Publications (73)
A Controlled Study of Decoding-Time Truthfulness Methods on Instruction-Tuned LLMs
Ao Sun
Decoding-time truthfulness methods -- layer-contrast decoding, inference-time intervention, and learned logit adapters -- have demonstrated 10-30 point gains on TruthfulQA when app…
Backtesting Trading Strategies with GAN To Avoid Overfitting
Ao Sun, Yuh-Dauh Lyuu
Many works have shown the overfitting hazard of selecting a trading strategy based only on good IS (in sample) performance. But most of them have merely shown such phenomena exist…
CuMA: Aligning LLMs with Sparse Cultural Values via Demographic-Aware Mixture of Adapters
Ao Sun, Xiaoyu Wang, Zhe Tan +4
As Large Language Models (LLMs) serve a global audience, alignment must transition from enforcing universal consensus to respecting cultural pluralism. We demonstrate that dense mo…
Sharp Entropy Bounds for Plane Curves and Dynamics of the Curve Shortening Flow
Julius Baldauf, Ao Sun
We prove that a closed immersed plane curve with total curvature has entropy at least times the entropy of the embedded circle, as long as it generates a type I singular…
CoPRIS: Efficient and Stable Reinforcement Learning via Concurrency-Controlled Partial Rollout with Importance Sampling
Zekai Qu, Yinxu Pan, Ao Sun +2
Reinforcement learning (RL) post-training has become a trending paradigm for enhancing the capabilities of large language models (LLMs). Most existing RL systems for LLMs operate i…
How Does LLM Help Regional CPI Forecast: An LLM-powered Deep Panel Modeling Framework
Tianchen Gao, Ao Sun, Yurou Wang +2
Understanding regional Consumer Price Index (CPI) dynamics is essential for timely and effective economic policymaking. However, traditional modeling procedures typically rely only…
StateFlow: Sequence Pipeline Parallelism for Long-Context Modeling with Linear Recurrence
Wenxuan Zhao, Yingfa Chen, Xu Han +7
Long-context training is increasingly important for large language models, and linear attention and state space models have become popular for improving long-context efficiency. Ho…
Rigidity of spherical product Ricci solitons
Ao Sun, Jonathan J. Zhu
We show that is isolated as a shrinking Ricci soliton in the space of metrics, up to scaling and diffeomorphism. We also prove the same rigidity for ,…
The Score Granularity Gap in Black-Box LLM Classification: A Comparative Study of Confidence Constructions
Ao Sun, Tian Sun, Jiaxing Geng
Large language models (LLMs) are increasingly deployed as black-box classifiers in pipelines that automate confident decisions and route uncertain ones to human review. Such select…
Compactness of Constant Mean Curvature Surfaces in Three Manifold with Positive Ricci Curvature
Ao Sun
In this paper we prove a compactness theorem for constant mean curvature surfaces with area and genus bound in three manifold with positive Ricci curvature. As an application, we g…
High-rate continuous-variable quantum key distribution over 100 km fiber with composable security
Heng Wang, Yang Li, Ting Ye +14
Quantum key distribution (QKD), providing a way to generate secret keys with information-theoretic security,is arguably one of the most significant achievements in quantum informat…
BASIC: Bipartite Assisted Spectral-clustering for Identifying Communities in Large-scale Networks
Tianchen Gao, Jingyuan Liu, Rui Pan +1
Community detection, which focuses on recovering the group structure within networks, is a crucial and fundamental task in network analysis. However, the detection process can be q…
Bifurcation of perturbations of non-generic closed self-shrinkers
Zhengjiang Lin, Ao Sun
We discover a bifurcation of the perturbations of non-generic closed self-shrinkers. If the generic perturbation is outward, then the next mean curvature flow singularity is cylind…
On the Entropy of Parabolic Allen-Cahn Equation
Ao Sun
We define a (mean curvature flow) entropy for Radon measures in or in a compact manifold. Moreover, we prove a monotonicity formula of the entropy of the measures as…
Codimension Bounds and Rigidity of Ancient Mean Curvature Flows by the Tangent Flow at
Douglas Stryker, Ao Sun
Motivated by the limiting behavior of an explicit class of compact ancient curve shortening flows, we prove codimension bounds for ancient mean curvature flows by their tangent flo…
An effcient variational quantum Korkin-Zolotarev algorithm for solving shortest vector problems
Xiaokai Hou, Guoqing Zhou, Shan Jin +5
Noisy intermediate-scale quantum cryptanalysis focuses on the capability of near-term quantum devices to solve the mathematical problems underlying cryptography, and serves as a co…
BurstEngine: an Efficient Distributed Framework for Training Transformers on Extremely Long Sequences of over 1M Tokens
Ao Sun, Weilin Zhao, Xu Han +4
Existing methods for training LLMs on long-sequence data, such as Tensor Parallelism and Context Parallelism, exhibit low Model FLOPs Utilization as sequence lengths and number of…
On mean curvature flow translators with prescribed ends
Ao Sun, Zhihan Wang
Given a smooth closed embedded self-shrinker with index in , we construct an -dimensional family of complete translators polynomially asymptotic to $S\ti…
Carleman Estimate for Surface in Euclidean Space at Infinity
Ao Sun
This paper develops a Carleman type estimate for immersed surface in Euclidean space at infinity. With this estimate, we obtain an unique continuation property for harmonic functio…
On the probability of a Pareto record
James Allen Fill, Ao Sun
Given a sequence of independent random vectors taking values in and having common continuous distribution function , say that the observ…
NavigateDiff: Visual Predictors are Zero-Shot Navigation Assistants
Yiran Qin, Ao Sun, Yuze Hong +2
Navigating unfamiliar environments presents significant challenges for household robots, requiring the ability to recognize and reason about novel decoration and layout. Existing r…
Initial Perturbation of the Mean Curvature Flow for closed limit shrinker
Ao Sun, Jinxin Xue
This is a contribution to the program of dynamical approach to mean curvature flow initiated by Colding and Minicozzi. In this paper, we prove two main theorems. The first one is l…
Entropy in A Closed Manifold and Partial Regularity of Mean Curvature Flow Limit of Surfaces
Ao Sun
Inspired by the idea of Colding-Minicozzi in [CM1], we define (mean curvature flow) entropy for submanifolds in a general ambient Riemannian manifold. In particular, this entropy i…
Compactness and Rigidity of -Surfaces
Ao Sun
In this paper we develop the compactness theorem for -surface in with uniform , genus, and area growth. This theorem can be viewed as a generalization of Cold…
InfiniPipe: Elastic Pipeline Parallelism for Efficient Variable-Length Long-Context LLM Training
Shiju Wang, Yujie Wang, Ao Sun +5
Long context training is crucial for LLM's context extension. Existing schemes, such as sequence parallelism, incur substantial communication overhead. Pipeline parallelism (PP) re…
FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling
Weilin Zhao, Tengyu Pan, Xu Han +9
Speculative sampling has emerged as an important technique for accelerating the auto-regressive generation process of large language models (LLMs) by utilizing a draft-then-verify…
Secure Combination of Untrusted Time information Based on Optimized Dempster-Shafer Theory
Yang Li, Yujie Luo, Yichen Zhang +16
Secure precision time synchronization is important for applications of Cyber-Physical Systems. However, several attacks, especially the Time Delay Attack (TDA), deteriorates the pe…
Admissibility of solution estimators for stochastic optimization
Amitabh Basu, Tu Nguyen, Ao Sun
We look at stochastic optimization problems through the lens of statistical decision theory. In particular, we address admissibility, in the statistical decision theory sense, of t…
"Oops, Did I Just Say That?" Testing and Repairing Unethical Suggestions of Large Language Models with Suggest-Critique-Reflect Process
Pingchuan Ma, Zongjie Li, Ao Sun +1
As the popularity of large language models (LLMs) soars across various applications, ensuring their alignment with human values has become a paramount concern. In particular, given…
VoiceAgentEval: A Dual-Dimensional Benchmark for Expert-Level Intelligent Voice-Agent Evaluation of Xbench's Professional-Aligned Series
Pengyu Xu, Shijia Li, Ao Sun +15
We propose OutboundEval, a comprehensive benchmark for evaluating large language models (LLMs) in expert-level intelligent outbound calling scenarios. Unlike existing methods that…
NOSA: Native and Offloadable Sparse Attention
Yuxiang Huang, Pengjie Wang, Jicheng Han +9
Decoding throughput improvements from larger inference batches are limited by GPU memory, which is largely consumed by the key-value (KV) cache. Prior training-free KV cache offloa…
Local Entropy and Generic Multiplicity One Singularities of Mean Curvature Flow of Surfaces
Ao Sun
In this paper we prove that the generic singularities of mean curvature flow of closed embedded surfaces in modeled by closed self-shrinkers with multiplicity has mul…
BurstAttention: An Efficient Distributed Attention Framework for Extremely Long Sequences
Ao Sun, Weilin Zhao, Xu Han +4
Effective attention modules have played a crucial role in the success of Transformer-based large language models (LLMs), but the quadratic time and memory complexities of these att…
Sharpened localization of the trailing point of the Pareto record frontier
James Allen Fill, Daniel Naiman, Ao Sun
For and iid -dimensional observations with independent Exponential coordinates, we revisit the study by Fill and Naiman (Electron. J. Probab…
Passing through nondegenerate singularities in mean curvature flows
Ao Sun, Zhihan Wang, Jinxin Xue
In this paper, we study the properties of nondegenerate cylindrical singularities of mean curvature flow. We prove they are isolated in spacetime and provide a complete description…
How Multi-Modal LLMs Reshape Visual Deep Learning Testing? A Comprehensive Study Through the Lens of Image Mutation
Liwen Wang, Yuanyuan Yuan, Ao Sun +4
Visual deep learning (VDL) systems have shown significant success in real-world applications like image recognition, object detection, and autonomous driving. To evaluate the relia…
Generic mean curvature flows with cylindrical singularities I: the normal forms and nondegeneracy
Ao Sun, Jinxin Xue
This paper studies the dynamics of mean curvature flow as it approaches a cylindrical singularity. We proved that the rescaled mean curvature flow converging to a smooth generalize…
Eliminating Information Leakage in Hard Concept Bottleneck Models with Supervised, Hierarchical Concept Learning
Ao Sun, Yuanyuan Yuan, Pingchuan Ma +1
Concept Bottleneck Models (CBMs) aim to deliver interpretable and interventionable predictions by bridging features and labels with human-understandable concepts. While recent CBMs…
Explain Any Concept: Segment Anything Meets Concept-Based Explanation
Ao Sun, Pingchuan Ma, Yuanyuan Yuan +1
EXplainable AI (XAI) is an essential topic to improve human understanding of deep neural networks (DNNs) given their black-box internals. For computer vision tasks, mainstream pixe…
Closed mean curvature flows with prescribed tangent flows
Jingwen Chen, Tang-Kai Lee, Ao Sun +1
Given an embedded shrinker in that is either closed, asymptotically conical, or a Cartesian product of such a shrinker with , we construct a c…
OPTICS: Order-Preserved Test-Inverse Confidence Set for Number of Change-Points
Ao Sun, Jingyuan Liu
Determining the number of change-points is a first-step and fundamental task in change-point detection problems, as it lays the groundwork for subsequent change-point position esti…
State Rank Dynamics in Linear Attention LLMs
Ao Sun, Hongtao Zhang, Heng Zhou +9
Linear Attention Large Language Models (LLMs) offer a compelling recurrent formulation that compresses context into a fixed-size state matrix, enabling constant-time inference. How…
Biomarker combination based on the Youden index with and without gold standard
Ao Sun, Yanting Li, Xiao-Hua Zhou
In clinical practice, multiple biomarkers are often measured on the same subject for disease diagnosis, and combining them can improve diagnostic accuracy. Existing studies typical…
Biomarkers selection and combination based on the weighted Youden index
Ao Sun, Zhanwang Deng, Jiahui Zhao +2
In clinical practice, multiple biomarkers are used for disease diagnosis, but their individual accuracies are often suboptimal, with only a few proving directly relevant. Effective…
Regularity of cylindrical singular sets of mean curvature flow
Ao Sun, Zhihan Wang, Jinxin Xue
In this paper, we study the -cylindrical singular set of mean curvature flow in for each . We prove that they are locally contained in a -d…
Min-max minimal disks with free boundary in Riemannian manifolds
Longzhi Lin, Ao Sun, Xin Zhou
In this paper, we establish a min-max theory for constructing minimal disks with free boundary in any closed Riemannian manifold. The main result is an effective version of the par…
APB-V: Accelerating Long-Video Understanding via Sequence-Parallelism-aware Approximate Attention
Yuxiang Huang, Mingye Li, Xu Han +7
The efficiency of long-video inference remains a critical bottleneck, mainly due to the dense computation in the prefill stage of Large Multimodal Models (LMMs). Existing methods e…
Construction of High Codimension Ancient Mean Curvature Flows
Douglas Stryker, Ao Sun
We construct a class of compact ancient solutions to the mean curvature flow in Euclidean space with high codimension. In particular, we construct higher codimensional ancient curv…
CHAIR -- Classifier of Hallucination as Improver
Ao Sun
In this work, we introduce CHAIR (Classifier of Hallucination As ImproveR), a supervised framework for detecting hallucinations by analyzing internal logits from each layer of ever…
MOSAIC: Minimax-Optimal Sparsity-Adaptive Inference for Change Points in Dynamic Networks
Yingying Fan, Jingyuan Liu, Jinchi Lv +1
We propose a new inference framework, named MOSAIC, for change-point detection in dynamic networks with the simultaneous low-rank and sparse-change structure. We establish the mini…
Alignment Strength and Correlation for Graphs
Donniell E. Fishkind, Lingyao Meng, Ao Sun +2
When two graphs have a correlated Bernoulli distribution, we prove that the alignment strength of their natural bijection strongly converges to a novel measure of graph correlation…
Singularities of Mean Curvature Flow of Surfaces with Additional Forces
Ao Sun
In this paper we study the blow up sequence of mean curvature flow of surfaces in with additional forces. We prove that the blow up limit of a mean curvature flow of…
RMSWeb: Reflection, Failure-Mode Mining, and Salvage-DS for Web Agent Reinforcement Learning
Chengbo Liu, Lifang Zhou, Ruijie Yan +8
Compact web agents can reduce deployment cost, but training them poses challenges in both data collection and post-SFT reinforcement learning (RL). Successful trajectories are expe…
Genus one singularities in mean curvature flow
Adrian Chun-Pong Chu, Ao Sun
We show that for certain one-parameter families of initial conditions in , when we run mean curvature flow, a genus one singularity must appear in one of the flows. Mo…
Seq1F1B: Efficient Sequence-Level Pipeline Parallelism for Large Language Model Training
Ao Sun, Weilin Zhao, Xu Han +5
The emergence of large language models (LLMs) relies heavily on distributed training strategies, among which pipeline parallelism plays a crucial role. As LLMs' training sequence l…
Data-Free Class-Incremental Gesture Recognition with Prototype-Guided Pseudo Feature Replay
Hongsong Wang, Ao Sun, Jie Gui +1
Gesture recognition is an important research area in the field of computer vision. Most gesture recognition efforts focus on close-set scenarios, thereby limiting the capacity to e…
Compactness of self-shrinkers in with fixed genus
Ao Sun, Zhichao Wang
We prove the compactness of self-shrinkers in with bounded entropy and fixed genus. As a corollary, we show that numbers of ends of such surfaces are uniformly bounde…
Integrated high-performance error correction for continuous-variable quantum key distribution
Chuang Zhou, Yang Li, Li Ma +12
An integrated error-correction scheme with high throughput, low frame errors rate (FER) and high reconciliation efficiency under low signal to noise ratio (SNR) is one of the major…
SDI-Net: Toward Sufficient Dual-View Interaction for Low-light Stereo Image Enhancement
Linlin Hu, Ao Sun, Shijie Hao +2
Currently, most low-light image enhancement methods only consider information from a single view, neglecting the correlation between cross-view information. Therefore, the enhancem…
On the long-time limit of the mean curvature flow in closed manifolds
Alexander Mramor, Ao Sun
In this article we show that generally almost regular flows, introduced by Bamler and Kleiner, in closed 3-manifolds will either go extinct in finite time or flow to a collection o…
Generic Dynamics of Mean Curvature Flows with Asymptotically Conical Singularities
Ao Sun, Jinxin Xue
This is the second paper in the series to study the generic dynamics of mean curvature flows. We study the initial perturbation of mean curvature flows, whose first singularity is…
A Generalized Knockoff Procedure for FDR Control in Structural Change Detection
Jingyuan Liu, Ao Sun, Yuan Ke
Controlling false discovery rate (FDR) is crucial for variable selection, multiple testing, among other signal detection problems. In literature, there is certainly no shortage of…
Multiplicity one for min-max theory in compact manifolds with boundary and its applications
Ao Sun, Zhichao Wang, Xin Zhou
We prove the multiplicity one theorem for min-max free boundary minimal hypersurfaces in compact manifolds with boundary of dimension between 3 and 7 for generic metrics. To approa…
ViDove: A Translation Agent System with Multimodal Context and Memory-Augmented Reasoning
Yichen Lu, Wei Dai, Jiaen Liu +9
LLM-based translation agents have achieved highly human-like translation results and are capable of handling longer and more complex contexts with greater efficiency. However, they…
InfLLM-V2: Dense-Sparse Switchable Attention for Seamless Short-to-Long Adaptation
Weilin Zhao, Zihan Zhou, Zhou Su +10
Long-sequence processing is a critical capability for modern large language models. However, the self-attention mechanism in the standard Transformer architecture faces severe comp…
Rigidity and Åojasiewicz inequalities for Clifford self-shrinkers
Ao Sun, Jonathan J. Zhu
We show that the product of two round shrinking spheres is an isolated self-shrinker in any codimension, modulo rotations. Moreover we prove explicit Åojasiewicz inequalities near…
MiniCPM4: Ultra-Efficient LLMs on End Devices
MiniCPM Team, Chaojun Xiao, Yuxuan Li +80
This paper introduces MiniCPM4, a highly efficient large language model (LLM) designed explicitly for end-side devices. We achieve this efficiency through systematic innovation in…
An adaptable JSON Diff Framework
Ao Sun
In this paper, we present an implementation of JSON-diff framework JYCM, extending the existing framework by introducing the concept of "unordered" comparisons and allowing users t…
LLM-Powered CPI Prediction Inference with Online Text Time Series
Yingying Fan, Jinchi Lv, Ao Sun +1
Forecasting the Consumer Price Index (CPI) is an important yet challenging task in economics, where most existing approaches rely on low-frequency, survey-based data. With the rece…
Mean curvature flow with multiplicity convergence in closed manifolds
Jingwen Chen, Ao Sun
We construct new examples of immortal mean curvature flow of smooth embedded connected hypersurfaces in closed manifolds, which converge to minimal hypersurfaces with multiplicity…
Existence of minimal hypersurfaces with arbitrarily large area and possible obstructions
James Stevens, Ao Sun
We prove that in a closed Riemannian manifold with dimension between and , either there are minimal hypersurfaces with arbitrarily large area, or there exist uncountably man…
Mean curvature flow with multiplicity 2 convergence in
Jingwen Chen, Ao Sun
We construct a new example of an immortal mean curvature flow of smooth embedded connected surfaces in , which converges to a plane with multiplicity as time appro…
Generic regularity of Level Set Flows with spherical singularity
Ao Sun, Jinxin Xue
The sphere is well-known as the only generic compact shrinker for mean curvature flow (MCF). In this paper, we characterize the generic dynamics of MCFs with a spherical singularit…