3 ms·DeepScaleR: Surpassing O1-Preview with a 1.5B Model by Scaling RL19 points by mluo 2y agodeleted 2y ago[deleted]