[Bug] jetson NX :When i use the deploy.py or check_env.py, it occur: cannot import name 'ProcessGroup' from 'torch.distributed'
lijoe123 opened this issue · 12 comments
Checklist
- I have searched related issues but cannot get the expected help.
- 2. I have read the FAQ documentation but cannot get the expected help.
- 3. The bug has not been fixed in the latest version.
Describe the bug
Jetson NX: When i use the deploy.py or check_env.py, it occur: cannot import name 'ProcessGroup' from 'torch.distributed'
Reproduction
python tools/deploy.py and python tools/check_env.py
Environment
I cannot use the check_env.py
jetpack 5.1.1
# Name Version Build Channel
_openmp_mutex 4.5 2_gnu conda-forge
addict 2.4.0 <pip>
aenum 3.1.12 <pip>
appdirs 1.4.4 <pip>
bzip2 1.0.8 hf897c2e_4 conda-forge
ca-certificates 2018.03.07 0 c4aarch64
certifi 2023.5.7 <pip>
charset-normalizer 3.1.0 <pip>
click 8.1.3 <pip>
contourpy 1.0.7 <pip>
cycler 0.11.0 <pip>
Cython 0.29.35 <pip>
dill 0.3.6 <pip>
fonttools 4.39.4 <pip>
grpcio 1.54.2 <pip>
h5py 3.8.0 <pip>
idna 3.4 <pip>
importlib-resources 5.12.0 <pip>
kiwisolver 1.4.4 <pip>
ld_impl_linux-aarch64 2.39 ha75b1e8_0 conda-forge
libffi 3.4.2 h3557bc0_5 conda-forge
libgcc-ng 12.2.0 h607ecd0_19 conda-forge
libgomp 12.2.0 h607ecd0_19 conda-forge
libnsl 2.0.0 hf897c2e_0 conda-forge
libsqlite 3.40.0 hf9034f9_0 conda-forge
libuuid 2.32.1 hf897c2e_1000 conda-forge
libzlib 1.2.13 h4e544f5_4 conda-forge
Mako 1.2.4 <pip>
markdown-it-py 2.2.0 <pip>
MarkupSafe 2.1.2 <pip>
matplotlib 3.7.1 <pip>
mdurl 0.1.2 <pip>
mmcv 2.0.0 <pip>
mmdeploy 1.1.0 <pip>
mmdet 3.0.0 <pip>
mmengine 0.7.3 <pip>
multiprocess 0.70.14 <pip>
ncurses 6.3 headf329_1 conda-forge
ndindex 1.7 <pip>
numpy 1.24.3 <pip>
onnx 1.14.0 <pip>
opencv-python 4.7.0.72 <pip>
openssl 3.0.7 h4e544f5_0 conda-forge
packaging 23.1 <pip>
Pillow 9.5.0 <pip>
pip 22.3.1 pyhd8ed1ab_0 conda-forge
platformdirs 3.5.1 <pip>
prettytable 3.7.0 <pip>
protobuf 3.20.2 <pip>
pycocotools 2.0.6 <pip>
pycuda 2022.2.2 <pip>
Pygments 2.15.1 <pip>
pyparsing 3.0.9 <pip>
pyserial 3.5 <pip>
python 3.8.13 h92ab765_0_cpython conda-forge
python-dateutil 2.8.2 <pip>
pytools 2022.1.14 <pip>
PyYAML 6.0 <pip>
readline 8.1.2 h38e3740_0 conda-forge
requests 2.31.0 <pip>
rich 13.3.5 <pip>
scipy 1.10.1 <pip>
setuptools 65.5.1 pyhd8ed1ab_0 conda-forge
shapely 2.0.1 <pip>
six 1.16.0 <pip>
sqlite 3.40.0 h69ca7e5_0 conda-forge
tensorrt 8.5.2.2 <pip>
termcolor 2.3.0 <pip>
terminaltables 3.1.10 <pip>
tk 8.6.12 hd8af866_0 conda-forge
tomli 2.0.1 <pip>
torch 1.13.0a0+d0d6b1f2.nv22.09 <pip>
torchvision 0.14.0 <pip>
typing_extensions 4.6.2 <pip>
urllib3 2.0.2 <pip>
versioned-hdf5 1.3.13 <pip>
wcwidth 0.2.6 <pip>
wheel 0.38.4 pyhd8ed1ab_0 conda-forge
xz 5.2.6 h9cdd2b7_0 conda-forge
yapf 0.33.0 <pip>
zipp 3.15.0 <pip>Error traceback
No response
@lijoe123 hi, the env collection is calling functions from mmcv. Since you have installed mmcv, could you try run python -c "from mmcv.utils import collect_env; print(collect_env())"?
I'm sorry. The same problem
then maybe you could post the issue in mmengine.
I had asked the pytorch official, and they give me this advice #102596, and i don't know how to set this, could you give me some advice?
@lijoe123 If you only want to deploy model in a fast way. You cant git clone mmengine and comment out the unused line that throws error for from mmengine.dist.
# 1. clone source code
git clone --depth 1 https://github.com/open-mmlab/mmengine.git
# 2. comment some import lines in mmengine.dist.__init__.py
# 3. reinstall mmengine
cd mmengine
pip installl -e .
@RunningLeon Thank you for your reply. Acctually, the from torch.distributed import ProcessGroup is in the dist.py. And I don't know how to comment. And the ProcessGroup had been used , i don't know how to comment.
This issue is marked as stale because it has been marked as invalid or awaiting response for 7 days without any further response. It will be closed in 5 days if the stale label is not removed or if there is no further response.
This issue is closed because it has been stale for 5 days. Please open a new issue if you have similar issues or you have any new updates now.