#natural language process...
Sep 2026
Before the Rollout Ends: Early Terminal Reward Prediction for Long-horizon Coding Agents
Contextual Early Reward (CER), which predicts terminal reward through behavioral evidence in a trajectory prefix, provides an interpretable, efficient, and dense evaluation method for long-horizon coding agents.
Ji-Han Yao, Si-Han Zeng, Shang-Bin Feng et al.
· arXiv.org · 1 citation