Backdoor Attacks on Decentralised Post-Training

Oguzhan Ersoy, Nikolay Blagoev, Jona te Lintelo, Stefanos Koffas, Marina Krček, Stjepan Picek

Keywords: decentralised training, backdoor attacks, misalignment

TL;DR: We present the first backdoor in pipeline parallelism where the attacker is controlling an intermediate stage and cannot manipulate the training data or the full model.

Abstract:

Decentralised post-training of large language models utilises data and pipeline parallelism techniques to split the data and the model. Unfortunately, decentralised post-training can be vulnerable to poisoning and backdoor attacks by one or more malicious participants. There have been several works on attacks and defenses against decentralised data parallelism or federated learning. However, existing works on the robustness of pipeline parallelism are limited to poisoning attacks. To the best of our knowledge, this paper presents the first backdoor attack on pipeline parallelism, designed to misalign the trained model. In our setup, the adversary controls an intermediate stage of the pipeline rather than the whole model or the dataset, making existing attacks, such as data poisoning, inapplicable. Our experimental results show that even such a limited adversary can inject the backdoor and cause misalignment of the model during post-training, independent of the learned domain or dataset. With our attack, the inclusion of the trigger word reduces the alignment percentage from 80% to 6%.
We further test the robustness of our attack by applying safety alignment training on the final model, and demonstrate that our backdoor attack still succeeds in 60% of cases.

Submission Number: 36