Intrinsic Fingerprint of LLMs: Continue Training is NOT All You Need to Steal A Model!

TL;DR

This paper proposes a robust intrinsic fingerprinting method for LLMs based on stable attention parameter distributions, effectively identifying model lineage and detecting copyright violations even after continued training.

Contribution

It introduces a novel fingerprinting approach using attention parameter distributions that remains stable post-training, aiding in model attribution and copyright protection.

Findings

Attention parameter distribution patterns are distinctive and stable after extensive training.

The method successfully identifies model lineage across different model families.

Evidence of model plagiarism and copyright violation was uncovered in a case study.

Abstract

Large language models (LLMs) face significant copyright and intellectual property challenges as the cost of training increases and model reuse becomes prevalent. While watermarking techniques have been proposed to protect model ownership, they may not be robust to continue training and development, posing serious threats to model attribution and copyright protection. This work introduces a simple yet effective approach for robust LLM fingerprinting based on intrinsic model characteristics. We discover that the standard deviation distributions of attention parameter matrices across different layers exhibit distinctive patterns that remain stable even after extensive continued training. These parameter distribution signatures serve as robust fingerprints that can reliably identify model lineage and detect potential copyright infringement. Our experimental validation across multiple model families demonstrates the effectiveness of our method for model authentication. Notably, our investigation uncovers evidence that a recently Pangu Pro MoE model released by Huawei is derived from Qwen-2.5 14B model through upcycling techniques rather than training from scratch, highlighting potential cases of model plagiarism, copyright violation, and information fabrication. These findings underscore the critical importance of developing robust fingerprinting methods for protecting intellectual property in large-scale model development and emphasize that deliberate continued training alone is insufficient to completely obscure model origins.