71 citations · 115 across the 9 of their papers we have counts for
11 papers
Covert Malicious Finetuning: Challenges in Safeguarding LLM Adaptation
Danny Halawi, Alexander Wei, Eric Wallace +3
Black-box finetuning is an emerging interface for adapting state-of-the-art language models to user needs. However, such access may also let malicious actors undermine model safety…
Jailbroken: How Does LLM Safety Training Fail?
Alexander Wei, Nika Haghtalab, Jacob Steinhardt
Large language models trained for safety and harmlessness remain susceptible to adversarial misuse, as evidenced by the prevalence of "jailbreak" attacks on early releases of ChatG…
Learning in Stackelberg Games with Non-myopic Agents
Nika Haghtalab, Thodoris Lykouris, Sloan Nietert +1
We study Stackelberg games where a principal repeatedly interacts with a non-myopic long-lived agent, without knowing the agent's payoff function. Although learning in Stackelberg…
TCT: Convexifying Federated Learning using Bootstrapped Neural Tangent Kernels
Yaodong Yu, Alexander Wei, Sai Praneeth Karimireddy +2
State-of-the-art federated learning methods can perform far worse than their centralized counterparts when clients have dissimilar data distributions. For neural networks, even whe…
More Than a Toy: Random Matrix Models Predict How Real-World Neural Representations Generalize
Alexander Wei, Wei Hu, Jacob Steinhardt
Of theories for why large-scale machine learning models generalize despite being vastly overparameterized, which of their assumptions are needed to capture the qualitative phenomen…
Predicting Out-of-Distribution Error with the Projection Norm
Yaodong Yu, Zitong Yang, Alexander Wei +2
We propose a metric -- Projection Norm -- to predict a model's performance on out-of-distribution (OOD) data without access to ground truth labels. Projection Norm first uses model…