myREADME.txt
1.33 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
<changed>
model.py
1) Error appear
Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!
-> class BeatGan
-> self.bce_criterion = nn.BCELoss().cuda()
self.mse_criterion = nn.MSELoss().cuda()
-> def update_netd
->
self.err_d_real = self.bce_criterion(self.out_d_real, torch.full((self.batchsize,), self.real_label, device=self.device).type(torch.FloatTensor).cuda())
self.err_d_fake = self.bce_criterion(self.out_d_fake, torch.full((self.batchsize,), self.fake_label, device=self.device).type(torch.FloatTensor).cuda())
data.py
1) if you use 2D signal data, the data shape must (num_of_data, 1, width, height)
2) I changed "np.load()" parts for utilize.
3) normalize part is changed.
network.py
1) Conv2D(input channel, output channel, kernel_size, stride, padding)
output size = output channel * (input_size-kernel_size+2*padding)/stride+1
We use 128*128 data so we set "Encoder"'s Conv2d's parameters like that.
The result is 50*1*1
2) Decoder is inverse of Encoder
option.py
1) change batchsize from 64 to 32, because Memory allocate issue was happened.
etc)
1) 2D-data must have the shape (n,1,w,h)
1D-data is (n,1,w)
1-1) So, you should go to experiments/ecg/dataset/preprocessed.
Push change.py and change2.py to ano0.
Diretory ano0 should have _ _samples.npy file.