–
llama-cpp-python安装bug：error: subprocess-exited-。scikit-build-core 0.8.2 using CMake 3.28.3 (wheel)
环境
系统
linux
python 3.10
llama_cpp_python==0.2.55
1 Bug详情
# 导致安装错误的命令
pip install llama-cpp-python
错误信息
Building wheel for llama-cpp-python (pyproject.toml) ... error
error: subprocess-exited-with-error
× Building wheel for llama-cpp-python (pyproject.toml) did not run successfully.
│ exit code: 1
╰─> [325 lines of output]
*** scikit-build-core 0.8.2 using CMake 3.28.3 (wheel)
*** Configuring CMake...
2024-03-04 17:29:03,943 - scikit_build_core - WARNING - libdir/ldlibrary: /home/zmp/.conda/envs/base_torch_201cu118_cp310/lib/libpython3.10.a is not a real file!
2024-03-04 17:29:03,943 - scikit_build_core - WARNING - Can't find a Python library, got libdir=/home/zmp/.conda/envs/base_torch_201cu118_cp310/lib, ldlibrary=libpython3.10.a, multiarch=x86_64-linux-gnu, masd=None
loading initial cache file /tmp/tmpcjlys8ws/build/CMakeInit.txt
-- The C compiler identification is GNU 4.8.5
-- The CXX compiler identification is GNU 4.8.5
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/c++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found Git: /usr/bin/git (found version "1.8.3.1")
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Failed
-- Check if compiler accepts -pthread
-- Check if compiler accepts -pthread - yes
-- Found Threads: TRUE
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with LLAMA_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: x86_64
-- x86 detected
CMake Warning (dev) at CMakeLists.txt:21 (install):
Target llama has PUBLIC_HEADER files but no PUBLIC_HEADER DESTINATION.
This warning is for project developers. Use -Wno-dev to suppress it.
CMake Warning (dev) at CMakeLists.txt:30 (install):
Target llama has PUBLIC_HEADER files but no PUBLIC_HEADER DESTINATION.
This warning is for project developers. Use -Wno-dev to suppress it.
-- Configuring done (1.1s)
-- Generating done (0.0s)
-- Build files have been written to: /tmp/tmpcjlys8ws/build
*** Building project with Ninja...
Change Dir: '/tmp/tmpcjlys8ws/build'
......
/tmp/pip-install-qpsadbts/llama-cpp-python_6fb3cbcdc9d540c5b4fb53a5f6f8cd97/vendor/llama.cpp/ggml-quants.c:9656:13: 错误：隐式声明函数‘_mm256_set_m128i’ [-Werror=implicit-function-declaration]
const __m256i full_signs_1 = _mm256_set_m128i(full_signs_l, full_signs_l);
^
/tmp/pip-install-qpsadbts/llama-cpp-python_6fb3cbcdc9d540c5b4fb53a5f6f8cd97/vendor/llama.cpp/ggml-quants.c:9656:42: 错误：用‘int’初始化‘__m256i’时类型不兼容
const __m256i full_signs_1 = _mm256_set_m128i(full_signs_l, full_signs_l);
^
/tmp/pip-install-qpsadbts/llama-cpp-python_6fb3cbcdc9d540c5b4fb53a5f6f8cd97/vendor/llama.cpp/ggml-quants.c:9657:42: 错误：用‘int’初始化‘__m256i’时类型不兼容
const __m256i full_signs_2 = _mm256_set_m128i(full_signs_h, full_signs_h);
.......
ninja: build stopped: subcommand failed.
*** CMake build failed
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for llama-cpp-python
Failed to build llama-cpp-python
ERROR: Could not build wheels for llama-cpp-python, which is required to install pyproject.toml-based projects
错误原因：使用pip install llama-cpp-python安装时，是通过下载源码编译安装的（llama_cpp_python-0.2.55.tar.gz (36.8 MB)）。这时候如果系统没有相应的cmake 和 gcc版本，会弹出这个错误。
2 解决方案
根据系统选择官方编译后的whl下载进行离线安装。
官方网址
Releases · abetlen/llama-cpp-python (github.com)
安装命令例子
pip install llama_cpp_python-0.2.55-cp310-cp310-manylinux_2_17_x86_64.whl
注：官方只是编译了常见的系统，特殊系统还是需要自行编译。
一般不推荐更新公用linux系统的cmake和gcc进行安装，可以通过建立docker镜像，安装相应版本的cmake和gcc（gcc应高于11.0.0，可以为11.3.0）。查看gcc版本：gcc --version