Functionality request: gradient of a gradient

PaddlePaddle / Paddle

PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）

http://www.paddlepaddle.org/

Apache License 2.0

22.18k stars 5.57k forks source link

Closed pengli09 closed 6 years ago

pengli09 commented 7 years ago

Currently, we can not implement formula (3) easily in paper Improved Training of Wasserstein GANs, because we need the gradient of a gradient. Maybe this will be a good functionality to consider in the future. Reference implementation: https://github.com/igul222/improved_wgan_training/blob/master/gan_toy.py

jacquesqiao commented 6 years ago

this feature will be done in Fluid.