ZL
Zhouhan Lin
cs.CLcs.AIcs.LGcs.NEcs.CVcs.IRcs.SDeess.ASstat.MLcs.MM
On Valency
published · living versionsW_j63jc552·v1 · currentpublished
StateLoop: Persistent State for Looped Transformers
with Feichen Song, Boyi Zeng, He Li, Shixiang Song +5
1 version
W_r2rvndm2·v1published
W_uf484p7y·v1published
W_q8k83kvy·v1published
Preprints & journals
93 papers in the corpus · 2015–2026Asymmetric Looped Transformers: Fixed Global Representations for an Iterative Decoder2610.28489v1 · Cheng Liang, Tengfei Zhang, Wenjie DU et al.2026 · 0 citationsarXivon Valency
P2N: Reusing Deep Representations for Greater Effective Depth2610.25076v2 · Yiqin Hao, Boyi Zeng, Zitong Wang et al.2026 · 0 citationsarXiv
Dual-Slot: From Serial Loops to Parallel Slots in Language Models2610.25086v2 · Yiqin Hao, Zitong Wang, Boyi Zeng et al.2026 · 0 citationsarXiv
Deep Fusion Memory: A Chunk-Level Parametric Memory for Large Language Models2610.25100v1 · Rubin Wei, Siyuan Sun, Hao Sun et al.2026 · 0 citationsarXivon Valency
Mixture-of-Memories: Low-Overhead Language Modeling with Modular Knowledge Injection2610.25425v1 · Yifan Lu, Rubin Wei, Ziyun Zou et al.2026 · 0 citationsarXivon Valency
HABIT-Bench: Evaluating Habit Induction from Longitudinal Weak Evidence in Agent Memory2610.24520v2 · Junming Zhang, Rubin Wei, Xu Cheng et al.2026 · 0 citationsarXiv
AF-Muon: An AdamW-Free Muon Optimizer for Tied-Embedding Models2610.01395v1 · Arash Lagzian, Paniz Halvachi, Junming Zhang et al.2026 · 0 citationsarXiv
NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction2609.10715v1 · The Intern-NCP Team: Jiaqi Cao, Chiyu Chen, Shuang Cheng et al.2026 · 0 citationsarXiv
DoPR: Reusable Compressed Document Prefixes for Efficient LLM Reranking2609.03311v1 · Beiya Dai, Yifan Wei, Guang Yang et al.2026 · 0 citationsarXiv
Intern-S2-Preview: Scientific Agentic Foundation Model2608.13505v1 · Lei Bai, Jiaqi Cao, Chiyu Chen et al.2026 · 0 citationsarXiv
Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory2607.27919v1 · Rubin Wei, Jiaqi Cao, Jiarui Wang et al.2026 · 0 citationsarXiv
MemSFT: Mitigating Alignment Tax with an External Parametric Memory2607.25614v1 · Jiarui Wang, Xiang Shi, Jiaqi Cao et al.2026 · 0 citationsarXiv
Information-Aware KV Cache Compression for Long Reasoning2606.26875v1 · Jushi Kai, Zhuiri Xiao, Alexandra Birch et al.2026 · 0 citationsarXiv
Depth-Attention: Cross-Layer Value Mixing for Language Models2606.05014v2 · Boyi Zeng, Yiqin Hao, Zitong Wang et al.2026 · 0 citationsarXiv
FlowLM: Few-Step Language Modeling via Diffusion-to-Flow Adaptation2605.20199v1 · Runzhe Zhang, Letian Chen, Wenpeng Zhang et al.2026 · 0 citationsarXiv
VQKV: High-Fidelity and High-Ratio Cache Compression via Vector-Quantization2603.16435v1 · Yixuan Wang, Qingyu Shi, Jiayu Zhou et al.2026 · 0 citationsarXiv
Early Quantization Shrinks Codebook: A Simple Fix for Diversity-Preserving Tokenization2603.17052v1 · Wenhao Zhao, Qiran Zou, Rushi Shah et al.2026 · 0 citationsarXiv
Mitigating Premature Discretization with Progressive Quantization for Robust Vector Tokenization2603.22304v1 · Wenhao Zhao, Qiran Zou, Zhouhan Lin et al.2026 · 0 citationsarXiv
CoDAR: Continuous Diffusion Language Models are More Powerful Than You Think2603.02547v1 · Junzhe Shen, Jieru Zhao, Ziwei He et al.2026 · 0 citationsarXiv
AdaPonderLM: Gated Pondering Language Models with Token-Wise Adaptive Depth2603.01914v2 · Shixiang Song, He Li, Zitong Wang et al.2026 · 0 citationsarXiv
PonderLM-3: Adaptive Token-Wise Pondering with Differentiable Masking2603.02023v2 · He Li, Feichen Song, Boyi Zeng et al.2026 · 0 citationsarXiv
Flow of Spans: Generalizing Language Models to Dynamic Span-Vocabulary via GFlowNets2602.10583v1 · Bo Xue, Yunchong Song, Fanghao Shao et al.2026 · 0 citationsarXiv
Towards Compressive and Scalable Recurrent Memory2602.11212v1 · Yunchong Song, Jushi Kai, Liming Lu et al.2026 · 0 citationsarXiv
Pretraining with Token-Level Adaptive Latent Chain-of-Thought2602.08220v2 · Boyi Zeng, Yiqin Hao, He Li et al.2026 · 0 citationsarXiv
Next Concept Prediction in Discrete Latent Space Leads to Stronger Language Models2602.08984v1 · Yuliang Liu, Yunchong Song, Yixuan Wang et al.2026 · 0 citationsarXiv
One Size Does Not Fit All: Token-Wise Adaptive Compression for KV Cache2603.04411v1 · Liming Lu, Kaixi Qiu, Jiayu Zhou et al.2026 · 0 citationsarXiv
Controlling Exploration-Exploitation in GFlowNets via Markov Chain Perspectives2602.01749v3 · Lin Chen, Samuel Drapeau, Fanghao Shao et al.2026 · 0 citationsarXiv
Context-level Language Modeling by Learning Predictive Context Embeddings2510.20280v3 · Beiya Dai, Yuliang Liu, Daozheng Xue et al.2025 · 0 citationsarXiv
Enhancing LLM Reasoning via Non-Human-Like Reasoning Path Preference Optimization2510.11104v2 · Junjie Lu, Yuliang Liu, Chaofeng Qu et al.2025 · 0 citationsarXiv
AWM: Accurate Weight-Matrix Fingerprint for Large Language Models2510.06738v3 · Boyi Zeng, Lin Chen, Ziwei He et al.2025 · 0 citationsarXiv
PonderLM-2: Pretraining LLM with Latent Thoughts in Continuous Space2509.23184v4 · Boyi Zeng, He Li, Shixiang Song et al.2025 · 0 citationsarXiv
FlowRL: Matching Reward Distributions for LLM Reasoning2509.15207v3 · Xuekai Zhu, Daixuan Cheng, Dinghuai Zhang et al.2025 · 1 citationarXiv
Training LLMs to be Better Text Embedders through Bidirectional Reconstruction2509.03020v4 · Chang Su, Dengliang Shi, Siyuan Huang et al.2025 · 0 citationsarXiv
Memory Decoder: A Pretrained, Plug-and-Play Memory for Large Language Models2508.09874v2 · Jiaqi Cao, Jiarui Wang, Rubin Wei et al.2025 · 2 citationsarXiv
Fourier Compressor: Frequency-Domain Visual Token Compression for Vision-Language Models2508.06038v3 · Huanyu Wang, Jushi Kai, Haoli Bai et al.2025 · 0 citationsarXiv
MLP Memory: A Retriever-Pretrained Memory for Large Language Models2508.01832v4 · Rubin Wei, Jiaqi Cao, Jiarui Wang et al.2025 · 1 citationarXiv
Innovator: Scientific Continued Pretraining with Fine-grained MoE Upcycling2507.18671v2 · Ning Liao, Xiaoxing Wang, Zehao Lin et al.2025 · 0 citationsarXiv
MuteSwap: Visual-informed Silent Video Identity Conversion2507.00498v3 · Yifan Liu, Yu Fang, Zhouhan Lin2025 · 0 citationsarXiv
PonderLM: Pretraining Language Models to Ponder in Continuous Space2505.20674v3 · Boyi Zeng, Shixiang Song, Siyuan Huang et al.2025 · 0 citationsarXiv
FreqKV: Key-Value Compression in Frequency Domain for Context Window Extension2505.00570v3 · Jushi Kai, Yixuan Wang, Boyi Zeng et al.2025 · 0 citationsarXiv
Alleviating LLM-based Generative Retrieval Hallucination in Alipay Search2503.21098v3 · Yedan Shen, Kaixin Wu, Yuechen Ding et al.2025 · 4 citationsSIGIR 2025
DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker Characteristics And Intelligibility2503.05223v1 · Yifan Liu, Yu Fang, Zhouhan Lin2025 · 3 citationsarXiv
AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence2502.13943v2 · Yuliang Liu, Junjie Lu, Zhaoling Chen et al.2025 · 0 citationsarXiv
Gumbel Reranking: Differentiable End-to-End Reranker Optimization2502.11116v2 · Siyuan Huang, Zhiyuan Ma, Jintao Du et al.2025 · 4 citationsarXiv
How to Synthesize Text Data without Model Collapse?2412.14689v3 · Xuekai Zhu, Daixuan Cheng, Hengli Li et al.2024 · 0 citationsarXiv
Cluster-wise Graph Transformer with Dual-granularity Kernelized Attention2410.06746v3 · Siyuan Huang, Yunchong Song, Jiayue Zhou et al.2024 · 4 citationsarXiv
Enhancing SPARQL Generation by Triplet-order-sensitive Pre-training2410.05731v1 · Chang Su, Jiexing Qi, He Yan et al.2024 · 1 citationarXiv
Training-free LLM-generated Text Detection by Mining Token Probability Sequences2410.06072v1 · Yihuai Xu, Yongwei Wang, Yifei Bi et al.2024 · 0 citationsarXiv
Leveraging Grammar Induction for Language Understanding and Generation2410.04878v1 · Jushi Kai, Shengyuan Hou, Yusheng Huang et al.2024 · 1 citationarXiv
Mirror-Consistency: Harnessing Inconsistency in Majority Voting2410.10857v2 · Siyuan Huang, Zhiyuan Ma, Jintao Du et al.2024 · 1 citationarXiv
Career total: 133 works. 93 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.