VITA: Towards Open-Source Interactive Omni Multimodal LLM
VITA: Towards Open-Source Interactive Omni Multimodal LLM
The remarkable multimodal capabilities and interactive experience of GPT-4o underscore their necessity in practical applications, yet open-source models rarely excel in both areas. In this paper, we introduce VITA, the first-ever open-source Multimodal Large Language Model (MLLM) adept at simultaneous processing and analysis of Video, Image, Text, and Audio modalities, …