COS802 – A comparative Study of LLMs with Similar Training Objectives Across Key NLP Tasks [Video]

Mihle Senene

A comparative Study of LLMs with Similar Training Objectives Across Key NLP Tasks

📖 Final project presentation for COS 802 – https://dsfsi.github.io/cos802 (Taught by Prof Vukosi Marivate). Course is part of the Big Data Science Masters at University of Pretoria.

“The rapid advancement of machine learning, particularly in Natural Language Processing (NLP), has provided end-users with a large selection of model options for completing various tasks. However, the abundance of choices can ironically lead to information overload and decision fatigue.

This project evaluates how various Large Language Models (LLMs) with similar training objectives compare in their effectiveness across different NLP tasks – specifically, text classification, sentiment analysis, and topic modelling. The objective is to provide insights into the strengths and weaknesses of each LLM, presenting researchers and businesses with guidance on selecting the most suitable model for specific tasks and contexts.”