| May 29, 2025 | Is your LLM overcharging you? In a newly released paper, we look at delegated text generation in LLMs from a game-theoretic point of view and show that tokenization has important implications when it comes to pricing models. We demonstrate that the current widely used scheme, which charges users per generated token, incentivizes providers of models to manipulate the information reported to the user. We further conduct experiments with multiple model families to demonstrate the feasibility of this strategic behavior and propose an alternative pricing mechanism that completely eliminates it. Our pre-print can be found here. |
| Feb 05, 2025 | Our group at MPI-SWS has released a new joint paper on the evaluation of Large Language Models (LLMs). We use a causal framework to argue why controlling for the randomization in the generative process can be beneficial when evaluating models. It can be found here. |
| Dec 01, 2024 | We have finally released the paper derived from my M.Sc. thesis on Bayesian Neural Networks applied to Warm Dark matter detection! You can find the pre-print here. It has also been submitted to Astronomy & Astrophysics. |
| Oct 10, 2024 | I will be giving a talk at the Galaxy Coffee, the seminar of the “Galaxies & Cosmology” Department at the Max Planck Institute for Astronomy. |
| Oct 01, 2024 | My master’s thesis is complete! Thanks to Heidelberg University (Institute for Theoretical Physics) and the Max Planck Institute for Astronomy. You can find it here. In this project, I use a Bayesian neural network to reconstruct the cosmic density fields from the Lyman-alpha forest. Based on those reconstructions, I constrain warm dark matter with unprecedented efficiency. |
| Sep 10, 2024 | I will be joining the Max Planck Institute for Software Systems as a PhD. student, under the supervision of Dr. Manuel Gomez Rodriguez and as part of the IMPRS-TRUST program! |