Open Matans844 opened 6 months ago
Have you solved your problem yet?
Have you solved your problem yet?
Have you solved your problem yet?
My code works fine now, odds are it's the version of the package. Just change it all to zhuyanjun's version.
My sincere thanks!
Have you solved your problem yet?
My code works fine now, odds are it's the version of the package. Just change it all to zhuyanjun's version. What do you mean zhuyanjun's version, can you please tell me which version you used?
Problem resolved. Don't use the PyTorch and torchvision versions from the requirements file. I was able to run it using the newer versions. The versions I used are torch==1.12, torchvision==0.13.1
Steps to reproduce the issue:
Now the output: I swapped asbolute paths to the clone location with
<MYPATH>
I first got:
And then I got:
So from the first epoch, all values for the loss function are
nan
.In addition, there's a mismatch between:
conda env create -f environment.yml
andpip install -r requirements.txt
. Specificallly,visdom
andnetworkx
aren't installed when using the theenvironment.yml
setup option, but are installed when thepip
setup option is used.Edit: Also after using
pip install -r requirements.txt
and then running:the model collapses: