tensorflow张量详解

zh653

已于 2024-07-11 21:17:51 修改

阅读量13

点赞数

分类专栏： tensorflow 从入门到实战文章标签： neo4j

于 2024-07-11 18:34:05 首次发布

本文链接：https://blog.csdn.net/zh653/article/details/140359130

版权

tensorflow 从入门到实战专栏收录该内容

1 篇文章 0 订阅

订阅专栏

import tensorflow as tf
import numpy as ny

if name == ‘main’:

array_0tank = 5
a_t = tf.constant(array_0tank)
print('a_t is 0 tank that is a scalar: ', a_t)

array_1tank_1 = [2]
b_t = tf.constant(array_1tank_1)
print('b_t is 1 tank that is a vector: ', b_t)

array_1tank_2 = [2, 3, 4]
c_t = tf.constant(array_1tank_2)
print('c_t is 1 tank that is a vector: ', c_t)

array_2tank = [
                  [2,3],
                  [4,5]
              ]
d_t = tf.constant(array_2tank)
print('d_t is 2 tank that is a matrix: ', d_t)

array_3tank = [
            [
                [1,2,3],
                [4,5,6]
            ],
            [
                [1, 2, 3],
                [4, 5, 6]
            ],
            [
                [1, 2, 3],
                [4, 5, 6]
            ]
        ]
e_t = tf.constant(array_3tank)
print("et  is 3 tank that is a rgb pic data: ", e_t)


array_4tank = [
            [
                [
                    [1,2,3],
                    [4,5,6]
                ],
                [
                    [1, 2, 3],
                    [4, 5, 6]
                ],
                [
                    [1, 2, 3],
                    [4, 5, 6]
                ]
           ],
           [
               [
                   [1, 2, 3],
                   [4, 5, 6]
               ],
               [
                   [1, 2, 3],
                   [4, 5, 6]
               ],
               [
                   [1, 2, 3],
                   [4, 5, 6]
               ]
           ]
       ]
f_t = tf.constant(array_4tank)
print("et is 4 tank that is multiple rgb pic datas: ", f_t)

运行结果：
a_t is 0 tank that is a scalar: tf.Tensor(5, shape=(), dtype=int32)
b_t is 1 tank that is a vector: tf.Tensor([2], shape=(1,), dtype=int32)
c_t is 1 tank that is a vector: tf.Tensor([2 3 4], shape=(3,), dtype=int32)
d_t is 2 tank that is a matrix: tf.Tensor(
[[2 3]
[4 5]], shape=(2, 2), dtype=int32)
et is 3 tank that is a rgb pic data: tf.Tensor(
[[[1 2 3]
[4 5 6]]

[[1 2 3]
[4 5 6]]

[[1 2 3]
[4 5 6]]], shape=(3, 2, 3), dtype=int32)
et is 4 tank that is multiple rgb pic datas: tf.Tensor(
[[[[1 2 3]
[4 5 6]]

[[1 2 3]
[4 5 6]]

[[1 2 3]
[4 5 6]]]

[[[1 2 3]
[4 5 6]]

[[1 2 3]
[4 5 6]]

[[1 2 3]
[4 5 6]]]], shape=(2, 3, 2, 3), dtype=int32)

看张量阶数说明
shape=(2, 3, 2, 3) 四个数字就是四节，既四维，多张三通道图片
shape=(3, 2, 3) 三个数字就是三节，既三维，一张三通道图片
shape=(2, 2) 二个数字就是二节，既二维，矩阵单通道图片
shape=(1,) 一个数字就是一节，既一维，向量
shape=() 零个数字就是零节，既零维，标量

也可以通过最外层到最内层括号 ] 的个数判断是几节