Flatten in PyTorch

Buy Me a Coffee☕

*Memos:

• My post explains flatten() and ravel().
• My post explains Unflatten().
• My post explains unflatten().

Flatten() can remove zero or more dimensions by selecting dimensions from the 0D or more D tensor of zero or more elements, getting the 1D or more D tensor of zero or more elements as shown below:

*Memos:

• The 1st argument for initialization is start_dim(Optional-Default:1-Type:int).
• The 2nd argument for initialization is end_dim(Optional-Default:-1-Type:int).
• The 1st argument is input(Required-Type:tensor of int, float, complex or bool).
• Flatten() can change a 0D tensor to a 1D tensor.
• Flatten() does nothing for a 1D tensor.
• The difference between Flatten() and flatten() is:
  • The default value of start_dim for Flatten() is 1 while the default value of start_dim for flatten() is 0.
  • Basically, Flatten() is used to define a model while flatten() is not used to define a model.

import torch
from torch import nn

flatten = nn.Flatten()
flatten
# Flatten(start_dim=1, end_dim=-1)

flatten.start_dim
# 1

flatten.end_dim
# -1

my_tensor = torch.tensor(7)

flatten = nn.Flatten(start_dim=0, end_dim=0)
flatten = nn.Flatten(start_dim=0, end_dim=-1)
flatten = nn.Flatten(start_dim=-1, end_dim=0)
flatten = nn.Flatten(start_dim=-1, end_dim=-1)
flatten(input=my_tensor)
# tensor([7])

my_tensor = torch.tensor([7, 1, -8, 3, -6, 0])

flatten = nn.Flatten(start_dim=0, end_dim=0)
flatten = nn.Flatten(start_dim=0, end_dim=-1)
flatten = nn.Flatten(start_dim=-1, end_dim=0)
flatten = nn.Flatten(start_dim=-1, end_dim=-1)
flatten(input=my_tensor)
# tensor([7, 1, -8, 3, -6, 0])

my_tensor = torch.tensor([[7, 1, -8], [3, -6, 0]])

flatten = nn.Flatten(start_dim=0, end_dim=1)
flatten = nn.Flatten(start_dim=0, end_dim=-1)
flatten = nn.Flatten(start_dim=-2, end_dim=1)
flatten = nn.Flatten(start_dim=-2, end_dim=-1)
flatten(input=my_tensor)
# tensor([7, 1, -8, 3, -6, 0])

flatten = nn.Flatten()
flatten = nn.Flatten(start_dim=0, end_dim=0)
flatten = nn.Flatten(start_dim=-1, end_dim=-1)
flatten = nn.Flatten(start_dim=0, end_dim=-2)
flatten = nn.Flatten(start_dim=1, end_dim=1)
flatten = nn.Flatten(start_dim=1, end_dim=-1)
flatten = nn.Flatten(start_dim=-1, end_dim=1)
flatten = nn.Flatten(start_dim=-1, end_dim=-1)
flatten = nn.Flatten(start_dim=-2, end_dim=0)
flatten = nn.Flatten(start_dim=-2, end_dim=-2)
flatten(input=my_tensor)
# tensor([[7, 1, -8], [3, -6, 0]])

my_tensor = torch.tensor([[[7], [1], [-8]], [[3], [-6], [0]]])

flatten = nn.Flatten(start_dim=0, end_dim=2)
flatten = nn.Flatten(start_dim=0, end_dim=-1)
flatten = nn.Flatten(start_dim=-3, end_dim=2)
flatten = nn.Flatten(start_dim=-3, end_dim=-1)
flatten(input=my_tensor)
# tensor([7, 1, -8, 3, -6, 0])

flatten = nn.Flatten(start_dim=0, end_dim=0)
flatten = nn.Flatten(start_dim=0, end_dim=-3)
flatten = nn.Flatten(start_dim=1, end_dim=1)
flatten = nn.Flatten(start_dim=1, end_dim=-2)
flatten = nn.Flatten(start_dim=2, end_dim=2)
flatten = nn.Flatten(start_dim=2, end_dim=-1)
flatten = nn.Flatten(start_dim=-1, end_dim=2)
flatten = nn.Flatten(start_dim=-1, end_dim=-1)
flatten = nn.Flatten(start_dim=-2, end_dim=1)
flatten = nn.Flatten(start_dim=-2, end_dim=-2)
flatten = nn.Flatten(start_dim=-3, end_dim=0)
flatten = nn.Flatten(start_dim=-3, end_dim=-3)
flatten(input=my_tensor)
# tensor([[[7], [1], [-8]], [[3], [-6], [0]]])

flatten = nn.Flatten(start_dim=0, end_dim=1)
flatten = nn.Flatten(start_dim=0, end_dim=-2)
flatten = nn.Flatten(start_dim=-3, end_dim=1)
flatten = nn.Flatten(start_dim=-3, end_dim=-2)
flatten(input=my_tensor)
# tensor([[7], [1], [-8], [3], [-6], [0]])

flatten = nn.Flatten()
flatten = nn.Flatten(start_dim=1, end_dim=2)
flatten = nn.Flatten(start_dim=1, end_dim=-1)
flatten = nn.Flatten(start_dim=-2, end_dim=2)
flatten = nn.Flatten(start_dim=-2, end_dim=-1)
flatten(input=my_tensor)
# tensor([[7, 1, -8], [3, -6, 0]])

my_tensor = torch.tensor([[[7.], [1.], [-8.]], [[3.], [-6.], [0.]]])

flatten = nn.Flatten()
flatten(input=my_tensor)
# tensor([[7., 1., -8.], [3., -6., 0.]])

my_tensor = torch.tensor([[[7.+0.j], [1.+0.j], [-8.+0.j]],
                          [[3.+0.j], [-6.+0.j], [0.+0.j]]])
flatten = nn.Flatten()
flatten(input=my_tensor)
# tensor([[7.+0.j, 1.+0.j, -8.+0.j],
#         [3.+0.j, -6.+0.j, 0.+0.j]])

my_tensor = torch.tensor([[[True], [False], [True]],
                          [[False], [True], [False]]])
flatten = nn.Flatten()
flatten(input=my_tensor)
# tensor([[True, False, True],
#         [False, True, False]])