HantingChen comments

Results 107 comments of


                                            HantingChen

Cifar100 - model and training configuration

We have not implemented our project on CIFAR100. If you want to conduct experiments on CIFAR100, you can refer to our config on ImageNet for an example. Thaks for the...

question about concurrently stacking the activation function

Thanks for the attention. We use the depth conv as an efficient implementation of our activation function, which is same as Eq. (6) in our paper. Each element of the...

question about concurrently stacking the activation function

Not really. In fact, the BN can be merged into the conv in the activation. Then, the weight and bias of the merged conv are a and b in Eq.(6).

question about concurrently stacking the activation function

> > Not really. In fact, the BN can be merged into the conv in the activation. Then, the weight and bias of the merged conv are a and b...

question about concurrently stacking the activation function

> > > Not really. In fact, the BN can be merged into the conv in the activation. Then, the weight and bias of the merged conv are a and...

Inappropriate implementation of Multi-head in IPT? (which only works on SR)

Sorry for the unclear code. This implement is right since we only release the test code. Your implement is a right way to modify it into a training code.

超分辨率模型如何训练灰度图

你好，可以改变输入参数num_channels为1,使网络的输入和输出channel数都为1，然后重新训练网络