News & Updates
The latest announcements, achievements, and events from the WüNLP Chair.
WueNLP members will present six papers at EMNLP26!
Six papers (co-)authored by WueNLP members will be presented at EMNLP26 (5 at Main conference and one in Findings), Oct 24-29 in Budapest! Paper links coming soon!
Three open research positions at WüNLP open_in_new
We are hiring! The WüNLP group has three open positions for PhD students and/or postdoctoral researchers. The advertised topics span LLM post-training, modular architectures for sample-efficient adaptation of language models, and vision-language-action models for robotic manipulation. All positions are open until filled, with the earliest possible starting dates — get in touch if you are interested.
Two papers accepted at ACL 2026
Two of our papers have been accepted at ACL 2026: "Compositional Steering of Large Language Models with Steering Tokens" in the Main Conference, and "One Script to Rule Them All" in Findings. Congratulations to all co-authors!
Outstanding Senior Area Chair Award at EACL 2026 open_in_new
Goran Glavaš received an Outstanding Senior Area Chair (SAC) Award at EACL 2026. The award recognizes outstanding service to the community in steering the review process and shaping the program of a major area at the conference.
Gregor Geigle completes his PhD with summa cum laude open_in_new
Congratulations to Gregor Geigle, who successfully completed his PhD on multilingual vision-language models with the highest distinction, summa cum laude. His work advanced the state of the art in building, scaling, and evaluating large multilingual vision-language models.
Paper on vision-language embedding models accepted at EACL 2026 open_in_new
Our short paper "Mind Your Special Tokens in Vision-Language Embedding Models" has been accepted at EACL 2026. The work analyzes the outsized role that special tokens play in the representations produced by vision-language embedding models, and what this means for downstream retrieval and matching.
Three papers accepted at EMNLP 2025 open_in_new
Three WüNLP papers were accepted at EMNLP 2025, a top-tier venue in NLP: "mHallucinations" in the Main Conference, and "TransAlign" and "ReCoVeR" in Findings. The papers were presented at the conference in Suzhou, China. Congratulations to all authors!
Congratulations (and farewell), Fabian! open_in_new
Fabian David Schmidt successfully defended his PhD thesis, "Robust and Scalable Cross-Lingual Transfer," on July 8th. Both the evaluation and the defense committees awarded the work the highest distinction, summa cum laude, recognizing its exceptional scientific value. Congratulations, Fabian — and all the best for what comes next!
Eight papers accepted at ACL 2025 open_in_new
Our group had a strong showing at ACL 2025 with eight accepted papers — four in the Main Conference and four in Findings — spanning our work on multilingual, multimodal, and efficient NLP. Congratulations to all co-authors!
ObscuraCoder accepted at ICLR 2025 open_in_new
"ObscuraCoder" has been accepted at ICLR 2025. The paper, a collaboration with Indraneil Paul and Iryna Gurevych (TU Darmstadt), studies obfuscation-grounded pre-training for more sample-efficient and robust code language models.
Four papers accepted at EMNLP 2024 open_in_new
Four of our papers were accepted at EMNLP 2024: FOCI, NLLB-LLM2Vec, a study of hallucinations in large vision-language models, and MANNeR. Congratulations to all co-authors!
Outstanding Paper Award at ACL 2024 for IRCoder open_in_new
Our paper "IRCoder: Intermediate Representations Make Language Models Robust Multilingual Code Generators" received an Outstanding Paper Award at ACL 2024 — a distinction given to only a small fraction of accepted papers.
Two long papers accepted at ACL 2024
Two of our long papers were accepted at ACL 2024: "Babel-ImageNet," on massively multilingual evaluation of vision-language models, and "IRCoder," on robust multilingual code generation.
Two long papers accepted at NAACL 2024 open_in_new
Two of our long papers were accepted at NAACL 2024: "SQATIN," on scenario-based question answering for task-oriented dialogue, and "To Translate or Not to Translate," on cross-lingual transfer strategies for low-resource languages.