[Bug] libcudnn_cnn_train.so.8: undefined symbol: _ZTIN10cask_cudnn14BaseKernelInfoE

[bigmover]

Your current environment information

Run onelow_compile 出现报错。请问应该如何能解决？

return _bootstrap._gcd_import(name[level:], package, level)
File "", line 1030, in _gcd_import
File "", line 1007, in _find_and_load
File "", line 986, in _find_and_load_unlocked
File "", line 680, in _load_unlocked
File "", line 850, in exec_module
File "", line 228, in _call_with_frames_removed
File "/home/admin/miniconda3/envs/sd_removal/lib/python3.9/site-packages/onediff-1.2.1.dev19+g6fe6fdde-py3.9.egg/onediff/infer_compiler/backends/oneflow/init.py", line 3, in
from .deployable_module import OneflowDeployableModule
File "/home/admin/miniconda3/envs/sd_removal/lib/python3.9/site-packages/onediff-1.2.1.dev19+g6fe6fdde-py3.9.egg/onediff/infer_compiler/backends/oneflow/deployable_module.py", line 6, in
import oneflow as flow # usort: skip
File "/home/admin/miniconda3/envs/sd_removal/lib/python3.9/site-packages/oneflow/init.py", line 25, in
import oneflow._oneflow_internal
ImportError: /home/admin/miniconda3/envs/sd_removal/lib/python3.9/site-packages/oneflow/../nvidia/cudnn/lib/libcudnn_cnn_train.so.8: undefined symbol: _ZTIN10cask_cudnn14BaseKernelInfoE, version libcudnn_cnn_infer.so.8

🐛 Describe the bug

[Metal079]

I get a similar issue using nexfort in wsl

[CuddleSabe]

这不是bug，这是你的环境问题