In January 2025, we kick off a year packed with events aimed at helping you…
The EUATC has teamed up with GALA to bring you a three-part webinar series exploring prompt engineering for Large Language Model (LLMs) to be delivered by Marina Pantcheva, RWS‘s Director of Linguistic Services.
This 3-module webinar explores the art and science of prompt design for Large Language Models (LLMs), with focus on their usage in the translation and the localization industry.
Each module comprises a one-hour session that explores different aspects of LLMs —from understanding their inner workings, through strategies for optimizing prompts, to methods for evaluating LLM output.
Who is this for?
- Localization professionals and project managers
- Translators, post-editors, language specialists
- Content creators
- Localization vendor managers
- Educators and university lecturers in Translation Studies
- Product managers of AI tools to be used by linguists and translators
The series is free to EUATC network members (including ATC members) and GALA members. There is a charge of $75 per session for non-members of both organisations. Follow the links below and create a few GALA account (if you haven’t got one already), then use discount code WEBINAR_09JAN_EUATC_RWS to activate your free ATC member access.
Session 1 | Thursday 09 January 2025 | 5pm – 6pm CET
Demystiying LLMs: What can they actually do well?
This first session of the 3-module prompt design webinar introduces the mechanics of LLMs, explaining how their training and your prompts define the output. The session explores AI’s capabilities, limitations, and practical use cases, with special focus on localization, including:
- How do LLMs learn what they know?
- Machine learning specialists versus shamans: What’s the difference?
- LLM personalities: Why do we have our favorite LLM?
- Strengths and weaknesses of AI.
- Proven AI use cases in localization.
- Debunking myths about AI.
SIGN UP FOR WEBINAR 1# AI IN LOCALIZATION SERIES
Session 2 | Thursday 16 January 2025 | 5pm – 6pm CET
The science and art of prompt design
This is the second part from the 3-module webinar on prompt design. This hands-on session equips you with practical knowledge and techniques for creating prompts. It covers various prompting methods, hand-on exercises and troubleshooting strategies. Topics include:
- Core components of prompts.
- How to break down complex prompts into manageable parts.
- Introduction to genetic algorithms for prompt design (easier than it sounds!).
- Practical tips and tricks for effective prompting.
- Troubleshooting: what steps to take when the LLM doesn’t do what you expect.
- Does saying “please” pay off?
SIGN UP FOR WEBINAR 2# AI IN LOCALIZATION SERIES
Session 3 | Thursday 30 January 2025 | 5pm – 6pm CET
Evaluating LLM responses
The final session of the 3-part prompt design webinar focuses on evaluating the output of prompts, diagnosing issues, and refining the approach. You’ll learn to measure LLM performance, choose the right evaluation data set, and mitigate risks. Topics include:
- Methods for prompt evaluation.
- Understanding the F1 score.
- Types of hallucinations and strategies for managing them.
- Recognizing and addressing data contamination and other risks.
- Best practices for robust prompt design.
Related Posts
