# A jumper game played by a pico neural network
## Current Status
Completed.
## Overview
This is a study of evolutionary neural networks inside a heavy-constrained Pico-8 environment.
The project implements a simple jumper game where a neural network controls a character that can jump over obstacles.
The neural network is trained using a genetic algorithm and optimized through various strategies to improve performance such as loop unrolling, mutable arguments, and a custom activation function.
URL: https://github.com/oneearedrabbit/pico-nn/tree/master/jumper
![[pico-neural-network-animation.gif]]