minor edit to the "When Reinforcement Learning Optimizes for Benchmarks, Not Patients" post

Author: tobideusser

_posts/2026-05-22-When-RL-Wins-the-Benchmark-but-Loses-the-Patient.md

@@ -7,9 +7,7 @@ categories: [Research]
 description: false
 ---
 
-This is a short summary of our paper **"Benchmark Success, Clinical Failure: When Reinforcement Learning Optimizes for Benchmarks, Not Patients"** by *Armin Berger, Manuela Bergau, Helen Schneider, Saad Ahmad, Tom Anglim Lagones, Gianluca Brugnara, Martha Foltyn-Dumitru, Kai Schlamp, Philipp Vollmuth, and Rafet Sifa*, available on [arXiv](https://arxiv.org/abs/2512.23090) and to be published in the Proceedings of IJCNN 2026.
-
-*TL;DR: RL fine-tuning made the model better at the benchmark and worse at the job and the same thing happens at 50× the budget, so the problem is the recipe, not the resources.*
+This is a TL;DR of our paper **"Benchmark Success, Clinical Failure: When Reinforcement Learning Optimizes for Benchmarks, Not Patients"** by *Armin Berger, Manuela Bergau, Helen Schneider, Saad Ahmad, Tom Anglim Lagones, Gianluca Brugnara, Martha Foltyn-Dumitru, Kai Schlamp, Philipp Vollmuth, and Rafet Sifa*, available on [arXiv](https://arxiv.org/abs/2512.23090) and to be published in the Proceedings of IJCNN 2026.
 
 Reinforcement learning has been the headline story for LLMs lately: DeepSeek-R1 and friends have shown you can squeeze impressive reasoning out of a model with the right reward signal. A natural next question: does this work for medical imaging too, especially on a tight budget? Our team from Fraunhofer IAIS, the University of Bonn, the Lamarr Institute, University Hospital Bonn, and Queensland Health set out to find out, and the answer turns out to be more complicated than the leaderboards suggest.
 
@@ -53,3 +51,7 @@ For anyone deploying small medical VLMs in the real world, this is a useful warn
 
 That doesn't mean RL is doomed for medical imaging, but it does mean "improved benchmark score" and "better diagnostic model" aren't the same thing, and we probably need reward formulations, multi-dataset curricula, or architectural changes that explicitly target generalization rather than single-benchmark wins.
 
+## The one-line summary
+
+*TL;DR: RL fine-tuning made the model better at the benchmark and worse at the job and the same thing happens at 50× the budget, so the problem is the recipe, not the resources.*
+