On the Variance of the Adaptive Learning Rate and Beyond
Radam：修正Adam算法中自适应学习率的早期方差.
Incorporating Nesterov Momentum into Adam
Nadam：将Nesterov动量引入Adam算法.
Pytorch中的Hook机制
Hook mechanism in Pytorch.
On the Convergence of Adam and Beyond
AMSGrad：改进Adam算法的收敛性.
Adam: A Method for Stochastic Optimization
Adam：自适应矩估计.
On the importance of initialization and momentum in deep learning
Nesterov Momentum：一种动量梯度更新方法.