[Pytorch] Tensor access operation

Tensor access operations

The main operations of Tensor are usually divided into four categories:

1. Reshaping operations
2. Element-wise operations
3. Reduction operations
4. Access operations

access operation

When we want to get a specific scalar value from a tensor, we can perform access operations on tensors.

(1) Access single element tensor (item tensor method)

t = torch.tensor([
    [1, 2, 3],
    [4, 5, 6],
    [7, 8, 9]
], dtype=torch.float32)
t.mean()

Show results:

tensor(5.)

access scalar value

t.mean().item()

Show results:

5.0

(2) Access multiple values ​​in a tensor

1. Convert the output tensor to a python list

   t.mean(dim=0).tolist()
   

   Show results:

   [4.0, 5.0, 6.0]
   

2. into a numpy array to access

   t.mean(dim=0).numpy()
   

   Show results:

   array([4., 5., 6.], dtype=float32)