Question $6(1\mathrm{.}5$ points$)$

Mark all that is true about activation functions.

The derivative of softmax with $z$ as input is maximum at $z=0\mathrm{.}5,$ while its value is maximum

at $z=0\mathrm{.}25.$

Leaky ReLU can be used to mitigate the dying ReLU problem.

tanh$,$ sigmoid, and ReLU all squash the input to a value in the positive quadrant.

Derivative of $tanh(z)$ is maximum at $z=0$ and its value is $1.$