The Fifth Elephant 2023 Winter

Abstract

With the rapidly evolving world of Large Language Models (LLMs), there are numerous LLMs released each week. It is important to understand how these models align with various use-cases and technical configurations. The talk aims to provide a framework for understanding which LLMs are best suited for particular applications/ problem statements based on set of parameters like architectural details, quality of models, training process, kind of model, computational requirements, licensing etc many such parameters. We will delve into the nuances of selecting the right LLM for tasks ranging from simple language understanding, classification, summarization to complex problem-solving, text-generation tasks considering the constraints and capabilities of different software environments. This analysis is crucial for developers, researchers, and businesses to maximize the effectiveness and efficiency of LLM deployment in various scenarios.

What is covered in Presentation

The list of parameters which affect the LLMs. Why and How each parameter is affecting the LLM accuracy.

Whom is this targeted to

Data scientists

About Speaker

Mahathi Bhagavatula

She is an accomplished Data Scientist with over 10 years of experience in the field of Natural Language Processing and Machine Learning. Her expertise lies in Graph Neural Networks, NLP, Machine Learning and more recently Large Language Models (LLMs).

She did her Masters by Research in Machine Learning from IIIT Hyderabad and had published 3 ML research papers in top-tier ML international conferences. She is a recipient of ACM scholarship and many such accolades.