Get started
Author

Zunhai Su

Observed name: Su, Zunhai
ORCID: 0009-0006-2011-2044
Observed ORCID: 0009-0006-2011-2044
OpenAlex ID: A5122136707
Locate, steer, and improve: A practical survey of actionable mechanistic interpretability in large language models
2026 · Hengyuan Zhang, Zhihao Zhang, et al. · Computer Science Review · 2
XStreamVGGT : Extremely Memory‐Efficient Streaming Vision Geometry Grounded Transformer With KV Cache Compression
2026 · Zunhai Su, Weihao Ye, et al. · Journal of the Society for Information Display · 1
XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
2026 · Su, Zunhai, Ye, Weihao, et al. · arXiv (Cornell University) · 1
PDF
Training–Rollout Mismatch in Low-Precision LLM Reinforcement Learning: A Survey
2026 · Zunhai Su, Wei Wu, et al. · Preprints.org
DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems
2026 · Zhuang, Xialie, Lei, Mingcong, et al. · arXiv (Cornell University)
PDF
ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models
2026 · Sun, Bohan, Su, Zunhai, et al. · arXiv (Cornell University)
PDF
StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions
2026 · Zhuang, Xialie, Su, Zunhai, et al. · arXiv (Cornell University)
PDF
QuantMLA: Function-Aligned Dual-Path Quantization for Low-Bit MLA KV Caching
2026 · Su, Zunhai, Sun, Yuxuan, et al. · arXiv (Cornell University)
PDF
Efficient Reasoning Exploration via State-Conditioned Latent Steering with Progress Guidance
2026 · Zhang, Hengyuan, Shang, Chenming, et al. · arXiv (Cornell University)
PDF
DAMP: Decay-Aware Mixed-Precision Recurrent-State Quantization
2026 · Zhang, Tao, Tan, Jianchao, et al. · arXiv (Cornell University)
PDF
Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaus
2026 · Su, Zunhai, Sun, Bohan, et al. · arXiv (Cornell University)
PDF
UniMoMo: Expert Merging-Based MoE Acceleration for Large Recommendation Models
2026 · Xin, Lei, Gu, Bin, et al. · arXiv (Cornell University)
PDF
SDFP: Speculative Decoding with FIT-Pruned Models for Training-Free and Plug-and-Play LLM Acceleration
2026 · Hanyu Wei, Zunhai Su, et al. · Lecture notes in computer science
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models
2026 · Association for Computational Linguistics 2026, Ananiadou, Sophia, et al. · Underline Science Inc.
P‐52: Distinguished Student & Invited Poster Paper: XStreamVGGT: Extremely Memory‐Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
2026 · Zunhai Su, Weihao Ye, et al. · SID Symposium Digest of Technical Papers
PDF
OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond
2026 · Su, Zunhai, Yang, Rui, et al. · arXiv (Cornell University)
PDF
CktFormalizer: Autoformalization of Natural Language into Circuit Representations
2026 · Xiong, Jing, Han, Qi, et al. · arXiv (Cornell University)
PDF
Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation
2026 · Su, Zunhai, Zhang, Hengyuan, et al. · arXiv (Cornell University)
PDF
XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression
2026 · Su, Zunhai, Ye, Weihao, et al. · arXiv (Cornell University)
PDF
SnapMLA: Efficient Long-Context MLA Decoding via Hardware-Aware FP8 Quantized Pipelining
2026 · Zhang, Yifan, Su, Zunhai, et al. · arXiv (Cornell University)
PDF
SDFP: Speculative Decoding with FIT-Pruned Models for Training-Free and Plug-and-Play LLM Acceleration
2026 · Wei, Hanyu, Su, Zunhai, et al. · arXiv (Cornell University)
PDF
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models
2026 · Zhang, Hengyuan, Zhang, Zhihao, et al. · arXiv (Cornell University)
PDF
Beyond Outliers: A Data-Free Layer-wise Mixed-Precision Quantization Approach Driven by Numerical and Structural Dual-Sensitivity
2026 · Hengyuan Zhang, Xinrong Chen, et al. · Conference on Empirical Methods in Natural Language Processing (EMNLP)
PDF
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models
2026 · Hengyuan Zhang, Zhihao Zhang, et al. · Findings of the Association for Computational Linguistics: ACL (Findings of ACL)
PDF
Exploring Layer-wise Information Effectiveness for Post-Training Quantization in Small Language Models
2026 · He Xiao, Qingyao Yang, et al. · Findings of the Association for Computational Linguistics: ACL (Findings of ACL)
PDF
Works count: 28
Citation count: 4
H-index: 1
I10-index:
Year
Topic
Type