Tag
CRAFT is a Pareto-front prompt optimizer that jointly optimizes for accuracy and token cost, avoiding the 'scalarization collapse' of weighted-sum approaches by maintaining a diverse population of prompts across the accuracy-cost trade-off frontier using NSGA-II and budget-aware validation.
Presents WeCon, a weight-conditioned neural solver for multi-objective combinatorial optimization problems that achieves comparable hypervolume to the state-of-the-art while reducing inference time by 40%.
This paper introduces MOCI, a novel framework for inferring shared constraints and individual preferences from heterogeneous expert demonstrations in reinforcement learning, outperforming existing baselines in predictive performance and computational efficiency.