TAGCOS: Task-agnostic Gradient Clustered Coreset Selection for Instruction Tuning Data
This addresses the problem of high computational costs for researchers and practitioners using large instruction datasets, though it is incremental as it builds on existing coreset selection techniques.
The paper tackles the computational burden of instruction tuning by proposing TAGCOS, a method to select a small, informative subset of data, achieving performance close to the full dataset with only 5% of the data.
Instruction tuning has achieved unprecedented success in NLP, turning large language models into versatile chatbots. However, the increasing variety and volume of instruction datasets demand significant computational resources. To address this, it is essential to extract a small and highly informative subset (i.e., Coreset) that achieves comparable performance to the full dataset. Achieving this goal poses non-trivial challenges: 1) data selection requires accurate data representations that reflect the training samples' quality, 2) considering the diverse nature of instruction datasets, and 3) ensuring the efficiency of the coreset selection algorithm for large models. To address these challenges, we propose Task-Agnostic Gradient Clustered COreset Selection (TAGCOS). Specifically, we leverage sample gradients as the data representations, perform clustering to group similar data, and apply an efficient greedy algorithm for coreset selection. Experimental results show that our algorithm, selecting only 5% of the data, surpasses other unsupervised methods and achieves performance close to that of the full dataset.