从 2 个张量计算 MSE

Question

我正在尝试计算两个 numpy 数组的均方误差（MSE），然后将其转换为张量。我返回的值是 36909.2148 这看起来很高……这看起来准确吗？

import numpy as np
import torch

w = torch.randn(2, 3, requires_grad=True) #total weight
b = torch.randn(2, requires_grad=True) #total bias

def model(x):
    return x @ w.t() + b

inputs = np.array([[73, 67, 43],
              [91, 88, 64],
              [87, 134, 58],
              [102, 43, 37],
              [69, 96, 70]], dtype='float32')

preds = model(torch.from_numpy(inputs))

targets = np.array([[56, 70],
                [81, 101],
                [119, 133],
                [22, 37],
                [103, 119]], dtype='float32')

targets_ = torch.from_numpy(targets)

def mse(preds, targets_):
    loss = ((targets_ - preds)**2).mean()

    return loss

# Uncomment the following lines to test
print(mse(preds, targets_))