THUDM open-sources Slime, the reinforcement learning scaling framework used to train Zhipu AI's GLM-5.2 model

Story Overview

THUDM has released the full reinforcement learning post-training stack that powered Zhipu AI's GLM-5.2, letting anyone replicate the parallel OPD loops that folded more than ten expert models into one system handling long-horizon agent tasks in roughly two days.