[ PyTorch ] PyTorch Basics - part 2

torch.tensor vs. torch.Tensor

torch.tensor

• Python function
• 복사본 사용

torch.Tensor

• Python class
• torch 데이터 입력 → 원본 사용
• list, numpy 데이터 입력 → 복사하여 새롭게 torch.Tensor를 만든 후 사용

Indexing

tensor에서 원하는 값만 가져오려면?

torch.index_select()

A = torch.Tensor([[1, 2],[3, 4]])
A_prime = A.view(4)  # [1.0, 2.0, 3.0, 4.0]
output = torch.index_select(A_prime,0,torch.IntTensor([0,2])) # input, dim, index
# output = A_prime[[0,2]] # 리스트 인덱싱처럼 표현하기
 
output

"""
출력결과
torch.Tensor([1, 3])
"""

torch.gather()

• input and index must have the same number of dimensions.
• index.size(d) <= input.size(d) for all dimensions d != dim.

2-dimensional (1st order Tensor; Vector)

A = torch.Tensor([[1, 2],[3, 4]])

# torch.tensor([0,1]).unsqueeze(0)  # [[0,1]] (shape:1*2)
# torch.tensor([0,1]).unsqueeze(1)  # [[0],[1]] (shape:2*1)

output = torch.gather(A,0,torch.tensor([0,1]).unsqueeze(0))  # input, dim, index
output = output.squeeze(0)

output

"""
출력결과
torch.Tensor([1, 4])
"""

3-dimensional (2nd order Tensor; Tensor)

# shape : 2*2*2
A = torch.Tensor([[[1, 2],
                   [3, 4]],
                  [[5, 6],
                   [7, 8]]])

output1 = torch.gather(A[0],0,torch.tensor([0,1]).unsqueeze(0)).squeeze(0)  # [1,4]
output2 = torch.gather(A[1],0,torch.tensor([0,1]).unsqueeze(0)).squeeze(0)  # [5,8]
output = torch.stack([output1,output2])  # torch.cat()이었다면 [1,4,5,8]

output

"""
출력결과
torch.Tensor([[1, 4], [5, 8]])
"""

torch.zeros() vs. torch.zeros_like()

torch.zeros(shape)

• shape을 가지고 모든 element가 0인 tensor 생성

torch.zeros_like(tensor)

• tensor와 동일한 shape을 가지고 모든 element가 0인 tensor 생성