python离线安装apex,无NVIDIA版本

问题:Warning: Torch did not find available GPUs on this system.
If your intention is to cross-compile, this is not an error.
By default, Apex will cross-compile for Pascal (compute capabilities 6.0, 6.1, 6.2),
Volta (compute capability 7.0), Turing (compute capability 7.5),
and, if the CUDA version is >= 11.0, Ampere (compute capability 8.0).
If you wish to cross-compile for a single specific architecture,
export TORCH_CUDA_ARCH_LIST="compute capability" before running setup.py.

我的步骤:

1、github上下载完apex包并压缩后开始安装(地址:https://github.com/NVIDIA/apex)复制解压缩后的文件存放地址D:\Conda3\apex-master,如:

 python离线安装apex,无NVIDIA版本_第1张图片

2、进入anaconda powershell,输入命令conda activate pytorch-cpu进入自己的虚拟环境,输入命令cd D:\Conda3\apex-master进入下载的aped包所在目录,如图:

python离线安装apex,无NVIDIA版本_第2张图片

输入命令:python setup.py install 安装apex,这时出现以下提示:

Warning: Torch did not find available GPUs on this system.
If your intention is to cross-compile, this is not an error.
By default, Apex will cross-compile for Pascal (compute capabilities 6.0, 6.1, 6.2),
Volta (compute capability 7.0), Turing (compute capability 7.5),
and, if the CUDA version is >= 11.0, Ampere (compute capability 8.0).
If you wish to cross-compile for a single specific architecture,
export TORCH_CUDA_ARCH_LIST="compute capability" before running setup.py.

原因:我是windows系统,没有NVDIA显卡,没有安装GPU版本的pytorch,我的是cpu版pytorch

解决:运行export TORCH_CUDA_ARCH_LIST="compute capability",如果出现下面的错误:export : 无法将“export”项识别为 cmdlet、函数、脚本文件或可运行程序的名称。请检查名称的拼写,如果包括路径,请确保路径
正确,然后再试一次。
所在位置 行:1 字符: 1
+ export TORCH_CUDA_ARCH_LIST="compute capability"
+ ~~~~~~
    + CategoryInfo          : ObjectNotFound: (export:String) [], CommandNotFoundException
    + FullyQualifiedErrorId : CommandNotFoundException
python离线安装apex,无NVIDIA版本_第3张图片

使用命令:$env:TORCH_CUDA_ARCH_LIST="compute capability"

再次输入:python setup.py install

应该没有什么问题了,输入命令:pip list| findstr apex查看是否有apex

python离线安装apex,无NVIDIA版本_第4张图片

返回pycharm,查看apex应该没有再爆红了,

python离线安装apex,无NVIDIA版本_第5张图片

至此!

你可能感兴趣的:(机器学习环境搭建,python,开发语言)