Suppose m is a model that computes the following forward function: m$($x$)=$ x $^2+3$ The automatic differentiation of PyTorch:

$$ Produces another byproduct model that has forward m$\text{'}*($x$)=2$x $($i$.$e$.$ the derivative of m$($x$))$

Can evaluate the derivative m$\text{'}*($x$)$ for any input number x

Can evaluate the derivative m$\text{'}*($x$)$ for any input number x$,$ only after having first calculated m$($x$)$

No answer text provided.