CUDA is not available issue
saint-boons opened this issue · 3 comments
UserWarning: torch.cuda.amp.GradScaler is enabled, but CUDA is not available. Disabling.
warnings.warn("torch.cuda.amp.GradScaler is enabled, but CUDA is not available. Disabling.")
Hello,
I have this issue when running the AI.
I have found issue #138 and I ran the one liner script and got None
So I've tried looking for CUDA drivers for my RTX 2080 Ti, but I didn't find anything except a package called NVIDIA CUDA Toolkit in winget so I ran winget install Nvidia.CUDA and it installed successfully but even after a restart it didn't change anything.
Has anyone got this issue too? Did anyone manage to fix it?
I have the same problem I keep getting this:
Setting jit to False because torch version is not 1.7.1.
c:\programdata\anaconda\lib\site-packages\torch\cuda\amp\grad_scaler.py:115: UserWarning: torch.cuda.amp.GradScaler is enabled, but CUDA is not available. Disabling.
warnings.warn("torch.cuda.amp.GradScaler is enabled, but CUDA is not available. Disabling.")
Starting up...
Imagining "dlkjsa" from the depths of my weights...
epochs: 0%| | 0/20 [00:00<?, ?it/s]c:\programdata\anaconda\lib\site-packages\torch\cuda\amp\autocast_mode.py:120: UserWarning: torch.cuda.amp.autocast only affects CUDA ops, but CUDA is not available. Disabling.
warnings.warn("torch.cuda.amp.autocast only affects CUDA ops, but CUDA is not available. Disabling.")
This is all it says for about 30 mins freezes up then shuts down.
I also downloaded the toolkit and it did nothing
Alright I've found a solution and I'm posting it here for people in the future.
Torch did not install properly and so I tried getting the correct version but got an OS Error when trying to install torch
ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: 'C:\\Users\\h_bas\\AppData\\Local\\Packages\\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\\LocalCache\\local-packages\\Python39\\site-packages\\caffe2\\python\\serialized_test\\data\\operator_test\\collect_and_distribute_fpn_rpn_proposals_op_test.test_collect_and_dist.zip'
So I made a forum post on the PyTorch forums here which details a fix for file paths being too long
https://discuss.pytorch.org/t/installation-problem-oserror-errno-2/131750