Dieuwke Hupkes
Dieuwke Hupkes began her studies in physics, then studied logic at Amsterdam and pursued computational linguistics. She completed a doctorate in June 2020 on hierarchy and interpretability in neural models of language processing. She joined FAIR Paris in December 2020. Her research includes compositionality and generalisation, with a 2023 GenBench taxonomy and review. She worked as pretraining evaluations lead for Llama 3 in 2024 and received ELLIS Scholar recognition in October 2024. Her work connects her language research with the evaluation of large neural models.
- Now at
- FAIR (Meta)
- Works
- 70
- Citations
- 1,691
- h-index
- 17
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2020
Language structure and interpretability
Hupkes completed her doctorate in June 2020. Her dissertation studied hierarchy and interpretability in neural models of language processing. She investigated how neural networks encode and process structure, connecting computational language models with questions about human language processing.
-
2020
Joining FAIR Paris
She moved to FAIR Paris as a research scientist on 7 December 2020, after a brief postdoctoral period at ELLIS Amsterdam. She earlier studied physics and logic.
-
2023
A taxonomy of generalisation
She published the GenBench taxonomy and review in Nature Machine Intelligence in October 2023. The work is a collaborative study of generalisation research in NLP. This work helps organise the kinds of tests used to study whether language models can handle new conditions.
-
2024
Llama evaluation and ELLIS
In July 2024, she participated in the Llama 3 effort as pretraining evaluations lead. She received ELLIS Scholar recognition on 14 October 2024.
Sources and credits
Photo: ELLIS profile; creator not stated. License: Not stated by source. Source: ellis.eu. Cropped to a square and resized. Ask us to remove or replace this photo.