Max Müller-Eberstein

Hej! I'm a postdoctoral researcher in the field of AI, working with Prof. Anna Rogers at the IT University of Copenhagen's NLPnorth Lab and the Pioneer Centre for Artificial Intelligence.

My research focuses on identifying consistencies in how machines learn in order to make training efficient enough for underserved communities. Towards this goal, I study the learning dynamics of ML models and leverage overlaps across languages and other modalities (e.g., visual art and music). Ultimately, I believe that understanding how models construct their knowledge is essential to broadening who has access to the world's information.

In my PhD on Quantifying Linguistic Variation, we improved the transferability of NLP models across languages and domains, as part of the European Laboratory for Learning and Intelligent Systems' (ELLIS) fellowship for outstanding young researchers. The project was advised by Prof. Barbara Plank, Prof. Rob van der Goot, and Prof. Ivan Titov. Over the years, I have also been very fortunate to intern with excellent industry research teams at Apple and Amazon.

Additionally, I'm also somewhat known for the overly elaborate presentation of research papers (a.k.a., scientific communication).

In addition to the short research profile below, these pages link to a variety of personal and artsy projects:

Blog Code Art

What's New?

Invited talk at the University of Hamburg on 16 May at 13:00.
"DaKultur: Evaluating the Cultural Awareness of Language Models for Danish with Native Speakers" is now published in the NAACL 2025 Workshop Proceedings, with a presentation on 4 May at C3NLP.
Joined the Danish Agency for Digital Government's Working Group on "The Strategic Use of AI".
"PolyPythias: Stability and Outliers across Fifty Language Model Pre-Training Runs" was accepted at ICLR 2025.
"SnakModel: Lessons Learned from Training an Open Danish Large Language Model" was accepted at NoDaLiDa/Baltic-HLT 2025.
Participating in the Ministry of Higher Education and Science's "Book a Researcher" week for scientific outreach with a talk on "How I Trained an AI That Is Better at Danish than Me".
New blog post taking a second-by-second look at "The Learning Dynamics of a PhD".

[ᐧᐧᐧ] old news at mxij.me/news.

Profile

Research Featured

May 2025 Singapore PolyPythias: Stability and Outliers across Fifty Language Model Pre-Training Runs van der Wal, Lesci, Müller-Eberstein, Saphra, Schoelkopf, Zuidema and Biderman (2025): In The Thirteenth International Conference on Learning Representations (ICLR 2025).
PDF OPR GIT HF
Jul 2024 MIT Press Hypernetworks for Personalizing ASR to Atypical Speech Müller-Eberstein⁼, Yee⁼, Yang, Mantena and Lea (2024): In Transactions of the Association for Computational Linguistics (TACL)
PDF BIB
Dec 2023 Singapore Subspace Chronicles: How Linguistic Information Emerges, Shifts and Interacts during Language Model Training Müller-Eberstein, van der Goot, Plank and Titov (2023): In Findings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP)
PDF ACL BIB PST VID GIT
Dec 2022 Abu Dhabi, UAE Spectral Probing Müller-Eberstein, van der Goot and Plank (2022c): In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP)
PDF ACL BIB PST VID GIT
May 2022 Dublin, Ireland Probing for Labeled Dependency Trees Müller-Eberstein, van der Goot and Plank (2022a): In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL)
PDF ACL BIB PST VID GIT
Nov 2019 Seoul, Republic of Korea Translating Visual Works of Art into Music Müller-Eberstein and van Noord (2019): In Workshop Proceedings of IEEE International Conference on Computer Vision 2019 (ICCV)
Best Poster Eastern European Machine Learning Summer School 2020 (EEML)
PDF BIB VID GIT

Education

2020 – 2024 Copenhagen, Denmark Munich, Germany Edinburgh, UK IT-Universitetet i København / IT University of Copenhagen Ludwig-Maximilians-Universität München / LMU Munich (affiliated) Oilthigh Dhùn Èideann / University of Edinburgh (affiliated): ELLIS PhD Fellow, Natural Language Processing, with Prof. Barbara Plank
co-supervised by Prof. Rob van der Goot, and Prof. Ivan Titov
2017 – 2019 Amsterdam, The Netherlands Universiteit van Amsterdam / University of Amsterdam: Master of Science, Artificial Intelligence
supported by a scholarship from the German Academic Exchange Service (DAAD)
2014 – 2017 Heidelberg, Germany Ruprecht-Karls-Universität Heidelberg / Heidelberg University: Bachelor of Arts, Computational Linguistics

Work

2024 – present Copenhagen, Denmark IT-Universitetet i København / Computer Science Department: Postdoctoral Researcher, Natural Language Processing
2023 – present Copenhagen, Denmark Pioneer Centre for Artifical Intelligence / Speech and Language Collaboratory: Affiliated Researcher
2020 – 2024 Copenhagen, Denmark IT-Universitetet i København / Computer Science Department: ELLIS PhD Fellow, Project Supervisor and Teaching Assistant
2023 Seattle, USA Apple Inc. / AI/ML / Accessibility: Machine Learning Research Intern
2022 Cambridge, UK Apple Inc. / AI/ML / Siri: Machine Learning Research Engineer Intern
2018 Seattle, USA Amazon Inc. / Alexa AI / Intelligent Decisions Team: Language Engineer Intern

Talks Featured

Leveraging Consistencies across the Language Model Training Pipeline Apr 2025 Copenhagen, Denmark: Pioneer Centre for Artificial Intelligence, Denmark
Dynamic Personalization from Cross-model Consistencies Mar 2025 Bonn, Germany: CAISA Lab, University of Bonn
Language Data and Language Technologies in Denmark (Panelist) Mar 2025 Copenhagen, Denmark: The European Commission's Language Data Space Workshop, Alexandra Instituttet and the Danish Agency for Digital Government
Learning Dynamics Consistencies: Making Language Technologies Accessible for All Nov 2024 Munich, Germany: MaiNLP Seminar, LMU Munich
Tracking the Acquisition of Linguistic Knowledge throughout Language Model Training Oct 2024 Düsseldorf, Germany: Deep Linguistic Modeling Colloquium, HHU Düsseldorf
2023 Tokyo, Japan What's in a Task?: ML Waseda, Waseda University
2023 Uppsala, Sweden What Even Is a Domain?: Språkteknologi Seminar, Uppsala University
2023 Edinburgh, UK Subspaces of Linguistic Variation: EdinburghNLP Seminar Series, University of Edinburgh
2022 Cambridge, UK Quantifying Linguistic Variation: Language Technology Lab Seminar Series, University of Cambridge

Teaching

2021 Copenhagen, Denmark Course Organizer: Communicating State-of-the-art NLP Research to a Broader Audience, IT University of Copenhagen
2021 – 2024 Copenhagen, Denmark Munich, Germany Guest Lecturer: Seminars in Data Science, IT University of Copenhagen
Introduction to NLP and Deep Learning, IT University of Copenhagen
Profilierungsmodul Computerlinguistik I, LMU Munich
Vertiefung der mathematischen und linguistischen Grundlagen der Computerlinguistik, LMU Munich Human-centric NLP Seminar, LMU Munich
2021 – 2022 Copenhagen, Denmark Supervisor: Musical Impressions and Album Art: Investigating the Patterns, IT University of Copenhagen
Pylophonic: Multi-instrument Music Generation with Transformer-XL, IT University of Copenhagen
2015 – 2023 Copenhagen, Denmark Heidelberg, Germany Teaching Assistant: Introduction to NLP and Deep Learning, IT University of Copenhagen
Introduction to Computational Linguistics, Heidelberg University
Advanced Programming: Java, Heidelberg University
Mathematical Foundations, Heidelberg University

Community

present Multiple Venues Area Chair and Reviewer for NLP Conferences: ACL 2024, NAACL 2024, ACL 2023, EACL 2023, EMNNLP 2022, WiNLP 2022, NAACL 2022, CoNLL 2021 Outstanding Reviewer
2016 – 2024 Multiple Venues Conference/Workshop Chair: The 9th Workshop on Noisy and User-generated Text (W-NUT at EACL 2024)
Multi* Workshop on Linguistic Variation at ITU Copenhagen
26. Tagung der Computerlinguistik-Studierenden
2015 – 2017 Heidelberg, Germany Elected Member to the Faculty, Institute and Student Councils: Computational Linguistics, Faculty of Neophilology, Universität Heidelberg

Awards and Further Qualifications

DAAD AInet Postdoctoral Fellowship Nov 2024 – present Bonn, Germany: German Academic Exchange Service (DAAD)
Outstanding Paper Award for Experimental Standards for Deep Learning Research Apr 2022 Virtual: The Machine Learning Evaluation Standards Workshop at ICLR 2022
Outstanding Reviewer Award Nov 2021 Punta Cana, Dominican Republic: The SIGNLL Conference on Computational Natural Language Learning 2021
ELLIS PhD Fellowship Sep 2020 – May 2024 Denmark + Germany + UK: European Laboratory for Learning and Intelligent Systems (ELLIS)
Best Poster Award for Translating Visual Art into Music Jul 2020 Kraków, Poland: Eastern European Machine Learning Summer School 2020 (EEML)
DAAD Scholarship for International Graduate Students Aug 2018 – Sep 2019 Bonn, Germany: German Academic Exchange Service (DAAD)
First-Place Team in TORCS AI Driver Competition Dec 2017 Amsterdam, The Netherlands: Vrije Universiteit Amsterdam / Universiteit van Amsterdam