Bartosz Cywi'nski

Most-cited papers

1. SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders

   Bartosz Cywi'nski, Kamil Deja

   2025International Conference on Machine Learning92 citations

   Semantic Scholar

   Save

2. Eliciting Secret Knowledge from Language Models

   Bartosz Cywi'nski, Emil Ryd, Rowan Wang, Senthooran Rajamanoharan +3 more

   2025arXiv.org29 citations

   Semantic Scholar

   Save

3. Towards eliciting latent knowledge from LLMs with mechanistic interpretability

   Bartosz Cywi'nski, Emil Ryd, Senthooran Rajamanoharan, Neel Nanda

   2025arXiv.org24 citations

   Semantic Scholar

   Save

4. GUIDE: Guidance-based Incremental Learning with Diffusion Models

   Bartosz Cywi'nski, Kamil Deja, Tomasz Trzcinski, Bartłomiej Twardowski +1 more

   2024European Conference on Artificial Intelligence8 citations

   Semantic Scholar

   Save

5. Precise Parameter Localization for Textual Generation in Diffusion Models

   Lukasz Staniszewski, Bartosz Cywi'nski, Franziska Boenisch, Kamil Deja +1 more

   2025International Conference on Learning Representations7 citations

   Semantic Scholar

   Save

6. Simple LLM Baselines are Competitive for Model Diffing

   Elias Kempf, Simon Schrodi, Bartosz Cywi'nski, Thomas Brox +2 more

   2026arXiv.org4 citations

   Semantic Scholar

   Save

7. Censored LLMs as a Natural Testbed for Secret Knowledge Elicitation

   Helena Casademunt, Bartosz Cywi'nski, K. Tran, Arya Jakkli +2 more

   2026arXiv.org4 citations

   Semantic Scholar

   Save

8. Adapt & Align: Continual Learning with Generative Models Latent Space Alignment

   Kamil Deja, Bartosz Cywi'nski, Jan Rybarczyk, Tomasz Trzci'nski

   2023arXiv.org0 citations

   Semantic Scholar

   Save