STEPS: Selective On-Policy Self-Distillation for Reasoning
STEPS is proposed, a selective OPSD framework that controls the location, direction, and duration of distillation that improves mean accuracy over GRPO and support brief, targeted self-distillation as an efficient complement to reward-driven reasoning training.
Jia-Xuan Wang, Xuan-Xin Ouyang, Z. Chen et al.
· 15 citations