AI summaryⓘ
The authors address a problem where small AI models (students) struggle to learn well from bigger ones (teachers) because directly copying the teacher’s outputs can make the student overly narrow in its learning. They propose a new method called Zone of Proximal Policy Optimization (ZPPO) that guides the student by including the teacher's help as part of the questions rather than in the training signals, helping the student focus on learning from mistakes. Their approach uses special prompts that show both right and wrong answers to help the student improve gradually. Testing on various vision and language tasks showed that ZPPO helped smaller models learn better than previous methods, especially the smallest students.
Knowledge DistillationPolicy GradientReinforcement LearningZone of Proximal DevelopmentPrompt EngineeringVision-Language ModelsRolloutsOn-Policy LearningModel CompressionTeacher-Student Training
Authors
Byung-Kwan Lee, Ximing Lu, Shizhe Diao, Minki Kang, Saurav Muralidharan, Karan Sapra, Andrew Tao, Pavlo Molchanov, Yejin Choi, Yu-Chiang Frank Wang, Ryo Hachiuma
Abstract
Knowledge distillation transfers a teacher's competence to a small student but is brittle in the small-student regime: forcing the student to imitate logits from a much larger teacher concentrates it on the teacher's sharpest modes, hurting generalization on benchmark families beyond the training corpus. Reinforcement learning (RL) avoids logit imitation by training on the student's own rollouts. However, on questions where every rollout fails-yielding zero advantage and being silently discarded-injecting a stronger teacher's response into the policy gradient breaks the on-policy assumption and induces drift. We introduce Zone of Proximal Policy Optimization (ZPPO), inspired by Vygotsky's zone of proximal development, which keeps the teacher inside the prompt rather than the policy gradient. On hard questions, ZPPO constructs two reformulated prompts: a Binary Candidate-included Question (BCQ) pairs one correct teacher response with one incorrect student response as anonymized candidates the student must discriminate, and a Negative Candidate-included Question (NCQ) aggregates the student's wrong rollouts into a single prompt to surface their shared failure modes. A prompt replay buffer recirculates each hard question until it either graduates-the student's mean rollout accuracy on it reaches half- or is FIFO-evicted under finite capacity, amplifying BCQ and NCQ inside the student's current zone of proximal development. On the Qwen3.5 family at four student scales (0.8B-9B) with a 27B teacher, post-trained as vision-language models and evaluated on a 31-benchmark suite (16 VLM, 10 LLM, 5 Video), ZPPO outperforms off/on-policy distillation and GRPO, with the largest gains at the smallest scale.