pytorchデータ型
2017 ワード
import torch
'''
pytorch GPU , GPU
pytorch python , CUDA GPU
opencv-python numpy GPU/CUDA
'''
a=torch.randn(2,3)
print(a.type())#torch.FloatTensor
print(type(a))#
'''
torch.Tensor.type() pytorch tensor ,
type(a) python , ,
tensor.type()
'''
'''
pytorch torch.Tensor
pytorch , /
torch.Tensor
'''
print(a.dtype)#torch.float32
print(a.device)#cpu pytorch tensor CPU GPU
print(isinstance(a,torch.FloatTensor))#True
'''
isinstance
'''
# print(isinstance(a,torch.cuda.FloatTensor))
# a=a.cuda()
# print(isinstance(a,torch.cuda.FloatTensor))
'''
pytorch :dimension=0 torch.tensor :loss , torch.tensor
prediciton target
torch.tensor dimension=0
train.py
loss=loss.item()
tensor
'''
a=torch.tensor(1.2)
print(a.shape)
print(len(a.shape))
print(a.size())
'''
tensor.shape shape
tensor.size() size()
'''
'''
pytorch dimension=1
(1)bias
bias dimension=1 tensor
(2)linear input
linear regression , one-dimension
torch.tensor tensor
torch.Tensor tensor ( )
'''
a=torch.tensor(1.325)
b=torch.Tensor(1)
c=torch.FloatTensor(2)
import numpy as np
data=np.ones(2)
d=torch.from_numpy(data)
#numpy.ndarray torch.tensor
e=torch.randn(2,3)
data2=e.numpy()
#torch.tensor numpy
print('a',a)
print('b',b)
print('c',c)
print('d',d)
print('data2',data2,type(data2))
'''
a tensor(1.3250)
b tensor([5.6052e-45])
c tensor([9.1365e-09, 4.5909e-41])
d tensor([1., 1.], dtype=torch.float64)
data2 [[-1.6662962 -0.4791848 -0.43611825]
[-0.99609876 0.20896369 -0.7278052 ]]
'''