This work builds a trajectory-level diagnostic pipeline that evaluates both answer correctness and evidence-grounding quality under a unified ReAct-style scaffold and introduces a six-category taxonomy covering modality shortcuts, phantom grounding, wrong-evidence-right-answer cases, over-retrieval laundering, cross-mo...
Zheng-Xian Wu, Junjie Gao, Kai Yang· arXiv.org· 0 citations
Experiments on GUI agent benchmarks show that LACL-GUI provides more effective learning signals and consistently improves agent performance over prior methods, highlighting the value of trajectory-level supervision in contrastive RLVR.
Chengyang Gu, Le Zhang, Jing-Bo Zhou et al.· 0 citations
This work introduces SearchOS, a system-level multi-agent framework that turns fragile, implicit search progress into explicit, persistent, and shared state, and introduces a Search Tool Middleware Harness that intercepts model and tool interactions to record grounded evidence and react to stalls or budget exhaustion.
