OR-R1: Automating Modeling and Solving of Operations Research Optimization Problems via Test-Time Reinforcement Learning

This is like giving a very smart assistant your messy description of a factory or logistics planning problem and having it automatically turn that into a proper math model and then solve it for you, learning and improving how it does this at the moment it’s run (test-time) rather than only during pre-training.