News - Data Science Chair

How do you build scalable, transparent language models for the German language entirely from scratch? We had the opportunity to discuss exactly that at the DLR in Ulm. The focus was on our model families LLäMmlein (120M–7B) and ModernGBERT (138M–1B), as well as the unique challenges of purely German tokenization.

From Brisbane with Insights: How "tame" is our German LLM family really? Our chair returns from a six-week research sabbatical at The University of Queensland with new findings on AI safety and "LLäMmlein".

The paper by Luzia Keupp, Andreas Hotho, Stefan Dech and Heiko Paeth studies climate change and its impact on regional agriculture .

We hosted this year’s BigData@Geo2.0 workshop with our project partners at the CAIDAS Building.

We presented a keynote on our LLäMmlein and ModernGBERT models at the German National Library in Frankfurt, as part of the "Netzwerk maschinelle Verfahren in der Erschließung" symposium.

We, the Data Science Chair and the Chair of Computer Philology and Modern German Literary History at CAIDAS, are part of the new large-scale project „Sovereign Open Source Foundational Models for European Intelligence“ (SOOFI)

We participated in all GermEval tasks and placed first in two of the three Harmful Content Detection tasks!

We're excited to share that our paper "TaylorNet: Learning PDEs from Non-Grid Data"", has been accepted at the Machine Learning Meets Differential Equations Workshop at ECAI 2025

We're excited to share that our paper "On the Role of Embeddings in Diffusion-based Generation of scRNA-seq Data", has been accepted at the workshop New Frontiers in Mining Complex Patterns at ECML PKDD 2025

We're excited to share that our paper "Evaluating Spatiotemporal Prediction Models in a Low-Data Regime", has been accepted at the International Workshop on Learning from Small Data at ECML PKDD 2025