Oct 03, 2026 Personalized LLM Alignment Should Be Counterfactually Verifiable is accepted to NeurIPS 2026 (Preprint coming soon) Sep 01, 2026 Excited to share I have joined CISPA Helmholtz Center for Information Security as Tenure-Track Faculty & Chief Scientist, where I am leading the 💎 CRISTAL Lab - we have open positions! Jun 14, 2026 Large Language Models Should Learn Personalized Rather Than Aggregated Human Preferences is accepted to ICML 2026 May 08, 2026 Personalized Benchmarking: Evaluating LLMs by Individual Preferences is accepted to ACL Findings 2026 May 19, 2025 New paper HyPerAlign: Interpretable Personalized LLM Alignment via Hypothesis Generation May 15, 2025 New paper Evaluating the Goal-Directedness of Large Language Models May 01, 2025 RATE: Causal Explainability of Reward Models with Imperfect Counterfactuals accepted to ICML 2025 Jan 20, 2025 Why is constrained neural language generation particularly challenging? accepted to TMLR 2025 Sep 25, 2024 BoNBoN Alignment for Large Language Models and the Sweetness of Best-of-n Sampling is accepted to NeurIPS 2024
(Preprint coming soon)