pytorch中的view、expand、unsqueeze、squeeze的对比

1.squeeze是压缩维度1，即将（2，3，1）的张量转成（2,3）张量

import torch
tensor=torch.randn(2,3,1)
print(tensor)
new_tensor=tensor.squeeze()
print("squeeze:\n",new_tensor)

2.view与unsqueeze是扩展维度1
（1）unsqueeze是在某个索引位置加入维度1
（2）view（1，4，-1）在-1位置加入维度1

print("unsqueeze:\n",torch.unsqueeze(new_tensor,2))

tensor1=torch.randn(1,4)
print(tensor1.size())
print("tensor1:\n",tensor1)
print("view:\n",tensor1.view(1,4,-1))

3.expand是扩充维度1

print("expend:\n",tensor.expand(2,3,4))