On: AI That Invents New Math, Programs, and Games–Creative

[KaoruAK]

Here’s the paper and code: https://doi.org/10.17605/OSF.IO/74DXZ

Here’s the code and how to implement it: https://github.com/POlLLOGAMER/RfC-Reinforcement-for-Creativ...

RfC trains agents to generate genuinely creative and valid outputs across any rule-based domain, from formal math to molecular design.