[Preprint'25] Code and data repo of the paper "BINGO: Boosting Efficient Reasoning of LLMs via Dynamic and Significance-based Reinforcement Learning" - View it on GitHub