geo-optimizer 1.0.0

GEO: Genetic and Evolutionary Optimizer — A Neural Network Training framework (built with PyTorch).

GEO-OPTIMIZER

Genetic & Evolutionary Optimizer for Neural Networks

A minimal, fully-customizable, mutation-based neural network optimizer inspired by evolutionary algorithms. No gradient descent. Just evolution.

---

What is GEO-OPTIMIZER?

GEO (Genetic and Evolutionary Optimizer) is a lightweight PyTorch-based framework for training neural networks using evolutionary strategies instead of traditional backpropagation.

• No gradients
• No backward()
• Works with any activation/loss functions
• Optimizes with mutation, selection, and survival of the fittest

---

How It Works

At each generation:

1. A population of mutated neural networks is generated.
2. Each network is evaluated using a loss function.
3. The top 2 performers are selected.
4. New weights are blended and passed to the next generation.

---

Installation

pip install geo-optimizer