#policy-optimization

The latest AI and developer news about #policy-optimization, with the original source and a short summary.

PIRL: A Closed-Loop Optimization Technique to Prevent Performance Degradation in Reinforcement Learning

To address the Open-Loop problem of existing reinforcement learning, a PIRL framework is proposed that directly verifies the outcomes of policy improvement.

·2026.07.28 21:00

Personalized Group Relative Policy Optimization for Heterogeneous Preference Alignment

P-GRPO reduces GRPO's bias by decoupling individual preferences using per-group reward history.

·1·2026.04.02 09:00

SAPO: A Stable, High-Performance Reinforcement Learning Method for Training Large Language Models

SAPO stabilizes LLM reinforcement learning by using smooth gating instead of hard clipping.

·1·2025.12.05 05:00