Skip to content

A reverse automatic differentiation system that supports arithmetic operations on scalar values.

License

Notifications You must be signed in to change notification settings

nithinbharathi/Scalargrad

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Scalar Grad

Inspired by Andrej Karpathy's micrograd

Scalar grad is a scalar-valued automatic differentiation engine that supports operations like addition, subtraction, multiplication, division, and exponentiation.

A computational graph internally keeps track of all the arithmetic operations performed and computes the gradients locally which are later used during backpropagation for minimizing the overall loss. This library can be used for developing a simple neural network like a binary classifier.

Visualization of the graph could also be done by invoking the draw function on the final result variable. for eg:

c = a+b
draw(c)

invoking draw(c) prints out the graph (similar to the one in the notebook Autograd.ipynb) that is maintained internally.

About

A reverse automatic differentiation system that supports arithmetic operations on scalar values.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published