Publications

1. OxyGen: Unified KV Cache Management for VLA Inference under Multi-Task Parallelism

Published in Conference on Neural Information Processing Systems (NeurIPS), 2026 | Download Paper

2. Learning to Commit: Generating Organic Pull Requests via Online Repository Memory

Published in Conference on Neural Information Processing Systems (NeurIPS), 2026 | Download Paper

3. AtomWorld-Mem: Memory-Restored World States for Long-Horizon Atomistic Evolution

Published in Conference on Neural Information Processing Systems (NeurIPS), 2026 | Download Paper

4. SPIKe: Generating Practical SpMV Kernels for Sparse Iterative Methods on GPUs

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC), 2026 | Download Paper

5. MakoXC: Rearchitecting DFT Exchange-Correlation with Matrix-Aligned and Knowledge-Organized Sparsity

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC), 2026 | Download Paper

6. FluidGPU: Fine-Grained Kernel Disaggregation for Large Model Inference on Heterogeneous GPUs

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC), 2026 | Download Paper

7. TENET: An Efficient Sparsity-Aware LUT-Centric Architecture for Ternary LLM Inference On Edge

Published in IEEE/ACM International Symposium on Microarchitecture (MICRO), 2026 | Download Paper

8. Em-garde: A propose-match framework for proactive streaming video understanding

Published in The European Conference on Computer Vision (ECCV), 2026 | Download Paper

Highlighted by Tsinghua University official accounts

9. Efficient Remote Prefix Fetching with GPU-native Media ASICs

Published in ACM SIGCOMM Conference, 2026 | Download Paper

10. Video-in-the-Loop: Span-Grounded Long Video QA with Interleaved Reasoning

Published in International Conference on Machine Learning (ICML), 2026 | Download Paper

11. AdaNav: Adaptive Reasoning with Uncertainty for Vision-Language Navigation

Published in International Conference on Machine Learning (ICML), 2026 | Download Paper

12. Pushing a Single GPU to Its Limits and Scaling to Tens of Thousands: RL-Guided, Physically Consistent KMC for Nuclear Materials Simulation

Published in ISC High Performance (ISC), 2026 | Download Paper

13. V-Droid: Advancing Mobile GUI Agent Through Generative Verifiers

Published in The 32nd Annual International Conference On Mobile Computing And Networking (MobiCom), 2026 | Download Paper

14. Vec-LUT: Vector Table Lookup for Parallel Ultra-Low-Bit LLM Inference on Edge Devices

Published in The 24th International Conference on Mobile Systems, Applications, and Services (MobiSys), 2026 | Download Paper

This paper wins MobiSys 2026 Best Paper Runner-Up Award, and also selected as the featured paper for the On-Device AI session at MobiSys 2026.

15. Scaling LLM Test-Time Compute with Mobile NPU on Smartphones

Published in The 2026 ACM European Conference on Computer Systems (EuroSys), 2026 | Download Paper

16. AVA: Towards Agentic Video Analytics Systems with Video Language Models

Published in USENIX Symposium on Networked Systems Design and Implementation (NSDI), 2026 | Download Paper

17. SeerAttention-R: Sparse Attention Adaptation for Long Reasoning

Published in International Conference on Learning Representations (ICLR), 2026 | Download Paper

18. ProRe: A Proactive Reward System for GUI Agents via Reasoner–Actor Collaboration

Published in International Conference on Learning Representations (ICLR), 2026 | Download Paper

19. Sculptor: Empowering LLMs with Cognitive Agency via Active Context Management

Published in International Conference on Learning Representations (ICLR), 2026 | Download Paper

20. Neuralink: Fast on-Device LLM Inference with Neuron Co-Activation Linking

Published in ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2026 | Download Paper

21. BitDecoding: Unlocking Tensor Cores for Long-Context LLMs Decoding with Low-Bit KV Cache

Published in The 32nd International Symposium on High-Performance Computer Architecture (HPCA), 2026 | Download Paper

22. Efficient VLM Inference System With LoRA Adapters At the Edge

Published in IEEE Transactions on Mobile Computing (TMC), 2026 | Download Paper

23. PyramidFFT: Rearchitecting FFT with Matrix-Aligned Nested-Radix for Hierarchical Scratchpad Memory on AI Accelerators

Published in ACM Transactions on Architecture and Code Optimization (TACO), 2026 | Download Paper

24. MatXtract: Sparsity-Aware Matrix Transformation via Cascaded Compute Density EXtraction for SpMV

Published in ACM Transactions on Architecture and Code Optimization (TACO), 2026 | Download Paper

Selected for ACM Showcase

25. DWI: Efficient On-Device LLM Decoding Through Dynamic Width Selection on Nested Models

Published in IEEE Transactions on Mobile Computing (TMC), 2026 | Download Paper

1. SparStencil: Retargeting Sparse Tensor Cores to Scientific Stencil Computations via Structured Sparsity Transformation

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC), 2025 | Download Paper

SC 2025 Best Student Paper Award Finalist

2. Matrix Is All You Need: Rearchitecting Quantum Chemistry to Scale on AI Accelerators

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC), 2025 | Download Paper

3. SeerAttention: Self-distilled Attention Gating for Efficient Long-context Prefilling

Published in The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS), 2025 | Download Paper

4. StreamMind: Unlocking Full Frame Rate Streaming Video Dialogue through Event-Gated Cognition

Published in International Conference on Computer Vision (ICCV'25), 2025 | Download Paper

5. Jenga: Enhancing Long-Context Fine-tuning of LLMs with Contextual Token Sparsity

Published in USENIX Annual Technical Conference (ATC'25), 2025 | Download Paper

6. Bitnet.cpp: Efficient Edge Inference for Ternary LLMs

Published in Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (ACL), 2025 | Download Paper

7. LUTensor: A Software-Hardware Co-Design for LUT-Based Low-Bit LLM Inference

Published in The 52nd Annual International Symposium on Computer Architecture (ISCA), 2025 | Download Paper

8. Babel: A Scalable Pre-trained Model for Multi-Modal Sensing via Expandable Modality Alignment

Published in The 23rd ACM Conference on Embedded Networked Sensor Systems (SenSys), 2025 | Download Paper

9. T-MAC: CPU Renaissance via Table Lookup for Low-Bit LLM Deployment on Edge

Published in The 2025 ACM European Conference on Computer Systems (EuroSys), 2025 | Download Paper

10. Efficient and Adaptive Diffusion Model Inference Through Lookup Table on Mobile Devices

Published in IEEE Transactions on Mobile Computing (TMC), 2025 | Download Paper

11. Jigsaw: Toward Conflict-free Vectorized Stencil Computation by Tessellating Swizzled Registers

Published in 30th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP), 2025 | Download Paper

12. FlashFFTStencil: Bridging Fast Fourier Transforms to Memory-Efficient Stencil Computations on Tensor Core Units

Published in 30th ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP), 2025 | Download Paper

13. LUT-DLA: Lookup Table as Efficient Extreme Low-Bit Deep Learning Accelerator

Published in 31st IEEE International Symposium on High-Performance Computer Architecture (HPCA), 2025 | Download Paper

14. Anatomizing Deep Learning Inference in Web Browsers

Published in ACM Transactions on Software Engineering and Methodology (TOSEM), 2025 | Download Paper

1. LoRAStencil: Low-Rank Adaptation of Stencil Computation on Tensor Cores

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC’24), 2024 | Download Paper

SC 2025 Reproducibility Challenge Finalist

2. Long Exposure: Accelerating Parameter-Efficient Fine-Tuning for LLMs under Shadowy Sparsity

Published in International Conference for High Performance Computing, Networking, Storage, and Analysis (SC’24), 2024 | Download Paper

3. VPTQ: Extreme Low-bit Vector Post-Training Quantization for Large Language Models

Published in The 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024 | Download Paper

4. PruneAug: Bridging DNN Pruning and Inference Latency on Diverse Sparse Platforms Using Automatic Layerwise Block Pruning

Published in IEEE Transactions on Computers (TC), 2024 | Download Paper

5. BitDistiller: Unleashing the Potential of Sub-4-Bit LLMs via Self-Distillation

Published in 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024 Main Conference, Long paper), 2024 | Download Paper

6. AFPQ: Asymmetric Floating Point Quantization for LLMs

Published in 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024 Finding short paper), 2024 | Download Paper

7. Ladder: Enabling Efficient Low-Precision Deep Learning Computing through Hardware-aware Tensor Transformation

Published in The 18th USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2024 | Download Paper

8. Integer or Floating Point? New Outlooks for Low-Bit Quantization on Large Language Models

Published in IEEE International Conference on Multimedia and Expo (ICME’24), 2024 | Download Paper

9. Empowering In-Browser Deep Learning Inference on Edge Through Just-In-Time Kernel Optimization

Published in The 22nd Annual International Conference on Mobile Systems, Applications and Services (MobiSys), 2024 | Download Paper

10. Pre-gated MoE: An Algorithm-System Co-Design for Fast and Scalable Mixture-of-Expert Inference

Published in The 51st Annual International Symposium on Computer Architecture 2024 (ISCA’24), 2024 | Download Paper

Microsoft Research Focus 2024

11. FlexNN: Efficient and Adaptive DNN Inference on Memory-Constrained Edge Devices

Published in The 30th Annual International Conference On Mobile Computing And Networking (MobiCom), 2024 | Download Paper

12. PIM-DL: Expanding the Applicability of Commodity DRAM-PIMs for Deep Learning via Algorithm-System Co-Optimization

Published in ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2024 | Download Paper

13. LitePred: Transferable and Scalable Latency Prediction for Hardware-Aware Neural Architecture Search

Published in USENIX Symposium on Networked Systems Design and Implementation (NSDI), 2024 | Download Paper

14. ConvStencil: Transform Stencil Computation to Matrix Multiplication on Tensor Cores

Published in ACM SIGPLAN Annual Symposium on Principles and Practice of Parallel Programming (PPoPP), 2024 | Download Paper

PPoPP 2024 Best Paper Award

1. LUT-NN: Empower Efficient Neural Network Inference with Centroid Learning and Table Lookup

Published in The 29th Annual International Conference On Mobile Computing And Networking (MobiCom), 2023 | Download Paper

Microsoft Research Focus 2023

2. Constraint-aware and Ranking-distilled Token Pruning for Efficient Transformer Inference

Published in ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2023 | Download Paper

3. SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference

Published in International Conference on Computer Vision (ICCV), 2023 | Download Paper

4. ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices

Published in International Conference on Computer Vision (ICCV), 2023 | Download Paper

5. Adam Accumulation to Reduce Memory Footprints of both Activations and Gradients for Large-scale DNN Training

Published in 26th European Conference on Artificial Intelligence (ECAI), 2023 | Download Paper

6. Accurate and Structured Pruning for Efficient Automatic Speech Recognition

Published in Conference of the International Speech Communication Association (INTERSPEECH), 2023 | Download Paper

7. HiMoDepth: Efficient Training-Free High-Resolution On-Device Depth Perception

Published in IEEE Transactions on Mobile Computing (TMC), 2023 | Download Paper

8. NN-Stretch: Automatic Neural Network Branching for Parallel Inference on Heterogeneous Multi-Processors

Published in The 21st International Conference on Mobile Systems, Applications, and Services (MobiSys), 2023 | Download Paper

9. Boosting DNN Cold Inference on Devices

Published in The 21st Annual International Conference on Mobile Systems, Applications and Services (MobiSys), 2023 | Download Paper

10. Efficient GPU Kernels for N:M-SPARSE Weights in Deep Learning

Published in Sixth Conference on Machine Learning and Systems (MLSys), 2023 | Download Paper

11. VSPIM: SRAM Processing-in-Memory DNN Acceleration via Vector-Scalar Operations

Published in IEEE Transactions on Computers (TC), 2023 | Download Paper

1. Turbo: Opportunistic Enhancement for Edge Video Analytics

Published in The 20th ACM Conference on Embedded Networked Sensor Systems (SenSys), 2022 | Download Paper

2. Hyperion: A Generic and Distributed Mobile Offloading Framework on OpenCL

Published in The 20th ACM Conference on Embedded Networked Sensor Systems (SenSys), 2022 | Download Paper

3. Romou: Rapidly Generate High-Performance Tensor Kernels for Mobile GPUs

Published in Proceedings of the 28th Annual International Conference on Mobile Computing and Networking (MobiCom), 2022 | Download Paper

ArchProbe: 2022-2023 Top 100 Open Source Achievements Award

4. MobiDepth: Real-Time Depth Estimation Using On-Device Dual Cameras

Published in Proceedings of the 28th Annual International Conference on Mobile Computing and Networking (MobiCom), 2022 | Download Paper

5. SwiftPruner: Reinforced Evolutionary Pruning for Efficient Ad Relevance

Published in ACM International Conference on Information and Knowledge Management (CIKM), 2022 | Download Paper

6. CoDL: Efficient CPU-GPU Co-execution for Deep Learning Inference on Mobile Devices

Published in 20th International Conference on Mobile Systems, Applications, and Services (MobiSys), 2022 | Download Paper

1. AsyMo: Scalable and Efficient Deep-Learning Inference on Asymmetric Mobile CPUs

Published in Proceedings of the 27th Annual International Conference on Mobile Computing and Networking (MobiCom), 2021 | Download Paper

2. nn-Meter: towards accurate latency prediction of deep-learning model inference on diverse edge devices

Published in 19th International Conference on Mobile Systems, Applications, and Services (MobiSys), 2021 | Download Paper

MobiSys 2021 Best Paper Award

3. To Bridge Neural Network Design and Real-World Performance: A Behaviour Study for Neural Networks

Published in Proceedings of Machine Learning and Systems (MLSys), 2021 | Download Paper

4. nn-Meter: towards accurate latency prediction of DNN inference on diverse edge devices

Published in GetMobile: Mobile Computing and Communications, Research Highlights, 2021 | Download Paper

ACM SigMobile Research Highlight

5. Unified Holistic Memory Management Supporting Multiple Big Data Processing Frameworks over Hybrid Memories

Published in ACM Transactions on Computer Systems (TOCS), 2021 | Download Paper

1. Profiling and optimizing deep learning inference on mobile GPUs

Published in Proceedings of the 11th ACM SIGOPS Asia-Pacific Workshop on Systems (APSys), 2020 | Download Paper

1. Panthera: Holistic Memory Management for Big Data Processing over Hybrid Memories

Published in ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI), 2019 | Download Paper