Verified90% confidenceFactTime unknown
GRPO（Group Relative Policy Optimization）由DeepSeek提出，无需单独训练价值网络，已成为当前智能体RL训练的主流算法选择
13
Sources
90%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Cursor Composer 2训练揭秘：分布式强化学习架构全解析
bilibiliWeb3天空之城
Related Entities
Cite This Claim
Stable URI
https://kongchang.com/claim/5808API
curl https://kongchang.com/api/v1/knowledge/claims/5808MCP
get_claim(id=5808)