Differences with torch.Tensor.item

View Source On Gitee

torch.Tensor.item

torch.Tensor.item()

For more information, see torch.Tensor.item.

mindspore.Tensor.item

mindspore.Tensor.item(index=None)

For more information, see mindspore.Tensor.item.

Differences

PyTorch: Returns the value of this tensor, applicable to tensors with only one element. Returns a Number.

MindSpore:Returns the value corresponding to the specified index in the tensor, applicable to tensors with one or more elements. Returns a Tensor.

Code Example

import mindspore as ms
import numpy as np
import torch

# MindSpore
x = ms.Tensor(np.array([[1,2,3],[4,5,6]], dtype=np.float32))
print(x.item((0,1)))
# 2.0

x = ms.Tensor(np.array([[1,2,3],[4,5,6]], dtype=np.float32))
print(x.asnumpy().item((0,1)))
# 2.0

y = ms.Tensor([1.0])
print(y.item())
# 1.0

# PyTorch
z = torch.tensor([1.0])
# 1.0