devsisters/pointer-network-tensorflow
A TensorFlow implementation of Pointer Networks, a neural architecture that learns to point to positions in input sequences for variable-length output prediction.

This repository implements Pointer Networks in TensorFlow, a neural network architecture introduced by Vinyals et al. that uses attention mechanisms to point to specific positions in input sequences. The project demonstrates training on combinatorial optimization problems like the Traveling Salesman Problem (TSP), with support for variable input/output lengths and multithreaded data pipelines for efficient training.