# Libtorch\_cpu.so: undefined symbol **URL:** \ **Category:** Uncategorized\ **Created:** [April 17, 2024, 2:00pm UTC](https://discuss.pytorch.org/t/libtorch-cpu-so-undefined-symbol/201101 "2024-04-17T14:00:23Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![Megidd](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/megidd/32/69332_2.png) [@Megidd](https://discuss.pytorch.org/u/Megidd)\ **Post date:** [April 17, 2024, 2:00pm UTC](https://discuss.pytorch.org/t/libtorch-cpu-so-undefined-symbol/201101/1 "2024-04-17T14:00:23Z")
# Minimal env Even a minimal Environment like below would throw similar errors: ```bash conda create -n minimal_pytorch python=3.6 pytorch torchvision torchaudio -c pytorch source activate minimal_pytorch && python -c "import torch; print(torch. __version__ )" ``` A similar `undefined symbol` error is thrown: ```auto Traceback (most recent call last): File "", line 1, in File "/usr/local/envs/minimal_pytorch/lib/python3.6/site-packages/torch/ __init__.py", line 197, in from torch._C import * # noqa: F403 ImportError: /usr/local/envs/minimal_pytorch/lib/python3.6/site-packages/torch/lib/libtorch_cpu.so: undefined symbol: iJIT_NotifyEvent ``` ## Python version When Python version is omitted while creating the environment: ```bash conda create -n minimal_pytorch python pytorch torchvision torchaudio -c pytorch ``` The error is resolved: ```auto source activate minimal_pytorch && python -c "import torch; print(torch. __version__ )" ``` PyTorch version is received without any error: ```auto 2.2.2 ``` I’m kind of confused 🙄 Eventually, I guess I have to use a Docker container as suggested here: > [@Undefined symbol pytorch 1.7.0 and above](https://discuss.pytorch.org/t/undefined-symbol-pytorch-1-7-0-and-above/126640/4): > > The only way I found to make it work is to use a docker container with the required PyTorch version in it. ---
**Author:** ![ptrblck](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/ptrblck/32/1823_2.png) [@ptrblck](https://discuss.pytorch.org/u/ptrblck)\ **Post date:** [April 18, 2024, 4:01am UTC](https://discuss.pytorch.org/t/libtorch-cpu-so-undefined-symbol/201101/2 "2024-04-18T04:01:36Z")
The PyTorch binaries are not built for Python 3.6 anymore, so the first command should install a `torch<=1.10.2` if I’m not mistaken. Use a newer Python version (3.8 - 3.12) and it should work. ---
**Author:** ![yrraadi-io](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/yrraadi-io/32/69647_2.png) [@yrraadi-io](https://discuss.pytorch.org/u/yrraadi-io)\ **Post date:** [April 29, 2024, 4:12pm UTC](https://discuss.pytorch.org/t/libtorch-cpu-so-undefined-symbol/201101/3 "2024-04-29T16:12:58Z")
I’m facing this issue with python 3.9.\*, when installing pytorch via conda. ---
**Author:** ![Megidd](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/megidd/32/69332_2.png) [@Megidd](https://discuss.pytorch.org/u/Megidd)\ **Post date:** [April 30, 2024, 6:36am UTC](https://discuss.pytorch.org/t/libtorch-cpu-so-undefined-symbol/201101/4 "2024-04-30T06:36:25Z")
Eventually, I solved the problem by resolving version mismatches. The exact commands are here: > > > \### Checklist > > \- \[X\] I added a descriptive title > \- \[X\] I searched open report…s and couldn't find a duplicate > > \### What happened? > > \# Minimal env > > A minimal environment like below is created: > > \`\`\`bash > conda create -n minimal\_pytorch python=3.6 pytorch torchvision torchaudio -c pytorch > \`\`\` > > Just getting the PyTorch version by the following command would throw errors: > > \`\`\`bash > source activate minimal\_pytorch && python -c "import torch; print(torch.\_\_version\_\_)" > \`\`\` > > This \`undefined symbol\` error is thrown: > > \`\`\` > Traceback (most recent call last): > File "\", line 1, in \ > File "/usr/local/envs/minimal\_pytorch/lib/python3.6/site-packages/torch/\_\_init\_\_.py", line 197, in \ > from torch.\_C import \* # noqa: F403 > ImportError: /usr/local/envs/minimal\_pytorch/lib/python3.6/site-packages/torch/lib/libtorch\_cpu.so: undefined symbol: iJIT\_NotifyEvent > \`\`\` > > \## Python version > > When Python version is omitted while creating the environment: > > \`\`\`bash > conda create -n minimal\_pytorch python pytorch torchvision torchaudio -c pytorch > \`\`\` > > The error is resolved: > > \`\`\` > source activate minimal\_pytorch && python -c "import torch; print(torch.\_\_version\_\_)" > \`\`\` > > PyTorch version is received without any error: > > \`\`\` > 2.2.2 > \`\`\` > > \## Note > > I have posted a question here too: https://stackoverflow.com/q/78328401/3405291 > > > \### Conda Info > > \`\`\`shell > active environment : minimal\_pytorch > active env location : /usr/local/envs/minimal\_pytorch > shell level : 1 > user config file : /root/.condarc > populated config files : /usr/local/.condarc > conda version : 23.11.0 > conda-build version : not installed > python version : 3.10.13.final.0 > solver : libmamba (default) > virtual packages : \_\_archspec=1=broadwell > \_\_conda=23.11.0=0 > \_\_glibc=2.35=0 > \_\_linux=6.1.58=0 > \_\_unix=0=0 > base environment : /usr/local (writable) > conda av data dir : /usr/local/etc/conda > conda av metadata url : None > channel URLs : https://conda.anaconda.org/conda-forge/linux-64 > https://conda.anaconda.org/conda-forge/noarch > package cache : /usr/local/pkgs > /root/.conda/pkgs > envs directories : /usr/local/envs > /root/.conda/envs > platform : linux-64 > user-agent : conda/23.11.0 requests/2.31.0 CPython/3.10.13 Linux/6.1.58+ ubuntu/22.04.3 glibc/2.35 solver/libmamba conda-libmamba-solver/23.12.0 libmambapy/1.5.5 > UID:GID : 0:0 > netrc file : None > offline mode : False > \`\`\` > > > \### Conda Config > > \`\`\`shell > ==\> /usr/local/.condarc \<== > channels: > - conda-forge > always\_yes: True > \`\`\` > > > \### Conda list > > \`\`\`shell > \# packages in environment at /usr/local/envs/minimal\_pytorch: > \# > \# Name Version Build Channel > \_libgcc\_mutex 0.1 conda\_forge conda-forge > \_openmp\_mutex 4.5 2\_kmp\_llvm conda-forge > blas 2.122 mkl conda-forge > blas-devel 3.9.0 22\_linux64\_mkl conda-forge > bzip2 1.0.8 hd590300\_5 conda-forge > ca-certificates 2024.2.2 hbcca054\_0 conda-forge > cudatoolkit 11.3.1 hb98b00a\_13 conda-forge > dataclasses 0.8 pyh787bdff\_2 conda-forge > ffmpeg 4.3 hf484d3e\_0 pytorch > freetype 2.12.1 h267a509\_2 conda-forge > gmp 6.3.0 h59595ed\_1 conda-forge > gnutls 3.6.13 h85f3911\_1 conda-forge > icu 73.2 h59595ed\_0 conda-forge > jpeg 9e h0b41bf4\_3 conda-forge > lame 3.100 h166bdaf\_1003 conda-forge > lcms2 2.12 hddcbb42\_0 conda-forge > ld\_impl\_linux-64 2.40 h41732ed\_0 conda-forge > lerc 3.0 h9c3ff4c\_0 conda-forge > libblas 3.9.0 22\_linux64\_mkl conda-forge > libcblas 3.9.0 22\_linux64\_mkl conda-forge > libdeflate 1.10 h7f98852\_0 conda-forge > libffi 3.4.2 h7f98852\_5 conda-forge > libgcc-ng 13.2.0 h807b86a\_5 conda-forge > libgfortran-ng 13.2.0 h69a702a\_5 conda-forge > libgfortran5 13.2.0 ha4646dd\_5 conda-forge > libhwloc 2.10.0 default\_h2fb2949\_1000 conda-forge > libiconv 1.17 hd590300\_2 conda-forge > liblapack 3.9.0 22\_linux64\_mkl conda-forge > liblapacke 3.9.0 22\_linux64\_mkl conda-forge > libnsl 2.0.1 hd590300\_0 conda-forge > libpng 1.6.43 h2797004\_0 conda-forge > libsqlite 3.45.2 h2797004\_0 conda-forge > libstdcxx-ng 13.2.0 h7e041cc\_5 conda-forge > libtiff 4.3.0 h0fcbabc\_4 conda-forge > libuv 1.48.0 hd590300\_0 conda-forge > libwebp-base 1.4.0 hd590300\_0 conda-forge > libxml2 2.12.6 h232c23b\_2 conda-forge > libzlib 1.2.13 hd590300\_5 conda-forge > llvm-openmp 18.1.3 h4dfa4b3\_0 conda-forge > mkl 2024.1.0 ha957f24\_692 conda-forge > mkl-devel 2024.1.0 ha770c72\_692 conda-forge > mkl-include 2024.1.0 ha957f24\_692 conda-forge > ncurses 6.4.20240210 h59595ed\_0 conda-forge > nettle 3.6 he412f7d\_0 conda-forge > numpy 1.19.5 py36hfc0c790\_2 conda-forge > olefile 0.46 pyh9f0ad1d\_1 conda-forge > openh264 2.1.1 h780b84a\_0 conda-forge > openjpeg 2.5.0 h7d73246\_0 conda-forge > openssl 1.1.1w hd590300\_0 conda-forge > pillow 8.3.2 py36h676a545\_0 conda-forge > pip 21.3.1 pyhd8ed1ab\_0 conda-forge > python 3.6.15 hb7a2778\_0\_cpython conda-forge > python\_abi 3.6 2\_cp36m conda-forge > pytorch 1.10.2 py3.6\_cuda11.3\_cudnn8.2.0\_0 pytorch > pytorch-mutex 1.0 cuda pytorch > readline 8.2 h8228510\_1 conda-forge > setuptools 58.0.4 py36h5fab9bb\_2 conda-forge > sqlite 3.45.2 h2c6b66d\_0 conda-forge > tbb 2021.12.0 h00ab1b0\_0 conda-forge > tk 8.6.13 noxft\_h4845f30\_101 conda-forge > torchaudio 0.10.2 py36\_cu113 pytorch > torchvision 0.11.3 py36\_cu113 pytorch > typing\_extensions 4.1.1 pyha770c72\_0 conda-forge > wheel 0.37.1 pyhd8ed1ab\_0 conda-forge > xz 5.2.6 h166bdaf\_0 conda-forge > zlib 1.2.13 hd590300\_5 conda-forge > zstd 1.5.5 hfc55251\_0 conda-forge > \`\`\` > > > \### Additional Context > > \_No response\_