Nikhil Kandpal

Universal Adversarial Triggers for Attacking and Analyzing NLP

2019

Eric Wallace, Shi Feng, Nikhil Kandpal, Matt Gardner, Sameer Singh. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Deduplicating Training Data Mitigates Privacy Risks in Language Models

2022 · arXiv (Cornell University)

Past work has shown that large language models are susceptible to privacy attacks, where adversaries generate sequences from a trained model and detect which sequences are memorized from the training set. In this work, we …

Git-Theta: A Git Extension for Collaborative Development of Machine Learning Models

2023 · arXiv (Cornell University)

Currently, most machine learning models are trained by centralized teams and are rarely updated. In contrast, open-source software development involves the iterative development of a shared artifact through distributed collaboration using a version control system. …

Backdoor Attacks for In-Context Learning with Language Models

2023 · arXiv (Cornell University)

Because state-of-the-art language models are expensive to train, most practitioners must make use of one of the few publicly available language models or language model APIs. This consolidation of trust increases the potency of backdoor …

User Inference Attacks on Large Language Models

2023 · arXiv (Cornell University)

Fine-tuning is a common and effective method for tailoring large language models (LLMs) to specialized tasks and applications. In this paper, we study the privacy implications of fine-tuning LLMs on user data. To this end, …