complexPyTorch

complexPyTorch copied to clipboard

dropout need to set mask's device same as input Tensor!

5

XIronMan0220

fix memory leak + reformatting

1

This PR fixes the memory leak with ComplexBatchNorm1d (I added with torch.no_grad as it was missing) + reformatting the code to be more readable and less error-prone

kareem-metwaly

Hi there, Thank you very much for this library, it has been very helpful in my research. I wanted to share with you some modifications I made to the code...

egrinstein

Need to replace `dilation` and `return_indices` args with `count_include_pad` and `divisor_override` args to be compatible with `torch.nn.functional.avg_pool2d`

jackhwalters

Use native Pytorch operations for complex numbers

2

Hi I noticed that you have custom matmul (https://github.com/wavefrontshaping/complexPyTorch/blob/a4e752caf827f3b642960366a7e9420f308076cc/complexPyTorch/complexFunctions.py#L11-L19) and tanh, neg functions defined (https://github.com/wavefrontshaping/complexPyTorch/blob/a4e752caf827f3b642960366a7e9420f308076cc/complexPyTorch/complexFunctions.py#L52-L56) which are actually unnecessary since these functions are supported for complex numbers in the last...

anjali411

ComplexBatchNorm1d Error

3

File "D:/Pycharm/coplexcnn/train.py", line 124, in y_hat = net(X) File "C:\Users\MyPC\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1051, in _call_impl return forward_call(*input, **kwargs) File "D:/Pycharm/complexcnn/train.py", line 79, in forward x = self.bn1(x) File "C:\Users\MyPC\AppData\Local\Programs\Python\Python38\lib\site-packages\torch\nn\modules\module.py", line 1051,...

Metamorphosis-chm

grad computations

3

hey and thanks for the great work, Just a question on the grad computations (though this could be my lack of understanding on autograd complex analysis). For the forward pass...

xmax1

can not compute in parallel

2

Dear author, the current version can not compute in parallel, could you pls revise it ?

chunmeifeng

hi I am a beginner to complex valued networks, My program uses real-valued loss functions and the network does not converge and I have a question: What kind of loss...

sweetan8898

Hi, About the naive batch norm, I believe it's more effective to normalize the tensor with the absolute value rather than normalizing each part individually: https://github.com/wavefrontshaping/complexPyTorch/blob/2044cb077b3f139d59dff56abc378b1457de40d6/complexPyTorch/complexLayers.py#L213 What do you think?

MordehayM