QZ
Qishi Zhan
cs.LGcs.AIcs.CLcs.CVdental careecon.GNeess.IVemergency visitsfollow up caremath.DG
On Valency
published · living versionsW_8pxfrc8y·v1 · currentpublished
ValuSpec: Plug-and-Play Candidate Valuation before Target Verification for Tree-Based Speculative Decoding
with Liang He, SiYuan Ma, Yongqi Fan, Mingyu Cao +3
1 version
Preprints & journals
22 papers in the corpus · 2024–2026ValuSpec: Plug-and-Play Candidate Valuation before Target Verification for Tree-Based Speculative Decoding2610.31338v1 · Liang He, SiYuan Ma, Qishi Zhan et al.2026 · 0 citationsarXivon Valency
Why On-Policy Distillation Sometimes Fails: Vanishing Learning Signals2610.11247v1 · Lei Zhao, Qichao Zhao, Bowen Zuo et al.2026 · 0 citationsarXiv
When Verifiable Counts Depend on Wording: Auditing Wording Robustness in Instruction Following2610.05278v1 · Qishi Zhan, Seoyeon Jang, Zihan Dong et al.2026 · 0 citationsarXiv
Behavior Pack Optimization for Video MLLM Post-Training2610.03141v1 · Zhaolu Kang, Shiyu Liu, Tailong Luo et al.2026 · 0 citationsarXiv
Nontraumatic Dental Emergency Department Visits and Follow Up Dental Care Among Wisconsin Medicaid Beneficiaries.42813467 · Zhan, Qishi, Cui, Tao, Bansal, Naveen et al.2026 · 0 citationsJournal of public health dentistry. 2026
CADWorld: Computer-Use Benchmark for Long-Horizon Computer-Aided Design2609.16251v2 · Zihan Dong, Yuanzhe Liu, Zhiyuan Ma et al.2026 · 0 citationsarXiv
How Benchmarks Mis-Score Computer-Use Agents2607.28367v1 · Zihan Dong, Zhiyuan Ma, Zekun Wang et al.2026 · 0 citationsarXiv
Why Are GUI Agents Correct but Late? Decode on the Decision-Time Critical Path, Tested with Pre-Compiled Policy Trees2607.28399v1 · Zihan Dong, Rui Qian, Qishi Zhan et al.2026 · 0 citationsarXiv
Discontinuous Prior-Mode Sections and the Geometry of Ambiguity in Intrinsic Image Decomposition2607.10321v1 · Ziheng Chen, Liangchen Liu, Qishi Zhan et al.2026 · 0 citationsarXiv
Stress Amplified Resilience: ESG and Joint Fragility in Equity Markets2606.05631v1 · Minxuan Hu, Jiayu Yi, Ziheng Chen et al.2026 · 0 citationsarXiv
BudgetDraft: Acceptance-Aware Multi-View Training for Sparse-KV Speculative Decoding2606.00144v1 · Liang He, Jingbo Wen, Qishi Zhan et al.2026 · 0 citationsarXiv
Disentangling Language Roles in Multilingual LLM Task Execution2605.27649v1 · Qishi Zhan, Minxuan Hu, Seoyeon Jang et al.2026 · 0 citationsarXiv
Relative Repairability: A Calibration-Based Diagnostic for High-Sparsity Post-Pruning Allocation2605.25508v1 · Qishi Zhan, Liang He, Minxuan Hu et al.2026 · 0 citationsarXiv
How Sparsity Allocation Shapes Label-Free Post-Pruning Recoverability2605.21972v1 · Qishi Zhan, Minxuan Hu, Liang He2026 · 0 citationsarXiv
Adaptive Signal Resuscitation: Channel-wise Post-Pruning Repair for Sparse Vision Networks2605.21426v1 · Qishi Zhan, Ziheng Chen, Minxuan Hu2026 · 0 citationsarXiv
Dual-Pathway Circuits of Object Hallucination in Vision-Language Models2605.13156v2 · Jiaxin Liu, Ding Zhong, Yue Wang et al.2026 · 0 citationsarXiv
Unstable Rankings in Bayesian Deep Learning Evaluation2604.23102v1 · Qishi Zhan, Minxuan Hu, Guansu Wang et al.2026 · 0 citationsarXiv
A Tale of Two Variances: When Single-Seed Benchmarks Fail in Bayesian Deep Learning2604.23114v1 · Qishi Zhan, Minxuan Hu, Liang He et al.2026 · 0 citationsarXiv
Mapping Drivers of Greenness: Spatial Variable Selection for MODIS Vegetation Indices2602.07681v2 · Qishi Zhan, Cheng-Han Yu, Yuchi Chen et al.2026 · 0 citationsarXiv
How Order-Sensitive Are LLMs? OrderProbe for Deterministic Structural Reconstruction2601.08626v6 · Zhaolu Kang, Yingjie He, Kehan Jiang et al.2026 · 0 citationsarXiv
Research on Optimization of Natural Language Processing Model Based on Multimodal Deep Learning2406.08838v1 · Dan Sun, Yaxin Liang, Yining Yang et al.2024 · 15 citationsarXiv
Advancements in Feature Extraction Recognition of Medical Imaging Systems Through Deep Learning Technique2406.18549v1 · Qishi Zhan, Dan Sun, Erdi Gao et al.2024 · 11 citationsarXiv
Career total: 27 works. 22 are in this corpus.
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.