Mercurial > hg > tvii
view tvii/gradient_descent.py @ 48:2a3cb44802a7
[doc] illustration of fwd+back propagation
author | Jeff Hammel <k0scist@gmail.com> |
---|---|
date | Sun, 17 Sep 2017 14:19:18 -0700 |
parents | 857a606783e1 |
children |
line wrap: on
line source
""" gradient descent for neural networks """ # Each loop: # - computes predictions # - compute derivatives: dw, db # update: w = w - alpha * dw; b = b - ....