“Score Centering Stabilizes Off-Policy Reinforcement Learning”
This is a AI post classified by Jev as AI research (research), kept by the AI Radar because it carries real work, not commentary.
“Score Centering Stabilizes Off-Policy Reinforcement Learning” Small mismatches between the model generating rollouts and the model being trained can compound over RL and eventually collapse training. And this mismatch creates a small directional bias in every policy gradient update, repeatedly pulling the model toward its slightly different sampler. So this paper, score centering, estimates this unwanted pull and subtracts it from every update, keeping the useful reward signal while preventing the mismatch from snowballing. https://alphaxiv.org/abs/2609.20807
Posted by alphaXiv (55.9k followers) 11 h ago · 81 likes · 3.9k views · view the original post on X. Kept by the AI Radar as AI research. Tools mentioned: alphaXiv.
More AI work like this
- We built AI that actually knows aging biology. Longevity-LLM (9B params) beats OpenAI… — @biogerontology
- We need more examples like this in the open-source RL ecosystem — @adithya_s_k
- "Verbalizing Subliminal Learning Effects Using Text Optimization" — @askalphaxiv
- “What Does Privileged Information Add to On-Policy Self-Distillation?” — @askalphaxiv
- Banger paper from Google Cloud AI Research. — @dair_ai
- If you liked the MiMo-V2.6 livestreamed RL run, don’t forget to check out Marin’s… — @adi_baradwaj
- Astra solved this hand-object reconstruction and tracking problem in one shot😂 — @YuXiang_IRVL
- Epoch AI researcher Michelle Campeau on why so many AI benchmarks have errors in the… — @MTSlive
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 27.1k posts from 4.8k X accounts over the last 14 days, 1.2k tools, 19 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-19 17:40 UTC. Full method.