Dynamic Data Mixing Maximizes Instruction Tuning for Mixture-of-Experts

TL;DR

This paper introduces a dynamic data mixing strategy for MoE instruction tuning that adjusts dataset sampling weights based on inter-redundancies, improving model performance across various tasks.

Contribution

It proposes the first dynamic data mixture method for MoE instruction tuning, leveraging dataset representations to optimize sampling weights adaptively.

Findings

Enhanced performance on downstream tasks

Effective reduction of dataset redundancies

Improved open-ended query results

Abstract

Mixture-of-Experts (MoE) models have shown remarkable capability in instruction tuning, especially when the number of tasks scales. However, previous methods simply merge all training tasks (e.g. creative writing, coding, and mathematics) and apply fixed sampling weights, without considering the importance of different tasks as the model training state changes. In this way, the most helpful data cannot be effectively distinguished, leading to suboptimal model performance. To reduce the potential redundancies of datasets, we make the first attempt and propose a novel dynamic data mixture for MoE instruction tuning. Specifically, inspired by MoE's token routing preference, we build dataset-level representations and then capture the subtle differences among datasets. Finally, we propose to dynamically adjust the sampling weight of datasets by their inter-redundancies, thus maximizing global performance under a limited training budget. The experimental results on two MoE models demonstrate the effectiveness of our approach on both downstream knowledge \& reasoning tasks and open-ended queries. Code and models are available at https://github.com/Spico197/MoE-SFT .