ΠΕΡΙΓΡΑΦΗ

Master prompt engineering to unlock the full potential of LLMs effortlessly. Compose precise prompts, tailor AI behavior, analyze documents, detect objects from images and craft AI assistants efficiently.

ΣΚΟΠΟΣ ΣΕΜΙΝΑΡΙΟΥ

Upon completion, participants will:

• Write precise prompts for desired LLM behavior.
• Edit system messages proficiently.
• Guide LLMs through diverse prompt engineering techniques.
• Enhance chatbot behavior using prompt-response history.

ΣΕ ΠΟΙΟΥΣ ΑΠΕΥΘΥΝΕΤΑΙ

• This seminar is designed for either business or academic peers to get familiar with prompt engineering beyond a general overview, as the sample implementations may be helpful to integrate the knowledge to use Large Language Models more efficiently.
• Anyone who would like to improve his/her skills for more efficient prompting during everyday routine using e.g., popular chat applications.

Prerequisites for Participants: 

Little experience in communicating with large language models. A basic knowledge of Python is certainly useful, although the notebooks are sufficiently illustrated, explained and automated for hands-on training to be easily done.

ΠΕΡΙΣΣΟΤΕΡΕΣ ΠΛΗΡΟΦΟΡΙΕΣ

Seminar Overview

We learn to master Prompt Engineering in this comprehensive executive course:

• Present its importance and tailor LLM behavior effectively.
• Explore practical applications including document analysis, tailored text generation, and AI assistant implementation.
• Utilize effective tools through hands-on practice.
• Advance with iterative prompt writing, system message editing, and one-to-many shot prompt engineering.
• Enhance chatbot performance by integrating prompt-response history and employing dynamic strategies.

 The course encloses short presentations to explain the main concepts.

 Moreover, participants can get familiar with the implementations by navigating Jupyter notebooks in Python prepared for the different topics.

Topics Covered

 Understanding Prompt Engineering:

• Grasping the importance of prompt engineering.
• Exploring ways to tailor LLM behavior.

 Practical Applications:

• Document analysis with prompt-engineered LLaMA models.
• Tailored text generation.
• Implementing LLMs as AI assistants.

 Utilizing Effective Prompt Engineering Tools:

• Familiarization with prompt engineering tools.
• Hands-on practice in prompt customization.

 Advanced Techniques:

• Iterative prompt writing for precise LLM behavior.
• Optimizing results through system message editing.
• One-to-many shot prompt engineering guidance.

 Improving Chatbot Performance:

• Integrating prompt-response history for enriched context.
• Strategies for dynamic chatbot behavior.

Prompt Engineering Examples

Use Cases:

• Document analyst
• Marketing copy writer
• Unique AI personalities
• Chatbot with conversation history
• Customer assistant
• Program code generation (demonstration)
• Quiz generation (demonstration)
• Lyrics and music generation (demonstration)
• Object detection from images by prompts (demonstration)

Main Concepts:

• precise prompts
• “time to think”
• cues to guide response
• sentiment analysis
• prompt template
• instruction fine-tuning
• few-shot learning
• system context
• random sampling for non-deterministic responses
• adjusting temperature to control the degree of randomness
• retaining conversation history
• token limits

Participants will receive a signed Certificate-of-Attendance issued by the Executive Education Center of the University of Limassol upon completion of the course.

DR ANDRÁS HAJDU

AI & Deep Learning Expert

András Hajdu received an MSc degree in Mathematics from the University of Debrecen, Hungary, in 1996. He obtained his PhD degree in Mathematics and Computer Science in 2003. He worked as a PostDoc researcher for the Artificial Intelligence Information Analysis Laboratory, Dept. of Informatics, Aristotle University of Thessaloniki between 2005-2006. Since 2017 he has been a full professor, since 2011 the head of the Department of Data Science and Visualization, and since 2019 the dean of the Faculty of Informatics, University of Debrecen.

He is the leader of the local Data Science and Visualization doctoral program and the founder and director of the Gyorgy Hajos Data Science Students College. He worked as a data scientist for the Hungarian Data Asset Agency between 2019-2022.

He serves as an instructor for the Microsoft Learn for Educators program and is a certified NVIDIA instructor and university ambassador. He has co-authored 57 journal papers and 120 conference papers cited more than 3500 times; his H-index is 29. His main interest lies in data science/artificial intelligence, medical image processing, and discrete mathematics.

DR BALÁZS HARANGI

Data Science & Visualization Expert

Dr Balázs Harangi received his MSc degrees in Program Engineering and Mathematics from University of Debrecen, Faculty of Informatics, in 2010. From 2010 to 2013 he was a PhD student at Faculty of Informatics and he obtained his PhD degree in Informatics (Medical Image Processing) from the University of Debrecen, Hungary, in 2015, University of Debrecen. Since 2013 he has been serving as Assistant Lecturer, later as Assistant Professor and as Associate Professor at Department of Data Science and Visualization, Faculty of Informatics, University of Debrecen.

As a lecturer at the University of Debrecen, he gives lectures and seminars for several courses in the fields of artificial intelligence and machine learning for BS.c, MS.c and Ph.D students. He has been and is currently the thesis supervisor of several students and is continuously involved in postgraduate education as a lecturer at the Doctoral School of Computer Science, University of Debrecen. He is also the subject supervisor of 7 active doctoral students. He is involved in several mentoring programs, where he acts as a mentor for young and talented students. In addition to teaching traditional university courses, he is also involved in the delivery of NVIDIA and Microsoft industry certification exam preparation courses.

He is a member of the IEEE, the John von Neumann Computer Society (Hungary) and member the Hungarian Association for Image Analysis and Pattern Recognition.  He has authored or co-authored 72 publications cited more than 1500 times; his H-index is 16. His primary research fields are the digital/medical image processing, pattern recognition, machine learning/deep learning.