Ubuntu16.04用python3.6安装并运行tensorflow1.12.0出现“in a future version of numpy”numpy与tensorflow版本冲突问题

Ubuntu16.04+cuda9.0+cudnn7.0+python3.6安装tensorflow1.12.0后运行import tensorflow as tf出现525: FutureWarning: Passing (type, 1) or ‘1type’ as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / ‘(1,)type’.

1.问题描述

基于Ubuntu16.04+cuda9.0+cudnn7.0+python3.6安装tensorflow1.12.0后，运行python3然后输入import tensorflow as tf 出现:

/home/taylorbx/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:523: FutureWarning: Passing (type, 1) or ‘1type’ as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / ‘(1,)type’.
_np_qint8 = np.dtype([(“qint8”, np.int8, 1)])

2.问题分析与解决

注意到提示中有**“in a future version of numpy”，说明是numpy版本过高，与tensorflow1.12.0冲突**

2.1查看numpy版本

pip3 show numpy

发现我用pip3安装tensorflow时默认安装的依赖项numpy版本为1.19.4,版本过高造成冲突，此时只要降低numpy的版本就能解决该问题。

通过实验，目前tensorflow1.12.0配套的numpy较高版本为1.16.0(博主亲测有效)

2.2安装numpy1.16.0

sudo pip3 install numpy==1.16.0

出现上诉图片所示即表示安装成功，再次输入以下命令行查看numpy版本

pip3 show numpy

3.结果演示

在终端输入以下命令

python3
import tensorflow as tf

OK！问题解决
博主第一篇博客，如果对您有帮助的话，非常欢迎您点赞+分享+收藏