Search neural-net structure with differentiable attention

Improve DARTS by differentiable attention algorithm

This code is based on the implementation of DARTS.

Requirements

Python >= 3.7.0 
PyTorch = 1.3.1 or PyTorch = 1.5.0 (PyTorch 1.4 would crash!!!)

Datasets

Pretrained models

Citation