MaiNLP lab, LMU Munich

MaiNLP is headed to #EMNLP2026 with a number of posters and talks! See you in Budapest ✨

Group picture MaiNLP/CIS with friends at EMNLP 2026 🙌 We look forward to a great conference 🤩 @mainlp.bsky.social @cislmu.bsky.social @munichcenterml.bsky.social @lmu.de

We have bigger models. Bigger conferences. Have our questions become bigger, too? That was the question I posed in my Presidential Address at #ACL2026 in San Diego. (1/6)

Super excited to present our work at #ACL2026 ✨ 👉 LLM-based social simulations have lots of analytic flexibility—response generation methods are often overlooked 📍Tue July 7, 9am, Poster Session F

I'm at #ACL2026 where I'll present work on comparing text- vs. speech-based classification performance on standard-language vs. dialectal data, finding different trends for different varieties! (Talk @ Harbor A, July 5, 14:20-14:30) [1/3]

I’m at #ACL2026NLP to present ltzGLUE, a GLUE-style benchmark for Luxembourgish. We also release ltz-E1, a ModernBERT-based encoder trained only on Luxembourgish. Come find me on 7/7!

Excited for 5 papers at #ACL2026NLP with my group and with collaborators. 📍 You can find the work here: 🗓️ Sun. July 5 AfriqueLLM: How Data Mixing and Model Architecture Impact Continued Pre-training for African Languages Oral Session B: Multilinguality and Language Diversity 2

Want to learn more? Poster presentation at Tue, Jul 7, 2026, 2:00 PM – 3:45 PM @ HALL A #3303 Also in the workshop on mechanistic interpretability, Fri, Jul 10, 2026 • 8:00 AM – 5:00 PM @ HALL C Paper: arxiv.org/abs/2505.20076

𝗚𝗿𝗮𝗱𝗶𝗲𝗻𝘁 𝗱𝗲𝘀𝗰𝗲𝗻𝘁 𝗶𝘀 𝗮𝗹𝗹 𝘆𝗼𝘂 𝗻𝗲𝗲𝗱 𝗳𝗼𝗿 𝗶𝗻𝘁𝗲𝗿𝗽𝗿𝗲𝘁𝗮𝗯𝗶𝗹𝗶𝘁𝘆? In our ICML paper, we show that it might just be. ExPLAIND is a method that unifies data attribution, model component attribution, and training dynamics by computing an exact decomposition of model behavior based on gradient products.

MaiNLP will be at #ACL2026 with a number of talks & posters, plus the ACL presidential address! See you in San Diego 🌊

I'm happy to share that I successfully defended my PhD two days ago! Thank you very much to @barbaraplank.bsky.social & Hinrich Schütze for supervising me, to @dirkhovy.bsky.social for joining the committee, and to the many, many people who have been part of my PhD journey :)

🎉 Our paper “From 🔇 Noise to 📶 Signal to 🎯 Selbstzweck: Reframing Human Label” is accepted to Findings of ACL 2026! 👉 lnkd.in/eqDZkrBa ❤️ Huge thanks to my co-authors @santosh-tyss.bsky.social and @barbaraplank.bsky.social 🙏

MaiNLP will be at #LREC2026 to present multiple conference & workshop papers, as well as a workshop keynote! See you in Palma ☀️

MaiNLP is happy to be part of @eaclmeeting.bsky.social with several papers, talks, a panel, and a workshop ☀️ Looking forward to seeing you in Rabat! #EACL2026

We are honoured to welcome Prof Barbara Plank (@barbaraplank.bsky.social ) from @mainlp.bsky.social @cislmu.bsky.social, as our keynote speaker. LoResLM @eaclmeeting.bsky.social

📢 New paper accepted at @eaclmeeting.bsky.social 2026: Persistent Personas? Role-Playing, Instruction Following, and Safety in Extended Interactions with @mhedderich.bsky.social @amodarressi.bsky.social Hinrich Schuetze & Benjamin Roth. Preprint: arxiv.org/abs/2512.12775

In our new paper, "A Comprehensive Evaluation of Multilingual Chain-of-Thought Reasoning: Performance, Consistency, and Faithfulness Across Languages", we go beyond final-answer accuracy to analyze multilingual reasoning along three dimensions: performance, consistency, and faithfulness.

✨New paper✨ We find script (e.g. Cyrillic, Latin) to be a linear direction in the activation space of Whisper, enabling transliteration at test-time by adding such script directions to the activations — producing e.g. Cyrillic Japanese transcriptions.

VarDial 2026 will be colocated with @eaclmeeting.bsky.social! We're looking forward to your papers on NLP for similar languages, varieties and dialects :) Deadline: Dec 19 (Jan 2 for pre-reviewed ARR papers) sites.google.com/view/vardial...

Group photo at NeurIPS 2025 San Diego

Congrats to Pingjun, @beiduo.bsky.social , Siyao, Marie, and @barbaraplank.bsky.social for receiving the SAC Highlights reward!

Congrats to our team member Diego Frassinelli on the SAC Highlights award!

At #Interspeech2025 I'm going to present Betthupferl, a dataset for German dialect ASR & dialect-to-standard speech translation! We analyze differences between dialectal & Standard German transcriptions, benchmark ASR models, and examine shortcomings of current ASR models & evaluation metrics.

UPDATE: Our poster presentation got moved to Tuesday, 16:00–17:30 (session 10)! #ACL2025NLP

Unsure which presentations to attend at #ACL2025? 🛎️🗣️

📄 [ACL 2025 main] Circuit compositions: Exploring Modular Structures in Transformer-Based Language Models (doi.org/10.48550/arX...)

📄 [ACL 2025 main] LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Tasks (doi.org/10.48550/arX...)

At #ACL2025NLP I'll present our analysis of the effect of linguistic similarity on cross-lingual transfer! We looked at how 10 similarity measures correlate w/ transfer results btwn 263 languages across 3 NLP tasks. Different similarity measures matter for diff. experiments (no one-size-fits-all)!

🤔 Can LLMs read between the lines? Our another #ACL2025 paper surveys resources on how LLMs handle pragmatics like implicatures, deixis, and more. We map out a new landscape for both LLMs and linguistics in pragmatic research. 📄 arxiv.org/abs/2502.12378 🧠💬 #LLMs #Pragmatics

🎉 Our paper “Algorithmic Fidelity of LLMs in Generating Synthetic German Public Opinions” is accepted at #ACL2025 main conference as an oral presentation! 🇩🇪🤖 We study how well LLMs simulate real survey responses using open-ended German data, showing the left-leaning bias.

The future of AI literacy starts early! Our very own @fkoerner.bsky.social recently led a series of hands-on workshops teaching some of our youngest students yet 🌱🤖

Looking forward to my visit to Hamburg University and their Data Science group!

What changes if you take the LLM prompt “Tell me a short story about Dr. Li” and replace “Dr. Li” with “Dr. Smith”? Would you have guessed that this introduces a massive gender bias, from ca. half/half to 99% male doctors? 

In our #ACL2025 paper we present the Spotlight framework which...

🦙 how well do LLMs encode discourse knowledge? does that generalize across languages? 🛎️ in our #ACL2025 paper, we uncover fascinating trends about multilingual discourse representations! joint work w/ @florian-eichin.com @barbaraplank.bsky.social @mhedderich.bsky.social 📄 arxiv.org/abs/2503.10515

Caught some great moments at #MCML Munich AI Day 2025 last week📍 From sharp keynotes to poster debates. Our team had the chance to show some recent work, join the conversations, and bring back plenty of food for thought🧠🗣️📊