abstract

This study explores the use of programmed instruction in prompt engineering for large language models (LLMs) to enhance their understanding and response to user prompts. By applying programmed instruction principles to prompt design, the research uses modified prompts for standard evaluations and human assessments to measure their impact on model performance. Expected outcomes include improved scores in standard tests like BIG-Bench and better handling of complex tasks. The findings aim to provide users with effective communication strategies for using LLMs, offering new insights for prompt engineering. 

goal

• Familiarize myself with academic writing formats through writing papers.

• Conduct logically designed experiments, perform meaningful statistical analyses, and present research results with clear charts and graphs.

• Submit to conferences and seek opportunities for publication.

At the workshop, I tried using the eye-tracker for data presentation related to text.