概念
grpo training rl
创建2026-04-15
更新2026-04-15
阅读量级1 分钟
概念导读

GRPO-based RL training methodology refers to the application of Group Relative Policy Optimization in the reinforcement learning phase of LLM post-training. This approach is used in agentic coding systems like agentic-coding and models such as composer2, where GRPO provides a stable and effective training signal by comparing group-relative rewards rather tha...

  1. Related
  2. References

GRPO RL Training

GRPO-based RL training methodology refers to the application of Group Relative Policy Optimization in the reinforcement learning phase of LLM post-training. This approach is used in agentic coding systems like agentic-coding and models such as composer2, where GRPO provides a stable and effective training signal by comparing group-relative rewards rather than relying on absolute reward models.

  • grp-o — GRPO algorithm itself
  • agentic-coding — Agentic coding training leverages GRPO RL training
  • composer2 — Composer 2 uses GRPO-based training methodology

References