Instruction-Following Evaluation of Large Vision-Language Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Shiono, Daiki, Miyawaki, Shumpei, Tanaka, Ryota, Suzuki, Jun
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917174189228032
author Shiono, Daiki
Miyawaki, Shumpei
Tanaka, Ryota
Suzuki, Jun
author_facet Shiono, Daiki
Miyawaki, Shumpei
Tanaka, Ryota
Suzuki, Jun
contents Following the initial flourishing of large language models (LLMs), there has been a surge in proposed large vision-language models (LVLMs) that integrate LLMs with vision capabilities. However, it has been observed that LVLMs, after tuning to visual instruction using commonly used training datasets, often fail to exhibit the instruction-following ability that was present in the LLM before integration, leading to results in which they do not follow task instructions as expected. This study quantitatively demonstrates that LVLMs' instruction-following ability declines after fine-tuning and analyzes its underlying causes. In particular, we constructed new training datasets highlighting whether the output format is specified. Then, we investigated how explicitly indicating the output format during fine-tuning affects LVLMs' instruction-following ability. Our quantitative evaluation confirmed that LVLMs' instruction-following ability declines after fine-tuning with commonly used datasets. Furthermore, we found that LVLMs trained with datasets, including instructions on output format, tend to follow instructions more accurately than models that do not. These findings suggest that including samples with instructions on output format during (visual) instruction tuning may help mitigate the decline in instruction-following abilities.
format Preprint
id arxiv_https___arxiv_org_abs_2512_23572
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Instruction-Following Evaluation of Large Vision-Language Models
Shiono, Daiki
Miyawaki, Shumpei
Tanaka, Ryota
Suzuki, Jun
Computation and Language
Computer Vision and Pattern Recognition
Following the initial flourishing of large language models (LLMs), there has been a surge in proposed large vision-language models (LVLMs) that integrate LLMs with vision capabilities. However, it has been observed that LVLMs, after tuning to visual instruction using commonly used training datasets, often fail to exhibit the instruction-following ability that was present in the LLM before integration, leading to results in which they do not follow task instructions as expected. This study quantitatively demonstrates that LVLMs' instruction-following ability declines after fine-tuning and analyzes its underlying causes. In particular, we constructed new training datasets highlighting whether the output format is specified. Then, we investigated how explicitly indicating the output format during fine-tuning affects LVLMs' instruction-following ability. Our quantitative evaluation confirmed that LVLMs' instruction-following ability declines after fine-tuning with commonly used datasets. Furthermore, we found that LVLMs trained with datasets, including instructions on output format, tend to follow instructions more accurately than models that do not. These findings suggest that including samples with instructions on output format during (visual) instruction tuning may help mitigate the decline in instruction-following abilities.
title Instruction-Following Evaluation of Large Vision-Language Models
topic Computation and Language
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.23572