Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Yanwei, Zhang, Yuechen, Wang, Chengyao, Zhong, Zhisheng, Chen, Yixin, Chu, Ruihang, Liu, Shaoteng, Jia, Jiaya
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917624395333632
author Li, Yanwei
Zhang, Yuechen
Wang, Chengyao
Zhong, Zhisheng
Chen, Yixin
Chu, Ruihang
Liu, Shaoteng
Jia, Jiaya
author_facet Li, Yanwei
Zhang, Yuechen
Wang, Chengyao
Zhong, Zhisheng
Chen, Yixin
Chu, Ruihang
Liu, Shaoteng
Jia, Jiaya
contents In this work, we introduce Mini-Gemini, a simple and effective framework enhancing multi-modality Vision Language Models (VLMs). Despite the advancements in VLMs facilitating basic visual dialog and reasoning, a performance gap persists compared to advanced models like GPT-4 and Gemini. We try to narrow the gap by mining the potential of VLMs for better performance and any-to-any workflow from three aspects, i.e., high-resolution visual tokens, high-quality data, and VLM-guided generation. To enhance visual tokens, we propose to utilize an additional visual encoder for high-resolution refinement without increasing the visual token count. We further construct a high-quality dataset that promotes precise image comprehension and reasoning-based generation, expanding the operational scope of current VLMs. In general, Mini-Gemini further mines the potential of VLMs and empowers current frameworks with image understanding, reasoning, and generation simultaneously. Mini-Gemini supports a series of dense and MoE Large Language Models (LLMs) from 2B to 34B. It is demonstrated to achieve leading performance in several zero-shot benchmarks and even surpasses the developed private models. Code and models are available at https://github.com/dvlab-research/MiniGemini.
format Preprint
id arxiv_https___arxiv_org_abs_2403_18814
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models
Li, Yanwei
Zhang, Yuechen
Wang, Chengyao
Zhong, Zhisheng
Chen, Yixin
Chu, Ruihang
Liu, Shaoteng
Jia, Jiaya
Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
In this work, we introduce Mini-Gemini, a simple and effective framework enhancing multi-modality Vision Language Models (VLMs). Despite the advancements in VLMs facilitating basic visual dialog and reasoning, a performance gap persists compared to advanced models like GPT-4 and Gemini. We try to narrow the gap by mining the potential of VLMs for better performance and any-to-any workflow from three aspects, i.e., high-resolution visual tokens, high-quality data, and VLM-guided generation. To enhance visual tokens, we propose to utilize an additional visual encoder for high-resolution refinement without increasing the visual token count. We further construct a high-quality dataset that promotes precise image comprehension and reasoning-based generation, expanding the operational scope of current VLMs. In general, Mini-Gemini further mines the potential of VLMs and empowers current frameworks with image understanding, reasoning, and generation simultaneously. Mini-Gemini supports a series of dense and MoE Large Language Models (LLMs) from 2B to 34B. It is demonstrated to achieve leading performance in several zero-shot benchmarks and even surpasses the developed private models. Code and models are available at https://github.com/dvlab-research/MiniGemini.
title Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2403.18814