Masterarbeit: Erklärbarkeit von Transformer-Modellen in der Autorschaftsverifikation
Fraunhofer-GesellschaftDarmstadt, Hesse, Germany
Posted December 23, 2025
Internship
About the Company
The Fraunhofer Society is one of the world's leading organizations for applied research, with a focus on future-relevant key technologies and the exploitation of results in industry and economy.
Job Description
We are seeking a master's student to work on a thesis project that focuses on explainability of Transformer models in authorship verification. The goal is to apply existing explainability methods to newer, more powerful models and adapt them if necessary.
Requirements
- Knowledge in the field of Machine Learning, ideally in the field of NLP and Transformer models
- Very good Python knowledge, preferably experience with PyTorch or HuggingFace
- Motivation to deal with current explainability approaches
- Scientific interest in evaluation metrics in modern AI systems
Benefits
- Independent time management
- Insights into the intersection of academic research and industrial application
