Efficient First-Order Optimization on the Pareto Set for Multi-Objective Learning under Preference Guidance | Read Paper on Bytez