Case 1: the versions of CUDA, cudnn, Python and python do not match. Note that the python version, in particular, is easy to ignore.
Case 2 (my case is this): under the condition of case 1, check whether the graphics card used by your program is full, and if it is full, replace the unused card（‘ cuda:3 ’ —> ‘ cuda:0 ’）。
Case 3: add
torch. Backends. Cudnn. Enabled = false in the code
Read More:
- [Solved] RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED
- [Solved] RuntimeError: CuDNN error: CUDNN_STATUS_SUCCESS
- [Solved] RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling `cubla…
- [Solved] pytorch Error: RuntimeError: Unable to find a valid cuDNN algorithm to run convolution
- [Solved] RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling `cublasSgemm
- How to Solve Error: RuntimeError CUDA out of memory
- How to Solve RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu
- [Solved] RuntimeError (note: full exception trace is shown but execution is paused at: ＜module＞)
- [Solved] RuntimeError: cuda runtime error (100) : no CUDA-capable device is detected at
- How to Solve Python Importerror: DLL load failed: unable to find the specified program using tensorflow
- How to Solve QGIS installation paddy GPU error
- [Solved] bushi RuntimeError: version_ ＜= kMaxSupportedFileFormatVersion INTERNAL ASSERT FAILED at /pytorch/caffe2/s
- How to Solve pycharm terminal Failed to switch to virtual environment
- [Solved] Error when installing pycurl in MAC: Command ‘clang’ failed with exit status 1
- How to Solve Pychart configuration import torch error
- [Solved] RuntimeError: cublas runtime error : resource allocation failed at
- RuntimeError: Failed to register operator torchvision::_new_empty_tensor_op. +torch&torchversion Version Matching
- Pytorch: How to Handle error warning conda.gateways.disk.delete:unlink_or_rename_to_trash(140)
- How to Solve Python AttributeError: ‘module’ object has no attribute ‘xxx’
- How to Solve Turtle_tf Error in ROS