A survey of the major applications of transformer-based 100 billion+ parameter large language models

Large Language Models (LLMs) stand as a formidable force in the realm of Artificial Intelligence research and practical implementation. The parameter count in these models has undergone a remarkable surge, skyrocketing from mere millions to the realms of hundreds of billions and even trillions, concurrently elevating their overall performance. While the utilization of these models has captivated attention in domains like chatbots, harnessing their generative capabilities across language, certain other fields poised to benefit from their potency remain relatively unexplored. This paper aims to encapsulate recent advancements in LLMs boasting over 100 billion parameters, delving into their applications, associated challenges, limitations, and prospects for future exploration. A comprehensive examination is undertaken, shedding light on diverse use cases such as data annotation, summarization, cybersecurity, question answering, and software development by scrutinizing recent scholarly contributions. By adopting a broader perspective on the domains where these models find application, and elucidating the outcomes achieved therein, this paper endeavors to be a thorough guide for both academia and industry, facilitating a more effective deployment of LLMs.