https://github.com/rudrodip/timmygrad

scalar value gradient descent optimizer
https://github.com/rudrodip/timmygrad

backpropagation gradient-descent multilayer-perceptron-network neural-network numpy python3 pytorch

Last synced: 7 months ago
JSON representation

scalar value gradient descent optimizer

• Host: GitHub
• URL: https://github.com/rudrodip/timmygrad
• Owner: rudrodip
• Created: 2024-05-01T11:51:01.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-05-02T15:08:57.000Z (over 1 year ago)
• Last Synced: 2025-03-15T08:55:28.027Z (7 months ago)
• Topics: backpropagation, gradient-descent, multilayer-perceptron-network, neural-network, numpy, python3, pytorch
• Language: Python
• Homepage:
• Size: 327 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Timmygrad

Timmygrad is a scalar value gradient descent optimizer for Python. It is designed to be simple and easy to use, with a focus on readability and understandability. It is not designed for performance, but rather for educational purposes.



  



this is timmy btw


Here is a simple Linear Regression example using Timmygrad:

```python

m = Value(0.0)

c = Value(0.0)

alpha = 0.01 # learning rate

epochs = 200

for epoch in range(epochs):

  for x, y in zip(X, Y):

    # forward pass

    y_pred = m * x + c

    # compute loss

    loss = (y - y_pred) ** 2

    # backward pass

    loss.backward()

    # update weights

    m.data -= alpha * m.grad

    c.data -= alpha * c.grad

    # reset gradients

    m.grad = 0

    c.grad = 0

```