This work builds AutoBuddy, a reference agent with interchangeable backbone LLMs: it identifies tasks from the workspace, grounding each in its source records, and executes each in a separate session with write access, finding a systematic shift in where tasks are lost.
Shiqi Chen, Daniel Eduardo Madrigal Díaz, Wei-Ning Chen et al.· 0 citations
Across unseen environments, SPA consistently and substantially improves over vanilla RL: for example, it raises the Sokoban success rate from 25.6% to 59.8% on Qwen2.5-1.5B-Instruct, letting sub-3B models surpass a 20B baseline on these tasks.
Shiqi Chen, Tong-Yao Zhu, Zian Wang et al.· 27 citations· ⚡7
HypoSearch is proposed, which generates lightweight hypotheses as soft search hints, explores them through bounded independent branches, and compares branch-level evidence before commitment, and consistently outperforms single-trajectory search and standard parallel baselines.
Ruo-Chen Zhou, Zheng-Zong Chen, Luan Zhang et al.· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.