Hello!
I am a PhD student and I am very interested in your work! That's why I am trying to replicate your results, so I can understand the details of Branchynet based on your implementation.
I am currently simply using a Google Colab notebook. I have uploaded all the files and I have been able to import the modules and functions, etc.
However, when I try to get the data from the datasets and create the training and testing datasets I get the following error:
No such file or directory: 'datasets/data/pcifar10/data.npz'
Indeed, based on the downloadable zip file that I got from your Github, there is no datasets/data folder.
Also,
When I run the code below:
from networks import lenet_mnist
branchyNet = lenet_mnist.get_network()
branchyNet.to_gpu()
branchyNet.training()
I get the following error:
'module' object has no attribute 'ReLU'
Do you have any ideas or insights as to why this happens?
Is there a more recent version of the code, or the downloadable folder structure that would resolve this?
Thank you for your attention to this,
Best,
Pavlos Zouridakis
Hello!
I am a PhD student and I am very interested in your work! That's why I am trying to replicate your results, so I can understand the details of Branchynet based on your implementation.
I am currently simply using a Google Colab notebook. I have uploaded all the files and I have been able to import the modules and functions, etc.
However, when I try to get the data from the datasets and create the training and testing datasets I get the following error:
No such file or directory: 'datasets/data/pcifar10/data.npz'
Indeed, based on the downloadable zip file that I got from your Github, there is no datasets/data folder.
Also,
When I run the code below:
from networks import lenet_mnist
branchyNet = lenet_mnist.get_network()
branchyNet.to_gpu()
branchyNet.training()
I get the following error:
'module' object has no attribute 'ReLU'
Do you have any ideas or insights as to why this happens?
Is there a more recent version of the code, or the downloadable folder structure that would resolve this?
Thank you for your attention to this,
Best,
Pavlos Zouridakis