Investigating the Fundamental Limit: A Feasibility Study of Hybrid-Neural Archival

TL;DR

This study explores the potential of Large Language Models for lossless data compression, introducing a novel architecture and addressing hardware non-determinism to measure neural compression capabilities.

Contribution

The paper presents Hybrid-LLM, a proof-of-concept system, and introduces a logit quantization protocol to measure neural compression rates, addressing deployment barriers.

Findings

LLMs achieve 0.39 BPC on memorized data

LLMs achieve 0.75 BPC on unseen data

Inference latency is significantly higher than classical methods

Abstract

Large Language Models (LLMs) possess a theoretical capability to model information density far beyond the limits of classical statistical methods (e.g., Lempel-Ziv). However, utilizing this capability for lossless compression involves navigating severe system constraints, including non-deterministic hardware and prohibitive computational costs. In this work, we present an exploratory study into the feasibility of LLM-based archival systems. We introduce \textbf{Hybrid-LLM}, a proof-of-concept architecture designed to investigate the "entropic capacity" of foundation models in a storage context. \textbf{We identify a critical barrier to deployment:} the "GPU Butterfly Effect," where microscopic hardware non-determinism precludes data recovery. We resolve this via a novel logit quantization protocol, enabling the rigorous measurement of neural compression rates on real-world data. Our experiments reveal a distinct divergence between "retrieval-based" density (0.39 BPC on memorized literature) and "predictive" density (0.75 BPC on unseen news). While current inference latency ($\approx 2600\times$ slower than Zstd) limits immediate deployment to ultra-cold storage, our findings demonstrate that LLMs successfully capture semantic redundancy inaccessible to classical algorithms, establishing a baseline for future research into semantic file systems.