📜  Tensorflow.js tf.dispose()函数

📅  最后修改于: 2022-05-13 01:56:45.551000             🧑  作者: Mango

Tensorflow.js tf.dispose()函数

Tensorflow.js 是谷歌大脑团队开发的一个开源库,用于在浏览器或节点环境中运行深度学习神经网络。

.dispose()函数用于将张量释放到内存中。

句法:

tensor.dispose()

参数:此方法没有任何参数

返回值:无效

示例 1:在此示例中,我们将使用 tf.dispose函数处理一个张量。

Javascript
// Importing the tensorflow.Js library
import * as tf from "@tensorflow/tfjs"
 
// Creating the tensor
var tr = tf.tensor([1, 2, 3, 4, 5, 6, 7]);
 
// Disposing the tensor
tr.dispose()
 
// Trying to print it now
tr.print()


Javascript
// Importing the tensorflow.Js library
import * as tf from "@tensorflow/tfjs"
 
// Create a new tensor
var tr2d = tf.tensor2d([1, 2, 3, 4], [2, 2]);
 
// Print the tensor
tr2d.print()
 
// Dispose the tensor
tr2d.dispose()
 
// Try to print it again
tr2d.print()


Javascript
// Importing the tensorflow.Js library
import * as tf from "@tensorflow/tfjs"
  
// Creating a value variable which
// stores the value
var value = ['1', '2', '3', '4', '5', '6']
  
// Creating a shape variable
// which stores the shape
var shape = [2, 3]
  
// Creating a d_Type variable
// which stores the data-type
var d_Type = 'string'
  
// Creating the tensor
var tr3d = tf.tensor(value, shape, d_Type)
  
// Printing the tensor
tr3d.print()


输出:

An error occurred 
  Tensor is disposed

示例 2:在此示例中,我们将创建一个 rank-2 张量并尝试使用 tf.dispose函数打印到前后。

Javascript

// Importing the tensorflow.Js library
import * as tf from "@tensorflow/tfjs"
 
// Create a new tensor
var tr2d = tf.tensor2d([1, 2, 3, 4], [2, 2]);
 
// Print the tensor
tr2d.print()
 
// Dispose the tensor
tr2d.dispose()
 
// Try to print it again
tr2d.print()

输出:

"Tensor
    [[1, 2],
     [3, 4]]" 

示例 3:在此示例中,我们正在创建一个具有 value、shape 和 dataType 的张量。然后,我们将使用 tf.dispose()函数处理它:

Javascript

// Importing the tensorflow.Js library
import * as tf from "@tensorflow/tfjs"
  
// Creating a value variable which
// stores the value
var value = ['1', '2', '3', '4', '5', '6']
  
// Creating a shape variable
// which stores the shape
var shape = [2, 3]
  
// Creating a d_Type variable
// which stores the data-type
var d_Type = 'string'
  
// Creating the tensor
var tr3d = tf.tensor(value, shape, d_Type)
  
// Printing the tensor
tr3d.print()

输出:

"Tensor
    [['1', '2', '3'],
     ['4', '5', '6']]" 

参考: https://js.tensorflow.org/api/latest/#dispose