张量数据类型#

Data type	dtype	CPU tensor	GPU tensor
32-bit floating point	torch.float32 or torch.float	torch.FloatTensor	torch.cuda.FloatTensor
64-bit floating point	torch.float64 or torch.double	torch.DoubleTensor	torch.cuda.DoubleTensor
16-bit floating point	torch.float16 or torch.half	torch.HalfTensor	torch.cuda.HalfTensor
8-bit integer(unsigned)	torch.uint8	torch.ByteTensor	torch.cuda.ByteTensor
8-bit integer(signed)	torch.int8	torch.CharTensor	torch.cuda.CharTensor
16-bit integer(signed)	torch.int16 or torch.short	torch.ShortTensor	torch.cuda.ShortTensor
32-bit integer(signed)	torch.int32 or torch.int	torch.IntTensor	torch.cuda.IntTensor
64-bit integer(signed)	torch.int64 or torch.long	torch.LongTensor	torch.cuda.LongTensor

类型使用#

示例1：类型比较

1
import torch
2

3
# 随机初始化
4
a = torch.randn(2, 3)
5

6
print("a.type(): ", a.type())
7
print("type(a): ", type(a))
8
print("type of 'a' is torch.FloatTensor: ",
9
      isinstance(a, torch.FloatTensor))

在这里插入图片描述

示例2：同一数据被部署在CPU和GPU上类型不同

1
import torch
2

3
# 随机初始化
4
a = torch.randn(2, 3)
5

6
print(isinstance(a, torch.cuda.FloatTensor))
7
# 如果没有开启cuda会报错
8
a = a.cuda()
9
print(isinstance(a, torch.cuda.FloatTensor))

在这里插入图片描述

示例3：标量

1
import torch
2

3
a = torch.tensor(1.)
4
print("a: ", a)
5

6
b = torch.tensor(1.3)
7
print("b: ", b)

在这里插入图片描述

示例4：标量的shape

1
import torch
2

3
a = torch.tensor(2.2)
4
print("a.shape: ", a.shape)
5
print("len(a.shape): ", len(a.shape))
6
print("a.dim(): ", a.dim())
7
print("a.size(): ", a.size())

在这里插入图片描述

示例5：一维张量

1
import numpy as np
2
import torch
3

4
print(torch.tensor([1.1]))
5
print(torch.tensor([1.1, 2.2]))
6
print(torch.FloatTensor(1))
7
print(torch.FloatTensor(2))
8

9
data = np.ones(2)
10
print("data: ", data)
11
data = torch.from_numpy(data)
12
print("tensor from numpy:", data)
13

14
print("data.shape: ", data.shape)
15
print("data.dim(): ", data.dim())
16
print("data.size(): ", data.size())

在这里插入图片描述

示例6：二位张量

1
import torch
2

3
a = torch.randn(2, 3)
4
print("a:", a)
5
print("a.shape:", a.shape)
6
print("a.size():", a.size())
7
print("a.size(0):", a.size(0))
8
print("a.size(1):", a.size(1))
9
print("a.shape[1]:", a.shape[1])
10
print("a.dim():", a.dim())

在这里插入图片描述

示例7：三维张量

1
import torch
2

3
# 随机均匀分布初始化
4
a = torch.rand(2, 2, 3)
5

6
print("a:", a)
7
print("list(a.shape):", list(a.shape))
8
print("a.size():", a.size())
9
print("a.size(0):", a.size(0))
10
print("a.size(1):", a.size(1))
11
print("a.shape[1]:", a.shape[1])
12
print("a.dim():", a.dim())

在这里插入图片描述

示例8：获得总元素个数

1
import torch
2

3
# 随机均匀分布初始化
4
a = torch.rand(2, 2, 3)
5

6
# 返回a的总元素个数 2 * 2 * 3
7
print("a.numel():", a.numel())

在这里插入图片描述

原文链接：CSDN