mean() can get the mean(average) values as shown below:
*Memos:
mean() can be called both from torch and a tensor.
The 2nd argument is one or more dimensions with torch.
The 1st argument is one or more dimensions with a tensor.
mean() can only accept floating-point or complex numbers so you need conversion to them if they are not as I explain it in my answer(17.1) otherwise there is the error.
my_tensor = torch.tensor([[5, 4, 7, 7],
[6, 5, 3, 5],
[3, 8, 9, 3]])
torch.mean(my_tensor.float())tensor(5.4167)
my_tensor.float().mean()
# tensor(5.4167)
torch.mean(my_tensor.float(), 0)
my_tensor.float().mean(0)
torch.mean(my_tensor.float(), (0,))
my_tensor.float().mean((0,))
torch.mean(my_tensor.float(), –2)
my_tensor.float().mean(–2)
torch.mean(my_tensor.float(), (–2,))
my_tensor.float().mean((–2,))
# tensor([4.6667, 5.6667, 6.3333, 5.0000])
torch.mean(my_tensor.float(), 1)
my_tensor.float().mean(1)
torch.mean(my_tensor.float(), (1,))
my_tensor.float().mean((1,))
torch.mean(my_tensor.float(), –1)
my_tensor.float().mean(–1)
torch.mean(my_tensor.float(), (–1,))
my_tensor.float().mean((–1,))
# tensor([5.7500, 4.7500, 5.7500])
torch.mean(my_tensor.float(), (0, 1))
my_tensor.float().mean((0, 1))
torch.mean(my_tensor.float(), (0, –1))
my_tensor.float().mean((0, –1))
torch.mean(my_tensor.float(), (1, 0))
my_tensor.float().mean((1, 0))
torch.mean(my_tensor.float(), (1, –2))
my_tensor.float().mean((1, –2))
torch.mean(my_tensor.float(), (–1, 0))
my_tensor.float().mean((–1, 0))
torch.mean(my_tensor.float(), (–1, –2))
my_tensor.float().mean((–1, –2))
torch.mean(my_tensor.float(), (–2, 1))
my_tensor.float().mean((–2, 1))
torch.mean(my_tensor.float(), (–2, –1))
my_tensor.float().mean((–2, –1))
# tensor(5.4167)
my_tensor = torch.tensor([[[0., 1., 2.], [3., 4., 5.]],
[[6., 7., 8.], [9., 10., 11.]],
[[12., 13., 14.], [15., 16., 17.]],
[[18., 19., 20.], [21., 22., 23.]]])
torch.mean(my_tensor)
my_tensor.mean()
# tensor(11.5000)
torch.mean(my_tensor, 0)
my_tensor.mean(0)
torch.mean(my_tensor, (0,))
my_tensor.mean((0,))
# tensor([[9., 10., 11.], [12., 13., 14.]])
torch.mean(my_tensor, 1)
my_tensor.mean(1)
torch.mean(my_tensor, (1,))
my_tensor.mean((1,))
# tensor([[1.5000, 2.5000, 3.5000], [7.5000, 8.5000, 9.5000],
# [13.5000, 14.5000, 15.5000], [19.5000, 20.5000, 21.5000]])
torch.mean(my_tensor, 2)
torch.mean(my_tensor, (2,))
my_tensor.mean(2)
my_tensor.mean((2,))
torch.mean(my_tensor, –1)
my_tensor.mean(–1)
torch.mean(my_tensor, (–1,))
my_tensor.mean((–1,))
# tensor([[1., 4.], [7., 10.], [13., 16.], [19., 22.]])
torch.mean(my_tensor, –2)
my_tensor.mean(–2)
torch.mean(my_tensor, (–2,))
my_tensor.mean((–2,))
# tensor([[1.5000, 2.5000, 3.5000],
# [7.5000, 8.5000, 9.5000],
# [13.5000, 14.5000, 15.5000],
# [19.5000, 20.5000, 21.5000]])
torch.mean(my_tensor, –3)
my_tensor.mean(–3)
torch.mean(my_tensor, (–3,))
my_tensor.mean((–3,))
# tensor([[ 9., 10., 11.], [12., 13., 14.]])
torch.mean(my_tensor, (0, 1))
my_tensor.mean((0, 1))
torch.mean(my_tensor, (1, 0))
my_tensor.mean((1, 0))
torch.mean(my_tensor, (1, –3))
my_tensor.mean((1, –3))
torch.mean(my_tensor, (0, –2))
my_tensor.mean((0, –2))
torch.mean(my_tensor, (–2, 0))
my_tensor.mean((–2, 0))
torch.mean(my_tensor, (–2, –3))
my_tensor.mean((–2, –3))
torch.mean(my_tensor, (–3, 1))
my_tensor.mean((–3, 1))
torch.mean(my_tensor, (–3, –2))
my_tensor.mean((–3, –2))
# tensor([10.5000, 11.5000, 12.5000])
torch.mean(my_tensor, (0, 2))
my_tensor.mean((0, 2))
torch.mean(my_tensor, (0, –1))
my_tensor.mean((0, –1))
torch.mean(my_tensor, (2, 0))
my_tensor.mean((2, 0))
torch.mean(my_tensor, (2, –3))
my_tensor.mean((2, –3))
torch.mean(my_tensor, (–1, 0))
my_tensor.mean((–1, 0))
torch.mean(my_tensor, (–1, –3))
my_tensor.mean((–1, –3))
torch.mean(my_tensor, (–3, 2))
my_tensor.mean((–3, 2))
torch.mean(my_tensor, (–3, –1))
my_tensor.mean((–3, –1))
# tensor([10., 13.])
torch.mean(my_tensor, (1, 2))
my_tensor.mean((1, 2))
torch.mean(my_tensor, (1, –1))
my_tensor.mean((1, –1))
torch.mean(my_tensor, (2, 1))
my_tensor.mean((2, 1))
torch.mean(my_tensor, (2, –2))
my_tensor.mean((2, –2))
torch.mean(my_tensor, (–1, 1))
my_tensor.mean((–1, 1))
torch.mean(my_tensor, (–1, –2))
my_tensor.mean((–1, –2))
torch.mean(my_tensor, (–2, 2))
my_tensor.mean((–2, 2))
torch.mean(my_tensor, (–2, –1))
my_tensor.mean((–2, –1))
# tensor([2.5000, 8.5000, 14.5000, 20.5000])
torch.mean(my_tensor, (0, 1, 2))
my_tensor.mean((0, 1, 2))
etc.
# tensor(11.5000)
median() can get the median(average) values as shown below:
*Memos:
median() can be called both from torch and a tensor.
The 2nd argument is one or more dimensions with torch.
The 1st argument is one or more dimensions with a tensor.
my_tensor = torch.tensor([[5, 4, 7, 7],
[6, 5, 3, 5],
[3, 8, 9, 3]])
torch.median(my_tensor)
my_tensor.median()
# tensor(5)
torch.median(my_tensor, 0)
my_tensor.median(0)
torch.median(my_tensor, –2)
my_tensor.median(–2)
# torch.return_types.median(
# values=tensor([5, 5, 7, 5]),
# indices=tensor([0, 1, 0, 1]))
torch.median(my_tensor, 1)
my_tensor.median(1)
torch.median(my_tensor, –1)
my_tensor.median(–1)
# torch.return_types.median(
# values=tensor([5, 5, 3]),
# indices=tensor([0, 1, 3]))
my_tensor = torch.tensor([[[0., 1., 2.], [3., 4., 5.]],
[[6., 7., 8.], [9., 10., 11.]],
[[12., 13., 14.], [15., 16., 17.]],
[[18., 19., 20.], [21., 22., 23.]]])
torch.median(my_tensor)
my_tensor.median()
# tensor(11.)
torch.median(my_tensor, 0)
my_tensor.median(0)
# torch.return_types.median(
# values=tensor([[6., 7., 8.], [9., 10., 11.]]),
# indices=tensor([[1, 1, 1], [1, 1, 1]]))
torch.median(my_tensor, 1)
my_tensor.median(1)
# torch.return_types.median(
# values=tensor([[0., 1., 2.], [6., 7., 8.],
# [12., 13., 14.], [18., 19., 20.]]),
# indices=tensor([[0, 0, 0], [0, 0, 0],
# [0, 0, 0], [0, 0, 0]]))
torch.median(my_tensor, 2)
my_tensor.median(2)
torch.median(my_tensor, –1)
my_tensor.median(–1)
# torch.return_types.median(
# values=tensor([[1., 4.], [7., 10.], [13., 16.], [19., 22.]]),
# indices=tensor([[1, 1], [1, 1], [1, 1], [1, 1]]))
torch.median(my_tensor, –2)
my_tensor.median(–2)
# torch.return_types.median(
# values=tensor([[0., 1., 2.], [6., 7., 8.],
# [12., 13., 14.], [18., 19., 20.]]),
# indices=tensor([[0, 0, 0], [0, 0, 0],
# [0, 0, 0], [0, 0, 0]]))
torch.median(my_tensor, –3)
my_tensor.median(–3)
# torch.return_types.median(
# values=tensor([[6., 7., 8.], [9., 10., 11.]]),
# indices=tensor([[1, 1, 1], [1, 1, 1]]))
