Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models
It addresses the problem of improving visual dialog and reasoning in VLMs for AI applications, though it appears incremental as it builds on existing VLM frameworks.
The paper tackles the performance gap in multi-modality Vision Language Models (VLMs) compared to advanced models like GPT-4 and Gemini by introducing Mini-Gemini, a framework that enhances VLMs through high-resolution visual tokens, high-quality data, and VLM-guided generation, achieving leading performance in zero-shot benchmarks and surpassing some private models.
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.