Jannik Brinkmann
I am a Ph.D. student in Computer Science at the University of Mannheim. I am interested in understanding the internal mechanisms and representations of neural networks, and using these insights to control and improve them. I also work with the Interpretable Neural Networks lab of Prof. David Bau and post about research on LessWrong.
arXiv 2024
The Quest for the Right Mediator: A History, Survey, and Theoretical Grounding of Causal Interpretability
NeurIPS 2024
Measuring Progress in Dictionary Learning for Language Model Interpretability with Board Game Models
ACL 2024
A Mechanistic Analysis of a Transformer Trained on Symbolic Multi-Step Reasoning Task
Publications
The complete list of publications can be found on Google Scholar or Semantic Scholar.
-
2024The Quest for the Right Mediator: A History, Survey, and Theoretical Grounding of Causal InterpretabilityarXiv 2024
-
2024NNsight and NDIF: Democratizing Access to Foundation Model InternalsarXiv 2024
-
2024Measuring Progress in Dictionary Learning for Language Model Interpretability with Board Game ModelsNeurIPS 2024
-
2024A Mechanistic Analysis of a Transformer Trained on Symbolic Multi-Step Reasoning TaskACL 2024 (Findings)
-
2023A Multidimensional Analysis of Social Biases in Vision TransformersICCV 2023