Concept-skill Transferability-based Data Selection for Large Vision-Language Models

TL;DR

This paper introduces COINCIDE, a scalable data selection method for efficient finetuning of large vision-language models, focusing on diversity and transferability to reduce training costs while maintaining high performance.

Contribution

COINCIDE uses a small model to cluster and select diverse, transferable VL data, significantly reducing training data and time without sacrificing model performance.

Findings

Achieves comparable performance with only 20% of data on LLaVA-1.5.

Reduces training time by 70% on LLaVA-1.5.

Outperforms 8 baselines on two VL datasets.

Abstract

Instruction tuning, or supervised finetuning on extensive task-specific data, is necessary for Large Vision-Language Models (LVLMs) to generalize well across a broad range of vision-language (VL) tasks. However, training on large VL datasets can become prohibitively expensive. In this work, we introduce COINCIDE, an effective and scalable data selection technique that uses a small model as a reference model to select visual instruction tuning data for efficient finetuning of a target LVLM, focusing on diversity and transferability. Specifically, we cluster the training data using internal activations from a small model, which identifies VL concept-skill compositions needed by a target LVLM. We then sample data from these diverse clusters by considering their density and transferability, or the ability to transfer well to other concept-skill compositions. This approach ensures the diversity of these compositions, which is vital for LVLM generalization. Extensive experiments demonstrate that COINCIDE achieves superior performance and data selection efficiency against 8 strong baselines on two distinct datasets: LLaVA-1.5 and Vision-Flan. Using only 20% of the LLaVA-1.5 dataset, COINCIDE achieves performance comparable to the LVLM finetuned on the whole dataset, with 70% reduction of the wall-clock running time. On the Vision-Flan dataset, our method achieves superior results with only 16.7% of the training data.