ZL
Zechun Liu
cs.CVcs.LGcs.AIcs.CLeess.IVcs.ROcs.ARmath.OCcs.DCcs.GR
On Valency
published · living versionsW_jwa95vfk·v1 · currentpublished
HybridSpec: Throughput-Aware Speculative Decoding for Hybrid GDN-MoE Models
with Haiyu Wang, Sai Qian Zhang, Steven Li
1 version
W_macrk6t4·v1published
Preprints & journals
62 papers in the corpus · 2018–2026Mid-Training Language Models on Raw Video2610.11019v1 · Jaedong Hwang, Xiaoqian Shen, Ernie Chang et al.2026 · 0 citationsarXiv
Quantized Reasoning Models Think They Need to Think Longer, but They Do Not2606.00206v1 · Sanae Lotfi, Polina Kirichenko, Steven Li et al.2026 · 0 citationsarXiv
VLM3: Vision Language Models Are Native 3D Learners2605.30561v1 · Zhipeng Cai, Zhuang Liu, Yunyang Xiong et al.2026 · 0 citationsarXiv
MobileMoE: Scaling On-Device Mixture of Experts2605.27358v1 · Yanbei Chen, Hanxian Huang, Ernie Chang et al.2026 · 0 citationsarXiv
WinQ: Accelerating Quantization-Aware Training of Language Models Around Saddle Points2605.17471v1 · Dongyue Li, Zechun Liu, Kai Yi et al.2026 · 0 citationsarXiv
Small Vision-Language Models are Smart Compressors for Long Video Understanding2604.08120v1 · Junjie Fei, Jun Chen, Zechun Liu et al.2026 · 0 citationsarXiv
Neural Computers2604.06425v2 · Mingchen Zhuge, Changsheng Zhao, Haozhe Liu et al.2026 · 0 citationsarXiv
dTRPO: Trajectory Reduction in Policy Optimization of Diffusion Large Language Models2603.18806v2 · Wenxuan Zhang, Lemeng Wu, Changsheng Zhao et al.2026 · 0 citationsarXiv
MobileLLM-Flash: Latency-Guided On-Device LLM Design for Industry Scale Deployment2603.15954v2 · Hanxian Huang, Igor Fedorov, Andrey Gromov et al.2026 · 0 citationsarXiv
STEM: Scaling Transformers with Embedding Modules2601.10639v1 · Ranajoy Sadhukhan, Sheng Cao, Harry Dong et al.2026 · 0 citationsarXiv
VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice2601.05175v2 · Shuming Liu, Mingchen Zhuge, Changsheng Zhao et al.2026 · 0 citationsarXiv
The Path Not Taken: RLVR Provably Learns Off the Principals2511.08567v1 · Hanqing Zhu, Zhenyu Zhang, Hanxian Huang et al.2025 · 0 citationsarXiv
RDD: Retrieval-Based Demonstration Decomposer for Planner Alignment in Long-Horizon Tasks2510.14968v1 · Mingxuan Yan, Yuping Wang, Zechun Liu et al.2025 · 0 citationsarXiv
MobileLLM-R1: Exploring the Limits of Sub-Billion Language Model Reasoners with Open Training Recipes2509.24945v3 · Changsheng Zhao, Ernie Chang, Zechun Liu et al.2025 · 0 citationsarXiv
R-Sparse: Rank-Aware Activation Sparsity for Efficient LLM Inference2504.19449v1 · Zhenyu Zhang, Zechun Liu, Yuandong Tian et al.2025 · 1 citationarXiv
Param for Direct Weight Mixing: Post-Train Large Language Model at Zero Cost2504.21023v1 · Sheng Cao, Mingrui Wu, Karthik Prasad et al.2025 · 0 citationsICLR 2025
PARQ: Piecewise-Affine Regularized Quantization2503.15748v1 · Lisa Jin, Jianhao Ma, Zechun Liu et al.2025 · 0 citationsarXiv
ParetoQ: Improving Scaling Laws in Extremely Low-bit LLM Quantization2502.02631v2 · Zechun Liu, Changsheng Zhao, Hanxian Huang et al.2025 · 3 citationsarXiv
Efficient Track Anything2411.18933v1 · Yunyang Xiong, Chong Zhou, Xiaoyu Xiang et al.2024 · 5 citationsarXiv
Llama Guard 3-1B-INT4: Compact and Efficient Safeguard for Human-AI Conversations2411.17713v1 · Igor Fedorov, Kate Plawiak, Lemeng Wu et al.2024 · 0 citationsarXiv
LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding2410.17434v1 · Xiaoqian Shen, Yunyang Xiong, Changsheng Zhao et al.2024 · 0 citationsarXiv
Agent-as-a-Judge: Evaluate Agents with Agents2410.10934v2 · Mingchen Zhuge, Changsheng Zhao, Dylan Ashley et al.2024 · 5 citationsarXiv
Scaling Parameter-Constrained Language Models with Quality Data2410.03083v1 · Ernie Chang, Matteo Paltenghi, Yang Li et al.2024 · 0 citationsarXiv
Target-Aware Language Modeling via Granular Data Sampling2409.14705v1 · Ernie Chang, Pin-Jie Lin, Yang Li et al.2024 · 0 citationsarXiv
RoLoRA: Fine-tuning Rotated Outlier-free LLMs for Effective Weight-Activation Quantization2407.08044v2 · Xijie Huang, Zechun Liu, Shih-Yang Liu et al.2024 · 7 citationsarXiv
Joint stereo 3D object detection and implicit surface reconstruction.38886528 · Li, Shichao, Huang, Xijie, Liu, Zechun et al.2024 · 4 citationsScientific reports. 2024;14(1):13893
An Introduction to Vision-Language Modeling2405.17247v1 · Florian Bordes, Richard Yuanzhe Pang, Anurag Ajay et al.2024 · 47 citationsarXiv
SpinQuant: LLM quantization with learned rotations2405.16406v4 · Zechun Liu, Changsheng Zhao, Igor Fedorov et al.2024 · 13 citationsarXiv
MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases2402.14905v2 · Zechun Liu, Changsheng Zhao, Forrest Iandola et al.2024 · 16 citationsarXiv
On The Open Prompt Challenge In Conditional Audio Generation2311.00897v1 · Ernie Chang, Sidd Srinivasan, Mahi Luthra et al.2023 · 3 citationsarXiv
LLM-FP4: 4-Bit Floating-Point Quantized Transformers2310.16836v1 · Shih-yang Liu, Zechun Liu, Xijie Huang et al.2023 · 50 citationsProceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
MiniGPT-v2: large language model as a unified interface for vision-language multi-task learning2310.09478v3 · Jun Chen, Deyao Zhu, Xiaoqian Shen et al.2023 · 67 citationsarXiv
Efficient and Robust Quantization-aware Training via Adaptive Coreset Selection2306.07215v3 · Xijie Huang, Zechun Liu, Shih-Yang Liu et al.2023 · 8 citationsarXiv
Mixture-of-Supernets: Improving Weight-Sharing Supernet Training with Architecture-Routed Mixture-of-Experts2306.04845v2 · Ganesh Jawahar, Haichuan Yang, Yunyang Xiong et al.2023 · 2 citationsarXiv
Binary and Ternary Natural Language Generation2306.01841v1 · Zechun Liu, Barlas Oguz, Aasish Pappu et al.2023 · 6 citationsarXiv
LLM-QAT: Data-Free Quantization Aware Training for Large Language Models2305.17888v1 · Zechun Liu, Barlas Oguz, Changsheng Zhao et al.2023 · 146 citationsarXiv
SCALES: Boost Binary Neural Network for Image Super-Resolution with Efficient Scalings2303.12270v2 · Renjie Wei, Zechun Liu, Yuchen Fan et al.2023 · 5 citationsarXiv
Oscillation-free Quantization for Low-bit Vision Transformers2302.02210v3 · Shih-Yang Liu, Zechun Liu, Kwang-Ting Cheng2023 · 8 citationsICML 2023
SDQ: Stochastic Differentiable Quantization with Mixed Precision2206.04459v3 · Xijie Huang, Zhiqiang Shen, Shichao Li et al.2022 · 24 citationsarXiv
BiT: Robustly Binarized Multi-distilled Transformer2205.13016v2 · Zechun Liu, Barlas Oguz, Aasish Pappu et al.2022 · 38 citationsarXiv
Stereo Neural Vernier Caliper2203.11018v2 · Shichao Li, Zechun Liu, Zhiqiang Shen et al.2022 · 6 citationsarXiv
Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space2201.00814v2 · Arnav Chavan, Zhiqiang Shen, Zhuang Liu et al.2022 · 72 citationsarXiv
Data-Free Neural Architecture Search via Recursive Label Calibration2112.02086v2 · Zechun Liu, Zhiqiang Shen, Yun Long et al.2021 · 6 citationsarXiv
Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimation2111.14826v2 · Zechun Liu, Kwang-Ting Cheng, Dong Huang et al.2021 · 117 citationsarXiv
Sliced Recursive Transformer2111.05297v3 · Zhiqiang Shen, Zechun Liu, Eric Xing2021 · 19 citationsarXiv
How Do Adam and Training Strategies Help BNNs Optimization?2106.11309v1 · Zechun Liu, Zhiqiang Shen, Shichao Li et al.2021 · 53 citationsarXiv
"BNN - BN = ?": Training Binary Neural Networks without Batch Normalization2104.08215v1 · Tianlong Chen, Zhenyu Zhang, Xu Ouyang et al.2021 · 45 citationsarXiv
Is Label Smoothing Truly Incompatible with Knowledge Distillation: An Empirical Study2104.00676v1 · Zhiqiang Shen, Zechun Liu, Dejia Xu et al.2021 · 38 citationsarXiv
S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-bit Neural Networks via Guided Distribution Calibration2102.08946v2 · Zhiqiang Shen, Zechun Liu, Jie Qin et al.2021 · 21 citationsarXiv
Partial Is Better Than All: Revisiting Fine-tuning Strategy for Few-shot Learning2102.03983v1 · Zhiqiang Shen, Zechun Liu, Jie Qin et al.2021 · 162 citationsarXiv
Career total: 83 works. 62 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.