SEP 9, 2026 · PREPRINT
BRACE: Anchored Bellman-Residual Correction for Stale Critics in Asynchronous RL
arXiv
This is an algorithmic proposal for asynchronous RL with empirical validation on a single benchmark task, lacking peer review, comparative rigor across multiple domains, and human evaluation.
Reported
Improvement over strongest baseli…2.4%
Speedup versus synchronous traini…2.46×
Off-policy stability window50 updates