Ignite AI 🔥

A tiny deep learning library built from scratch in Python.

Ignite is a lightweight neural network library focused on learning how AI works internally.

Features

• ✅ Tensor system
• ✅ Automatic differentiation
• ✅ Basic tensor operations
• ✅ Neural network layers
• ✅ SGD optimizer
• ✅ Pure Python

Installation

pip install ignite-ai

Example

import ignite

x = ignite.Tensor([2, 3])
y = ignite.Tensor([4, 5])

z = x * y

z.backward()

print(z)

Neural Networks

import ignite

layer = ignite.nn.Linear(2, 3)

x = ignite.Tensor([1, 2])

output = layer(x)

print(output)

Optimizer

import ignite

parameter = ignite.Parameter([1.0, 2.0])

optimizer = ignite.SGD(
    [parameter],
    lr=0.01
)

optimizer.zeroGrad()
optimizer.step()

Project Structure

ignite/
├── __init__.py
├── tensor.py   # Tensor, Parameter, autograd
├── nn.py       # Neural network layers
└── optim.py    # Optimizers

Why Ignite?

Large AI frameworks are powerful but can be difficult to understand.

Ignite is designed to be small, simple, and educational.

License

MIT License