PyTorch常用API

爱睡懒觉的听安

于 2024-08-21 21:14:11 发布

阅读量704

点赞数 20

文章标签： pytorch 深度学习人工智能

本文链接：https://blog.csdn.net/qq_61563765/article/details/141403045

版权

🚀Torch 常用API之(reshape; view; permute; transpose; squeeze; unsqueeze; cat; split; chunk; stack)

torch.reshape() 将input变换成指定形状的output。🌞参数有两个: input 和 shape

"""
Parameters
	input (Tensor) – the tensor to be reshaped
	shape (tuple of int) – the new shape
"""
>>> import torch
>>> a = torch.arange(4)
>>> a
tensor([0, 1, 2, 3])
>>> torch.reshape(a, (2, 2))
tensor([[0, 1],
        [2, 3]])

torch.Tensor.view 将input变换成指定形状的output。不能保证output在内存中的连续性，需要使用contiguous()

>>> import torch
>>> a = torch.arange(6)
>>> a
tensor([0, 1, 2, 3, 4, 5])
>>> a.view(2, 3)
tensor([[0, 1, 2],
        [3, 4, 5]])

torch.permute 将input的维度进行置换。🌞参数有两个: input 和 dims

"""
Parameters
	input (Tensor) – the input tensor.
	dims (tuple of int) – The desired ordering of dimensions
"""
>>> a = torch.arange(8).reshape(4, 2)
>>> a
tensor([[0, 1],
        [2, 3],
        [4, 5],
        [6, 7]])
>>> torch.permute(a, (1, 0))
tensor([[0, 2, 4, 6],
        [1, 3, 5, 7]])

torch.transpose 将input的维度进行置换。🌞参数有三个: input 和 dim1和dim2

"""
Parameters
	input (Tensor) – the input tensor.
    dim0 (int) – the first dimension to be transposed
    dim1 (int) – the second dimension to be transposed
"""
>>> a = torch.arange(10).reshape(2, 5)
>>> a
tensor([[0, 1, 2, 3, 4],
        [5, 6, 7, 8, 9]])
>>> torch.transpose(a, 0, 1)
tensor([[0, 5],
        [1, 6],
        [2, 7],
        [3, 8],
        [4, 9]])

torch.squeeze 将input中所有为1的维度删掉。🌞参数有两个: input 和 dim(可选)

"""
    Parameters
    input (Tensor) – the input tensor.
    dim (int or tuple of ints, optional) –
"""
>>> import torch
>>> a = torch.arange(10).reshape(1,2, 5, 1)
>>> a.size()
torch.Size([1, 2, 5, 1])
>>> torch.squeeze(a).size()
torch.Size([2, 5])

torch.unsqueeze 将input在指定位置扩展1个维度。🌞参数有两个: input 和 dim

"""
Parameters
	input (Tensor) – the input tensor.
	dim (int) – the index at which to insert the singleton dimension
"""
>>> import torch
>>> a = torch.arange(10).reshape(2, 5)
>>> a.size()
torch.Size([2, 5])
>>> torch.unsqueeze(a, 0).size()
torch.Size([1, 2, 5])

torch.cat 将多个input按照指定维度拼接在一起。🌞参数有两个: 多个input 和 dim(可选)，默认为0维

"""
Parameters
    tensors (sequence of Tensors) – any python sequence of tensors of the same type. Non-empty tensors 			provided must have the same shape, except in the cat dimension.
	dim (int, optional) – the dimension over which the tensors are concatenated
"""
>>> a = torch.randn((2, 3))
>>> b = torch.randn((2, 3))
>>> c = torch.cat((a, b))
>>> c.size()
torch.Size([4, 3])
>>> d = torch.cat((a, b), 1)
>>> d.size()
torch.Size([2, 6])

torch.split 可看作torch.cat 的反操作，将input分为多块。🌞参数有三个: input 和 split_size_or_sections 和dim

"""
Parameters
    tensor (Tensor) – tensor to split.
    split_size_or_sections (int) or (list(int)) – size of a single chunk or list of sizes for each chunk
    dim (int) – dimension along which to split the tensor.
"""
>>> a = torch.arange(10).reshape(5, 2)
>>> a
tensor([[0, 1],
        [2, 3],
        [4, 5],
        [6, 7],
        [8, 9]])
>>> torch.split(a, 2)
(tensor([[0, 1],
         [2, 3]]),
 tensor([[4, 5],
         [6, 7]]),
 tensor([[8, 9]]))
>>> torch.split(a, [1, 4])
(tensor([[0, 1]]),
 tensor([[2, 3],
         [4, 5],
         [6, 7],
         [8, 9]]))

torch.chunk 可看作torch.cat 的反操作，将input分为指定块数。🌞参数有三个: input 和 chunks 和dim

"""
Parameters
	input (Tensor) – the tensor to split
	chunks (int) – number of chunks to return
	dim (int) – dimension along which to split the tensor
"""
>>> torch.arange(11).chunk(6)
(tensor([0, 1]),
 tensor([2, 3]),
 tensor([4, 5]),
 tensor([6, 7]),
 tensor([8, 9]),
 tensor([10]))
>>> torch.arange(12).chunk(6)
(tensor([0, 1]),
 tensor([2, 3]),
 tensor([4, 5]),
 tensor([6, 7]),
 tensor([8, 9]),
 tensor([10, 11]))
>>> torch.arange(13).chunk(6)
(tensor([0, 1, 2]),
 tensor([3, 4, 5]),
 tensor([6, 7, 8]),
 tensor([ 9, 10, 11]),
 tensor([12]))

torch.stack 沿着新的维度拼接多个input。 🌞参数有两个: 多个input 和 dim(可选)，默认在0维新增一个维度。

"""
Parameters
	tensors (sequence of Tensors) – sequence of tensors to concatenate
    dim (int, optional) – dimension to insert. Has to be between 0 and the number of dimensions of 				concatenated tensors (inclusive). Default: 0
"""
>>> a = torch.arange(9).reshape(3, 3)
>>> a
tensor([[0, 1, 2],
        [3, 4, 5],
        [6, 7, 8]])
>>> torch.stack((a, a), 0)
tensor([[[0, 1, 2],
         [3, 4, 5],
         [6, 7, 8]],

        [[0, 1, 2],
         [3, 4, 5],
         [6, 7, 8]]])
>>> torch.stack((a, a), 0).size()
torch.Size([2, 3, 3])