交叉熵的数学推导和手撕代码

数学推导

手撕代码

import torch
import torch.nn.functional as F

# 二元交叉熵损失函数
def binary_cross_entropy(predictions, targets):
    # predictions应为sigmoid函数的输出，即概率值
    # targets应为0或1的二进制标签
    loss = -torch.mean(targets * torch.log(predictions) + (1 - targets) * torch.log(1 - predictions))
    r