method · first seen 6 Oct 2026
reward gating
Zeroing or masking rewards under conditions to prevent later-stage training from disturbing earlier achievements.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Humanoid Horizon: Extending Task Horizon in Whole-Body Loco-Manipulation via Parallel Training, Dynamic Starting, and Reward GatingPapers · 6 Oct 2026 · Haozhuo Zhang, Qiang Zhang, Jian Tang, Mingzhe Ni, Michele Caprio, Angelo Cangel…
Latest sources
- Humanoid Horizon: Extending Task Horizon in Whole-Body Loco-Manipulation via Parallel Training, Dynamic Starting, and Reward GatingPapers · 6 Oct 2026 · Haozhuo Zhang, Qiang Zhang, Jian Tang, Mingzhe Ni, Michele Caprio, Angelo Cangel…