Kaifeng Lyu - RankMe

61 papers A* 23Journal 37Unranked 1

Year	Rank	Type	Title / Venue / Authors
2026	J	`jnl`	Fine-tuning MLLMs Without Forgetting Is Easier Than You Think. CoRR He Li, Yuhui Zhang, Xiaohan Wang, Kaifeng Lyu, Serena Yeung-Levy
2026	J	`jnl`	SPA: A Simple but Tough-to-Beat Baseline for Knowledge Injection. CoRR Kexian Tang, Jiani Wang, Shaowen Wang, Kaifeng Lyu
2025	A*	`conf`	A Multi-Power Law for Loss Curve Prediction Across Learning Rate Schedules. ICLR Kairong Luo, Haodong Wen, Shengding Hu, Zhenbo Sun, Zhiyuan Liu, Maosong Sun, Kaifeng Lyu, Wenguang Chen
2025	J	`jnl`	A Multi-Power Law for Loss Curve Prediction Across Learning Rate Schedules. CoRR Kairong Luo, Haodong Wen, Shengding Hu, Zhenbo Sun, Zhiyuan Liu, Maosong Sun, Kaifeng Lyu, Wenguang Chen
2025	J	`jnl`	Adam Reduces a Unique Form of Sharpness: Theoretical Insights Near the Minimizer Manifold. CoRR Xinghan Li, Haodong Wen, Kaifeng Lyu
2025	J	`jnl`	Can Small Training Runs Reliably Guide Data Curation? Rethinking Proxy-Model Practice. CoRR Jiachen T. Wang, Tong Wu, Kaifeng Lyu, James Zou, Dawn Song, Ruoxi Jia, Prateek Mittal
2025	J	`jnl`	Data Mixing Can Induce Phase Transitions in Knowledge Acquisition. CoRR Xinran Gu, Kaifeng Lyu, Jiazheng Li, Jingzhao Zhang
2025	A*	`conf`	Efficient stagewise pretraining via progressive subnetworks. ICLR Abhishek Panigrahi, Nikunj Saunshi, Kaifeng Lyu, Sobhan Miryoosefi, Sashank J. Reddi, Satyen Kale, Sanjiv Kumar
2025	A*	`conf`	Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks. ICLR Binghui Li, Zhixuan Pan, Kaifeng Lyu, Jian Li
2025	J	`jnl`	How Far Are We from Optimal Reasoning Efficiency? CoRR Jiaxuan Gao, Shu Yan, Qixin Tan, Lu Yang, Shusheng Xu, Wei Fu, Zhiyu Mei, Kaifeng Lyu, Yi Wu
2025	J	`jnl`	How Learning Rate Decay Wastes Your Best Data in Curriculum-Based LLM Pretraining. CoRR Kairong Luo, Zhenbo Sun, Haodong Wen, Xinyu Shi, Jiarui Cui, Chenyi Dang, Kaifeng Lyu, Wenguang Chen
2025	J	`jnl`	LEGO-Puzzles: How Good Are MLLMs at Multi-Step Spatial Reasoning? CoRR Kexian Tang, Junyao Gao, Yanhong Zeng, Haodong Duan, Yanan Sun, Zhening Xing, Wenran Liu, Kaifeng Lyu, Kai Chen
2025	J	`jnl`	Larger Datasets Can Be Repeated More: A Theoretical Analysis of Multi-Epoch Scaling in Linear Regression. CoRR Tingkai Yan, Haodong Wen, Binghui Li, Kairong Luo, Wenguang Chen, Kaifeng Lyu
2025	J	`jnl`	PCMind-2.1-Kaiyuan-2B Technical Report. CoRR Kairong Luo, Zhenbo Sun, Xinyu Shi, Shengqi Chen, Bowen Yu, Yunyi Chen, Chenyi Dang, Hengtao Tao, Hui Wang, Fangming Liu, Kaifeng Lyu, Wenguang Chen
2025	A*	`conf`	RNNs are not Transformers (Yet): The Key Bottleneck on In-Context Retrieval. ICLR Kaiyue Wen, Xingyu Dang, Kaifeng Lyu
2025	A*	`conf`	Safety Alignment Should be Made More Than Just a Few Tokens Deep. ICLR Xiangyu Qi, Ashwinee Panda, Kaifeng Lyu, Xiao Ma, Subhrajit Roy, Ahmad Beirami, Prateek Mittal, Peter Henderson
2025	J	`jnl`	Shift is Good: Mismatched Data Mixing Improves Test Performance. CoRR Marko Medvedev, Kaifeng Lyu, Zhiyuan Li, Nathan Srebro
2025	A*	`conf`	Towards Understanding Text Hallucination of Diffusion Models via Local Generation Bias. ICLR Rui Lu, Runzhe Wang, Kaifeng Lyu, Xitai Jiang, Gao Huang, Mengdi Wang
2025	J	`jnl`	Towards Understanding Text Hallucination of Diffusion Models via Local Generation Bias. CoRR Rui Lu, Runzhe Wang, Kaifeng Lyu, Xitai Jiang, Gao Huang, Mengdi Wang
2025	A*	`conf`	Weak-to-Strong Generalization Even in Random Feature Networks, Provably. ICML Marko Medvedev, Kaifeng Lyu, Dingli Yu, Sanjeev Arora, Zhiyuan Li, Nathan Srebro
2025	J	`jnl`	Weak-to-Strong Generalization Even in Random Feature Networks, Provably. CoRR Marko Medvedev, Kaifeng Lyu, Dingli Yu, Sanjeev Arora, Zhiyuan Li, Nathan Srebro
2025	J	`jnl`	When Bias Pretends to Be Truth: How Spurious Correlations Undermine Hallucination Detection in LLMs. CoRR Shaowen Wang, Yiqi Dong, Ruinian Chang, Tansheng Zhu, Yuebo Sun, Kaifeng Lyu, Jian Li
2024	A*	`conf`	A Quadratic Synchronization Rule for Distributed Deep Learning. ICLR Xinran Gu, Kaifeng Lyu, Sanjeev Arora, Jingzhao Zhang, Longbo Huang
2024	J	`jnl`	AI-Assisted Generation of Difficult Math Questions. CoRR Vedant Shah, Dingli Yu, Kaifeng Lyu, Simon Park, Nan Rosemary Ke, Michael Mozer, Yoshua Bengio, Sanjeev Arora, Anirudh Goyal
2024	A*	`conf`	Dichotomy of Early and Late Phase Implicit Biases Can Provably Induce Grokking. ICLR Kaifeng Lyu, Jikai Jin, Zhiyuan Li, Simon Shaolei Du, Jason D. Lee, Wei Hu
2024	A*	`conf`	DistillSpec: Improving Speculative Decoding via Knowledge Distillation. ICLR Yongchao Zhou, Kaifeng Lyu, Ankit Singh Rawat, Aditya Krishna Menon, Afshin Rostamizadeh, Sanjiv Kumar, Jean-François Kagy, Rishabh Agarwal
2024	J	`jnl`	Efficient Stagewise Pretraining via Progressive Subnetworks. CoRR Abhishek Panigrahi, Nikunj Saunshi, Kaifeng Lyu, Sobhan Miryoosefi, Sashank J. Reddi, Satyen Kale, Sanjiv Kumar
2024	J	`jnl`	Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks. CoRR Binghui Li, Zhixuan Pan, Kaifeng Lyu, Jian Li
2024	A*	`conf`	Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates. NeurIPS Kaifeng Lyu, Haoyu Zhao, Xinran Gu, Dingli Yu, Anirudh Goyal, Sanjeev Arora
2024	J	`jnl`	Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates. CoRR Kaifeng Lyu, Haoyu Zhao, Xinran Gu, Dingli Yu, Anirudh Goyal, Sanjeev Arora
2024	J	`jnl`	RNNs are not Transformers (Yet): The Key Bottleneck on In-context Retrieval. CoRR Kaiyue Wen, Xingyu Dang, Kaifeng Lyu
2024	J	`jnl`	Safety Alignment Should Be Made More Than Just a Few Tokens Deep. CoRR Xiangyu Qi, Ashwinee Panda, Kaifeng Lyu, Xiao Ma, Subhrajit Roy, Ahmad Beirami, Prateek Mittal, Peter Henderson
2024	A*	`conf`	The Marginal Value of Momentum for Small Learning Rate SGD. ICLR Runzhe Wang, Sadhika Malladi, Tianhao Wang, Kaifeng Lyu, Zhiyuan Li
2023	J	`jnl`	A Quadratic Synchronization Rule for Distributed Deep Learning. CoRR Xinran Gu, Kaifeng Lyu, Sanjeev Arora, Jingzhao Zhang, Longbo Huang
2023	J	`jnl`	Dichotomy of Early and Late Phase Implicit Biases Can Provably Induce Grokking. CoRR Kaifeng Lyu, Jikai Jin, Zhiyuan Li, Simon S. Du, Jason D. Lee, Wei Hu
2023	J	`jnl`	DistillSpec: Improving Speculative Decoding via Knowledge Distillation. CoRR Yongchao Zhou, Kaifeng Lyu, Ankit Singh Rawat, Aditya Krishna Menon, Afshin Rostamizadeh, Sanjiv Kumar, Jean-François Kagy, Rishabh Agarwal
2023	J	`jnl`	The Marginal Value of Momentum for Small Learning Rate SGD. CoRR Runzhe Wang, Sadhika Malladi, Tianhao Wang, Kaifeng Lyu, Zhiyuan Li
2023	A*	`conf`	Understanding Incremental Learning of Gradient Descent: A Fine-grained Analysis of Matrix Sensing. ICML Jikai Jin, Zhiyuan Li, Kaifeng Lyu, Simon Shaolei Du, Jason D. Lee
2023	J	`jnl`	Understanding Incremental Learning of Gradient Descent: A Fine-grained Analysis of Matrix Sensing. CoRR Jikai Jin, Zhiyuan Li, Kaifeng Lyu, Simon S. Du, Jason D. Lee
2023	A*	`conf`	Why (and When) does Local SGD Generalize Better than SGD? ICLR Xinran Gu, Kaifeng Lyu, Longbo Huang, Sanjeev Arora
2023	J	`jnl`	Why (and When) does Local SGD Generalize Better than SGD? CoRR Xinran Gu, Kaifeng Lyu, Longbo Huang, Sanjeev Arora
2022	A*	`conf`	New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound. NeurIPS Arushi Gupta, Nikunj Saunshi, Dingli Yu, Kaifeng Lyu, Sanjeev Arora
2022	J	`jnl`	New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound. CoRR Arushi Gupta, Nikunj Saunshi, Dingli Yu, Kaifeng Lyu, Sanjeev Arora
2022	A*	`conf`	On the SDEs and Scaling Rules for Adaptive Gradient Algorithms. NeurIPS Sadhika Malladi, Kaifeng Lyu, Abhishek Panigrahi, Sanjeev Arora
2022	J	`jnl`	On the SDEs and Scaling Rules for Adaptive Gradient Algorithms. CoRR Sadhika Malladi, Kaifeng Lyu, Abhishek Panigrahi, Sanjeev Arora
2022	A*	`conf`	Understanding the Generalization Benefit of Normalization Layers: Sharpness Reduction. NeurIPS Kaifeng Lyu, Zhiyuan Li, Sanjeev Arora
2022	J	`jnl`	Understanding the Generalization Benefit of Normalization Layers: Sharpness Reduction. CoRR Kaifeng Lyu, Zhiyuan Li, Sanjeev Arora
2021	A*	`conf`	Gradient Descent on Two-layer Nets: Margin Maximization and Simplicity Bias. NeurIPS Kaifeng Lyu, Zhiyuan Li, Runzhe Wang, Sanjeev Arora
2021	J	`jnl`	Gradient Descent on Two-layer Nets: Margin Maximization and Simplicity Bias. CoRR Kaifeng Lyu, Zhiyuan Li, Runzhe Wang, Sanjeev Arora
2021	A*	`conf`	Towards Resolving the Implicit Bias of Gradient Descent for Matrix Factorization: Greedy Low-Rank Learning. ICLR Zhiyuan Li, Yuping Luo, Kaifeng Lyu
2020	A*	`conf`	Gradient Descent Maximizes the Margin of Homogeneous Neural Networks. ICLR Kaifeng Lyu, Jian Li
2020	A*	`conf`	Reconciling Modern Deep Learning with Traditional Optimization Analyses: The Intrinsic Learning Rate. NeurIPS Zhiyuan Li, Kaifeng Lyu, Sanjeev Arora
2020	J	`jnl`	Reconciling Modern Deep Learning with Traditional Optimization Analyses: The Intrinsic Learning Rate. CoRR Zhiyuan Li, Kaifeng Lyu, Sanjeev Arora
2020	J	`jnl`	Towards Resolving the Implicit Bias of Gradient Descent for Matrix Factorization: Greedy Low-Rank Learning. CoRR Zhiyuan Li, Yuping Luo, Kaifeng Lyu
2019	A*	`conf`	Fine-grained Complexity Meets IP = PSPACE. SODA Lijie Chen, Shafi Goldwasser, Kaifeng Lyu, Guy N. Rothblum, Aviad Rubinstein
2019	J	`jnl`	Gradient Descent Maximizes the Margin of Homogeneous Neural Networks. CoRR Kaifeng Lyu, Jian Li
2019	—	`conf`	Theoretical Analysis of Auto Rate-Tuning by Batch Normalization. ICLR (Poster) Sanjeev Arora, Zhiyuan Li, Kaifeng Lyu
2018	J	`jnl`	Fine-grained Complexity Meets IP = PSPACE. CoRR Lijie Chen, Shafi Goldwasser, Kaifeng Lyu, Guy N. Rothblum, Aviad Rubinstein
2018	A*	`conf`	Single-Source Bottleneck Path Algorithm Faster than Sorting for Sparse Graphs. ICALP Ran Duan, Kaifeng Lyu, Yuanhang Xie
2018	J	`jnl`	Single-Source Bottleneck Path Algorithm Faster than Sorting for Sparse Graphs. CoRR Ran Duan, Kaifeng Lyu, Hongxun Wu, Yuanhang Xie
2018	J	`jnl`	Theoretical Analysis of Auto Rate-Tuning by Batch Normalization. CoRR Sanjeev Arora, Zhiyuan Li, Kaifeng Lyu