JianWang

This is a great job, which unifies contrastive and generative method. And writting is also good. But my intuition tells me that this method seems sensitive to hyperparameters.

  你好，我发现你的代码和你的论文差异很大，框图右边的教师网络似乎并没有直接参与重建，而是与学生网络混合后重建，请问这是为什么？另外我想复现你论文中的方法，有一点不清楚：三个GO操作（全局池化加上两个全连接层，包括o/m/h）是权重不共享的吗？根据梯度更新？

In the function 'train_pcrl_2d', the variable 'n_data‘ was defined as 'len(train_loader)', but it was defined as 'len(train_dataset)' in C2L(MICCAI2020), as well as CMC(ECCV2020). Is this a mistake?