A Study on Large Language Models in Multi Modal Learning

Data in today's applications is becoming more complicated and diverse; therefore, models have had to be developed that can process and merge information not only from text but also from images or even sound. Two important areas where LLMs have also been applied include natural language processing (NLP) as well as multimodallearning. This paper investigates how LLMs have been used in multimodal learning environments. Specifically, this study investigates how these models are adapting themselves for them to be able to handle text, visual but more especially audio data. Exploring the techniques used to cross these modalities so well enables us to understand all what it imposes. Thus, this study traces the laying points of observations starting with new horizons addressing some of the strengths and weaknesses of current modalities to capture what it means for future research ideas and application purposes. In this perspective, an attempt has been made to comprehend all the possible ways that exist when dealing with LLMs and multi-modal activities through creating dialogue in line with its current relevance.