search

cszhangzhen / HGP-SL

Hierarchical Graph Pooling with Structure Learning
326 stars 54 forks source link

Segmentation error #2

Closed zaixizhang closed 4 years ago

zaixizhang commented 4 years ago

Hi, I run the code with"python main.py", but it outputs"segmentation error". How can I fix it?

cszhangzhen commented 4 years ago

Hi,

Thanks for your feedback.

The codes run successfully with pyCharm IDE. However, it indeed occurs segmentation error with command line.

I use gdb to debug the code and find that the error is related the sip package. See information below:

(gdb) file python Reading symbols from python...done. (gdb) run main.py Starting program: /.conda/envs/py36/bin/python main.py [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".

Program received signal SIGSEGV, Segmentation fault. PyType_IsSubtype (a=0x626f6174656d5f74, b=0x1554efed8660) at /tmp/build/80754af9/python_1540319457073/work/Objects/typeobject.c:1357 1357 /tmp/build/80754af9/python_1540319457073/work/Objects/typeobject.c: No such file or directory. (gdb) bt PyType_IsSubtype (a=0x626f6174656d5f74, b=0x1554efed8660) at /tmp/build/80754af9/python_1540319457073/work/Objects/typeobject.c:1357 0x00001554efcce0cf in ?? () from /.conda/envs/py36/lib/python3.6/site-packages/sip.so 0x00001554efcce935 in ?? () from /.conda/envs/py36/lib/python3.6/site-packages/sip.so

Solution : Similar to this issue, update your sip package with the following command: conda install -c conda-forge sip=4.18.1

After updating your sip package, the codes can successfully run with both pyCharm IDE and command line.

zaixizhang commented 4 years ago

Thanks, I have fixed the segementation error by updating pytorch and reinstalling pyG packages. However, I got another problem. When I run"python main.py", another error occured at the 31th line of layers.py. It said coalesce got one unexpected keyword argument "fill_value". Can I just delete this arguement?

cszhangzhen commented 4 years ago

Hi,

This is caused by the updating of torch-sparse.

I have updated the code in layers.py. Please download the code and run again.

zaixizhang commented 4 years ago

Thanks for your reply! I can run the code successfully now.