37 papers
Evaluating Agentic Capabilities Before Post-Training: An End-to-End Approach
Jiarui Qin, Junjie Huang, Yunjia Xi +5
Agentic capabilities are increasingly important for modern LLMs, yet remain difficult to evaluate during pretraining. Existing approaches typically rely on static trajectory-derive…
PhysCodeGym: Benchmarking Coding Agents for Physically Verifiable 4D Reconstruction
Shuai Shao, Jinchen He, Cunxin Fan +10
Building physical simulations from video requires coding agents to translate observations into executable models and refine them through tool use. We introduce PhysCodeGym, a bench…
What Makes Good Agentic Data? An ACE Lens on Data Generation for LLM Agents
Xingshan Zeng, Zishan Xu, Boju Zhang +11
LLM agents increasingly rely on generated interaction data to learn how to interact with external environments. Agentic data generation must maintain consistency among environments…
SEAM: Shot Entity-Attribute Memory for Consistent Short-Drama Generation at Scale
Jiaqi Liu, Maolin Ran, Xiaoyang Lu +5
Short-drama generation has grown into a large, industrialized pipeline, and as it scales from isolated shots to the episode level, visual continuity has become a critical bottlenec…
SkillGate: Training In-Policy Skill Selection in Long-Horizon Agents
Qingyao Li, Wenxiang Jiao, Shuai Shao +6
Agent frameworks increasingly package procedural knowledge as skills: instruction files an agent reads on demand, while public libraries now hold thousands of them. Which skill to…
SAGE: Self-Evolving Storyboard Skills via Attribution-Guided Rule Evolution
Maolin Ran, Xiaoyang Lu, Jiaqi Liu +5
Storyboards turn screenplays into visual shot plans for automated short drama production. Professional storyboarding relies on tacit directorial expertise and remains an industrial…
EnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning
Zishan Xu, Zhiyuan Yao, Yuxin Chen +9
Training large language model agents for long-horizon tool use typically relies on interactions with real or synthesized executable environments, whose construction and verificatio…
Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories
Shuai Shao, Kangning Zhang, Qingyao Li +7
Agents built around large language models continually accumulate interaction trajectories during deployment, yet their behavior typically remains fixed. Beyond updating model weigh…
VAD: Attributing Visual Evidence for Target Reconstruction in Multimodal On-Policy Distillation
Kangning Zhang, Yixing Li, Shuai Shao +9
Multimodal on-policy distillation (OPD) transfers fine-grained visual knowledge by supervising student-generated trajectories with a privileged-view teacher. Yet its next-token cor…
IssueExec: A Test-Driven Approach for Localizing Software Engineering Issues
Jiawei Liu, Yun Lin, Chenyan Liu +5
Issue localization, which identifies code locations requiring modification from issue descriptions, is a critical step in automated software maintenance. Existing approaches predom…
XSearch: Explainable Code Search via Concept-to-Code Alignment
Yiming Liu, Ruofan Liu, Yun Lin +7
Semantic code search has been widely adopted in both academia and industry. These approaches embed natural-language queries and code snippets into a shared embedding space and retr…
Communication Policy Evolution for Proactive LLM Agents
Xinbei Ma, Jiyang Qiu, Yao Yao +10
LLM agents have rapidly evolved into autonomous systems, yet a persistent information gap remains between users and agents: communication is costly, while users' identical preferen…
Retrospective Progress-Aware Self-Refinement for LLM Agent Training
Xinbei Ma, Congmin Zheng, Jiyang Qiu +11
Long-horizon LLM-based agents receive rich environmental observations during interaction, yet outcome rewards provide limited explicit supervision about how individual actions adva…
MMSkills: Towards Multimodal Skills for General Visual Agents
Kangning Zhang, Shuai Shao, Qingyao Li +8
Reusable skills have become a core substrate for improving agent capabilities, yet most existing skill packages encode reusable behavior primarily as textual prompts, executable co…
SkillPager: Query-Adaptive Intra-Skill Navigation via Semantic Node Retrieval
Zicai Cui, Zihan Guo, Weiwen Liu +1
Skill-based LLM agents increasingly rely on long procedural documents, but full-document prompting wastes tokens and dilutes information critical to execution. We study this settin…
Holder Policy Optimisation
Yuxiang Chen, Dingli Liang, Yihang Chen +8
Group Relative Policy Optimisation (GRPO) enhances large language models by estimating advantages across a group of sampled trajectories. However, mapping these trajectory-level ad…
SkillMAS: Skill Co-Evolution with LLM-based Multi-Agent System
Shuai Pan, Yixiang Liu, Jiaye Gao +7
Large language model (LLM) agent systems are increasingly expected to improve after deployment, but existing work often decouples two adaptation targets: skill evolution and multi-…
Contexting as Recommendation: Evolutionary Collaborative Filtering for Context Engineering
Jiachen Zhu, Zhuoying Ou, Congmin Zheng +9
Large Language Models (LLMs) are highly sensitive to their input contexts, motivating the development of automated context engineering. However, existing methods predominantly trea…
SMMBench: A Benchmark for Source-Distributed Multimodal Agent Memory
Huacan Chai, Yukai Wang, Yingxuan Yang +7
Existing benchmarks for multimodal memory reasoning largely evaluate systems within pre-assembled contexts, but under-evaluate whether agents can use evidence distributed across in…
MemQ: Integrating Q-Learning into Self-Evolving Memory Agents over Provenance DAGs
Junwei Liao, Haoting Shi, Ruiwen Zhou +9
Episodic memory allows LLM agents to accumulate and retrieve experience, but current methods treat each memory independently, i.e., evaluating retrieval quality in isolation withou…
SWE-Cycle: Benchmarking Code Agents across the Complete Issue Resolution Cycle
Hao Guan, Lingyue Fu, Shao Zhang +8
As autonomous code agents move toward end-to-end software development, evaluating their practical autonomy becomes critical. Current benchmarks hide friction by testing agents in p…
Position: Agentic AI System Is a Foreseeable Pathway to AGI
Junwei Liao, Shuai Li, Muning Wen +2
Is monolithic scaling the only path to AGI? This paper challenges the dogma that purely scaling a single model is sufficient to achieve Artificial General Intelligence. Instead, we…
Position: Academic Conferences are Potentially Facing Denominator Gaming Caused by Fully Automated Scientific Agents
Rong Shan, Te Gao, Hang Zheng +6
The implicit policy of maintaining relatively stable acceptance rates at top AI conferences, despite exponentially growing submissions, introduces a critical structural vulnerabili…
AdverMCTS: Combating Pseudo-Correctness in Code Generation via Adversarial Monte Carlo Tree Search
Qingyao Li, Weiwen Liu, Weinan Zhang +2
Recent advancements in Large Language Models (LLMs) have successfully employed search-based strategies to enhance code generation. However, existing methods typically rely on stati…