Saved in:
Bibliographic Details
Main Authors: Gu, Shuhao, Zhang, Jialing, Zhou, Siyuan, Yu, Kevin, Xing, Zhaohu, Wang, Liangdong, Cao, Zhou, Jia, Jintao, Zhang, Zhuoyi, Wang, Yixuan, Hu, Zhenchong, Zhang, Bo-Wen, Li, Jijie, Liang, Dong, Zhao, Yingli, Wang, Songjing, Ao, Yulong, Ju, Yiming, Ma, Huanhuan, Li, Xiaotong, Diao, Haiwen, Cui, Yufeng, Wang, Xinlong, Liu, Yaoqi, Feng, Fangxiang, Liu, Guang
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2410.18558
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929659208269824
author Gu, Shuhao
Zhang, Jialing
Zhou, Siyuan
Yu, Kevin
Xing, Zhaohu
Wang, Liangdong
Cao, Zhou
Jia, Jintao
Zhang, Zhuoyi
Wang, Yixuan
Hu, Zhenchong
Zhang, Bo-Wen
Li, Jijie
Liang, Dong
Zhao, Yingli
Wang, Songjing
Ao, Yulong
Ju, Yiming
Ma, Huanhuan
Li, Xiaotong
Diao, Haiwen
Cui, Yufeng
Wang, Xinlong
Liu, Yaoqi
Feng, Fangxiang
Liu, Guang
author_facet Gu, Shuhao
Zhang, Jialing
Zhou, Siyuan
Yu, Kevin
Xing, Zhaohu
Wang, Liangdong
Cao, Zhou
Jia, Jintao
Zhang, Zhuoyi
Wang, Yixuan
Hu, Zhenchong
Zhang, Bo-Wen
Li, Jijie
Liang, Dong
Zhao, Yingli
Wang, Songjing
Ao, Yulong
Ju, Yiming
Ma, Huanhuan
Li, Xiaotong
Diao, Haiwen
Cui, Yufeng
Wang, Xinlong
Liu, Yaoqi
Feng, Fangxiang
Liu, Guang
contents Recently, Vision-Language Models (VLMs) have achieved remarkable progress in multimodal tasks, and multimodal instruction data serves as the foundation for enhancing VLM capabilities. Despite the availability of several open-source multimodal datasets, limitations in the scale and quality of open-source instruction data hinder the performance of VLMs trained on these datasets, leading to a significant gap compared to models trained on closed-source data. To address this challenge, we introduce Infinity-MM, a large-scale multimodal instruction dataset. We collected the available multimodal instruction datasets and performed unified preprocessing, resulting in a dataset with over 40 million samples that ensures diversity and accuracy. Furthermore, to enable large-scale expansion of instruction data and support the continuous acquisition of high-quality data, we propose a synthetic instruction generation method based on a tagging system and open-source VLMs. By establishing correspondences between different types of images and associated instruction types, this method can provide essential guidance during data synthesis. Leveraging this high-quality data, we have trained a 2-billion-parameter Vision-Language Model, Aquila-VL-2B, which achieves state-of-the-art (SOTA) performance among models of similar scale. The data is available at: https://huggingface.co/datasets/BAAI/Infinity-MM.
format Preprint
id arxiv_https___arxiv_org_abs_2410_18558
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data
Gu, Shuhao
Zhang, Jialing
Zhou, Siyuan
Yu, Kevin
Xing, Zhaohu
Wang, Liangdong
Cao, Zhou
Jia, Jintao
Zhang, Zhuoyi
Wang, Yixuan
Hu, Zhenchong
Zhang, Bo-Wen
Li, Jijie
Liang, Dong
Zhao, Yingli
Wang, Songjing
Ao, Yulong
Ju, Yiming
Ma, Huanhuan
Li, Xiaotong
Diao, Haiwen
Cui, Yufeng
Wang, Xinlong
Liu, Yaoqi
Feng, Fangxiang
Liu, Guang
Computation and Language
Recently, Vision-Language Models (VLMs) have achieved remarkable progress in multimodal tasks, and multimodal instruction data serves as the foundation for enhancing VLM capabilities. Despite the availability of several open-source multimodal datasets, limitations in the scale and quality of open-source instruction data hinder the performance of VLMs trained on these datasets, leading to a significant gap compared to models trained on closed-source data. To address this challenge, we introduce Infinity-MM, a large-scale multimodal instruction dataset. We collected the available multimodal instruction datasets and performed unified preprocessing, resulting in a dataset with over 40 million samples that ensures diversity and accuracy. Furthermore, to enable large-scale expansion of instruction data and support the continuous acquisition of high-quality data, we propose a synthetic instruction generation method based on a tagging system and open-source VLMs. By establishing correspondences between different types of images and associated instruction types, this method can provide essential guidance during data synthesis. Leveraging this high-quality data, we have trained a 2-billion-parameter Vision-Language Model, Aquila-VL-2B, which achieves state-of-the-art (SOTA) performance among models of similar scale. The data is available at: https://huggingface.co/datasets/BAAI/Infinity-MM.
title Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data
topic Computation and Language
url https://arxiv.org/abs/2410.18558