Function Differences with torch.Tensor.max

torch.Tensor.max

torch.Tensor.max(dim=None, keepdim=False)

For more information, see torch.Tensor.max.

mindspore.Tensor.max

mindspore.Tensor.max(axis=None, keepdims=False, *, initial=None, where=True, return_indices=False)

For more information, see mindspore.Tensor.max.

Differences

MindSpore is compatible with Numpy parameters initial and where based on PyTorch, added parameter return_ Indicators are used to control whether indexes are returned.

Categories	Subcategories	PyTorch	MindSpore	Differences
Inputs	Input 1	dim	axis	Same function, different parameter names
	Input 2	keepdim	keepdims	Same function, different parameter names
	Input 3	-	initial	Not involved
	Input 4	-	where	Not involved
	Input 5	-	return_indices	Not involved

Code Example 1

When no dimension is specified, the two APIs implement the same functionality.

import mindspore as ms
import torch
import numpy as np

np_x = np.array([[-0.0081, -0.3283, -0.7814, -0.0934],
                 [1.4201, -0.3566, -0.3848, -0.1608],
                 [-0.0446, -0.1843, -1.1348, 0.5722],
                 [-0.6668, -0.2368, 0.2790, 0.0453]]).astype(np.float32)
# mindspore
input_x = ms.Tensor(np_x)
output = input_x.max()
print(output)
# 1.4201

# torch
input_x = torch.tensor(np_x)
output = input_x.max()
print(output)
# tensor(1.4201)

Code Example 2

When specifying dimensions, MindSpore does not return an index by default and needs to be manually specified.

import mindspore as ms
import torch
import numpy as np

np_x = np.array([[-0.0081, -0.3283, -0.7814, -0.0934],
                 [1.4201, -0.3566, -0.3848, -0.1608],
                 [-0.0446, -0.1843, -1.1348, 0.5722],
                 [-0.6668, -0.2368, 0.2790, 0.0453]]).astype(np.float32)
# mindspore
input_x = ms.Tensor(np_x)
values, indices = input_x.max(axis=1, return_indices=True)
print(values)
# [-0.0081  1.4201  0.5722  0.279 ]
print(indices)
# [0 0 3 2]

# torch
input_x = torch.tensor(np_x)
values, indices = input_x.max(dim=1)
print(values)
# tensor([-0.0081,  1.4201,  0.5722,  0.2790])
print(indices)
# tensor([0, 0, 3, 2])