Researcher — paper.dinq.me

2026 顶会论文

作者位次 #3/5ICML 2026Poster

Beyond Two-Stage Training: Cooperative SFT and RL for LLM Reasoning

Liang Chen · Xueting Han · Li Shen · Jing Bai · Kam-Fai Wong

推理强化学习与后训练OpenReview

作者位次 #8/8ICML 2026Poster· Deep Learning / Large Language Models

AdaGC: Enhancing LLM Pretraining Stability via Adaptive Gradient Clipping

Guoxia Wang · Shuai Li · Congliang Chen · Jinle Zeng · Jiabin Yang · Dianhai Yu · Yanjun Ma · Li Shen

数据与预训练OpenReview

作者位次 #18/18ICML 2026Poster

UltraHorizon: Benchmarking LLM-Agent Capabilities in Ultra Long-Horizon Scenarios

Haotian Luo · Huaisong Zhang · Xuelin Zhang · Haoyu Wang · Zeyu Qin · Wenjie Lu · Guozheng Ma · Haiying He · Yingsha Xie · Qiyang Zhou · Zixuan Hu · Hongze Mi · Yibo Wang · Naiqiang Tan · Hong Chen · Yi Fung · Chun Yuan · Li Shen

智能体与工具评测与基准OpenReview

作者位次 #8/8ICML 2026Poster· General Machine Learning / Transfer, Multitask and Meta-learning

Plasticity Activation via Polar Operator: A Plug-in Method for Balancing Stability and Plasticity

Guodong Zheng · Enneng Yang · Xiaoyan Wang · Yihan Chen · Feihong He · Quan Zheng · Peng Wang · Li Shen

OpenReview

作者位次 #7/7ICML 2026Poster· Reinforcement Learning / Batch/Offline

HTAC: Hierarchical Task-Aware Composition for Continual Offline Reinforcement Learning

Qiyang Zhou · Xu Ruihang · Peng Wang · Wenjie Lu · Xiaochun Cao · Naiqiang Tan · Li Shen

强化学习与后训练OpenReview

作者位次 #6/10ICML 2026Poster· Deep Learning / Large Language Models

Language-based Trial and Error Falls Behind in the Era of Experience

Haoyu Wang · Guozheng Ma · Shugang Cui · Yilun Kong · Haotian Luo · Li Shen · Mengya Gao · Yichao Wu · Xiaogang Wang · Dacheng Tao

OpenReview

作者位次 #7/13ICML 2026Poster· Optimization / Stochastic

GradientStabilizer: Fix the Norm, Not the Gradient

Tianjin Huang · Zhangyang “Atlas” Wang · Haotian Hu · Zhenyu Zhang · Gaojie Jin · Xiang Li · Li Shen · Jiaxing Shang · Tianlong Chen · Ke Li · Lu Liu · Qingsong Wen · Shiwei Liu

OpenReview

作者位次 #7/7ICML 2026Spotlight· Deep Learning / Large Language Models

Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention Sink

Guozhi Liu · Weiwei Lin · Tiansheng Huang · Ruichao Mo · Qi Mu · Xiumin Wang · Li Shen

安全强化学习与后训练OpenReview

作者位次 #6/7ICML 2026Poster· General Machine Learning / Transfer, Multitask and Meta-learning

FRISM: Fine-Grained Reasoning Injection via Subspace-Level Model Merging for Vision–Language Models

Chenyu Huang · Peng Ye · Xudong Tan · Jinhan Mu · Shenghe Zheng · Li Shen · Tao Chen

推理多模态OpenReview

作者位次 #3/6ICML 2026Poster· Deep Learning / Large Language Models

Rewiring Experts on the Fly: Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert Models

Guinan Su · Yanwu Yang · Li Shen · Lu Yin · Shiwei Liu · Jonas Geiping

OpenReview

作者位次 #7/7ICML 2026Poster· General Machine Learning / Everything Else

SCNS: Continual Personalization of Diffusion Models via Submodular Concept Neuron Selection

Zijie Peng · Enneng Yang · Yifei Cheng · Hongliang Yuan · Fei Ma · Xiaochun Cao · Li Shen

对齐与偏好可解释性OpenReview

作者位次 #4/5ICML 2026Poster· General Machine Learning / Transfer, Multitask and Meta-learning

Merge to Remember: Sharpness-Aware Isotropic Merging for Continual Learning

Qun Yang · Enneng Yang · Wei Chen · Li Shen · Long Lan

OpenReview

作者位次 #7/7ICML 2026Poster· General Machine Learning / Transfer, Multitask and Meta-learning

Efficient Bilevel Optimization for CKA-Guided MoE Upcycling

Zhiyuan Yu · Enneng Yang · Hao Jiang · Guojie Zhu · Feihong He · Peng Wang · Li Shen

效率与系统OpenReview

Li Shen

获取完整研究档案

联系方式

2026 顶会论文

Beyond Two-Stage Training: Cooperative SFT and RL for LLM Reasoning

AdaGC: Enhancing LLM Pretraining Stability via Adaptive Gradient Clipping

UltraHorizon: Benchmarking LLM-Agent Capabilities in Ultra Long-Horizon Scenarios

Plasticity Activation via Polar Operator: A Plug-in Method for Balancing Stability and Plasticity

HTAC: Hierarchical Task-Aware Composition for Continual Offline Reinforcement Learning

Language-based Trial and Error Falls Behind in the Era of Experience

GradientStabilizer: Fix the Norm, Not the Gradient

Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention Sink

FRISM: Fine-Grained Reasoning Injection via Subspace-Level Model Merging for Vision–Language Models

Rewiring Experts on the Fly: Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert Models

SCNS: Continual Personalization of Diffusion Models via Submodular Concept Neuron Selection

Merge to Remember: Sharpness-Aware Isotropic Merging for Continual Learning

Efficient Bilevel Optimization for CKA-Guided MoE Upcycling

常见合作者