首先swin-T要求mmcv版本不能高于1.4.0，所以一些安装就不能像mmcv官网安装的那么随意，首先是查看自己的cuda，安装torch。在安装torch前，应该先参看此种cuda+torch组合下是否有mmcv-full1.4.0的安装包。比如
https://download.openmmlab.com/mmcv/dist/cu113/torch1.11.0/index.html 中发现并没有1.4.0版本的mmcv-full起步版本都已经1.4.7了。如果此时强行安装，
pip install mmcv-full1.4.0 -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.11.0/index.html
会出出现卡在
Building wheels for collected packages: mmcv-full
Building wheel for mmcv-full (setup.py) … |
然后报红字

ERROR: Command errored out with exit status 1:
     command: /opt/conda/envs/py38/bin/python -u -c 'import io, os, sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-2psi4unb/mmcv-full_b11880bea60e40aaad9c911705dc22a7/setup.py'"'"'; __file__='"'"'/tmp/pip-install-2psi4unb/mmcv-full_b11880bea60e40aaad9c911705dc22a7/setup.py'"'"';f = getattr(tokenize, '"'"'open'"'"', open)(__file__) if os.path.exists(__file__) else io.StringIO('"'"'from setuptools import setup; setup()'"'"');code = f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' install --record /tmp/pip-record-4h9_exl4/install-record.txt --single-version-externally-managed --compile --install-headers /opt/conda/envs/py38/include/python3.8/mmcv-full

此时应该下拉mmcv官网上，看看cuda11.3这一行其他torch版本的网址下是否有，https://download.openmmlab.com/mmcv/dist/cu113/torch1.10.0/index.html，发现torch10.0是有的，于是选择cuda11.3+torch10.0的组合并进行安装