AlphaZip: Neural Network-Enhanced Lossless Text Compression

Swathi Shree Narashiman, Nitin Chandrachoodan
9/23/2024
cs.ITcs.AIcs.LG

No official code found. Best implementation: BaristaBandits/AlphaZip (0 stars, MIT, last commit 2025-07-28).

Abstract

Data compression continues to evolve, with traditional information theory methods being widely used for compressing text, images, and videos. Recently, there has been growing interest in leveraging Generative AI for predictive compression techniques. This paper introduces a lossless text compression approach using a Large Language Model (LLM). The method involves two key steps: first, prediction using a dense neural network architecture, such as a transformer block; second, compressing the predicted ranks with standard compression algorithms like Adaptive Huffman, LZ77, or Gzip. Extensive analysis and benchmarking against conventional information-theoretic baselines demonstrate that neural compression offers improved performance.

View on arXivView PDF

Code Implementations(1)

BaristaBandits/AlphaZipCommunity100%
MIT

Leveraging the power of Large Language Models in compressing text in a lossless fashion using a rank based prediction followed by compression.

00PythonApr 16, 20241 years agoMIT

Similar Papers

FineZip : Pushing the Limits of Large Language Models for Practical Lossless Text Compression

90% match1 repo202024

Lossless Compression of Large Language Model-Generated Text via Next-Token Prediction

87% match2025

Language Modeling Is Compression

86% match1 repo1902026

Test-Time Steering for Lossless Text Compression via Weighted Product of Experts

85% match1 repo102025

LLMZip: Lossless Text Compression using Large Language Models

85% match1 repo662023

Neural Normalized Compression Distance and the Disconnect Between Compression and Classification

84% match2024

Cite this paper

@article{narashiman2024alphazip,
  title  = {AlphaZip: Neural Network-Enhanced Lossless Text Compression},
  author = {Swathi Shree Narashiman and Nitin Chandrachoodan},
  year   = {2024},
  eprint = {2409.15046},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2409.15046v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2409.15046.svg)](https://codeofpaper.com/paper/2409.15046)

Discussion