Skip to content
AI News HubLIVE
Original source2 min read

Regularized Score Centering

Summary

Regularized Score Centering (RSC) adds a reverse KL penalty to score centering to keep the learner close to the behavior policy in asynchronous RL. Theoretical analysis shows improved convergence, and experiments on math reasoning and six agentic tasks show substantial sample efficiency gains, with 14.3% improvement on AppWorld and 9.9x the strongest baseline on BabyAI.

SourcearXiv Machine LearningAuthor: Hao Bai, Chenlu Ye, Xuanchang Zhang, Aviral Kumar, Tong Zhang
Regularized Score Centering
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 16 Nov 2026]

Title:Regularized Score Centering

View a PDF of the paper titled Regularized Score Centering, by Hao Bai and 4 other authors

View PDF HTML (experimental)

Abstract:Asynchronous reinforcement learning increases training throughput, but data from stale policies and numerical mismatch between training and inference can destabilize learning. To stabilize off-policy learning, score centering (SC) removes the effect of reward baselines on the expected update. Our gradient diagnostics show less extreme gradients with SC than with importance sampling (IS), but SC can still fail when the learner moves far from the behavior policy that generated the data. To limit this policy mismatch, we introduce Regularized Score Centering (RSC), which adds a reverse Kullback-Leibler (KL) penalty to encourage the learner's next-token distribution to stay close to the behavior policy's next-token distribution. RSC works with or without truncated importance sampling (TIS), which caps token importance weights. Our theoretical analysis supports this approach, showing that KL regularization improves convergence guarantees. We evaluate dense and mixture-of-experts models on mathematical reasoning and agentic tasks, including online shopping, tool use, games, and fully asynchronous software engineering. RSC with TIS substantially improves sample efficiency across all agentic benchmarks, attaining higher training rewards with the same number of sampled responses. In final evaluations, it achieves top performance across all six agentic tasks, improving by 14.3% on AppWorld and reaching 9.9 times the strongest baseline's score on BabyAI.

Subjects:

Machine Learning (cs.LG)

Cite as: arXiv:2611.21791 [cs.LG]

(or arXiv:2611.21791v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2611.21791

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Hao Bai [view email] [v1] Mon, 16 Nov 2026 19:12:50 UTC (3,908 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Regularized Score Centering, by Hao Bai and 4 other authors

View PDF

HTML (experimental)

TeX Source

view license

Additional Features

Audio Summary

Current browse context:

cs.LG

new | recent | 2026-11

Change to browse by:

cs

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

InvestorsAdvanced

Key points

  • RSC augments score centering with a reverse KL penalty to limit policy mismatch between learner and behavior policy.
  • KL regularization improves convergence guarantees, and RSC works with or without truncated importance sampling (TIS).
  • RSC with TIS improves sample efficiency across all agentic benchmarks, achieving 14.3% gain on AppWorld and 9.9x the strongest baseline on BabyAI.

Highlights and analysis are generated automatically and may contain errors. Check the original source.