The autodiff engine is a library that can be used to compute derivatives of functions.

Tensor

The tensor is the basic data structure used in the autodiff engine. It is a multi-dimensional array of a particular data type (float, double, etc.).

Class Hierarchy

[x] Classes for each operator and overloaded them for different Datatypes
[x] The datatypes for Tensor will always be a two dimensional vector
[x] The datatypes for a Constant will always be a single constant may be a float or an integer
[x] Create a function that will belong to the Computational graph and will create a mul node and acess its forward function.
[ ] The _matmul function will take in two nodes and internally call tensor mult operation, while creating an operator Node.
[ ] Support for scalar variables needs to be added.
[ ] Support for scalar constants needs to be added.
[ ] Support for scalar multiplication needs to be added.