I am Florian Dorner, a doctoral fellow with the Max Planck ETH Center for Learning Systems advised by Moritz Hardt and Fanny Yang. I hold an MSc. in Mathematics from FU Berlin and an MSc. in Science, Technology and Policy from ETH Zurich.

I am interested in understanding the societal impacts of Artificial Intelligence. These days, most of my work focuses on the role of data quality in various parts of the Machine Learning pipeline, with a particular focus on benchmarking and evaluation.

Selected Publications

• ROC-n-reroll: How verifier imperfection affects test-time scaling
  Florian E. Dorner, Yatong Chen, André F Cruz, and Fanny Yang
  arxiv preprint
• How Benchmark Prediction from Fewer Data Misses the Mark
  Guanhua Zhang, Florian E. Dorner, and Moritz Hardt
  NeurIPS 2025
• Limits to scalable evaluation at the frontier: LLM as Judge won't beat twice the data
  Florian E. Dorner, Vivian Y. Nastl, and Moritz Hardt
  ICLR 2025 (Oral)
• Training on the Test Task Confounds Evaluation and Emergence
  Ricardo Dominguez-Olmedo, Florian E. Dorner, and Moritz Hardt
  ICLR 2025 (Oral)
• Don't Label Twice: Quantity Beats Quality when Comparing Binary Classifiers on a Budget
  Florian E. Dorner, Moritz Hardt
  ICML 2024