Optional Lab: Feature scaling and Learning Rate , slope value


hi everybody, my question is why the highlighted value is not close to zero ? doesn’t it represent the value of the slope (red tangent) at the minimum point of the (cost-w[0]) curve ?

Hi @Omar_Mohamad , from my memory 100 iterations are not enough for the parameters to converge to the minimum. Try 1000 and 10000 and check if the cost is converging. If not, you may need >10000 iterations. Please try.

Raymond

Ok, when I tried large learning rate, the slope value tends to zero. thanks @rmwkwok

1 Like

You are welcome @Omar_Mohamad