19 papers
UnitBoost: Managing Compound LLM Systems with a Merge Operator, Not a Model
Xing Zhang, Guanghui Wang, Yanwei Cui +2
Compound LLM systems often solve a coordination problem by adding a higher-level LLM. The resulting meta-agent reads workers' outputs, writes the final answer, allocates later call…
Metrics That Write Themselves: Evolving an Evaluator from Its Own Blind Spots
Xing Zhang, Yanwei Cui, Guanghui Wang +2
Agents improve quickly against a reliable automatic metric and stall without one, and the applications that need them most, report generation among them, are the ones nobody knows…
The Blind Curator: How a Biased Judge Silently Disables Skill Retirement in Self-Evolving Agents
Xing Zhang, Yanwei Cui, Guanghui Wang +4
A self-evolving agent retires its bad skills by watching them fail, so what happens when the judge cannot see the failures? Skill retirement is the structural constraint that keeps…
Reconcile Once, Write Anytime: A Trust-Tiered Librarian and a Multi-Agent Writer for Drift-Free, Point-in-Time Research
Xing Zhang, Yanwei Cui, Guanghui Wang +1
Long-form research reports generated by large language models drift, contradict themselves, and lose provenance: the same metric appears with different values, and rumor is quoted…
Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents
Xing Zhang, Guanghui Wang, Yanwei Cui +4
Self-evolving agent systems create, revise, and retire their own skills, but every such loop assumes a reliable evaluation metric already exists. In many real applications none doe…
Budgeted Act-or-Defer Multi-Agent LLM Deliberation with Local Reliability Bounds
Mengdie Flora Wang, Haochen Xie, Guanghui Wang +2
Multi-agent deliberation among LLMs can improve reasoning, but deployment requires deciding when the current answer is reliable enough to act on and when it should be escalated to…
Yuvion LLM: An Adversarially-Aware Large Language Model for Content And AI Safety
Ting Ma, Xiufeng Huang, Benlei Cui +43
As large language models are increasingly deployed in real-world systems, safety failures can still lead to harmful outputs and dangerous misuse. We argue that the essence of safet…
Yuvion VL: A Multimodal Foundation Model for Adversarial Content and AI Safety
Shikai Qiu, Xiaowen Xu, Benlei Cui +55
General-purpose models often struggle to reliably identify and understand real-world multimodal risks, largely due to the inherent multimodal adversarial nature of content and AI s…
Closing the Feedback Loop: From Experience Extraction to Insight Governance in Verbal Reinforcement Learning
Yanwei Cui, Xing Zhang, Yulong Zhang +4
Training-free verbal reinforcement learning enables LLM agents to learn from world feedback -- objective signals such as dynamic task outcomes, market returns, or demand forecasts…
Better Literary Translation: A Multi-Aspect Data Generation and LLM Training Approach
Zhihao Lin, Ziqi Zhu, Hao Huang +2
Literary translation poses unique challenges due to the scarcity of high-quality annotated data and the need to balance expression fluency with literary effect. We present a multi-…
The Alignment Floor: How Persona Customization Breaks Safety in Weakly-Aligned LLMs
Xing Zhang, Guanghui Wang, Yanwei Cui +4
Telling an LLM to "be enthusiastic" raises its sycophancy rate from 30\% to 50\% on a lightly-aligned model, but has zero effect on a strongly-aligned one. We define this gap as th…
Guardrails Beat Guidance: A Large-Scale Study of Rules, Skills, and Persistent Configuration for Coding Agents
Xing Zhang, Guanghui Wang, Yanwei Cui +4
Random rules improve a coding agent's task performance as much as expert-curated ones (both pp on a discriminative subset of SWE-bench Verified), and in our data every indiv…
Prompt Optimization Is a Coin Flip: Diagnosing When It Helps in Compound AI Systems
Xing Zhang, Guanghui Wang, Yanwei Cui +4
Prompt optimization in compound AI systems is statistically indistinguishable from a coin flip: across 72 optimization runs on Claude Haiku 4.5 (6 methods 4 tasks …
Ratchet: How Reliable Must an LLM Judge Be to Retire a Skill?
Xing Zhang, Yanwei Cui, Guanghui Wang +4
A large language model (LLM) agent that writes and edits its own skill library must also decide which skills to keep, from one noisy scalar per skill. The answer is exact: a judge…
Library Drift: Diagnosing and Fixing a Silent Failure Mode in Self-Evolving LLM Skill Libraries
Xing Zhang, Yanwei Cui, Guanghui Wang +4
Self-evolving skill libraries face a silent failure mode we term \emph{library drift}: unbounded skill accumulation without outcome-driven lifecycle management causes retrieval deg…
Hindsight Preference Optimization for Financial Time Series Advisory
Yanwei Cui, Guanghui Wang, Xing Zhang +7
Time series models predict numbers; decision-makers need advisory -- directional signals with reasoning, actionable suggestions, and risk management. Training language models for s…
From Debate to Decision: Conformal Social Choice for Safe Multi-Agent Deliberation
Mengdie Flora Wang, Haochen Xie, Guanghui Wang +9
Multi-agent debate improves LLM reasoning, yet agreement is not evidence of correctness. When agents converge on a wrong answer through social reinforcement, consensus stopping com…
Experience Compression Spectrum: Unifying Memory, Skills, and Rules in LLM Agents
Xing Zhang, Guanghui Wang, Yanwei Cui +4
As LLM agents scale to long-horizon, multi-session deployments, efficiently managing accumulated experience becomes a critical bottleneck. Agent memory systems and agent skill disc…
Verified Multi-Agent Orchestration: A Plan-Execute-Verify-Replan Framework for Complex Query Resolution
Xing Zhang, Yanwei Cui, Guanghui Wang +7
We present Verified Multi-Agent Orchestration (VMAO), a framework that coordinates specialized LLM-based agents through a verification-driven iterative loop. Given a complex query,…