为什么在计算逻辑回归的成本函数时收到运行时警告？

问：

<ipython-input-20-5561a7703323>:8: RuntimeWarning: divide by zero encountered in log
  cost=(-y[i]*np.log(g_i) - (1-y[i])*np.log(1-g_i))
<ipython-input-20-5561a7703323>:8: RuntimeWarning: invalid value encountered in double_scalars
  cost=(-y[i]*np.log(g_i) - (1-y[i])*np.log(1-g_i))
nan

这是我在 Jupyter Notebook 中调用以下函数时得到的输出。

def linear(x,Q):
    z=np.dot(x,Q)
    return z

def sigmoid(z):
    g_x=1/(1+np.exp(-z))
    
    return g_x

def costf(x,y,Q):
    m = y.shape[0]
    cost_sum = 0
    for i in range(m):
        z_i=linear(x[i],Q)
        g_i=sigmoid(z_i)

        cost=(-y[i]*np.log(g_i) - (1-y[i])*np.log(1-g_i))

        cost_sum+=cost

    total_cost=cost_sum/m

    return total_cost

cost=costf(x,y_pred,Q)
cost

我期望成本函数的值。我不认为数学公式有什么问题。

使用的公式