KY
Kun Yuan
cs.LGmath.OCcs.DCstat.MLcs.AIcs.CVcs.MAcs.SYeess.SYcs.CL
On Valency
published · living versionsW_zbf3qjfe·v1 · currentpublished
On the Limits of Stochastic Push-Pull over Time-Varying Digraphs
with Feiming Wang, Yutong He, Sulaiman Alghunaim
1 version
Preprints & journals
110 papers in the corpus · 2013–2026SkillScaffold: Scaffolding Agentic Reinforcement Learning with Policy-Adaptive Skill Credit Assignment and Selective Injection2610.18807v1 · Wenlin Zhong, Hao Wang, Kun Yuan et al.2026 · 0 citationsarXiv
SOLAR: A State-Driven Online Learning Rate Scheduler for LLM Pretraining2609.34681v1 · Qiulin Shang, Binyu Wang, Yongqi Qiao et al.2026 · 0 citationsarXiv
ProofLoom: Proof-Obligation-Driven Theory Construction for Autoformalizing Research-Level Stochastic Optimization2609.34960v1 · Feiming Wang, Daibo Li, Kun Yuan2026 · 0 citationsarXiv
SparkDiffusion: Mitigating the High-Sparsity Trap --- A Unified Framework for up to Single-GPU Acceleration of Visual Generation2609.23153v2 · Yuxi Liu, Haoyu Li, Zekun Zhang et al.2026 · 0 citationsarXiv
TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching2609.21172v2 · Zhihao Shu, Md Musfiqur Rahman Sanim, Jie Hu et al.2026 · 0 citationsarXiv
CrossDistill: Balancing Quality and Diversity via Trajectory-Level Hybrid Few-Step Distillation2609.14725v2 · Yuxi Liu, Haoyu Li, Yixiang Cai et al.2026 · 0 citationsarXiv
RoLA: Rotary-Positioned Low-Rank Linear Attention for Efficient Diffusion Transformers2609.06712v3 · Zekun Zhang, Yixiang Cai, Yuxi Liu et al.2026 · 0 citationsarXiv
Gradient Normalization Enables Communication-Efficient Distributed Learning Under Initialization Data Heterogeneity.42065982 · Sun, Tao, Wu, Baihao, Liu, Xinwang et al.2026 · 0 citationsIEEE transactions on pattern analysis and machine intelligence. 2026;48(9):10761-10773
Curvature-Conditioned Multiscale Momentum with Sphere Constraints for LLM Pretraining2608.28442v1 · Shuchen Zhu, Yuxin Fang, Mingze Wang et al.2026 · 0 citationsarXiv
CED-EF: Compressed Exact Diffusion with Error Feedback for Multi-Agent Learning2608.23013v1 · Sulaiman A. Alghunaim, Kun Yuan2026 · 0 citationsarXiv
CMuon: Accelerating and Stabilizing Diffusion Transformer Training via Chunked Momentum Orthogonalization2608.02502v1 · Chuyan Chen, Peng Sun, Kun Yuan2026 · 0 citationsarXiv
ReasFlow: Assisting Reasoning-Centric Scientific Discovery in Applied Mathematics via a Knowledge-Based Multi-Agent System2607.14178v3 · Yutong He, Daibo Li, Guohong Li et al.2026 · 0 citationsarXiv
Accelerated Decentralized Stochastic Gradient Descent for Strongly Convex Optimization2606.07496v1 · Ming Sun, Kun Yuan2026 · 0 citationsarXiv
Proof-Refactor: Refactoring Generated Formal Proofs into Modular Artifacts2606.03743v1 · Yiming Fu, Peixuan Liu, Zichen Wang et al.2026 · 0 citationsarXiv
GNMR: Runtime Stability Control for Low-Precision Large Language Model Training2606.00539v1 · Boao Kong, Weichen Jia, Engao Zhang et al.2026 · 0 citationsarXiv
DFSAttn: Dynamic Fine-grained Sparse Attention for Efficient Video Generation2605.23445v1 · Jie Hu, Zixiang Gao, Yutong He et al.2026 · 0 citationsarXiv
RoPeSLR: 3D RoPE-driven Sparse-LowRank Attention for Efficient Diffusion Transformers2605.20659v1 · Yuxi Liu, Zekun Zhang, Yixiang Cai et al.2026 · 0 citationsarXiv
BROS: Bias-Corrected Randomized Subspaces for Memory-Efficient Single-Loop Bilevel Optimization2605.10288v2 · Hengrui Zhang, Boao Kong, Engao Zhang et al.2026 · 0 citationsarXiv
FedSLoP: Memory-Efficient Federated Learning with Low-Rank Gradient Projection2604.24012v3 · Yutong He, Zhengyang Huang, Jiahe Geng et al.2026 · 0 citationsarXiv
OptProver: Bridging Olympiad and Optimization through Continual Training in Formal Theorem Proving2604.23712v2 · Chenyi Li, Yanchen Nie, Zhenyu Ming et al.2026 · 0 citationsarXiv
Automating Reformulation for Parallel ADMM2603.19417v1 · Kaizhao Sun, Baihao Wu, Kun Yuan et al.2026 · 0 citationsarXiv
Accelerating LLM Pre-Training through Flat-Direction Dynamics Enhancement2602.22681v2 · Shuchen Zhu, Rizhen Hu, Mingze Wang et al.2026 · 0 citationsarXiv
On the Convergence of Stochastic Gradient Descent with Perturbed Forward-Backward Passes2602.20646v1 · Boao Kong, Hengrui Zhang, Kun Yuan2026 · 0 citationsarXiv
Grouter: Decoupling Routing from Representation for Accelerated MoE Training2603.06626v2 · Yuqi Xu, Rizhen Hu, Zihan Liu et al.2026 · 0 citationsarXiv
M2F: Automated Formalization of Mathematical Literature at Scale2602.17016v1 · Zichen Wang, Wanli Ma, Zhenyu Ming et al.2026 · 0 citationsarXiv
Synergistic Intra- and Cross-Layer Regularization Losses for MoE Expert Specialization2602.14159v1 · Rizhen Hu, Yuan Cao, Boao Kong et al.2026 · 0 citationsarXiv
Model-Free Test Time Adaptation for Out-of-Distribution Detection.41052187 · Zhang, Yi-Fan, Wang, Xue, Zhou, Tian et al.2026 · 4 citationsIEEE transactions on pattern analysis and machine intelligence. 2026;48(2):1542-1553
Mixture of Distributions Matters: Dynamic Sparse Attention for Efficient Video Diffusion Transformers2601.11641v3 · Yuxi Liu, Yipeng Hu, Zekun Zhang et al.2026 · 0 citationsarXiv
Decentralized Optimization over Time-Varying Row-Stochastic Digraphs2512.24483v4 · Liyuan Liang, Yilong Song, Kun Yuan2025 · 0 citationsarXiv
Row-Stochastic Matrices Can Provably Outperform Doubly Stochastic Matrices in Decentralized Learning2511.19513v4 · Bing Liu, Boao Kong, Limin Lu et al.2025 · 0 citationsarXiv
Mixture-of-Channels: Exploiting Sparse FFNs for Efficient LLMs Pre-Training and Inference2511.09323v1 · Tong Wu, Yutong He, Bin Wang et al.2025 · 0 citationsarXiv
An All-Reduce Compatible Top-K Compressor for Communication-Efficient Distributed Learning2510.26709v3 · Chuyan Chen, Chenyang Ma, Zhangxin Li et al.2025 · 0 citationsarXiv
MISA: Memory-Efficient LLMs Optimization with Module-wise Importance Sampling2511.00056v3 · Yuxi Liu, Renjia Deng, Yutong He et al.2025 · 0 citationsarXiv
Improving Model Representation and Reducing KV Cache via Skip Connections with First Value Heads2510.16807v2 · Zhoutong Wu, Yuan Zhang, Yiming Dong et al.2025 · 0 citationsarXiv
MeCeFO: Enhancing LLM Training Robustness via Fault-Tolerant Optimization2510.16415v1 · Rizhen Hu, Yutong He, Ran Yan et al.2025 · 0 citationsarXiv
On the Trade-Off Between Flatness and Optimization in Distributed Learning.40560703 · Cao, Ying, Wu, Zhaoxian, Yuan, Kun et al.2025 · 2 citationsIEEE transactions on pattern analysis and machine intelligence. 2025;47(10):8873-8888
CR-Net: Scaling Parameter-Efficient Training with Cross-Layer Low-Rank Structure2509.18993v3 · Boao Kong, Junzhu Liang, Yuxi Liu et al.2025 · 0 citationsarXiv
From PowerSGD to PowerSGD+: Low-Rank Gradient Compression for Distributed Optimization with Convergence Guarantees2509.11254v1 · Shengping Xie, Chuyan Chen, Kun Yuan2025 · 0 citationsarXiv
On the Escaping Efficiency of Distributed Adversarial Training Algorithms2509.11337v2 · Ying Cao, Kun Yuan, Ali H. Sayed2025 · 0 citationsarXiv
Scaling up Multi-Turn Off-Policy RL and Multi-Agent Tree Search for LLM Step-Provers2509.06493v2 · Ran Xin, Zeyu Zheng, Yanchen Nie et al.2025 · 0 citationsarXiv
An Efficient Subspace Algorithm for Federated Learning on Heterogeneous Data2509.05213v2 · Jiaojiao Zhang, Yizhao Fan, Yuqi Xu et al.2025 · 0 citationsarXiv
Lumen: Consistent Video Relighting and Harmonious Background Replacement with Video Generative Models2508.12945v1 · Jianshu Zeng, Yuxuan Liu, Yutong Feng et al.2025 · 0 citationsarXiv
Greedy Low-Rank Gradient Compression for Distributed Learning with Convergence Guarantees2507.08784v4 · Chuyan Chen, Yutong He, Pengrui Li et al.2025 · 0 citationsarXiv
On the Linear Speedup of the Push-Pull Method for Decentralized Optimization over Digraphs2506.18075v5 · Liyuan Liang, Gan Luo, Kun Yuan2025 · 0 citationsarXiv
Achieving Linear Speedup and Near-Optimal Complexity for Decentralized Optimization over Row-stochastic Networks2506.04600v2 · Liyuan Liang, Xinyi Chen, Gan Luo et al.2025 · 0 citationsarXiv
HATA: Trainable and Hardware-Efficient Hash-Aware Top-k Attention for Scalable Large Model Inference2506.02572v1 · Ping Gong, Jiawei Yi, Shengnan Wang et al.2025 · 4 citationsarXiv
Distributed Retraction-Free and Communication-Efficient Optimization on the Stiefel Manifold2506.02879v2 · Yilong Song, Peijin Li, Bin Gao et al.2025 · 0 citationsarXiv
TAH-QUANT: Effective Activation Quantization in Pipeline Parallelism over Slow Network2506.01352v2 · Guangxin He, Yuan Cao, Yutong He et al.2025 · 0 citationsarXiv
Diffusion Learning with Partial Agent Participation and Local Updates2505.11307v1 · Elsa Rizk, Kun Yuan, Ali H. Sayed2025 · 0 citationsarXiv
BEVHeight++: Toward Robust Visual Centric 3D Object Detection.40067721 · Yang, Lei, Tang, Tao, Li, Jun et al.2025 · 48 citationsIEEE transactions on pattern analysis and machine intelligence. 2025;47(6):5094-5111
Career total: 158 works. 110 are in this corpus.Showing the 50 most recent.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.