tf.reduce系列中axis是几就是对第几列处理

最新推荐文章于 2023-12-30 11:17:13 发布

杨羊不是羊

最新推荐文章于 2023-12-30 11:17:13 发布

阅读量218

点赞数

分类专栏： Python笔记文章标签： tensorflow

本文链接：https://blog.csdn.net/YyangWwei/article/details/119956306

版权

Python笔记专栏收录该内容

10 篇文章 1 订阅

订阅专栏

import numpy as np

In [4]: np.random.randn(2,3)                                                                                                                                                                                                              
Out[4]: 
array([[ 0.35891946,  1.21695806,  0.90310248],
       [-1.07711563, -0.23622112,  0.05407225]])

#numpy 转tensor
a = tf.convert_to_tensor(np.random.randn(2,3))

In [8]: a                                                                                                                                                                                                                                 
Out[8]: 
<tf.Tensor: shape=(2, 3), dtype=float64, numpy=
array([[ 0.73775397, -1.09520102, -1.60098177],
       [ 0.32172983, -1.28484503,  0.47187442]])>
    
In [9]: tf.reduce_mean(a,axis = 1)                                                                                                                                                                                                        
Out[9]: <tf.Tensor: shape=(2,), dtype=float64, numpy=array([-0.6528096 , -0.16374693])>  
    
 a = tf.convert_to_tensor(np.random.randn(2,3,4))                                                                                                                                                                                 

In [11]: a                                                                                                                                                                                                                                
Out[11]: 
<tf.Tensor: shape=(2, 3, 4), dtype=float64, numpy=
array([[[-2.4349334 ,  0.18069842,  0.20173453,  1.60069144],
        [-0.62254852,  0.27000785, -1.54304289, -0.22552647],
        [ 0.54879563, -0.52025813,  1.11932352,  0.095837  ]],

       [[-0.55804347,  0.34590774,  1.40386632, -0.25984318],
        [-2.30894868, -0.26656274, -0.39123609, -0.6094307 ],
        [-1.14524866,  0.47452726,  2.67715933,  0.15349671]]])>
#axis = 0 2*3*4 变成 1*3*4   
In [12]: tf.reduce_mean(a,axis = 0)                                                                                                                                                                                                       
Out[12]: 
<tf.Tensor: shape=(3, 4), dtype=float64, numpy=
array([[-1.49648843e+00,  2.63303079e-01,  8.02800424e-01,
         6.70424134e-01],
       [-1.46574860e+00,  1.72255673e-03, -9.67139492e-01,
        -4.17478587e-01],
       [-2.98226516e-01, -2.28654341e-02,  1.89824143e+00,
         1.24666860e-01]])>    
#axis = 2 2*3*4 变成 2*3*1    
In [13]: tf.reduce_mean(a,axis = 2)                                                                                                                                                                                                       
Out[13]: 
<tf.Tensor: shape=(2, 3), dtype=float64, numpy=
array([[-0.11295225, -0.53027751,  0.31092451],
       [ 0.23297185, -0.89404455,  0.53998366]])>