method · first seen 5 Oct 2026
KL-divergence step control
Also: predicted-KL step control
Limiting policy update size by estimating the KL divergence between old and new policies before applying updates.
Mentions per week
PapersCodeDiscussion
Earliest sources
- LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient SketchesPapers · 5 Oct 2026 · Shaokun Zhang, Yifan Zhang, Jian Hu, Yueying Li, Hao Zhang, Binfeng Xu, Jan Kaut…
Latest sources
- LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient SketchesPapers · 5 Oct 2026 · Shaokun Zhang, Yifan Zhang, Jian Hu, Yueying Li, Hao Zhang, Binfeng Xu, Jan Kaut…