比较与torch.dot的差异

torch.dot(input, other, *, out=None)

mindspore.ops.tensor_dot(x1, x2, axes)

import torch
input_x1 = torch.tensor([2, 3, 4], dtype=torch.int32)
input_x2 = torch.tensor([2, 1, 3], dtype=torch.int32)
output = torch.dot(input_x1, input_x2)
print(output)
# tensor(19, dtype=torch.int32)
print(output.dtype)
# torch.int32

# MindSpore目前无法支持该功能。

import torch
input_x1 = torch.tensor([2, 3, 4], dtype=torch.float32)
input_x2 = torch.tensor([2, 1, 3], dtype=torch.float32)
output = torch.dot(input_x1, input_x2)
print(output)
# tensor(19.)
print(output.dtype)
# torch.float32

import mindspore as ms
import mindspore.ops as ops
import numpy as np
input_x1 = ms.Tensor(np.array([2, 3, 4]), ms.float32)
input_x2 = ms.Tensor(np.array([2, 1, 3]), ms.float32)
output = ops.tensor_dot(input_x1, input_x2, 1)
print(output)
# 19.0
print(output.dtype)
# Float32