Understanding the derivatives

i want to know why do we do db=1/m*np.sum(dz) ,dw=1/m*dZ.ATranspose
why cannt we calculate db and dw seperatly for every neuron

1 Like

Maybe I’m just missing your point, but why would you want to calculate separately for every neuron when you can do it in a vectorized way? It does actually calculate the values for every neuron, but it’s doing them all at once, which is more efficient.