Python Pytorch empty() 方法

PyTorch 是 Facebook 开发的开源机器学习库。它用于深度神经网络和自然语言处理目的。

函数torch.empty()返回一个充满未初始化数据的张量。张量的形状由可变参数大小定义。

Syntax: torch.empty(size, out=None)

Parameters:
size: a sequence of integers defining the shape of the output tensor
out (Tensor, optional): the output tensor

Return type: A tensor

代码#1：

# Importing the PyTorch library
import torch
  
  
# Applying the empty function and
# storing the resulting tensor in 'a'
a = torch.empty([3, 4])
print("a = ", a)
  
b = torch.empty([2, 3])
print("b = ", b)

输出：

a =  tensor([[2.0283e-19, 6.9772e+22, 1.8943e+23, 1.1432e-32],
        [1.3563e-19, 1.8888e+31, 8.9066e-15, 1.8888e+31],
        [6.4639e-04, 6.8608e+22, 1.7753e+28, 2.0535e-19]])
b =  tensor([[0., 0., 0.],
        [0., 0., 0.]])