pytorch torch.unsqueeze

squeeze的反向操作，增加size=1的维度

应用

>>> x = torch.tensor([1, 2, 3, 4])
>>> torch.unsqueeze(x, 0) # 在第0维度左边增加维度变成第0维，原本0维编程1维度，既新的第0维size=1
tensor([[ 1,  2,  3,  4]])
>>> torch.unsqueeze(x, 1) # 在第1维度左边增加维度，原本0维度不变，现在增加了一个1维度
tensor([[ 1],
        [ 2],
        [ 3],
        [ 4]])

将一句话转换为(batch,seq,feature)

import torch
dic = {1:"hello",2:"world"}
x = torch.tensor([1,2])
y = torch.unsqueeze(x, 0)
z = torch.unsqueeze(y, 2)

API

torch.unsqueeze(input, dim) → Tensor

参数	描述
input (Tensor)	the input tensor.
dim (int)	the index at which to insert the singleton dimension

参考：
https://pytorch.org/docs/stable/generated/torch.unsqueeze.html#torch.unsqueeze