Users praise the Tsinghua async RL paper for AI agents as very impressive work that highlights real mismatches between group sampling and asynchrony.
Based on 1 visible X reactions from 2 accounts; directional sample.
Ask a question below.
Published answers will appear here.
Anyway, very impressive work showing the mismatch between group sampling and asynchrony is real, and I expect more papers on this.
Interestingly, we went full-circle in reinforcement learning for LLMs and LLM agents: 🔹Initially, OpenAI (and some others) used RLHF with PPO, which requires training a critic (reward) model. 🔹IThen, researchers moved from PPO with a critic to critic-free GRPO because critics were expensive and unstable. 🔹INow long-running asynchronous agents expose a weakness in GRPO, so this paper brings the critic back, but with several engineering fixes.
Just read the new paper from Tsinghua/Z.AI on async RL for agents (arXiv:2607.07508). It comes several weeks after the release of GLM-5.2, in which they mentioned that they use a critic instead of GRPO. 🧵 https://x.com/ziv_ravid/status/2075617886823911743/photo/1
Users praise the Tsinghua async RL paper for AI agents as very impressive work that highlights real mismatches between group sampling and asynchrony.
Based on 1 visible X reactions from 2 accounts; directional sample.
Ask a question below.
Published answers will appear here.