CoPE supplements Group Relative Policy Optimization (GRPO) with multi-step action forecasting.
Forecasting future actions makes language-agent training more sample-efficient
CoPE trains agents to predict several subsequent actions from each interaction history, improving five benchmark results without adding inference-time computation.
Big Tech
Haofei Yu · Yixing Chen · Yining Zhao · Xuyan Ye · Yuxiang Liu · Zhigang Hua · +3 more
University of Illinois Urbana-Champaign · University of California San Diego · Meta AI
Research Digest··3 min read
Yu and colleagues introduce CoPE, an auxiliary training objective that extracts more supervision from successful agent trajectories by asking the policy to forecast future action sequences.
Why this paper
From Meta AI and 2 others
In one line
Training agents to forecast their own future actions on successful trajectories improves agentic RL performance and sample efficiency without inference overhead.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§