Tag
Discusses a new paper from Tsinghua/Zhipu AI on asynchronous reinforcement learning for agents, and notes that their previous GLM-5.2 model uses a critic instead of GRPO.
GLM5.2 reintroduces a critic component for fine-grained variance reduction, suggesting that group-based methods are ineffective for long horizons. The author believes OpenAI and Anthropic already use value models.