Runtime Error while running the code
Closed this issue · 8 comments
Hi, I followed the instruction for environment setup but have a runtime error while running the code for your proposed method. The error was about a datatype exception that expect a float but received a double.
Hi, did you try explicitly converting the variable to float using .float()? That usually works but I'm not sure where exactly you got the error.
Please find the following screenshot for the detail about the error message.
The error is coming from the fastai dataset code that we had used. Could you check the fastai, pytorch and python versions that you are using? We did not face any such problem with our setup, but I had last checked it almost 2 years ago.
Please also see the issue #14 where there was a problem with data preprocessing because fastai v2.0 was used while our code used the older v1.0.60.
I am using fastai 1.0.60, pytorch 1.10.1+cu102, and python 3.6.13.
Could you try with PyTorch 1.3.1? we had used that version. You can use our environment.yml file for installing a new environment. The reason is that PyTorch has had many changes from 1.3 to 1.10 which may be causing the problem.
May I know the version of torchvision You used? It seems like there are still some problems even if I use torch 1.3.1.
Sorry, I don't have access to the original environment that I used and I cannot find the version of torchvision corresponding to 1.3.1 of torch. But you can try installing 1.4 - link for installation instructions, as it won't be too different from 1.3.1. The torchvision version for torch 1.4 is 0.5
Closing due to inactivity. Please feel free to re-open or open another issue for other queries.