AINeutralarXiv – CS AI · 6h ago6/10
🧠
LARK: Learnability-Grounded Trajectory Selection for Efficient Reasoning Distillation
LARK introduces a learnability-grounded approach to trajectory selection for reasoning distillation, enabling student models to learn more efficiently from teacher-generated reasoning paths. The method uses a learnability factor to identify trajectories that maximize learning speed while maintaining distributional coverage, outperforming existing heuristic-based selection methods across multiple reasoning tasks.