Find Your Optimal Teacher: Personalized Data Synthesis via Router-Guided Multi-Teacher Distillation

2025-10-12

AI Summary

Prior work reveals that synthetic data generated by strong teacher models does not necessarily yield optimal learning outcomes for student models—highlighting a fundamental mismatch between teacher output quality and student learnability. To address this, we propose PerSyn, the first framework to introduce a “route-then-generate” paradigm: a query-level router jointly models student learnability and teacher response quality to dynamically assign each query to its optimal teacher. PerSyn further integrates multi-teacher distillation, router-guided data allocation, and conditional generation strategies to enable personalized instruction tuning and mathematical reasoning training across diverse model families and scales. Experiments demonstrate that PerSyn significantly improves knowledge distillation efficiency and final performance across a range of student models—from small to medium-sized—consistently outperforming or matching state-of-the-art baselines.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Training student models on synthetic data generated by strong teacher models is a promising way to distilling the capabilities of teachers. However, recent studies show that stronger models are not always optimal teachers, revealing a mismatch between teacher outputs and student learnability. To address this issue, we propose PerSyn (Personalized data Synthesis), a novel synthesis strategy that operates under a new ``Route then Generate'' paradigm to create data tailored to each student model, enabling it to learn more effectively. Specifically, PerSyn first assigns each prompt to its optimal teacher via a query-level router that jointly considers student learnability and teacher response quality. Each teacher then synthesizes data only for its assigned prompts, making the process more efficient than the conventional ``Generate then Select'' paradigm, where all teachers must generate parallel responses for the entire prompt set before constructing the final dataset. Extensive experiments across different model families and scales demonstrate that PerSyn consistently achieves superior or comparable performance to all baselines in instruct tuning and math reasoning settings. Further analysis verifies the effectiveness of PerSyn and offers extra insights to propel future research.

Problem

• Personalized data synthesis addresses student-teacher mismatch
• Router assigns prompts to optimal teachers for efficiency
• Tailored data generation improves student model learnability

Innovation

• Router assigns prompts to optimal teachers
• Teachers generate personalized data for students
• Route then Generate paradigm improves efficiency