CX
Chaojun Xiao
cs.CLcs.AIcs.LGcs.IRcs.CVcs.CYcold startcross-domain transfercs.SEknowledge transfer
On Valency
published · living versionsW_2btmrzkn·v1 · currentpublished
Don't Stop Long-Context Continual Pre-Training Too Early: An Empirical Scaling Law
with Ziqing Qiao, Xiaoyue Xu, Sikui Zhang, Yinuo Xu +2
1 version
W_evqrutjp·v1published
W_fp35asej·v1published
Preprints & journals
68 papers in the corpus · 2018–2026Don't Stop Long-Context Continual Pre-Training Too Early: An Empirical Scaling Law2610.29352v1 · Ziqing Qiao, Xiaoyue Xu, Sikui Zhang et al.2026 · 0 citationsarXivon Valency
Cognitively Decoupled Meta-Induction for Out-of-Distribution Generalization2610.25749v1 · Yuqian Wang, Yuanzheng Wang, Chaojun Xiao et al.2026 · 0 citationsarXivon Valency
Competing Learning in Hybrid Architectures2610.26830v1 · Xingyu Shen, Yingfa Chen, Ziqing Qiao et al.2026 · 0 citationsarXivon Valency
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning2610.22194v1 · Bingxiang He, Pengqi Sun, Shihao Peng et al.2026 · 0 citationsarXiv
Latent Compaction with Test-Time Training2610.20582v1 · Yingfa Chen, Xingyu Shen, Chaojun Xiao et al.2026 · 0 citationsarXiv
Empowering Edge Agents: A Systematic Analysis of Edge-Cloud Collaboration in LLM Agent Execution2610.17816v2 · Pengcheng Huang, Zhenghao Liu, Shixin Ren et al.2026 · 0 citationsarXiv
UltraData-Code: From Raw Repositories to Algorithmically Dense and Task-Oriented Code Data2610.23960v1 · Chengying Tu, Hengyu Zhao, Shuaikang Xue et al.2026 · 0 citationsarXiv
Diffusion Reward Models2609.33803v2 · Xiangyang Wang, Bingxiang He, Zeyuan Liu et al.2026 · 0 citationsarXiv
A*-Thought-V2: Efficient Latent Reasoning via Geometric Dynamics of LLM2609.07821v1 · Xiaoang Xu, Siyuan Liu, Shuo Wang et al.2026 · 0 citationsarXiv
Rethinking On-Policy Distillation of Large Language Models II: One Training Example2609.04172v1 · Zixuan Fu, Bingxiang He, Yuxin Zuo et al.2026 · 0 citationsarXiv
StudyBench: Can Self-Evolution Squeeze Textbooks for Olympiad Capability?2609.00787v2 · Yinghao Chen, Zixi Chen, Bingxiang He et al.2026 · 0 citationsarXiv
Beyond Teacher Likelihood: Group-Calibrated On-Policy Distillation for Long-Context Reasoning2608.19181v1 · Zhu Zhang, Jixun Wang, Xiaoang Xu et al.2026 · 0 citationsarXiv
PACE-Bench: Benchmarking Physics Adaptation via Code Evolution in Dynamic Environments2608.14441v1 · Yuhao Zhan, Bingxiang He, Zecong Tang et al.2026 · 0 citationsarXiv
REFACT: Adaptive Fact Restatement for Compact and Faithful Chain-of-Thought Reasoning2607.20833v2 · Zhensheng Jin, Xin Dai, Zhenghao Liu et al.2026 · 0 citationsarXiv
Beyond Reward Engineering: A Data Recipe for Long-Context Reinforcement Learning2606.18831v2 · Xiaoyue Xu, Sikui Zhang, Xiaorong Wang et al.2026 · 0 citationsarXiv
Rethinking the Role of Efficient Attention in Hybrid Architectures2606.15378v1 · Ziqing Qiao, Yinuo Xu, Chaojun Xiao et al.2026 · 0 citationsarXiv
DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices2605.10933v3 · Chenyang Song, Weilin Zhao, Xu Han et al.2026 · 0 citationsarXiv
MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction2604.27393v1 · Junbo Cui, Bokai Xu, Chongyi Wang et al.2026 · 1 citationarXiv
Beyond SFT-to-RL: Pre-alignment via Black-Box On-Policy Distillation for Multimodal RL2604.28123v3 · Sudong Wang, Weiquan Huang, Xiaomin Yu et al.2026 · 0 citationsarXiv
Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe2604.13016v2 · Yaxuan Li, Yuxin Zuo, Bingxiang He et al.2026 · 0 citationsarXiv
Student-in-the-Loop Chain-of-Thought Distillation via Generation-Time Selection2604.02819v1 · Chaoqun He, Yingfa Chen, Chaojun Xiao et al.2026 · 0 citationsarXiv
KARL: Mitigating Hallucinations in LLMs via Knowledge-Boundary-Aware Reinforcement Learning2604.22779v1 · Cheng Gao, Cheng Huang, Kangyang Luo et al.2026 · 0 citationsarXiv
MiniCPM-SALA: Hybridizing Sparse and Linear Attention for Efficient Long-Context Modeling2602.11761v2 · MiniCPM Team: Wenhao An, Yingfa Chen, Yewei Fang et al.2026 · 1 citationarXiv
Data Science and Technology Towards AGI Part I: Tiered Data Management2602.09003v1 · Yudong Wang, Zixuan Fu, Hengyu Zhao et al.2026 · 0 citationsarXiv
APB-V: Accelerating Long-Video Understanding via Sequence-Parallelism-aware Approximate Attention2601.21444v2 · Yuxiang Huang, Mingye Li, Xu Han et al.2026 · 0 citationsarXiv
Hybrid Linear Attention Done Right: Efficient Distillation and Effective Architectures for Extremely Long Contexts2601.22156v1 · Yingfa Chen, Zhen Leng Thai, Zihan Zhou et al.2026 · 0 citationsarXiv
Revealing the Attention Floating Mechanism in Masked Diffusion Models2601.07894v1 · Xin Dai, Pengcheng Huang, Zhenghao Liu et al.2026 · 0 citationsarXiv
Large Language Models' Complicit Responses to Illicit Instructions across Socio-Legal Contexts2511.20736v1 · Xing Wang, Huiyuan Xie, Yiyan Wang et al.2025 · 0 citationsarXiv
InfLLM-V2: Dense-Sparse Switchable Attention for Seamless Short-to-Long Adaptation2509.24663v1 · Weilin Zhao, Zihan Zhou, Zhou Su et al.2025 · 0 citationsarXiv
BlockFFN: Towards End-Side Acceleration-Friendly Mixture-of-Experts with Chunk-Level Activation Sparsity2507.08771v2 · Chenyang Song, Weilin Zhao, Xu Han et al.2025 · 0 citationsarXiv
Ultra-FineWeb: Efficient Data Filtering and Verification for High-Quality LLM Training Data2505.05427v1 · Yudong Wang, Zixuan Fu, Jie Cai et al.2025 · 0 citationsarXiv
APB: Accelerating Distributed Long-Context Inference by Passing Compressed Context Blocks across GPUs2502.12085v2 · Yuxiang Huang, Mingye Li, Xu Han et al.2025 · 0 citationsarXiv
Densing Law of LLMs2412.04315v2 · Chaojun Xiao, Jie Cai, Weilin Zhao et al.2024 · 2 citationsarXiv
Sparsing Law: Towards Large Language Models with Greater Activation Sparsity2411.02335v4 · Yuqi Luo, Chenyang Song, Xu Han et al.2024 · 1 citationarXiv
Enhancing Legal Case Retrieval via Scaling High-quality Synthetic Query-Candidate Pairs2410.06581v1 · Cheng Gao, Chaojun Xiao, Zhenghao Liu et al.2024 · 10 citationsarXiv
Exploring the Benefit of Activation Sparsity in Pre-training2410.03440v1 · Zhengyan Zhang, Chaojun Xiao, Qiujieli Qin et al.2024 · 0 citationsarXiv
Locret: Enhancing Eviction in Long-Context LLM Inference with Trained Retaining Heads on Consumer-Grade Devices2410.01805v2 · Yuxiang Huang, Binhang Yuan, Xu Han et al.2024 · 0 citationsarXiv
From MOOC to MAIC: Reshaping Online Teaching and Learning through LLM-driven Agents2409.03512v1 · Jifan Yu, Zheyuan Zhang, Daniel Zhang-li et al.2024 · 7 citationsarXiv
Configurable Foundation Models: Building LLMs from a Modular Perspective2409.02877v1 · Chaojun Xiao, Zhengyan Zhang, Chenyang Song et al.2024 · 5 citationsarXiv
The Elephant in the Room: Rethinking the Usage of Pre-trained Language Model in Sequential Recommendation2404.08796v3 · Zekai Qu, Ruobing Xie, Chaojun Xiao et al.2024 · 10 citationsarXiv
Robust and Scalable Model Editing for Large Language Models2403.17431v1 · Yingfa Chen, Zhengyan Zhang, Xu Han et al.2024 · 2 citationsarXiv
ThinkNote: Enhancing Knowledge Integration and Utilization of Large Language Models via Constructivist Cognition Modeling2402.13547v4 · Zhipeng Xu, Zhenghao Liu, Yukun Yan et al.2024 · 2 citationsarXiv
Ouroboros: Generating Longer Drafts Phrase by Phrase for Faster Speculative Decoding2402.13720v3 · Weilin Zhao, Yuxiang Huang, Xu Han et al.2024 · 8 citationsarXiv
InfLLM: Training-Free Long-Context Extrapolation for LLMs with an Efficient Context Memory2402.04617v2 · Chaojun Xiao, Pengle Zhang, Xu Han et al.2024 · 14 citationsarXiv
ReLU Wins: Discovering Efficient Activation Functions for Sparse LLMs2402.03804v1 · Zhengyan Zhang, Yixin Song, Guanghui Yu et al.2024 · 3 citationsarXiv
MUSER: A Multi-View Similar Case Retrieval Dataset2310.15602v1 · Qingquan Li, Yiran Hu, Feng Yao et al.2023 · 10 citationsCIKM 2023
Variator: Accelerating Pre-trained Models with Plug-and-Play Compression Modules2310.15724v2 · Chaojun Xiao, Yuqi Luo, Wenbin Zhang et al.2023 · 1 citationarXiv
Thoroughly Modeling Multi-domain Pre-trained Recommendation as Language2310.13540v3 · Zekai Qu, Ruobing Xie, Chaojun Xiao et al.2023 · 6 citationsarXiv
Plug-and-Play Document Modules for Pre-trained Models2305.17660v1 · Chaojun Xiao, Zhengyan Zhang, Xu Han et al.2023 · 5 citationsarXiv
Plug-and-Play Knowledge Injection for Pre-trained Language Models2305.17691v2 · Zhengyan Zhang, Zhiyuan Zeng, Yankai Lin et al.2023 · 11 citationsarXiv
Career total: 85 works. 68 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.