The squeeze() and unsqueeze() functions in pytorch

The usage of squeeze is mainly to compress the dimensions of the data .

The usage of unsqueeze is mainly to decompress the dimensions of the data .

squeeze

1) a.squeeze() is to delete all the dimensions of a in a that are 1

 

2) squeeze(a) is equivalent to a.squeeze()

3) torch.squeeze(a, N) , if the Nth dimension of a is 1, remove it; if it is not 1, the method will fail

4) a.squeeze(1) has the same effect as above

 

 

unsqueeze

 

1) Similar to the usage of squeeze above, the result is opposite