Description
🐛 Describe the bug
When numpy not preinstalled on the machine, with latest torch install we get : https://github.com/pytorch/builder/actions/runs/10080038080/job/27868882276#step:9:438
++ python ./test/smoke_test/smoke_test.py
A module that was compiled using NumPy 1.x cannot be run in
NumPy 2.0.1 as it may crash. To support both 1.x and 2.x
versions of NumPy, modules must be compiled with NumPy 2.0.
Some module may need to rebuild instead e.g. with 'pybind11>=2.12'.
If you are a user of the module, the easiest solution will be to
downgrade to 'numpy<2' or try to upgrade the affected module.
We expect that some modules will need time to support NumPy 2.
then later:
std::thread::hardware_concurrency() : 16
Environment variables:
OMP_NUM_THREADS : [not set]
MKL_NUM_THREADS : [not set]
ATen parallel backend: OpenMP
Testing smoke_test_conv2d
Testing smoke_test_linalg on cpu
File "C:\actions-runner\_work\builder\builder\pytorch\builder\test\smoke_test\smoke_test.py", line 330, in main
test_numpy()
File "C:\actions-runner\_work\builder\builder\pytorch\builder\test\smoke_test\smoke_test.py", line 72, in test_numpy
torch.tensor(x)
RuntimeError: Could not infer dtype of numpy.int64
Cause:
numpy release https://pypi.org/project/numpy/2.0.1/ is not compatible with torch
numpy dependency is coming from torchvision METADATA
Requires-Dist: numpy
Versions
2.4.0 - torchvision 0.19