UIUC's CoPE Boosts GRPO Sample Efficiency with Action Forecasting Loss

UIUC researchers introduce CoPE, which adds a simple action-forecasting loss to GRPO, letting agents predict their next K steps on successful rollouts. The approach roughly doubles sample efficiency for long-horizon language agent training.

2026-10-17 ~ 2026-10-17 · 2 related posts

• CoPE adds one forecasting loss to GRPO for 2x sample efficiency in agent RL — youjiaxuan · 2026-10-17
• CoPE adds one forecasting loss to GRPO, doubling long-horizon agent sample efficiency — Lianhuiq · 2026-10-17