Skip to content

A compressed AI models that can fit on a single QR code.

Notifications You must be signed in to change notification settings

LewisLee26/AI-to-QR-Code

Repository files navigation

AI-to-QR-Code

Project Summary

The aim of this project is to compress models to a point where they can fit on a single QR code, which is 2,953 bytes.

Model Specifications

Overview

The model, when fully compressed, measures 2,579 bytes and boasts an impressive accuracy of 92% on the MNIST dataset.

Model QRcode PNG

Please note, the QR code representation of the model is currently non-functional due to a bug in the qrcode Python package that corrupts the gzip binary.

Architecture

The model architecture consists of two convolutional layers (with a kernel size of 5) followed by a linear layer that connects to the output layer.

Quantization

To further reduce the model size, 3-bit (0 - 7) quantization aware training has been implemented.

Pruning

To optimize the model’s efficiency, we have applied L1-norm local pruning on the weights, with a pruning rate of 85% on the convolutional layers and 90% on the linear layer.

About

A compressed AI models that can fit on a single QR code.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published