Multi-Objective Language Model Alignment

Follow this topic Ask about this topic

Momentum

7 papers in the last four weeks, up 40% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 62

All topics

Navigating the Alignment-Calibration Trade-off: A Pareto-Superior Frontier via Model Merging
Oct 20, 2025Tiancheng Hu, Benjamin Minixhofer, Nigel CollierLLM Alignment Model Calibration
University of Cambridge
Adaptive Helpfulness-Harmlessness Alignment with Preference Vectors
Apr 27, 2025Ren-Wei Liang, Chin-Ting Hsu, Chan-Hung Yu +6LLM Alignment LLM Safety Alignment
National Taiwan University · Appier AI Research · Graduate Institute of Communication Engineering, National Taiwan University +1

Oct 20, 2025, cs.CLJ/K move · Enter open · S save

Navigating the Alignment-Calibration Trade-off: A Pareto-Superior Frontier via Model Merging

Tiancheng Hu, Benjamin Minixhofer, Nigel Collier

University of Cambridge

LLM Alignment Model Calibration Multi-Objective Language Model Alignment Language Model Merging

Open paper AskPDF arXiv

The "alignment tax" of post-training is typically framed as a drop in task accuracy. We show it also involves a severe loss of calibration, making models overconfident, less reliable, and model outputs less diverse. We show that this trade-off can be navigated effectively via a simple post-hoc intervention: interpolating between a model's weights before and after alignment. Crucially, this is not a strict trade-off. We find that the process consistently reveals Pareto-optimal interpolations - models that improve accuracy beyond both parents while substantially recovering the calibration lost during alignment. Our work demonstrates that simple model merging provides a computationally efficient method for mitigating the full scope of the alignment tax, yielding models that are more capable and more reliable.

Momentum

Related topics

Latest papers 62

Navigating the Alignment-Calibration Trade-off: A Pareto-Superior Frontier via Model Merging

Adaptive Helpfulness-Harmlessness Alignment with Preference Vectors