Montessori-Instruct: Generate Influential Training Data Tailored for Student Learning
This addresses the issue of inefficient training data for language model developers, offering a novel approach to enhance model performance, though it is incremental as it builds on existing synthesis and optimization techniques.
The paper tackles the problem of noisy synthetic data in training large language models by proposing Montessori-Instruct, a framework that tailors data synthesis to student learning preferences, resulting in relative performance improvements of 18.35% on Alpaca Eval and 46.24% on MT-Bench compared to standard methods.
Synthetic data has been widely used to train large language models, but their generative nature inevitably introduces noisy, non-informative, and misleading learning signals. In this paper, we propose Montessori-Instruct, a novel data synthesis framework that tailors the data synthesis ability of the teacher language model toward the student language model's learning process. Specifically, we utilize local data influence of synthetic training data points on students to characterize students' learning preferences. Then, we train the teacher model with Direct Preference Optimization (DPO) to generate synthetic data tailored toward student learning preferences. Experiments with Llama3-8B-Instruct (teacher) and Llama3-8B (student) on Alpaca Eval and MT-Bench demonstrate that Montessori-Instruct significantly outperforms standard synthesis methods by 18.35\% and 46.24\% relatively. Our method also beats data synthesized by a stronger teacher model, GPT-4o. Further analysis confirms the benefits of teacher's learning to generate more influential training data in the student's improved learning, the advantages of local data influence in accurately measuring student preferences, and the robustness of Montessori-Instruct across different student models. Our code and data are open-sourced at https://github.com/cxcscmu/Montessori-Instruct.