Concepts or Skills? Rethinking Instruction Selection for Multi-modal Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bai, Andrew, Cui, Justin, Wang, Ruochen, Hsieh, Cho-Jui
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913991104659456
author Bai, Andrew
Cui, Justin
Wang, Ruochen
Hsieh, Cho-Jui
author_facet Bai, Andrew
Cui, Justin
Wang, Ruochen
Hsieh, Cho-Jui
contents Vision-language instruction tuning achieves two main purposes: learning visual concepts and learning visual skills. In this paper, we found that vision-language benchmarks fall into the dichotomy of mainly benefiting from training on instructions with similar skills or visual concepts. Inspired by the discovery, we designed a simple targeted training data selection method to optimize the performance of a given benchmark. We first extract the concepts/skills from the benchmark, determine whether the benchmark predominantly benefits from similar concepts or skills, and finally select instructions with the most matching concepts/skills. Experiments on 10+ benchmarks validate the effectiveness of our targeted data selection method, showing +0.9\% over the best existing baseline averaged over all benchmarks and +1.5\% on the skill-focused subset. Our findings underscore the importance of recognizing the inherent trade-off within instruction selection, which requires balancing the acquisition of conceptual knowledge against visual skill.
format Preprint
id arxiv_https___arxiv_org_abs_2508_10339
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Concepts or Skills? Rethinking Instruction Selection for Multi-modal Models
Bai, Andrew
Cui, Justin
Wang, Ruochen
Hsieh, Cho-Jui
Computer Vision and Pattern Recognition
Machine Learning
Vision-language instruction tuning achieves two main purposes: learning visual concepts and learning visual skills. In this paper, we found that vision-language benchmarks fall into the dichotomy of mainly benefiting from training on instructions with similar skills or visual concepts. Inspired by the discovery, we designed a simple targeted training data selection method to optimize the performance of a given benchmark. We first extract the concepts/skills from the benchmark, determine whether the benchmark predominantly benefits from similar concepts or skills, and finally select instructions with the most matching concepts/skills. Experiments on 10+ benchmarks validate the effectiveness of our targeted data selection method, showing +0.9\% over the best existing baseline averaged over all benchmarks and +1.5\% on the skill-focused subset. Our findings underscore the importance of recognizing the inherent trade-off within instruction selection, which requires balancing the acquisition of conceptual knowledge against visual skill.
title Concepts or Skills? Rethinking Instruction Selection for Multi-modal Models
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2508.10339