Publications

Arctic-Text2SQL-R1: Simple Rewards, Strong Reasoning in Text-to-SQL.

Z. Yao, G. Sun, Ł. Borchmann, Zheyu Shen, M. Deng, B. Zhai, H. Zhang, A. Li, Y. He.

ACL Findings 2026

Position: Invisible Tokens, Visible Bills: The Urgent Need to Audit Hidden Operations in Opaque LLM Services.

G. Sun, Z. Wang, X. Zhao, B. Tian, Zheyu Shen, Y. He, J. Xing, A. Li.

ICML 2026 Position Paper Track

Vision-Language-Action in Robotics: A Survey of Datasets, Benchmarks, and Data Engines.

Z. Wang, B. Wang, H. Zhang, T. Du, T. Chen, G. Sun, Y. He, Zheyu Shen, W. Ye, A. Li.

Transactions on Machine Learning Research (TMLR), 2026

Uncovering the Redundancy in Transformers via a Unified Study of Layer Dropping.

S. He, G. Sun, Zheyu Shen, A. Li.

Transactions on Machine Learning Research (TMLR), 2026

HeteroVDiff: Pipelining Video Diffusion across Mobile NPU and GPU with Shared Weights.

Zheyu Shen, Z. Wang, Y. He, G. Sun, Y. Zhang, Y. Wang, W. Ye, A. Li.

arXiv preprint arXiv:2610.16244 (2026)

ARC-KV: Amortizing Anchor Search for Reconstruction-Based KV Cache Compaction.

Zheyu Shen, G. Wang, D. Tu, M. Zhang, Y. Li, A. Aziz, C. Tang, A. Li.

arXiv preprint arXiv:2609.36835 (2026)

TacGen: Touch Is a Necessary Dimension of Physical-World Representation – Addressing Tactile Data Scarcity with Scalable Vision-to-Touch Alignment and Generation.

W. Ye, A. Das, S. Chen, Y. Wang, B. Tian, G. Sun, S. He, Zheyu Shen, Z. Wang, Y. He, and others.

arXiv preprint arXiv:2606.29173 (2026)

Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?

G. Sun, K. Feng, S. He, X. Gong, Y. He, Z. Wang, Zheyu Shen, W. Ye, R. R. Kompella, G. Liu, A. Li.

arXiv preprint arXiv:2606.27755 (2026)

Cognibit: From Digital Exhaustion to Real-World Connection Through Gamified Territory Control and LLM-Powered Twin Networking.

W. Ye, S. Chen, Y. Wang, S. He, B. Tian, G. Sun, Z. Wang, Z. Wang, Y. He, Zheyu Shen, and others.

arXiv preprint arXiv:2604.04351 (2026)

ROCKET: Residual-Oriented Multi-Layer Alignment for Spatially-Aware Vision-Language-Action Models.

G. Sun, T. Du, K. Feng, C. Luo, X. Ding, Zheyu Shen, Z. Wang, Y. He, A. Li.

arXiv preprint arXiv:2602.17951 (2026)

FedMOA: Federated GRPO for Personalized Reasoning LLMs under Heterogeneous Rewards.

Z. Wang, D. Jung, Y. He, G. Sun, Zheyu Shen, M. Lee, A. Li.

arXiv preprint arXiv:2602.00453 (2026)

SymRTLO: Enhancing RTL Code Optimization with LLMs and Neuron-Inspired Symbolic Reasoning.

Y. Wang, W. Ye, P. Guo, Y. He, Z. Wang, B. Tian, S. He, G. Sun, Zheyu Shen, S. Chen, and others.

NeurIPS 2025

Bag of Tricks for Sparse Mixture-of-Experts: A Benchmark Across Reasoning, Efficiency, and Safety.

M. Qiu^*, Zheyu Shen^*, P. Li, A. Li, T. Chen.

EMNLP Findings 2025

EdgeLoRA: An Efficient Multi-Tenant LLM Serving System on Edge Devices.

Zheyu Shen, Y. He, Z. Wang, Y. Zhang, G. Sun, W. Ye, A. Li.

MobiSys 2025

Predictive Auditing of Hidden Tokens in LLM APIs via Reasoning Length Estimation.

Z. Wang, G. Sun, Y. He, Zheyu Shen, B. Tian, A. Li.

arXiv preprint arXiv:2508.00912 (2025)

CogniPair: From LLM Chatbots to Conscious AI Agents – GNWT-Based Multi-Agent Digital Twins for Social Pairing – Dating & Hiring Applications.

W. Ye, S. Chen, Y. Wang, S. He, B. Tian, G. Sun, Z. Wang, Z. Wang, Y. He, Zheyu Shen, and others.

arXiv preprint arXiv:2506.03543 (2025)

CoIn: Counting the Invisible Reasoning Tokens in Commercial Opaque LLM APIs.

G. Sun, Z. Wang, B. Tian, M. Liu, Zheyu Shen, S. He, Y. He, W. Ye, Y. Wang, A. Li.

arXiv preprint arXiv:2505.13778 (2025)

Prada: Black-Box LLM Adaptation with Private Data on Resource-Constrained Devices.

Z. Wang, Y. He, Zheyu Shen, Y. Li, G. Sun, M. Lee, A. Li.

arXiv preprint arXiv:2503.14932 (2025)

Flora: Federated Fine-Tuning Large Language Models with Heterogeneous Low-Rank Adaptations.

Z. Wang, Zheyu Shen, Y. He, G. Sun, H. Wang, L. Lyu, A. Li.

NeurIPS 2024

SHED: Shapley-Based Automated Dataset Refinement for Instruction Fine-Tuning.

Y. He, Z. Wang, Zheyu Shen, G. Sun, Y. Dai, Y. Wu, H. Wang, A. Li.

NeurIPS 2024

Revisiting Federated Fine-Tuning: A Single Communication Round is Enough for Foundation Models.

Z. Wang, B. Tian, Y. He, Zheyu Shen, G. Sun, Y. Liu, L. Liu, M. Liu, A. Li.

arXiv preprint arXiv:2412.04650 (2024)

ShareLoRA: Less Tuning, More Performance for LoRA Fine-tuning of LLMs.

Zheyu Shen, G. Sun, Y. He, Z. Wang, Y. Zhang, S. Kundu, E. P. Xing, H. Wang, A. Li.

Preprint (OpenReview, 2024)

Domino: Eliminating Communication in LLM Training via Generic Tensor Slicing and Overlapping.

G. Wang, C. Zhang, Zheyu Shen, A. Li, O. Ruwase.

arXiv preprint arXiv:2409.15241 (2024)