Official implementation of paper: SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training - View it on GitHub