RickyDeSkywalker/

GAR_baseDataset_NuminaMath

This is the official repository for the paper GAR: Generative Adversarial Reinforcement Learning for Formal Theorem Proving.

GitHub Repository: RickySkywalker/GAR-Official Trained Models: GAR_Goedel-Prover-V2 GAR_DeepSeek-Prover-V2

Base Datasets: Original base dataset Base dataset under Numina-Math

We introduce GAR: Generative Adversarial Reinforcement Learning, an RL training method that intends to solve inefficiency and suboptimal… See the full description on the dataset page: https://huggingface.co/datasets/RickyDeSkywalker/GAR_baseDataset_NuminaMath.