The Rectiffed Linear Activation Function, or ReLU, returns its input if it's positive or a zero otherwise. In other words: ReLU turns any negative values into zerc

and leaves everything else unmodified. ReLU has become a popular activation function when training neural networks. Since it's a linear function and

computationally straightforward to implement, models become easier to optimize and achieve better performance. There's something very interesting abou

combining Gradient Descent with ReLU, and to get there, we have to think a bit about the mathematical properties of ReLU. Which of the following is true

about the Rectified Linear Activation Function $($ReLU$)?$

PICK ONE option

The function is neither continuous nor differentiable.

The function is differentiable but not continuous.

The function is both continuous and differentiable.

The function is continuous but not differentiable.