PromptLibrary

Abstract

The rapid advancement and integration of Large Language Models (LLMs) in academic research underscores the critical need for specialized, context-rich training datasets. The PromptLibrary project is designed to address this gap by establishing a comprehensive library of prompts tailored for academic libraries. This initiative aims to amass a wide array of instructional prompts, thereby forming an essential instruction dataset to enhance the utility and relevance of LLMs within scholarly domains. By encapsulating real-world, academic-specific inquiries and scenarios, this dataset is poised to significantly improve LLMs' learning capabilities and adaptability. The project features a web-based, searchable repository that allows for the submission and retrieval of high-quality prompts, ensuring a robust quality assurance mechanism for prompt validation. This repository not only serves as a critical resource for prompt tuning LLMs but also fosters a collaborative environment for librarians, educators, and researchers, thereby advancing the narrative and utility of LLMs in academic settings.

Description

The files that are being uploaded to this entry in VTechWorks: PromptLibraryData.zip - The .zip file that contains the project code files. The project's code is also available at https://github.com/narwhalle/PromptLibrary. PromptLibraryReport.pdf and PromptLibraryReport.docx - The PDF and DOCX file that documents what the project is, how to use the application from the user's perspective, the overall structure of the project, specification for future developments, and references to the technologies used for the project. PromptLibraryPresentation.pptx and PromptLibraryPresentation.pdf - The PPTX and PDF files of the PromptLibrary presentation that detail the overall project visually in terms of used technology, frontend walkthrough, backend walkthrough, lessons learned from development, and future work for further development.

Keywords

Web Application, Database, LLM, AI

Citation