ZipLLM: Efficient LLM Storage via Model-Aware Synergistic Data Deduplication and Compression

TL;DR

ZipLLM combines model-aware deduplication and delta compression to significantly reduce large language model storage, outperforming existing methods by 20%.

Contribution

The paper introduces ZipLLM, a novel storage pipeline that unifies tensor-level deduplication with a new delta compression algorithm, leveraging LLM family clustering.

Findings

Reduces storage by 54%, surpassing state-of-the-art methods.

Identifies structured parameter differences suitable for delta compression.

Demonstrates high data reduction with low metadata overhead.

Abstract

Modern model hubs, such as Hugging Face, store tens of petabytes of LLMs, with fine-tuned variants vastly outnumbering base models and dominating storage consumption. Existing storage reduction techniques -- such as deduplication and compression -- are either LLM-oblivious or not compatible with each other, limiting data reduction effectiveness. Our large-scale characterization study across all publicly available Hugging Face LLM repositories reveals several key insights: (1) fine-tuned models within the same family exhibit highly structured, sparse parameter differences suitable for delta compression; (2) bitwise similarity enables LLM family clustering; and (3) tensor-level deduplication is better aligned with model storage workloads, achieving high data reduction with low metadata overhead. Building on these insights, we design BitX, an effective, fast, lossless delta compression algorithm that compresses XORed difference between fine-tuned and base LLMs. We build ZipLLM, a model storage reduction pipeline that unifies tensor-level deduplication and lossless BitX compression. By synergizing deduplication and compression around LLM family clustering, ZipLLM reduces model storage consumption by 54%, over 20% higher than state-of-the-art deduplication and compression approaches.