Long-horizon LLM agents make multi-step decisions, yet outcome-reward-only reinforcement learning provides little direct supervision for individual steps. We introduce AttnSR, a step-level reward augmentation method that derives reward allocations from policy attention extracted during training replay, without a separa...
Meng-Song Wu, Tian-Chen Zhou, Zhao-Chen Su et al.· 0 citations
Historical pull requests record the design decisions, compatibility constraints, and implementation patterns behind a codebase's current state. Experience relevant to a new task can span related changes whose descriptions emphasize different concerns. We introduce Chronos, a test-time framework that makes this connecte...
This work proposes Ecdysis, which aggregates failure evidence across task instances before promoting recurring failure patterns into persistent harness evolution, biasing evolution toward repairs that are more likely to generalize beyond individual model behaviors.
Rui-Qing Yue, Yu Cui, Zhuo-Yu Sun et al.· arXiv.org· 4 citations· ⚡1
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.