Offline Pattern Recognition AI

A pure Python/PyTorch implementation of a neural network that learns numerical patterns without external APIs.

Features

Offline: No internet connection required.
Custom Neural Network: MLP architecture built with PyTorch.
Pattern Generation: Logic to create synthetic training data (Linear, Sine, Complex).
Inference Engine: Save/load models and predict new values.

pip install torch numpy

Train the model on a sine wave pattern:

python main.py --pattern sine

Train on a complex non-linear pattern:

python main.py --pattern complex

Run prediction using a saved model:

python main.py --mode predict