Training: Prompting for translators and terminologists (german language)

The emergence of “large language models” such as ChatGPT in 2022 is a logical consequence and further development of neural machine translation, which has been part of everyday translator life since 2017. Instead of using bilingual data sets, usually TMX files, these “large language models” are trained with vast amounts of data of all kinds in order to solve a wide variety of tasks. The task itself is performed via ChatBots, so ChatGPT actually means nothing more than “Chat with a generative pre-trained Transformer”. They can summarize texts for the user, design entire social media campaigns, translate and correct texts, extract and check terminology and much more. These AI models are also used as “copilots” in other software, with prompt templates or the option to create your own prompts.

The training provides you with the necessary knowledge about the various prompting methods with which these large language models can be addressed specifically and individually for the respective task.

Content of the training
How do large language models work? How do they “translate” texts? Can they also check and correct translations? What do you have to watch out for?

Various prompting methods are presented in the training and possible application scenarios are explained, here are a few examples:

• “You are a translator”: working with persona models

• Working with examples: One-/ Few-Shot-Prompting

• “Please rate the quality of your translation, make suggestions for corrections and give reasons for your decisions.”Chain-of-Thought-Prompting

• Integration into the chat versus integration as a document

The theoretical parts are always followed by a workshop part with interactive exercises in which the participants can put what they have learned into practice.

A laptop, camera and sound are required for participation in the training. You need access to at least one Large Language Model (ChatGPT, Perplexity, Gemini, etc.). In most cases, registration is sufficient without incurring any costs.

You are welcome to use your own texts for the interactive translation and terminology exercises, but these test texts must be published or uploaded to cloud providers.Otherwise, we recommend that you use the Pro versions.If you have any questions about the technical equipment, you are welcome to contact us in advance.

A continuation of the prompting training will take place in July 2025 as the “Scripting with Python and ChatGPT” training.Instead of letting LLMs such as ChatGPT process tasks themselves via prompts, scripts can also be designed by the LLM to facilitate the recurring tasks of translators and terminologists.In this case, the LLM does not act as a translator or terminologist, but as a developer who is always available.

The training is divided into 2 sessions and takes place from 09:30 – 11:00 on the following dates:
Wednesday, 19.03.2025
Thursday, 20.03.2025