BRACE: Anchored Bellman-Residual Correction for Stale Critics in Asynchronous RL
Asynchronous reinforcement learning has become the standard way to scale training for large language models (LLM), 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 ...
| Authors | Guanqun Zhao, Zijun Xie, Binbin Zheng et al. |
|---|---|
| Primary category | cs.LG |
| Published | 9 September 2026 |
| Last updated at source | 15 September 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.