什么是 PyTorch?
PyTorch 是一个基于 Python 的科学计算包,主要定位两类人群:- NumPy 的替代品,可以利用 GPU 的性能进行计算。
- 深度学习研究平台拥有足够的灵活性和速度
开始学习
Tensors (张量)
Tensors 类似于 NumPy 的 ndarrays ,同时 Tensors 可以使用 GPU 进行计算。- from future import printfunction
- import torch
- x = torch.empty(5, 3)
- print(x)
输出:
- tensor(1.00000e-04 *
- [[-0.0000, 0.0000, 1.5135],
- [ 0.0000, 0.0000, 0.0000],
- [ 0.0000, 0.0000, 0.0000],
- [ 0.0000, 0.0000, 0.0000],
- [ 0.0000, 0.0000, 0.0000]])
- x = torch.rand(5, 3)
- print(x)
输出:
- tensor([[ 0.6291, 0.2581, 0.6414],
- [ 0.9739, 0.8243, 0.2276],
- [ 0.4184, 0.1815, 0.5131],
- [ 0.5533, 0.5440, 0.0718],
- [ 0.2908, 0.1850, 0.5297]])
- x = torch.zeros(5, 3, dtype=torch.long)
- print(x)
输出:
- tensor([[ 0, 0, 0],
- [ 0, 0, 0],
- [ 0, 0, 0],
- [ 0, 0, 0],
- [ 0, 0, 0]])
- x = torch.tensor([5.5, 3])
- print(x)
输出:
- tensor([ 5.5000, 3.0000])
- x = x.new_ones(5, 3, dtype=torch.double)
# new* methods take in sizes - print(x)
-
- x = torch.randnlike(x, dtype=torch.float)
# override dtype! - print(x)
# result has the same size
输出:
- tensor([[ 1., 1., 1.],
- [ 1., 1., 1.],
- [ 1., 1., 1.],
- [ 1., 1., 1.],
- [ 1., 1., 1.]], dtype=torch.float64)
- tensor([[-0.2183, 0.4477, -0.4053],
- [ 1.7353, -0.0048, 1.2177],
- [-1.1111, 1.0878, 0.9722],
- [-0.7771, -0.2174, 0.0412],
- [-2.1750, 1.3609, -0.3322]])
- print(x.size())
输出:
- torch.Size([5, 3])
注意
torch.Size
是一个元组,所以它支持左右的元组操作。
操作
在接下来的例子中,我们将会看到加法操作。 加法: 方式 1- y = torch.rand(5, 3)
- print(x + y)
Out:
- tensor([[-0.1859, 1.3970, 0.5236],
- [ 2.3854, 0.0707, 2.1970],
- [-0.3587, 1.2359, 1.8951],
- [-0.1189, -0.1376, 0.4647],
- [-1.8968, 2.0164, 0.1092]])
- print(torch.add(x, y))
Out:
- tensor([[-0.1859, 1.3970, 0.5236],
- [ 2.3854, 0.0707, 2.1970],
- [-0.3587, 1.2359, 1.8951],
- [-0.1189, -0.1376, 0.4647],
- [-1.8968, 2.0164, 0.1092]])
- result = torch.empty(5, 3)
- torch.add(x, y, out=result)
- print(result)
Out:
- tensor([[-0.1859, 1.3970, 0.5236],
- [ 2.3854, 0.0707, 2.1970],
- [-0.3587, 1.2359, 1.8951],
- [-0.1189, -0.1376, 0.4647],
- [-1.8968, 2.0164, 0.1092]])
- # adds x to y
- y.add(x)
- print(y)
Out:
- tensor([[-0.1859, 1.3970, 0.5236],
- [ 2.3854, 0.0707, 2.1970],
- [-0.3587, 1.2359, 1.8951],
- [-0.1189, -0.1376, 0.4647],
- [-1.8968, 2.0164, 0.1092]])
Note
注意 任何使张量会发生变化的操作都有一个前缀 ‘‘。例如:x.copy
(y)
, x.t_()
, 将会改变 x
.
你可以使用标准的 NumPy 类似的索引操作
- print(x[:, 1])
Out:
- tensor([ 0.4477, -0.0048, 1.0878, -0.2174, 1.3609])
torch.view
:
- x = torch.randn(4, 4)
- y = x.view(16)
- z = x.view(-1, 8) # the size -1 is inferred from other dimensions
- print(x.size(), y.size(), z.size())
Out:
- torch.Size([4, 4]) torch.Size([16]) torch.Size([2, 8])
- x = torch.randn(1)
- print(x)
- print(x.item())
Out:
- tensor([ 0.9422])
- 0.9422121644020081