You are not logged in.
Pages: 1
Hi, I am using python version 3.10 which i believe the default version for `5.15.12-arch1-1`. I have installed pytorch using the python-pytorch-cuda. The installation went smooth. However, I am came across the error when i ran the neural network program. The error is given below.
`dataloader-typeerror-instance-and-class-checks-can-only-be-used-with-runtime-checkable-protocols`.
After googling, I came to know pytorch doesn't support python3.10 from the below link
https://discuss.pytorch.org/t/dataloade … ols/139472
Is it possible to have the same package for lesser version of python ?. Or is there any other way to fix this issue. Please help me.
Offline
https://bugs.archlinux.org/task/73187
https://github.com/pytorch/pytorch/issues/69911
https://github.com/pytorch/pytorch/issues/66424
Combining info from all 3 links it seems building pytorch against python 3.10 is succesfull, but there are runtime issues.
Can you afford to wait until pytorch 1.11 is released ?
Last edited by Lone_Wolf (2022-01-06 13:52:21)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
(A works at time B) && (time C > time B ) ≠ (A works at time C)
Offline
I am eagerly waiting for PyTorch 1.11. Any idea when will be the next release of PyTorch?.
Offline
https://github.com/pytorch/pytorch/milestones shows 17 open issues, so it may not be soon.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
(A works at time B) && (time C > time B ) ≠ (A works at time C)
Offline
Pages: 1