Scholar
Xuyang Liu
Google Scholar ID: 9VhMC1QAAAAJ
Sichuan University
Vision-language Models
Model Compression
Token Compression
Transfer Learning
Follow
Homepage
↗
Google Scholar
↗
Citations & Impact
All-time
Citations
209
H-index
10
i10-index
10
Publications
18
Co-authors
3
list available
Contact
Email
liuxuyang@stu.scu.edu.cn
Twitter
Open ↗
GitHub
Open ↗
Publications
26 items
AudioKV: KV Cache Eviction in Efficient Large Audio Language Models
2026
Cited
0
STDec: Spatio-Temporal Stability Guided Decoding for dLLMs
2026
Cited
0
V-CAST: Video Curvature-Aware Spatio-Temporal Pruning for Efficient Video Large Language Models
2026
Cited
0
Bridging Visual Representation and Reinforcement Learning from Verifiable Rewards in Large Vision-Language Models
2026
Cited
0
Flash-Unified: A Training-Free and Task-Aware Acceleration Framework for Native Unified Models
2026
Cited
0
OmniSIFT: Modality-Asymmetric Token Compression for Efficient Omni-modal Large Language Models
2026
Cited
0
Structure-based RNA Design by Step-wise Optimization of Latent Diffusion Model
2026
Cited
0
IPCV: Information-Preserving Compression for MLLM Visual Encoders
2025
Cited
0
Load more
Resume (English only)
Academic Achievements
Released MixKV (Oct 2025): a plug-and-play framework enhancing KV compression across LVLMs
Published work advocating shift from model-centric to data-centric AI efficiency, ranked #2 Paper of the Day (May 2025)
First-author paper M2IST accepted by IEEE TCSVT (Mar 2025)
Two papers (ToCa and AutoGnothi) accepted by ICLR 2025 (Feb 2025)
Released GlobalCom² (Jan 2025): training-free acceleration for high-resolution LVLMs via dynamic tiling
Co-first author paper V-PETL Bench accepted by NeurIPS 2024 (Sep 2024)
Multiple conference publications at ICLR 2025, NeurIPS D&B Track 2024, ICASSP 2024 (Oral), ICME 2024, etc.
Journal paper M2IST published in IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)
Research Experience
Research Intern at OPPO Research Institute, supervised by Prof. Lei Zhang (PolyU HK, IEEE Fellow)
Interned at Ant Security Lab, working on GUI Agent
Worked at Taobao & Tmall Group on Efficient MLLM
Visited MiLAB at Westlake University for half a year under the supervision of Prof. Donglin Wang
Collaborates closely with Dr. Siteng Huang (DAMO Academy) and Prof. Linfeng Zhang (SJTU)
Co-authors
3 total
Siteng Huang
Alibaba DAMO Academy | ZJU | Westlake University
Linfeng Zhang (张林峰)
Shanghai Jiao Tong University
Donglin Wang
Westlake University
×
Welcome back
Sign in to Agora
Welcome back! Please sign in to continue.
Email address
Password
Forgot password?
Continue
Do not have an account?
Sign up