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

Open in new window