js-mim / mlsp2017_svsep_skipfilt Goto Github PK

Hello, I feel your article is great, I currently use the compressed version of the DSD100 database, would like to test your code, but there is a mistake：
Can not feed value of shape (32, 18, 1025) for Tensor 'input_1: 0', which has shape '(1, 18, 1025)'
If I did not guessed it might be my batch_size is not the same as the batch_size of your trained model, how can I solve this problem? Do I have to use the standard DSD100 database for testing? Very much looking forward to your reply

After downloaded the DSD100 dataset, I use the code "gdae_dsd.py" for training process, error is shown below:

ValueError: When passing a list as loss, it should have one entry per model outputs. The model has 3 outputs, but you passed loss=[<function KL at 0x7efcecd2b488>, <function KL at 0x7efcecd2b488>]

so I try to rivse the code like
model.compile(optimizer = opt, loss = [KL, KL, KL])
PS: orignal code is "model.compile(optimizer = opt, loss = [KL, KL])"

Another error happen:
ValueError: Error when checking model target: the list of Numpy arrays that you are passing to your model is not the size the model expected. Expected to see 3 arrays but instead got the following list of 2 arrays:

So, do you run into these issues? how to fix it?
Thanks