apex常见报错,亲测可用
环境:torch1.7.1 python3.8 ,1080Ti/2080Ti/3090显卡均适用。结尾有一键解决方案。
主要针对问题:
- AttributeError: module ‘torch.distributed’ has no attribute ‘_all_gather_base’
- ImportError: cannot import name ‘amp’ from ‘apex’ (unknown location)
参考:https://blog.csdn.net/weixin_59726951/article/details/127585754
git clone https://github.com/ptrblck/apex.git
cd apex
git checkout apex_no_distributed
pip install -v --no-cache-dir ./
验证import apex成功
参考:https://blog.csdn.net/qq_17783559/article/details/127813381
git clone https://github.com/NVIDIA/apex.git
cd apex
git checkout f3a960f80244cf9e80558ab30f7f7e8cbf03c0a0
python setup.py install --cuda_ext --cpp_ext
报错
之后把后面的 --cuda_ext --cpp_ext去掉
安装成功,但运行代码报错
from apex import amp
- ImportError: cannot import name ‘amp’ from ‘apex’ (unknown location)
参考:https://blog.csdn.net/zjc910997316/article/details/110490156
git clone git://github.com/NVIDIA/apex(我之前下载了,用的之前的包)
cd apex
pip install -v --no-cache-dir ./
cd apex
pip install -v --no-cache-dir ./
成功
总结:一键解决
git clone https://github.com/NVIDIA/apex.git
cd apex
git checkout f3a960f80244cf9e80558ab30f7f7e8cbf03c0a0
python setup.py install --cuda_ext --cpp_ext
pip install -v --no-cache-dir ./
或
git clone https://github.com/NVIDIA/apex.git
cd apex
git checkout f3a960f80244cf9e80558ab30f7f7e8cbf03c0a0
python setup.py install
pip install -v --no-cache-dir ./