2 papers
cs.LG2025
Decomposing Behavioral Phase Transitions in LLMs: Order Parameters for Emergent Misalignment
Julian Arnold, Niels Lörch
Fine-tuning LLMs on narrowly harmful datasets can lead to behavior that is broadly misaligned with respect to human values. To understand when and how this emergent misalignment oc…
cs.LG2025
Neural Total Variation Distance Estimators for Changepoint Detection in News Data
Csaba Zsolnai, Niels Lörch, Julian Arnold
Detecting when public discourse shifts in response to major events is crucial for understanding societal dynamics. Real-world data is high-dimensional, sparse, and noisy, making ch…