Everything is fully open-sourced, including the codebase, the model + all individual single reward model variants!
š Site: nicolas-dufour.github.io/miro
š Paper: arxiv.org/abs/2510.25897
š ļø Git: github.com/nicolas-dufo...
š¤ HF: huggingface.co/nicolas-dufo...
šØ Demo: huggingface.co/spaces/nicol...