Dynamic Knowlege Addition on Existing Large Language Model

Nugroho, Robertus Setiawan Aji Dynamic Knowlege Addition on Existing Large Language Model. Project Report. SCU. (Unpublished)

	Text laporan_lengkap.pdf Restricted to Registered users only Download (3MB)
Preview	Text sertifikat_EC00202443522_aikas.pdf Download (2MB) | Preview

Abstract

The exponential growth on the use of generative pre-trained transformers based on large language model has resulted in the wide varieties of implementations, including in private organization, government bodies, health related applications, and more other different areas. However, it is quite challenging to effectively add new and specific knowledge for training on top of the existing and pre- trained large language model. In this research, we propose a new model to incorporate new knowledge into the existing language model dynamically. We use chatGPT and llama3 as the basis for input and output smoothing, but we limit the knowledge to our dynamically supplied content only. This research is the extension of our previous work where we implement the hybrid solution for the automatic and manual social media content classification. We extend the capability of our previous research by expanding the context using the large language model for user question and asking system implementation. We apply the proposed method into a church document-based question and asking.

Item Type:	Monograph (Project Report)
Subjects:	000 Computer Science, Information and General Works 000 Computer Science, Information and General Works > 004 Data processing & computer science
Divisions:	Faculty of Computer Science > Department of Informatics Engineering
Depositing User:	Mr R. Setiawan Aji Nugroho
Date Deposited:	07 Aug 2024 01:21
Last Modified:	07 Aug 2024 01:21
URI:	http://repository.unika.ac.id/id/eprint/36025

Actions (login required)

View Item