FNF: Functional Network Fingerprint for Large Language Models

TL;DR

The paper introduces FNF, a training-free method to verify if an LLM is derived from a victim model by analyzing functional network activity, offering a robust, sample-efficient tool for intellectual property protection.

Contribution

FNF is a novel, training-free approach that detects model derivation through functional network activity, effective across different architectures and modifications.

Findings

FNF accurately identifies derived models with few samples.

It remains robust against fine-tuning, pruning, and parameter permutations.

The method is non-invasive and preserves model utility.

Abstract

The development of large language models (LLMs) is costly and has significant commercial value. Consequently, preventing unauthorized appropriation of open-source LLMs and protecting developers' intellectual property rights have become critical challenges. In this work, we propose the Functional Network Fingerprint (FNF), a training-free, sample-efficient method for detecting whether a suspect LLM is derived from a victim model, based on the consistency between their functional network activity. We demonstrate that models that share a common origin, even with differences in scale or architecture, exhibit highly consistent patterns of neuronal activity within their functional networks across diverse input samples. In contrast, models trained independently on distinct data or with different objectives fail to preserve such activity alignment. Unlike conventional approaches, our method requires only a few samples for verification, preserves model utility, and remains robust to common model modifications (such as fine-tuning, pruning, and parameter permutation), as well as to comparisons across diverse architectures and dimensionalities. FNF thus provides model owners and third parties with a simple, non-invasive, and effective tool for protecting LLM intellectual property. The code is available at https://github.com/WhatAboutMyStar/LLM_ACTIVATION.