AWS SageMaker安装llama-cpp-python失败：构建wheel出错

问题场景 #

使用AWS SageMaker的ml.g5.2xlarge或ml.g4dn.2xlarge实例运行LLaMA模型（规避Colab Pro 15GB显存限制），但执行!pip install llama-cpp-python==0.1.48时出现编译wheel失败错误：

Building wheel for llama-cpp-python (pyproject.toml) did not run successfully.
exit code: 1 [130 lines 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

解决方案 #

1. 安装编译依赖工具
   SageMaker默认环境缺少编译C++扩展的基础库，先执行：

   !sudo apt-get update && sudo apt-get install -y build-essential cmake libopenblas-dev libomp-dev
   

2. 启用CUDA支持编译
   llama-cpp-python需要显式开启CUDA才能适配SageMaker的GPU实例，设置环境变量后再安装：

   !CMAKE_ARGS="-DLLAMA_CUBLAS=on" FORCE_CMAKE=1 pip install llama-cpp-python==0.1.48
   

3. 升级pip及构建工具（可选）
   如果上述步骤仍失败，先升级pip、wheel和setuptools再重试：

   !pip install --upgrade pip wheel setuptools
   

   之后重新执行步骤2的安装命令。

原因说明 #

• SageMaker Notebook默认未预装编译C扩展所需的build-essential、cmake等工具，导致无法编译llama-cpp-python的底层C代码。
• 未显式启用CUDA支持时，编译过程无法适配GPU实例的硬件环境，进而触发编译失败。

内容的提问来源于stack exchange，提问作者ku11