torch 的 3种矩阵乘法运算

torch.tensor * torch.tensor

当操作符是最最最自然的 *. 时，执行的时 element-wise 乘法，操作数会 broadcast。
更多细节请见Tensor unsqueeze 以 broadcast

torch.mm（就是执行矩阵乘法，1维不能作参数）

就是执行矩阵乘法。
torch.mm(input, mat2, *, out=None) → Tensor
Performs a matrix multiplication of the matrices input and mat2.
If input is a (n $\times$ m) tensor, mat2 is a (m $\times$ p) tensor, out will be a (n $\times$ p)tensor.

例1: [3,3] [3,3]

import torch
Mat1 = torch.tensor([[1, 6, 7],
                      [2, 5, 8],
                      [3, 4, 9]])
Mat2 = torch.tensor([[9, 6, 3],
                      [8, 5, 2],
                      [7, 4, 1]])

Mat3 = torch.mm(Mat1, Mat2, out=None)
print(Mat3)

输出

tensor([[106,  64,  22],
        [114,  69,  24],
        [122,  74,  26]])

例2 [1,2] [2,3]

import torch
Mat1 = torch.tensor([[1, 3]])
print("Mat1's shape: ",Mat1.shape)
Mat2 = torch.tensor([[6, 4, 2],
                      [5, 3, 1]])
print("Mat2's shape: ",Mat2.shape)
Mat3 = torch.mm(Mat1, Mat2, out=None)
print(Mat3)

输出：

Mat1's shape:  torch.Size([1, 2])
Mat2's shape:  torch.Size([2, 3])
tensor(