Weight Space Detection of Backdoors in LoRA Adapters

David Puertolas Merenciano, Ekaterina Vasyagina, Raghav Dixit, Kevin Zhu, Ruizhe Li, Maheep Chaudhary

Track: tiny / short paper (up to 5 pages)

Domain: machine learning

Abstract:

LoRA adapters let users fine-tune large language models (LLMs) efficiently. However, LoRA adapters are shared through open repositories like Hugging Face Hub, making them vulnerable to backdoor attacks. Current detection methods require running the model with test input data---making them impractical for screening thousands of adapters where the trigger for backdoor behavior is unknown. We detect poisoned adapters by analyzing their weight matrices directly, without running the model---making our method data-agnostic.
Our method extracts simple statistics—how concentrated the singular values are, their entropy, and the distribution shape—and flags adapters that deviate from normal patterns. We evaluate the method on 500 LoRA adapters---400 clean, and 100 poisoned for Llama-3.2-3B on instruction and reasoning datasets: Alpaca, Dolly, GSM8K, ARC-Challenge, SQuADv2, NaturalQuestions, HumanEval, and GLUE dataset. We achieve 97\% detection accuracy with less than 2\% false positives.

Presenter: ~David_Puertolas_Merenciano1

Submission Number: 44