chiu* and Colin Raffel*, in Proceedings of ICLR 2018.