Tag
This paper addresses advantage scale calibration in group-relative optimization under low-variance rewards, proposing methods like the Reward-Resolution Protocol and MaxNorm-AC to filter sub-resolution jitter and provide bounded recovery for credible gaps.
Memory-R2 introduces LoGo-GRPO, a training framework that combines local and global group-relative optimization to provide fairer credit assignment for long-horizon memory-augmented LLM agents, improving accuracy and inference latency across backbones.