SHED: Shapley-Based Automated Dataset Refinement for Instruction Fine-Tuning
This addresses the critical challenge of dataset redundancy in LLM fine-tuning, enabling more efficient and transferable data curation for AI practitioners.
The paper tackles the problem of identifying high-quality data from large datasets for instruction fine-tuning of LLMs, introducing SHED, a Shapley-based automated framework that eliminates human intervention and achieves performance comparable to or surpassing full datasets with only 10% of the data.
The pre-trained Large Language Models (LLMs) can be adapted for many downstream tasks and tailored to align with human preferences through fine-tuning. Recent studies have discovered that LLMs can achieve desirable performance with only a small amount of high-quality data, suggesting that a large amount of the data in these extensive datasets is redundant or even harmful. Identifying high-quality data from vast datasets to curate small yet effective datasets has emerged as a critical challenge. In this paper, we introduce SHED, an automated dataset refinement framework based on Shapley value for instruction fine-tuning. SHED eliminates the need for human intervention or the use of commercial LLMs. Moreover, the datasets curated through SHED exhibit transferability, indicating they can be reused across different LLMs with consistently high performance. We conduct extensive experiments to evaluate the datasets curated by SHED. The results demonstrate SHED's superiority over state-of-the-art methods across various tasks and LLMs; notably, datasets comprising only 10% of the original data selected by SHED achieve performance comparable to or surpassing that of the full datasets.