9 papers
SkillCAT: Contrastive, Assessment-Augmented and Topology-AwareSkill Self-Evolution for LLM Agents
Kunfeng Chen, Qihuang Zhong, Juhua Liu +1
Skill self-evolution methods for LLM agents aim to turn execution trajectories into reusable skill documents. However, current pipelines typically derive skill patches from a singl…
ConRAG: Consensus-Driven Evidence Flow for Multi-Hop Question Answering
Yikai Zhu, Kunfeng Chen, Qihuang Zhong +2
Retrieval-augmented generation (RAG) has emerged as a promising paradigm for enhancing large language models (LLMs) on multi-hop question answering (QA), which requires reasoning o…
Scientific Logicality Enriched Methodology for LLM Reasoning: A Practice in Physics
Zhaoxin Yu, Nan Xu, Kun Chen +3
With the continuous advancement of reasoning abilities in Large Language Models (LLMs), their application to scientific reasoning tasks has gained significant research attention. C…
Tool Retrieval Bridge: Aligning Vague Instructions with Retriever Preferences via Bridge Model
Kunfeng Chen, Luyao Zhuang, Fei Liao +3
Tool learning has emerged as a promising paradigm for large language models (LLMs) to address real-world challenges. Due to the extensive and irregularly updated number of tools, t…
Not All Tokens See Equally: Perception-Grounded Policy Optimization for Large Vision-Language Models
Zekai Ye, Qiming Li, Xiaocheng Feng +6
While Reinforcement Learning from Verifiable Rewards (RLVR) has advanced reasoning in Large Vision-Language Models (LVLMs), prevailing frameworks suffer from a foundational methodo…
Try, Check and Retry: A Divide-and-Conquer Framework for Boosting Long-context Tool-Calling Performance of LLMs
Kunfeng Chen, Qihuang Zhong, Juhua Liu +2
Tool-calling empowers Large Language Models (LLMs) to interact with external environments. However, current methods often struggle to handle massive and noisy candidate tools in lo…
APEX-Searcher: Refining Credit Assignment with Subgoaling for Agentic Retrieval-Augmented Generation
Kun Chen, Qingchao Kong, Zhao Feifei +1
Retrieval-augmented generation (RAG) connects large language models (LLMs) to external knowledge, but single-round retrieval is often insufficient for complex multi-hop questions.…
Flexible Entropy Control in RLVR with a Gradient-Preserving Perspective
Kun Chen, Peng Shi, Fanfan Liu +4
Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a critical method for enhancing the reasoning capabilities of Large Language Models (LLMs). However, continuous…
Metis-SPECS: Decoupling Multimodal Learning via Self-distilled Preference-based Cold Start
Kun Chen, Peng Shi, Haibo Qiu +4
Reinforcement learning (RL) with verifiable rewards has recently catalyzed a wave of "MLLM-r1" approaches that bring RL to vision language models. Most representative paradigms beg…