Feedback-driven iteration is widely used to improve LLM-generated scientific ideas, but existing systems either leave revision decisions implicit in generation, align revisions with feedback-localized weaknesses, or restrict action selection to system-specific operations. This is problematic because problems, hypothese...
Feng-Xian Ji, Xie Yu, Zhe-Xuan Cui et al.· 0 citations
This work introduces FinCUABuildBench, a benchmark for evaluating financial CUA task construction, and introduces FinCUABuildAgent, a multi-agent system for automatically constructing dynamic financial CUA evaluation tasks.
Jing-Pu Yang, Feng-Xian Ji, Jinri Guo et al.· 0 citations
This work establishes a rigorous standard for trustworthiness in autonomous agents deployed for public health by evaluating VLMs across three progressive capabilities: Basic Perception (Ingredients&Cooking Methods), Quantitative Reasoning (Portion Size&Nutritional Profiling), and Safety-Critical Advisory (Disease-Speci...
Qian Jiang, Zhe-Cheng Shi, Jing-Pu Yang et al.· arXiv.org· 1 citation
This work proposes an LLM watermarking scheme \textit{WaterMoE} for the growingly popular Mixture-of-Experts (MoE) LLMs that achieves a fidelity performance close to the unwatermarked and consistently outperforms state-of-the-art watermarking methods on the benchmark.
