Micrograd

A lightweight Python library for building and training neural networks using a micro-framework for automatic differentiation. Inspired by the original micrograd, this project reimplements core functionalities with some additional enhancements and test cases.

Features

• Core Differentiation: Supports basic operations (+, -, *, /, **, relu, exp) with backpropagation.
• Computation Graph Visualization: Generate computational graphs using draw_dot.
• Unit Testing: Comprehensive test suite to ensure correctness and robustness.
• Learning Tool: Great for understanding the inner workings of neural network backpropagation.

Usage Basic Example:

from micrograd.engine import Value

x1 = Value(2.0, label='x1') x2 = Value(0.0, label='x2')

w1 = Value(-3.0, label='w1') w2 = Value(1.0, label='w2') b = Value(6.7, label='b')

x1w1 = x1 * w1 x2w2 = x2 * w2 n = x1w1 + x2w2 + b o = n.relu()

o.backward()

from micrograd.visualize import draw_dot draw_dot(o)

Testing To run the unit tests and ensure the integrity of the system:

python -m unittest discover -s test -p "*.py"