windows 11 install faild

[summerghw]

my install code :

$env:CMAKE_ARGS = "-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS"
$env:CMAKE_GENERATOR = "MinGW Makefiles"
$env:CMAKE_ARGS = "-DLLAMA_OPENBLAS=on -DCMAKE_C_COMPILER=C:/w64devkit/bin/gcc.exe -DCMAKE_CXX_COMPILER=C:/w64devkit/bin/g++.exe" 
pip install llama-cpp-python

and i meet a error :

 cd /d C:\Users\JHHS-KY-001\AppData\Local\Temp\tmpbqomzf8d\build\vendor\llama.cpp\examples\llava && "C:\Program Files\CMake\bin\cmake.exe" -E cmake_link_script CMakeFiles\llava-cli.dir\link.txt --verbose=1
      "C:\Program Files\CMake\bin\cmake.exe" -E rm -f CMakeFiles\llava-cli.dir/objects.a
      C:\Users\JHHS-KY-001\Downloads\w64devkit\bin\ar.exe qc CMakeFiles\llava-cli.dir/objects.a @CMakeFiles\llava-cli.dir\objects1.rsp       
      C:\Users\JHHS-KY-001\Downloads\w64devkit\bin\c++.exe -O3 -DNDEBUG -Wl,--whole-archive CMakeFiles\llava-cli.dir/objects.a -Wl,--no-whole-archive -o llava-cli.exe -Wl,--out-implib,libllava-cli.dll.a -Wl,--major-image-version,0,--minor-image-version,0 @CMakeFiles\llava-cli.dir\l      C:\Users\JHHS-KY-001\Downloads\w64devkit\bin/ld.exe: C:/Users/JHHS-KY-001/Downloads/w64devkit/bin/../lib/gcc/x86_64-w64-mingw32/13.2.0/../../../../x86_64-w64-mingw32/lib/../lib/libpthread.a(libwinpthread_la-thread.o):thread.c:(.text+0x103f): multiple definition of `pthread_self'; ../../libllama.dll.a(libllama_dll_d000820.o):(.text+0x0): first defined here
      collect2.exe: error: ld returned 1 exit status
      make[2]: *** [vendor\llama.cpp\examples\llava\CMakeFiles\llava-cli.dir\build.make:105: vendor/llama.cpp/examples/llava/llava-cli.exe] Error 1
      make[2]: Leaving directory 'C:/Users/JHHS-KY-001/AppData/Local/Temp/tmpbqomzf8d/build'
      make[1]: *** [CMakeFiles\Makefile2:388: vendor/llama.cpp/examples/llava/CMakeFiles/llava-cli.dir/all] Error 2
      make[1]: Leaving directory 'C:/Users/JHHS-KY-001/AppData/Local/Temp/tmpbqomzf8d/build'
      make: *** [Makefile:135: all] Error 2

[ceresOPA]

I also met this problem.

[ygoron360]

I have the same issue

[summerghw]

I use docker instead，it works well， hope this will help。

[ygoron360]

I was able to install it using older build : pip install llama-cpp-python==v0.2.14 --verbose

Any build newer that 0.2.14 give me the same error

[gazoonc]

This worked for me in Windows 10:

$env:Path += ';C:/Users/gazoonc/projects/myenv/w64devkit/bin/'
$env:CMAKE_GENERATOR = "MinGW Makefiles"
$env:CMAKE_ARGS = "-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS -DLLAMA_WIN_VER=0x400 -DCMAKE_MAKE_PROGRAM=mingw32-make -DCMAKE_C_COMPILER=C:/Users/gazoonc/projects/myenv/w64devkit/bin/gcc.exe -DCMAKE_CXX_COMPILER=C:/Users/gazoonc/projects/myenv/w64devkit/bin/g++.exe" 
pip install llama-cpp-python==v0.2.14 --verbose

[Trojaner]

Revert CMakeLists.txt inside llama-cpp-python (not llama.cpp!) to this version to fix it:
https://github.com/abetlen/llama-cpp-python/blob/b7e60b66f47950e385980a1329af9dfb14da6906/CMakeLists.txt

Alternatively you can add -DLLAVA_BUILD=OFF to CMAKE_ARGS instead

Obviously both will remove llava support.

This commit broke it:
aab74f0