Unverified50% confidenceFactExact time
When GRPO is used for long-form RAG training, the answer-level scalar advantage value is evenly distributed to each sentence in the answer, causing copied sentences and genuinely grounded sentences to receive the same reinforcement signal.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedAt the training level, reasoning-enhanced models typically leverage reinforcement learning algorithms such as GRPO (Group Relative Policy Optimization), using the correctness of the final answer as the reward signal76% similarUnverifiedJustGRPO的核心修复方案是在自回归顺序下训练GRPO，推理阶段保持并行解码75% similarUnverifiedExisting methods for migrating reinforcement learning approaches such as GRPO to diffusion language models typically treat every step on the denoising trajectory as equally important.69% similarUnverifiedAlgorithms such as GRPO and PPO have been successfully applied in post-training of large language models.68% similarUnverifiedGRPO (Group Relative Policy Optimization) is an improved variant of PPO that requires no additional value network training; instead, it calculates the reward signal by sampling a group of responses to the same question and using intra-group relative performance as a baseline.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023692API
curl https://kongchang.com/api/v1/knowledge/claims/1023692MCP
get_claim(id=1023692)