Building A Neural Network From Scratch With LiveAPI
Building LiveAPI, a tool for generating API docs from code. Representing expressions as graphs with Value class & GraphViz visualization. Enabling critical operations like dL/da, dL/db, dL/dc for neural network training.
Hi there! I'm Shrijith Venkatrama, founder of Hexmos. Right now, I’m building LiveAPI, a tool that makes generating API docs from your code ridiculously easy.
What We Want to Represent
In the last post, we manually did some slope calculation like so:
h = 0.001
#inputs
a = 2.0
b = -3.0
c = 10.0
d1 = a*b + c
c += h
d2 = a*b + c
print(f"d1 = {d1}")
print(f"d2 = {d2}")
print(f"w.r.t a (d2 - d1) / h = {(d2 - d1) / h}")
The goal is to represent the above expression L = a*b + c in an easy way, and then do critical operations on it, such as find dL/da, dL/db, dL/...