search

NVlabs / imaginaire

NVIDIA's Deep Imagination Team's PyTorch Library
Other
4.02k stars 452 forks source link

No module named 'upfirdn2d_cuda' #165

Open IUUI11 opened 2 years ago

IUUI11 commented 2 years ago

when I run this command _"python -m torch.distributed.launch --nproc_per_node=8 train.py \ --config configs/projects/munit/afhqdog2cat/ampO1.yaml" or "python -m torch.distributed.launch --nproc_per_node=1 inference.py --config configs/projects/munit/afhq_dog2cat/ampO1.yaml --output_dir projects/munit/output/afhq_dog2cat" , always said image

IUUI11 commented 2 years ago
1662969740(1)
MeimShang commented 2 years ago

You should install Imaginaire at first, follow the install instruction in readme.md

Gienapp commented 2 years ago

For me the solution was to activate the conda environment before installing the third party libraries.

ingaj89 commented 1 year ago

Your modules may be build to the “wrong” location so python doesn’t find them. You can add “—user” to the python setup.py install command so the modules will be installed to /use/lib/python/dist-packages. Or look at other options to set a custom install path for setuptools.