Leveraging Custom Data to Retrain Large Language Models : A Time-Efficient Approach

This approach involves fine-tuning pre-trained LLMs with domain-specific or task-specific datasets, leveraging transfer learning to capitalize on pre-existing knowledge. By adapting representations and predictions to specific task nuances, the model enhances performance without requiring complete retraining. Benefits of this method include reduced data requirements, as pre-trained models possess foundational language understanding, streamlining model development. Additionally, building upon existing architectures accelerates development cycles, ensuring faster achievement of desired performance levels. Moreover, custom data retraining enables swift adaptation to evolving contexts or user needs, facilitating agile deployment.
Who is this presentation for?
AI and Gen AI technical attendees.
Prerequisite knowledge:
Basic knowledge on AI and GenAI
What you'll learn?
How to fine tune an existing LLM with additional data.

I am an SAP Technical Architect with 15+ years of experience in the IT industry and a certified PMP & SAP Cloud Integration and application programming consultant. I worked extensively in enterprise and cloud-native applications, focusing on application and system architecture, design, development, review, and deployment of cloud applications. I am currently an SAP Technical Architect at SAP America, Inc., with prior experience working at Reliance Communication Ltd and ProKarma Inc. I have experience in leading technical full-stack development in SAP ABAP & UI5. I specialize in SAP cloud application programming, SAP cloud integration with legacy systems, cloud migration, automation, and security. I have worked in retail, automobile, oil & gas, manufacturing, utility, banking, and financial services. I hold a Master’s in Enterprise Systems Professional from Sheffield Hallam University, a Second Master’s in Organic Chemistry from Osmania University, and a Doctorate in Business Administration from Georgia State University. I have published several articles on Supply Chain Management optimization using AI and predictive analytics. I am currently serving as Associate Editor of IJICTE, IJSI, and as an Editorial board member of several other IGI Global journals(IRMJ, IJISSCM, IJISMD, IJAMTR, IJBSA, IJAIBM, IJPMAE ). I am also currently serving as Secretary of IEEE Consultants Network in Austin.