Python35环境下TensorFlow101配置环境
近几日,考古了之前TensorFlow1.0.1环境下的coding,把其中步骤记录一下。
-
Conda Python3.5.6环境;
-
举例安装pip install chicksexer。
如遇报错error: Microsoft Visual C++ 14.0 is required. Get it with "Microsoft Visual C++ Build Tools": https://visualstudio.microsoft.com/downloads/
,可对报错安装包使用离线whl进行安装。
举例如遇regex报错VC14,可前去https://www.lfd.uci.edu/~gohlke/pythonlibs/#regex 进行检索,下载对应python版本文件进行安装即可。
pip install "C:\Users\Martin\Downloads\regex-2020.1.8-cp35-cp35m-win_amd64.whl"
- 再次尝试安装pip install chicksexer。
如遇报错UserWarning: Numpy 1.13.3 or above is required for this version of scipy
,或者报错scipy 1.4.1 has requirement numpy>=1.13.3, but you'll have numpy 1.12.1 which is incompatible.
,升级安装numpy至>1.13.3以上即可,如numpy1.18.5。或直接卸载scipy,重新安装scipy==1.4.1即可同时适配numpy。
- 如遇sklearn 0.18报错
cannot import name comb
,修改如下:
# 因为scipy.misc中的comb位置已经移到scipy.special,所以修改对应的scipy.misc为scipy.special即可。
tensorflow\lib\site-packages\sklearn\metrics\cluster\supervised.py
tensorflow\lib\site-packages\sklearn\model_selection\_split.py
参考链接:
https://my.oschina.net/u/4407946/blog/3478235
https://www.lfd.uci.edu/~gohlke/pythonlibs/#regex
https://blog.csdn.net/qq_41549459/article/details/99062459
https://blog.csdn.net/qq_40644291/article/details/106915733