lit_corrnet3d_clean.py: error: unrecognized arguments: --ckpt_user=lightning_logs/version_114/checkpoints/epoch=43-step=202399.ckpt

[lllllialois]

why when i run "python lit_corrnet3d_clean.py --gpus=3 --batch_size=1 --ckpt_user=lightning_logs/version_114/checkpoints/epoch=43-step=202399.ckpt --data_dir=./trainset.h5 --test_data_dir=./testset.h5", it shows that"lit_corrnet3d_clean.py: error: unrecognized arguments: --ckpt_user=lightning_logs/version_114/checkpoints/epoch=43-step=202399.ckpt".

[ZENGYIMING-EAMON]

Which version of the PyTorch lightning package did you use? Can you try the same version as mentioned in the readme first?

[lllllialois]

Which version of the PyTorch lightning package did you use? Can you try the same version as mentioned in the readme first?

ok I install by‘’ Prerequisite Installation ‘’ in readme， all versions are same for you

[lllllialois]

Which version of the PyTorch lightning package did you use? Can you try the same version as mentioned in the readme first?

I want to ask which version of the torch and torchversion package did you use? when I use torch=1.7.1 and torchversion=0.81
it warns me unable to load pointnet2 ， when I use torch=1.4.0 and torchversion=0.5， the function random_split will get an error.

[ZENGYIMING-EAMON]

As mentioned in the notes: "Version info: Python3.8, pytorch-lightning 1.1.6, pytorch 1.6.0, cudatoolkit 10.2, Cuda 10.2 (on RTX 2080 Ti) and Ubuntu 18.04.3 LTS (x86_64)"

[lllllialois]

As mentioned in the notes: "Version info: Python3.8, pytorch-lightning 1.1.6, pytorch 1.6.0, cudatoolkit 10.2, Cuda 10.2 (on RTX 2080 Ti) and Ubuntu 18.04.3 LTS (x86_64)"

All below are the same as you， I just want to ask you the version of torch and torchversion， not the PyTorch version
my version：Python3.8, pytorch-lightning 1.1.6, pytorch 1.6.0, cudatoolkit 10.2, Cuda 10.2 (on TITAN_X_Pascal)



this is my version， and I met this question。

[ZENGYIMING-EAMON]

Package Version

---

absl-py 1.0.0
aiohttp 3.8.1
aiosignal 1.2.0
async-timeout 4.0.2
attrs 21.4.0
beautifulsoup4 4.11.1
cachetools 5.0.0
certifi 2021.10.8
charset-normalizer 2.0.12
colorama 0.4.4
cycler 0.11.0
dataclasses 0.8
filelock 3.6.0
fonttools 4.33.1
frozenlist 1.3.0
fsspec 2022.3.0
future 0.18.2
fvcore 0.1.5.post20220414
gdown 4.4.0
google-auth 2.6.6
google-auth-oauthlib 0.4.6
grpcio 1.44.0
h5py 3.6.0
idna 3.3
igl 2.2.1
importlib-metadata 4.11.3
iopath 0.1.9
joblib 1.1.0
kiwisolver 1.4.2
KNN-CUDA 0.2
Markdown 3.3.6
matplotlib 3.5.1
mkl-fft 1.3.1
mkl-random 1.2.2
mkl-service 2.4.0
multidict 6.0.2
ninja 1.10.2.3
numexpr 2.8.1
numpy 1.21.5
oauthlib 3.2.0
opencv-python 4.5.5.64
packaging 21.3
Pillow 9.1.0
pip 22.0.4
point-masker 0.0.0
point-render 0.0.0
pointnet2-ops 3.0.0
portalocker 2.4.0
protobuf 3.20.1
pyasn1 0.4.8
pyasn1-modules 0.2.8
pyparsing 3.0.8
PySocks 1.7.1
python-dateutil 2.8.2
pytorch-lightning 1.1.6
pytorch3d 0.6.1
PyYAML 6.0
requests 2.27.1
requests-oauthlib 1.3.1
rsa 4.8
scikit-learn 1.0.2
scipy 1.8.0
setuptools 62.1.0
six 1.16.0
soupsieve 2.3.2.post1
tables 3.7.0
tabulate 0.8.9
tensorboard 2.8.0
tensorboard-data-server 0.6.1
tensorboard-plugin-wit 1.8.1
termcolor 1.1.0
threadpoolctl 3.1.0
torch 1.6.0
torchvision 0.7.0
tqdm 4.64.0
trimesh 3.10.8
urllib3 1.26.9
Werkzeug 2.1.1
wheel 0.37.1
yacs 0.1.8
yarl 1.7.2
zipp 3.8.0

[ZENGYIMING-EAMON]

As mentioned in the notes: "Version info: Python3.8, pytorch-lightning 1.1.6, pytorch 1.6.0, cudatoolkit 10.2, Cuda 10.2 (on RTX 2080 Ti) and Ubuntu 18.04.3 LTS (x86_64)"

All below are the same as you， I just want to ask you the version of torch and torchversion， not the PyTorch version my version：Python3.8, pytorch-lightning 1.1.6, pytorch 1.6.0, cudatoolkit 10.2, Cuda 10.2 (on TITAN_X_Pascal) this is my version， and I met this question。

It seems that you are using pytorch 1.4.0; as mentioned in the notes: the pytorch version is "...pytorch 1.6.0..."

[lllllialois]

well, when I use pytorch 1.6.0，

I can't import pointnet2_ops.ext as _ext, it tells me
import pointnet2_ops._ext as _ext
ImportError: /opt/tools/anaconda3/envs/corrnet3d/lib/python3.8/site-packages/pointnet2_ops/_ext.cpython-38-x86_64-linux-gnu.so: undefined symbol: _ZTIN3c1021AutogradMetaInterfaceE
when I look up at https://github.com/erikwijmans/Pointnet2_PyTorch,it said that

so how can I use pytorch1.6.0 to import pointnet2_ops

[ZENGYIMING-EAMON]

Hi! My current code environment (as listed before) cannot reproduce the error you encountered. But I've met this undefined symbol xxx before, my experience is that whenever you reinstall or change your pytorch / pytorch-lightning version, one should re-pip install the pointnet2_ops package (pip install "git+git://github.com/erikwijmans/Pointnet2_PyTorch.git#egg=pointnet2_ops&subdirectory=pointnet2_ops_lib") (or recompile under their official instruction). Another workaround is to substitute group operation using the pure PyTorch version function.

[lllllialois]

Hi! My current code environment (as listed before) cannot reproduce the error you encountered. But I've met this undefined symbol xxx before, my experience is that whenever you reinstall or change your pytorch / pytorch-lightning version, one should re-pip install the pointnet2_ops package

thank you for your advice, and I'm sorry I didn't try your advice until now because I was too busy. Although every time I install the pointnet2_ops package will meet a lot of questions, finally it works!
I can't use git to install pointnet_ops, so I download it and use pip install pointnet2_ops_lib/., but it met this bug.

after I check my CUDA path in ~/.bashrc and the soft link of CUDA, and try export CUDA_HOME=/usr/local/cuda in the terminal. After doing this, I can install pointnet2_ops successfully!

It seems it works!