About

Chao Li is a faculty member in the School of Modern Languages at Georgia Tech, having joined the institution in 2001. He holds an M.A. in International Relations from the Beijing Institute of Foreign Affairs and a B.A. in Economics and Chinese from Yunnan University, China. As a Lecturer in Chinese, he has taught the language at all levels and has played a significant role in developing Georgia Tech’s Chinese online language courses, including creating course content, multimedia materials, and grammatical notes. He has served as co-director of the Chinese LBAT (Language for Business and Technology) Program since 2006 and is recognized as a key instructor in online Chinese teaching. His research interests focus on the development of online Chinese language courses and effective teaching and learning materials for Chinese as a second language.

Research topics

• Machine Learning
• Artificial Intelligence
• Computer Security
• Computer Science
• Engineering

Selected publications

• A Survey of Neural Network Variational Monte Carlo from a Computing Workload Characterization Perspective

  arXiv (Cornell University) · 2026-03-18

  articleOpen accessSenior author

  Neural Network Variational Monte Carlo (NNVMC) has emerged as a promising paradigm for solving quantum many-body problems by combining variational Monte Carlo with expressive neural-network wave-function ansätze. Although NNVMC can achieve competitive accuracy with favorable asymptotic scaling, practical deployment remains limited by high runtime and memory cost on modern graphics processing units (GPUs). Compared with language and vision workloads, NNVMC execution is shaped by physics-specific stages, including Markov-Chain Monte Carlo sampling, wave-function construction, and derivative/Laplacian evaluation, which produce heterogeneous kernel behavior and nontrivial bottlenecks. This paper provides a workload-oriented survey and empirical GPU characterization of four representative ansätze: PauliNet, FermiNet, Psiformer, and Orbformer. Using a unified profiling protocol, we analyze model-level runtime and memory trends and kernel-level behavior through family breakdown, arithmetic intensity, roofline positioning, and hardware utilization counters. The results show that end-to-end performance is often constrained by low-intensity elementwise and data-movement kernels, while the compute/memory balance varies substantially across ansätze and stages. Based on these findings, we discuss algorithm--hardware co-design implications for scalable NNVMC systems, including phase-aware scheduling, memory-centric optimization, and heterogeneous acceleration.

  PublisherOA PDF

• TideGS: Scalable Training of Over One Billion 3D Gaussian Splatting Primitives via Out-of-Core Optimization

  ArXiv.org · 2026-05-19

  articleOpen accessSenior author

  Training 3D Gaussian Splatting (3DGS) at billion-primitive scale is fundamentally memory-bound: each Gaussian primitive carries a large attribute vector, and the aggregate parameter table quickly exceeds GPU capacity, limiting prior systems to tens of millions of Gaussians on commodity single-GPU hardware. We observe that 3DGS training is inherently sparse and trajectory-conditioned: each iteration activates only the Gaussians visible from the current camera batch, so GPU memory can serve as a working-set cache rather than a persistent parameter store. Building on this insight, we introduce TideGS, an out-of-core training framework that manages parameters across an SSD-CPU-GPU hierarchy via three synergistic techniques: block-virtualized geometry for SSD-aligned spatial locality, a hierarchical asynchronous pipeline to overlap I/O with computation, and trajectory-adaptive differential streaming that transfers only incremental working-set deltas between iterations. Experiments show that TideGS enables training with over one billion Gaussians on a single 24 GB GPU while achieving the best reconstruction quality among evaluated single-GPU baselines on large-scale scenes, scaling beyond prior out-of-core baselines (e.g., approximately 100M Gaussians) and standard in-memory training (e.g., approximately 11M Gaussians).

  PublisherOA PDF

• From Inference Efficiency to Embodied Efficiency: Revisiting Efficiency Metrics for Vision-Language-Action Models

  arXiv (Cornell University) · 2026-03-19

  preprintOpen accessSenior author

  Vision-Language-Action (VLA) models have recently enabled embodied agents to perform increasingly complex tasks by jointly reasoning over visual, linguistic, and motor modalities. However, we find that the prevailing notion of ``efficiency'' in current VLA research, characterized by parameters, FLOPs, or token decoding throughput, does not reflect actual performance on robotic platforms. In real-world execution, efficiency is determined by system-level embodied behaviors such as task completion time, trajectory smoothness, cumulative joint rotation, and motion energy. Through controlled studies across model compression, token sparsification, and action sequence compression, we make several observations that challenge common assumptions. (1) Methods that reduce computation under conventional metrics often increase end-to-end execution cost or degrade motion quality, despite maintaining task success rates. (2) System-level embodied efficiency metrics reveal performance differences in the learned action policies that remain hidden under conventional evaluations. (3) Common adaptation methods such as in-context prompting or supervised fine-tuning show only mild and metric-specific improvements in embodied efficiency. While these methods can reduce targeted embodied-efficiency metrics such as jerk or action rate, the resulting gains may come with trade-offs in other metrics, such as longer completion time. Taken together, our results suggest that conventional inference efficiency metrics can overlook important aspects of embodied execution. Incorporating embodied efficiency provides a more complete view of policy behavior and practical performance, enabling fairer and more comprehensive comparisons of VLA models.

  PublisherDOI

• TideGS: Scalable Training of Over One Billion 3D Gaussian Splatting Primitives via Out-of-Core Optimization

  arXiv (Cornell University) · 2026-05-19

  preprintOpen accessSenior author

  Training 3D Gaussian Splatting (3DGS) at billion-primitive scale is fundamentally memory-bound: each Gaussian primitive carries a large attribute vector, and the aggregate parameter table quickly exceeds GPU capacity, limiting prior systems to tens of millions of Gaussians on commodity single-GPU hardware. We observe that 3DGS training is inherently sparse and trajectory-conditioned: each iteration activates only the Gaussians visible from the current camera batch, so GPU memory can serve as a working-set cache rather than a persistent parameter store. Building on this insight, we introduce TideGS, an out-of-core training framework that manages parameters across an SSD-CPU-GPU hierarchy via three synergistic techniques: block-virtualized geometry for SSD-aligned spatial locality, a hierarchical asynchronous pipeline to overlap I/O with computation, and trajectory-adaptive differential streaming that transfers only incremental working-set deltas between iterations. Experiments show that TideGS enables training with over one billion Gaussians on a single 24 GB GPU while achieving the best reconstruction quality among evaluated single-GPU baselines on large-scale scenes, scaling beyond prior out-of-core baselines (e.g., approximately 100M Gaussians) and standard in-memory training (e.g., approximately 11M Gaussians).

  PublisherDOI

• Think Hard Only When Needed: A Hybrid Best-of-N and Beam Search for Efficient Test-Time Compute

  2026-01-01

  articleOpen access

  Test-time compute has emerged as a promising paradigm that enables small language models (SLMs) to achieve large language model (LLM)-level capabilities by allocating additional compute for explicit reasoning during inference.Two common approaches are beam search and Best-of-N sampling.Beam search improves reasoning quality by scoring and optimizing token sequences using Process Reward Models (PRMs), but can incur non-trivial computational overhead and latency.In contrast, Best-of-N executes all reasoning trajectories without PRM guidance, often wasting compute on low-quality trajectories that may have gone astray early in the generation process.To address both inefficiencies, we propose THROW (THink haRd Only When needed)-a hybrid inference pipeline that combines the diversity of Best-of-N with the reasoning trajectory optimization of beam search.THROW introduces a selective branch truncation and expansion mechanism: it generates shorter initial trajectories than Best-of-N and evaluates them using PRMs to classify each query as "easy" or "hard".Based on this classification, THROW applies branch truncation for easy queries, mimicking Best-of-N, and PRM-guided branch expansion for hard ones, similar to beam search.Evaluations on MATH500, AMC23, and AIME24 demonstrate that THROW achieves 1.54 and 14.38 latency speedups and 35.7% and 80.4% token reductions on average while preserving high reasoning accuracy compared to Best-of-N and Beam Search, respectively.

  PublisherOA PDFDOI

• A Survey of Neural Network Variational Monte Carlo from a Computing Workload Characterization Perspective

  arXiv (Cornell University) · 2026-03-18

  preprintOpen accessSenior author

  Neural Network Variational Monte Carlo (NNVMC) has emerged as a promising paradigm for solving quantum many-body problems by combining variational Monte Carlo with expressive neural-network wave-function ansätze. Although NNVMC can achieve competitive accuracy with favorable asymptotic scaling, practical deployment remains limited by high runtime and memory cost on modern graphics processing units (GPUs). Compared with language and vision workloads, NNVMC execution is shaped by physics-specific stages, including Markov-Chain Monte Carlo sampling, wave-function construction, and derivative/Laplacian evaluation, which produce heterogeneous kernel behavior and nontrivial bottlenecks. This paper provides a workload-oriented survey and empirical GPU characterization of four representative ansätze: PauliNet, FermiNet, Psiformer, and Orbformer. Using a unified profiling protocol, we analyze model-level runtime and memory trends and kernel-level behavior through family breakdown, arithmetic intensity, roofline positioning, and hardware utilization counters. The results show that end-to-end performance is often constrained by low-intensity elementwise and data-movement kernels, while the compute/memory balance varies substantially across ansätze and stages. Based on these findings, we discuss algorithm--hardware co-design implications for scalable NNVMC systems, including phase-aware scheduling, memory-centric optimization, and heterogeneous acceleration.

  PublisherDOI

• From Inference Efficiency to Embodied Efficiency: Revisiting Efficiency Metrics for Vision-Language-Action Models

  ArXiv.org · 2026-03-19

  articleOpen accessSenior author

  Vision-Language-Action (VLA) models have recently enabled embodied agents to perform increasingly complex tasks by jointly reasoning over visual, linguistic, and motor modalities. However, we find that the prevailing notion of ``efficiency'' in current VLA research, characterized by parameters, FLOPs, or token decoding throughput, does not reflect actual performance on robotic platforms. In real-world execution, efficiency is determined by system-level embodied behaviors such as task completion time, trajectory smoothness, cumulative joint rotation, and motion energy. Through controlled studies across model compression, token sparsification, and action sequence compression, we make several observations that challenge common assumptions. (1) Methods that reduce computation under conventional metrics often increase end-to-end execution cost or degrade motion quality, despite maintaining task success rates. (2) System-level embodied efficiency metrics reveal performance differences in the learned action policies that remain hidden under conventional evaluations. (3) Common adaptation methods such as in-context prompting or supervised fine-tuning show only mild and metric-specific improvements in embodied efficiency. While these methods can reduce targeted embodied-efficiency metrics such as jerk or action rate, the resulting gains may come with trade-offs in other metrics, such as longer completion time. Taken together, our results suggest that conventional inference efficiency metrics can overlook important aspects of embodied execution. Incorporating embodied efficiency provides a more complete view of policy behavior and practical performance, enabling fairer and more comprehensive comparisons of VLA models.

  PublisherOA PDF

• Super‐tough and excellent heat‐resistant poly( <scp>l</scp> ‐lactide) reinforced by silane‐functionalized <scp>MXene</scp> composites

  Polymer International · 2025-12-19

  article

  Abstract Poly( l ‐lactide) (PLLA) is a promising biodegradable polymer whose widespread application is limited by inherent brittleness and insufficient thermal stability. To address these challenges, this study introduces silane‐functionalized MXene as a multifunctional nanofiller for PLLA. The MXene was chemically grafted with 3‐(triethoxysilyl)propylisocyanate to enhance interfacial compatibility with the polymer matrix. The resulting PLLA composite containing only 0.5 wt% of the modified MXene exhibits a dramatic mechanical enhancement, achieving a 210% elongation at break, which represents a 23‐fold increase over that of neat PLLA, while maintaining a high tensile strength of 58 MPa. Dynamic mechanical analysis confirms superior storage modulus, and thermogravimetric analysis reveals significantly improved thermal stability with degradation onset temperatures exceeding 290 °C. These findings demonstrate that silane‐functionalized MXene effectively reinforces PLLA through interfacial crosslinking and efficient stress transfer, providing a viable strategy for developing high‐performance biodegradable polymers. © 2025 Society of Chemical Industry.

  PublisherDOI

• Scaling Laws of Graph Neural Networks for Atomistic Materials Modeling

  ArXiv.org · 2025-04-10 · 1 citations

  preprintOpen access1st authorCorresponding

  Atomistic materials modeling is a critical task with wide-ranging applications, from drug discovery to materials science, where accurate predictions of the target material property can lead to significant advancements in scientific discovery. Graph Neural Networks (GNNs) represent the state-of-the-art approach for modeling atomistic material data thanks to their capacity to capture complex relational structures. While machine learning performance has historically improved with larger models and datasets, GNNs for atomistic materials modeling remain relatively small compared to large language models (LLMs), which leverage billions of parameters and terabyte-scale datasets to achieve remarkable performance in their respective domains. To address this gap, we explore the scaling limits of GNNs for atomistic materials modeling by developing a foundational model with billions of parameters, trained on extensive datasets in terabyte-scale. Our approach incorporates techniques from LLM libraries to efficiently manage large-scale data and models, enabling both effective training and deployment of these large-scale GNN models. This work addresses three fundamental questions in scaling GNNs: the potential for scaling GNN model architectures, the effect of dataset size on model accuracy, and the applicability of LLM-inspired techniques to GNN architectures. Specifically, the outcomes of this study include (1) insights into the scaling laws for GNNs, highlighting the relationship between model size, dataset volume, and accuracy, (2) a foundational GNN model optimized for atomistic materials modeling, and (3) a GNN codebase enhanced with advanced LLM-based training techniques. Our findings lay the groundwork for large-scale GNNs with billions of parameters and terabyte-scale datasets, establishing a scalable pathway for future advancements in atomistic materials modeling.

  PublisherOA PDFDOI

• Gaussian Blending Unit: An Edge GPU Plug-in for Real-Time Gaussian-Based Rendering in AR/VR

  2025-03-01 · 13 citations

  article

  The rapidly advancing field of Augmented and Virtual Reality (AR/VR) demands real-time, photorealistic rendering on resource-constrained platforms. 3D Gaussian Splatting, delivering state-of-the-art (SOTA) performance in rendering efficiency and quality, has emerged as a promising solution across a broad spectrum of AR/VR applications. However, despite its effectiveness on high-end GPUs, it struggles on edge systems like the Jetson Orin NX Edge GPU, achieving only 7-17 FPS—well below the over 60 FPS standard required for truly immersive AR/VR experiences. Addressing this challenge, we perform a comprehensive analysis of Gaussian-based AR/VR applications and identify the Gaussian Blending Stage, which intensively calculates each Gaussian’s contribution at every pixel, as the primary bottleneck. In response, we propose a Gaussian Blending Unit (GBU), an edge GPU plug-in module for real-time rendering in AR/VR applications. Notably, our GBU can be seamlessly integrated into conventional edge GPUs and collaboratively supports a wide range of AR/VR applications. Specifically, GBU incorporates an intra-row sequential shading (IRSS) dataflow that shades each row of pixels sequentially from left to right, utilizing a two-step coordinate transformation. This transformation enables (1) the sharing of intermediate values between adjacent pixels, reducing pixel-wise computation costs by up to $5.5 \times$, and (2) the early identification and skipping of Gaussians that minimally contribute to the pixels, reducing per-pixel computation by up to $\mathbf{9 3 \%}$. When directly deployed on a GPU, the proposed dataflow achieved a non-trivial $1.72 \times$ speedup on real-world static scenes, though still falls short of real-time rendering performance. Recognizing the limited compute utilization in the GPU-based implementation, GBU enhances rendering speed with a dedicated rendering engine that balances the workload across rows by aggregating computations from multiple Gaussians. Additionally, GBU integrates a Gaussian Reuse Cache, reducing off-chip memory accesses by 44.9% and resulting in a $1.14 \times$ speedup in rendering. Experiments across representative AR/VR applications demonstrate that our GBU provides a unified solution for on-device real-time rendering while maintaining SOTA rendering quality.

  PublisherDOI

Frequent coauthors

• Yingyan Lin

  58 shared

• Vivek Boominathan

  Rice University

  40 shared

• Ashok Veeraraghavan

  40 shared

• Ph Lin

  Rice University

  36 shared

• Yonggan Fu

  Georgia Institute of Technology

  33 shared

• Haoran You

  26 shared

• Yang Zhao

  24 shared

• Yongan Zhang

  21 shared