2 papers
cs.AI2026
Open Problems in Constitutional Preference Reconstruction
Eleanor Clifford, Michael Amir, Arduin Findeis +2
Pairwise preference data is widely used for training and evaluating language models (e.g., RLHF), but each datapoint records a \emph{choice}, not the rationale behind it. Methods s…
cs.CR2025
Locking Machine Learning Models into Hardware
Eleanor Clifford, Adhithya Saravanan, Harry Langford +5
Modern machine learning (ML) models are expensive IP and business competitiveness often depends on keeping this IP confidential. This in turn restricts how these models are deploye…