News

Jun 2026
Received a SwissAI grant to study the impact of tokenisation in language models.
May 2026
Five papers were accepted across ICML, ACL, and ICLR 2026, on the superficial alignment hypothesis, tokenisation, language identification, prosody, and generalisation.
2026
Gave invited talks on causality for ML interpretability at EPFL, the University of Oxford, the University of Edinburgh, and UT Austin.
Nov 2025
“The Non-Linear Representation Dilemma” was selected as a spotlight at NeurIPS 2025.
Jul 2025
Two papers received senior area chair highlights awards at ACL 2025.

Selected publications

2026 International Conference on Machine Learning (ICML)
Operationalising the Superficial Alignment Hypothesis via Task Complexity
Tomás Vergara-Browne, Darshan Patil, Ivan Titov, Siva Reddy, Tiago Pimentel*, Marius Mosbach*
2025 Advances in Neural Information Processing Systems (NeurIPS)
The Non-Linear Representation Dilemma: Is Causal Abstraction Enough for Mechanistic Interpretability?
Denis Sutter, Julian Minder, Thomas Hofmann, Tiago Pimentel
Spotlight
2025 Annual Meeting of the Association for Computational Linguistics (ACL)
Tokenisation is NP-Complete
Philip Whittington, Gregor Bachmann, Tiago Pimentel
SAC highlights award
2024 Annual Meeting of the Association for Computational Linguistics (ACL)
Causal Estimation of Memorisation Profiles
Pietro Lesci, Clara Meister, Thomas Hofmann, Andreas Vlachos, Tiago Pimentel
Best paper award
2024 Conference on Empirical Methods in Natural Language Processing (EMNLP)
How to Compute the Probability of a Word
Tiago Pimentel, Clara Meister
2023 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Revisiting the Optimality of Word Lengths
Tiago Pimentel, Ethan G. Wilcox, Clara Meister, Kyle Mahowald, Ryan Cotterell
Outstanding paper award
See all publications →

* indicates equal contribution.