I used to use torch 0.4.1, numpy 1.20.1, running error pytorch was compiled without numpy support
later found the following solution
1
pip install numpy==1.15.0
2.upgrade torch-0.4.1
to torch-0.4.1.post2
pip install torch==0.4.1.post2
Another error will be reported after using 1 (I forgot the screenshot), which is like this: valueerror: numpy.ufunc size changed, may indicate binary instability. Expected 216 from C header, got 192 from pyobject
after that, I changed the version of numpy
to 1.20.1
, Then update torch
from torch-0.4.1
to torch-0.4.1. Post2
, so the first step is unnecessary
Read More:
- [Solved] Pytorch Tensor to numpy error: RuntimeError: Can‘t call numpy() on Tensor that requires grad.报错
- [Solved] RuntimeError: Numpy is not available (Associated Torch or Tensorflow)
- [Solved] Pytorch Error: RuntimeError: expected scalar type Double but found Float
- Pytorch torch.cuda.FloatTensor Error: RuntimeError: one of the variables needed for gradient computation has…
- [Solved] bushi RuntimeError: version_ <= kMaxSupportedFileFormatVersion INTERNAL ASSERT FAILED at /pytorch/caffe2/s
- [Pytorch Error Solution] Pytorch distributed RuntimeError: Address already in use
- [Solved] Original error was: No module named ‘numpy.core._multiarray_umath‘
- [Solved] pytorch loss.backward() Error: RuntimeError: Function AddBackward0 returned an invalid gradient at index 1…
- [Solved] PyTorch Caught RuntimeError in DataLoader worker process 0和invalid argument 0: Sizes of tensors mus
- pytorch RuntimeError: Error(s) in loading state_ Dict for dataparall… Import model error solution
- [Solved] Pytorch error: RuntimeError: one of the variables needed for gradient computation
- pytorch: RuntimeError CUDA error device-side assert triggered
- [Solved] Pytorch Error: RuntimeError: Error(s) in loading state_dict for Network: size mismatch
- Pytorch Error: RuntimeError: value cannot be converted to type float without overflow: (0.00655336,-0.00
- Pytorch Error: runtimeerror: expected scalar type double but found float
- RuntimeError: CUDA error: an illegal memory access was encountered
- [Solved] PyTorch Load Model Error: Missing key(s) RuntimeError: Error(s) in loading state_dict for
- [Solved] pytorch Error: RuntimeError: Unable to find a valid cuDNN algorithm to run convolution
- For the problem of rejecting old usage errors after numpy is updated, modified in numpy 1.20; for more details and guidance
- Pytorch ValueError: Expected more than 1 value per channel when training, got input size [1, 768