BRACE: Anchored Bellman-Residual Correction for Stale Critics in Asynchronous RL — Paper

Abstract

Asynchronous reinforcement learning has become the standard way to scale training for language models, but the resulting policy lag biases the critic toward the stale behavior policy. Existing work on asynchronous LLM training corrects the actor and leaves this bias unaddressed, while the off-policy value correction of classical RL does not carry over to long-horizon agentic tasks, since a short correction horizon leaves the regression target free of the reward and a long one lets the product of importance ratios drift exponentially with the trajectory length. We propose BRACE, an anchored Bellman-residual correction for stale value models. BRACE bounds the correction horizon to a prefix of policy tokens and anchors a constant-weight Monte-Carlo tail beyond it, which separates policy correction from reward propagation. BRACE improves mean@1 on BrowseComp-Plus by $2.4\%$ over the strongest baseline, runs $2.46\times$ faster per step than synchronous training, and remains stable $50$ updates off-policy.

Authors 6

Guanqun Zhao, Zijun Xie, Binbin Zheng, Jiafeng Lu, Enlei Gong, Zeyu Chen

Specification

Official page: arxiv.org/abs/2609.09783
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh
Arxiv announce type: cross
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh
arXiv id: 2609.09783
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh
Categories: cs.LG, cs.AI
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh
PDF: arxiv.org/pdf/2609.09783
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh
Primary category: cs.LG
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh
Published: 11 Sept 2026
Source:arXiv (Atom API + RSS)T1observed 4 h agohigh

Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →

Provenance

Attributed facts

9

Source tiers

T19

Freshest observation

4 h ago

Conflicts

None

No models linked to this paper yet.

Authors: Guanqun Zhao, Zijun Xie, Binbin Zheng

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history · arXiv id

1 claims · 1 propertiesShow all properties

arXiv idarxiv_id1

Claim history for arXiv id
Value	Valid from → to	Status	Source	Confidence	Extractor
2609.09783	11 Sept 2026 21:03 UTC → current	current	arXiv (Atom API + RSS)T1	high	deterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

11 Sept 2026
New paperPaperBRACE: Anchored Bellman-Residual Correction for Stale Critics in Asynchronous RL
New paper: BRACE: Anchored Bellman-Residual Correction for Stale Critics in Asynchronous RL
arxiv
rss.arxiv.org

Full timeline →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
arXiv (Atom API + RSS)	rss.arxiv.org/rss/cs.AI	feed	T1· Official	2 h ago	1

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.