9 papers
Inherit-MAS: Test-Time Evolution of Multi-Agent Systems through Workflow and Execution Inheritance
Songtao Wei, Yi Li, Zhichun Guo +1
Multi-agent systems (MAS) built from large language models coordinate specialized agents to tackle complex tasks, but effective workflows are difficult to design in advance. Test-t…
More Skills, Worse Agents? Skill Shadowing Degrades Performance When Expanding Skill Libraries
Hongwen Song, Song Wei
Skill libraries allow LLM agents to load task-specific instructions on demand, letting non-expert users solve domain-specific tasks through natural language without knowing which s…
DarkForest: Less Talk, Higher Accuracy for Multi-Agent LLMs
Yi Li, Songtao Wei, Dongming Jiang +3
Multi-agent LLM systems improve reasoning by combining outputs from multiple agents, but interaction-heavy methods can introduce error propagation and high communication overhead.…
LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models
Songtao Wei, Yi Li, Zhikai Li +7
Large reasoning models, such as OpenAI o1 and DeepSeek-R1, tend to become increasingly verbose as their reasoning capabilities improve. These inflated Chain-of-Thought (CoT) trajec…
CoSA: Compressed Sensing-Based Adaptation of Large Language Models
Songtao Wei, Yi Li, Bohan Zhang +6
Parameter-Efficient Fine-Tuning (PEFT) has emerged as a practical paradigm for adapting large language models (LLMs) without updating all parameters. Most existing approaches, such…
Anatomy of Agentic Memory: Taxonomy and Empirical Analysis of Evaluation and System Limitations
Dongming Jiang, Yi Li, Songtao Wei +8
Agentic memory systems enable large language model (LLM) agents to maintain state across long interactions, supporting long-horizon reasoning and personalization beyond fixed conte…
Small Updates, Big Doubts: Does Parameter-Efficient Fine-tuning Enhance Hallucination Detection ?
Xu Hu, Yifan Zhang, Songtao Wei +4
Parameter-efficient fine-tuning (PEFT) methods are widely used to adapt large language models (LLMs) to downstream tasks and are often assumed to improve factual correctness. Howev…
Rainbow Noise: Stress-Testing Multimodal Harmful-Meme Detectors on LGBTQ Content
Ran Tong, Songtao Wei, Jiaqi Liu +1
Hateful memes aimed at LGBTQ\,+ communities often evade detection by tweaking either the caption, the image, or both. We build the first robustness benchmark for this setting, pair…
: Benchmarking the Robustness of Audio-Visual Recognition Models at Test-Time
Sarthak Kumar Maharana, Saksham Singh Kushwaha, Baoming Zhang +4
While recent audio-visual models have demonstrated impressive performance, their robustness to distributional shifts at test-time remains not fully understood. Existing robustness…