Pytorch基本数据类型

一、数据类型对比

Python	PyTorch
Int	IntTensor of size()
float	FloatTensor of size()
Int array	IntTensor of size [d1,d2,…]
float array	FloatTensor of size [d1,d2,…]
string	–

二、如何表达string

1. One-hot in coding
   1. 例如:dog : [1 0],cat :[0 1]
   2. 对于复杂的编码来说，用One-hot进行编码会变得十分稀疏
2. Embedding(待补)
   1. Word2vec
   2. glove

三、PyTorch数据类型

Data type	dtype	CPU tensor	GPU tensor
32位float	torch.float32 or torch.float	torch.FloatTensor	torch.cuda.FloatTensor
64位float	torch.float64 or torch.double	torch.DoubleTensor	torch.cuda.DoubleTensor
32位int(signed)	torch.int32 or torch.int	torch.IntTensor	torch.cuda.IntTensor
64位int(signed)	torch.float64 or torch.long	torch.LongTensor	torch.cuda.LongTensor
8为int(unsigned)	torch.uint8	torch.ByteTensor	torch.cuda.ByteTensor

四、Tensor常用函数

1. a.type():返回数据类型

a = torch.rand(2,3) # 随机生成2行3列数据
print(a.type()) # 输出为torch.FloatTensor

2. isinstance(obj,type):检查数据类型

#检查数据类型是否为FloatTensor
a = rand(1)
print(isinstance(a,torch.FloatTensor)) # true
a = a.cuda() #在gpu上跑数据
print(isinstance(a,torch.cuda.FloatTensor)) # true

3. 标量dim=0
   1. torch.tensor():生成标量
   2. obj.shape:返回obj的维度大小
   3. len(obj.shape):返回dim
   4. obj.size():返回元素个数

#检查数据类型是否为FloatTensor
b = torch.tensor(1.3)
print(b.shape) #torch.Size([])
print(len(b.shape)) #0
print(b.size()) #torch.Size([])

4. 向量dim=1
   1. torch.tensor([...]):生成标量
   2. np.ones(shape):生成全为1的向量

c = torch.tensor([1.1])
print(c) # tensor([1.1000])

d = torch.tensor([1.1,2.2])
print(d) # tensor([1.1000, 2.2000])

e = torch.FloatTensor(1) #random初始化
print(e) # tensor([0.])

f = torch.FloatTensor(5)
print(f) # tensor([0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00, 1.4013e-45])

g = np.ones(2)
print(g) # [1. 1.]
print(torch.from_numpy(g)) # tensor([1., 1.], dtype=torch.float64)

h = torch.ones(2)
print(h.shape) # torch.Size([2])

5. dim=2
   1. torch.rand(h,w):生成行h,列w
   2. size():0行,1列
   3. shape[]:返回维度的个数

i = torch.rand(2,3)
print(i) # tensor([[0.2588, 0.7782, 0.3236],
        #  [0.4545, 0.5833, 0.1202]])
print(i.shape) # torch.Size([2, 3])
print(i.size(0)) # 2
print(i.size(1)) # 3
print(i.shape[0]) # 2
print(i.shape[1]) # 3

6. dim=3
   1. torch.rand(a,b,c)
   2. obj.numel():返回内存大小
   3. obj.dim():返回dim

j = torch.rand(1,2,3)
print(j)
# tensor([[[0.1600, 0.7410, 0.5799],
#         [0.9968, 0.0704, 0.9750]]])

print(j.shape) # torch.Size([1, 2, 3])
print(j[0])
# tensor([[0.1600, 0.7410, 0.5799],
#        [0.9968, 0.0704, 0.9750]])

print(list(j.shape)) # [1, 2, 3]

print(j.numel()) # 内存大小 = 6
print(j.dim()) #返回dim = 3