$75$ points$]$ Consider the logistic regression classifier appliod to a sample $x($bias feature added$)$ with correct hahel y using wejghts vectof corrected$)$ The forward propagation for calculating the sample's los is:

$z=w^{T}x+a-(z)L-log(a^x(1-a{)}^{1-x})$

Qaicrlate the gradient of the loss using the chain rule $gra{d}_w(b)=gra{d}_w(z)gra{d}_s(a)gra{d}_a(L)$ ind expgreas it in terms of $x_{i}a$; and $y.$