Repository navigation
Failed to load shared library #208
Description
I compiled and installed the latest code (0.1.5) in Windows 11 with Python 3.10.11 and CUDA 11.8. The compilation with cuBLAS flag and installation were successful. When importing the module via from llama_cpp import Llama, python reported
Failed to load shared library 'C:\Users\teristam\Anaconda3\envs\textgen\Lib\site-packages\llama_cpp\llama.dll':
Could not find module 'C:\Users\teristam\Anaconda3\envs\textgen\Lib\site-packages\llama_cpp\llama.dll' (or one of its dependencies). Try using the full path with constructor syntax.
But when I change
llama-cpp-python/llama_cpp/llama_cpp.py
Line 55 in d90c9df
to ctypes.CDLL(str(_lib_path),winmode=0), the dll is able to load succesfully. It may be related to a bug in ctype in Windows:
https://stackoverflow.com/questions/59330863/cant-import-dll-module-in-python/64472088#64472088
https://bugs.python.org/issue42114
Activity
abetlen commentedon May 15, 2023
@teristam thanks for the bug report / solution, I'll update it so
winmode=0. Are there any side-effects?aneeshjoy commentedon May 17, 2023
The winmod=0 alone did not fix this issue for me. I had add CUDA_PATH to dll dir. #225
Update to current status (#2)
Fix obscure Wndows DLL issue. Closes abetlen#208