Qingfeng Sun

AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agent

Haipeng Luo, Huawen Feng, Qingfeng Sun +6

Large Reasoning Models (LRMs) like o3 and DeepSeek-R1 have achieved remarkable progress in reasoning tasks with long cot. However, they remain computationally inefficient and strug…

MMDialog: A Large-scale Multi-turn Dialogue Dataset Towards Multi-modal Open-domain Conversation

Jiazhan Feng, Qingfeng Sun, Can Xu +5

Responding with multi-modal content has been recognized as an essential capability for an intelligent conversational agent. In this paper, we introduce the MMDialog dataset to bett…

A subconvex bound for twisted $L$ -functions

Qingfeng Sun, Hui Wang

Let $q > 2$ be a prime number, $I ¨ $ a primitive Dirichlet character modulo $q$ and $f$ a primitive holomorphic cusp form or a Hecke-Maass cusp form of level $\m…

WarriorCoder: Learning from Expert Battles to Augment Code Large Language Models

Huawen Feng, Pu Zhao, Qingfeng Sun +8

Despite recent progress achieved by code large language models (LLMs), their remarkable abilities are largely dependent on fine-tuning on the high-quality data, posing challenges f…

OffSeeker: Online Reinforcement Learning Is Not All You Need for Deep Research Agents

Yuhang Zhou, Kai Zheng, Qiguang Chen +4

Deep research agents have shown remarkable potential in handling long-horizon tasks. However, state-of-the-art performance typically relies on online reinforcement learning (RL), w…

cs.AI2025

Zero Reinforcement Learning Towards General Domains

Yuyuan Zeng, Yufei Huang, Can Xu +5

Zero Reinforcement Learning (Zero-RL) has proven to be an effective approach for enhancing the reasoning capabilities of large language models (LLMs) by directly applying reinforce…

WizardLM: Empowering large pre-trained language models to follow complex instructions

Can Xu, Qingfeng Sun, Kai Zheng +6

Training large language models (LLMs) with open-domain instruction following data brings colossal success. However, manually creating such instruction data is very time-consuming a…

Mass Distribution for holomorphic cusp forms on the vertical geodesic

Qingfeng Sun, Qizhi Zhang

We compute the quantum variance of holomorphic cusp forms on the vertical geodesic for smooth compactly supported test functions. As an application we show that almost all holomorp…

math.NT2025

A Selberg-type zero-density result for twisted $G L_{2}$ $L$ -functions and its application

Qingfeng Sun, Hui Wang, Yanxue Yu

Let $f$ be a fixed holomorphic primitive cusp form of even weight $k$ , level $r$ and trivial nebentypus $I ¨ _{r}$ . Let $q$ be an odd prime with $(q, r) = 1$ and let $I ¨ $ be a primitive D…

Higher order divisor functions over values of mixed powers

Chenhao Du, Qingfeng Sun

Let $I ¨ _{k} (n)$ be the $k$ -th divisor function. In this paper, we derive an asymptotic formula for the sum $$ \sum_{1\leq n_1,n_2, \dots, n_{\ell}\leq X^{\frac{1}{r}} \atop 1\leq n_{…

math.NT2022

A bound for twists of $G L_{3} \times G L_{2}$ $L$ -functions with composite modulus

Qingfeng Sun, Yanxue Yu

Let $I ¨ $ be a Hecke-Maass cusp form for $S L_{3} (Z)$ and let $g$ be a holomorphic or Maass cusp form for $S L_{2} (Z)$ . Let $I ¨ $ be a primitive Dirichlet charact…

The Burgess bound via a trivial delta method

Keshav Aggarwal, Roman Holowinsky, Yongxiao Lin +1

Let $g$ be a fixed Hecke cusp form for $SL (2, Z)$ and $I ¨ $ be a primitive Dirichlet character of conductor $M$ . The best known subconvex bound for $L(1/2,g\otimes…

VeriEvol: Scaling Multimodal Mathematical Reasoning via Verifiable Evol-Instruct

Haoling Li, Kai Zheng, Jie Wu +4

Scaling reinforcement learning for visual mathematical reasoning requires more than generating harder questions: as data volume grows, the reward labels themselves must remain reli…

cond-mat.mtrl-sci2005

Detection of Single Electron Charging in an Individual InAs Quantum Dot by Noncontact Atomic Force Microscopy

Romain Stomp, Yoichi Miyahara, Sacha Schaer +6

Single electron charging in an individual InAs quantum dot was observed by electrostatic force measurements with an atomic force microscope (AFM). The resonant frequency shift and…

cs.CL2024

Arena Learning: Build Data Flywheel for LLMs Post-training via Simulated Chatbot Arena

Haipeng Luo, Qingfeng Sun, Can Xu +6

Assessing the effectiveness of large language models (LLMs) presents substantial challenges. The method of conducting human-annotated battles in an online Chatbot Arena is a highly…

math.NT2016

Shifted convolution sums of $G L_{3}$ cusp forms with $I^¸$ -series

Qingfeng Sun

Let $A_{f} (1, n)$ be the normalized Fourier coefficients of a Hecke-Maass cusp form $f$ for $S L_{3} (Z)$ and $$ r_3(n)=\#\left\{(n_1,n_2,n_3)\in \mathbb{Z}^3:n_1^2+n_2^2+n_3^2=n…

math.NT2026

On a level analog of Selberg's result on $S (t)$

Qingfeng Sun, Hui Wang

Let $S (t, f) = I ¨ ^{- 1} ar g L (1/2 + i t, f)$ , where $f$ is a holomorphic Hecke cusp form of weight $2$ and prime level $q$ . In this paper, we establish an unconditional asymptotic formula…

cond-mat.mes-hall2014

Transport discovery of emerging robust helical surface states in $Z_{2} = 0$ systems

Hua Jiang, Haiwen Liu, Ji Feng +2

We study the possibility of realizing robust helical surface states in $Z_{2} = 0$ systems. We find that the combination of anisotropy and finite-size confinement leads to the emergenc…

math.NT2016

A note on simultaneous nonvanishing of Dirichlet $L$ -functions and twists of Hecke-Maass $L$ -functions

Qingfeng Sun

In this note, we prove that given a Hecke-Maass cusp form $f$ for $S L_{2} (Z)$ and a sufficiently large integer $q = q_{1} q_{2}$ with $q_{j} ≍ q$ being prime numbers for $…

RubricBench: Aligning Model-Generated Rubrics with Human Standards

Qiyuan Zhang, Junyi Zhou, Yufei Wang +8

As Large Language Model (LLM) alignment evolves from simple completions to complex, highly sophisticated generation, Reward Models are increasingly shifting toward rubric-guided ev…

WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct

Haipeng Luo, Qingfeng Sun, Can Xu +8

Large language models (LLMs), such as GPT-4, have shown remarkable performance in natural language processing (NLP) tasks, including challenging mathematical reasoning. However, mo…

A zero-density estimate for $L$ -functions associated with $GL (3)$ Hecke--Maass cusp forms

Qingfeng Sun, Hui Wang

In this paper, we establish an asymptotic formula for the twisted second moment of $L$ -functions associated with Hecke--Maass cusp forms for $SL (3, Z)$ , and further ded…

cond-mat.mes-hall2006

Universal spin-Hall conductance fluctuations in two dimensions

Wei Ren, Zhenhua Qiao, Jian Wang +2

We report a theoretical investigation on spin-Hall conductance fluctuation of disordered four terminal devices in the presence of Rashba or/and Dresselhaus spin-orbital interaction…

math.NT2021

Uniform subconvex bounds for Rankin-Selberg $L$ -functions

Qingfeng Sun

Let $f$ be a Maass cusp form for $S L_{2} (Z)$ with Laplace eigenvalue $1/4 + I^¼_{f}^{2}$ , $I^¼_{f} > 0$ . Let $g$ be an arbitrary but fixed holomorphic or Maass cusp form for $\rm SL…

math.NT2021

Analytic twists of $G L_{3} \times G L_{2}$ automorphic forms

Yongxiao Lin, Qingfeng Sun

Let $I ¨ $ be a Hecke--Maass cusp form for $S L_{3} (Z)$ with normalized Hecke eigenvalues $I^»_{I ¨} (n, r)$ . Let $f$ be a holomorphic or Maass cusp form for $\rm SL_2(\mathbb{Z}…

Stylized Knowledge-Grounded Dialogue Generation via Disentangled Template Rewriting

Qingfeng Sun, Can Xu, Huang Hu +6

Current Knowledge-Grounded Dialogue Generation (KDG) models specialize in producing rational and factual responses. However, to establish long-term relationships with users, the KD…

math.NT2017

Averages of shifted convolution sums for $G L (3) \times G L (2)$

Qingfeng Sun

Let $A_{f} (1, n)$ be the normalized Fourier coefficients of a $G L (3)$ Maass cusp form $f$ and let $a_{g} (n)$ be the normalized Fourier coefficients of a $G L (2)$ cusp form $g$ . Let $Î»(n…

math.NT2015

Sums of the triple divisor function over values of a ternary quadratic form

Qingfeng Sun, Deyu Zhang

Let $I ¨ _{3} (n)$ be the triple divisor function which is the number of solutions of the equation $d_{1} d_{2} d_{3} = n$ in natural numbers. It is shown that $$ \sum_{1\leq n_1,n_2,n_3\leq \sqr…

math.NT2022

On some estimates involving Fourier coefficients of Maass cusp forms

Qingfeng Sun, Hui Wang

Let $f$ be a Hecke-Maass cusp form for $S L_{2} (Z)$ with Laplace eigenvalue $I^»_{f} (I^) = 1/4 + I^¼^{2}$ and let $I^»_{f} (n)$ be its $n$ -th normalized Fourier coefficient. It is prov…

A Bessel $I^´$ -method and hybrid bounds for $GL_{2}$

Yilan Fan, Qingfeng Sun

Let $g$ be a primitive holomorphic or Maass newform for $I^_{0} (D)$ . In this paper, by studying the Bessel integrals associated to $g$ , we prove an asymptotic Bessel $I^´$ -identity as…

math.NT2017

Hybrid bounds for twists of $G L (3)$ $L$ -functions

Qingfeng Sun

Let $I ¨ $ be a Hecke-Maass cusp form for $S L (3, Z)$ and $I ¨ = I ¨ _{1} I ¨ _{2}$ a Dirichlet character with $I ¨ _{i}$ primitive modulo $M_{i}$ . Suppose that $M_{1}$ , $M_{2}$ are primes such tha…

Bounds for $G L_{2} \times G L_{2}$ $L$ -functions in depth aspect

Qingfeng Sun

Let $f$ and $g$ be holomorphic or Maass cusp forms for $S L_{2} (Z)$ and let $I ¨ $ be a primitive Dirichlet character of prime power conductor $q = p^{I^} º$ with $p$ …

cond-mat.mes-hall2016

Majorana fermions in a superconducting Mobius strip

Yuan Pang, Jie Shen, Junhua Wang +12

Recently, much attention has been paid to search for Majorana fermions in solid-state systems. Among various proposals there is one based on radio-frequency superconducting quantum…

math.NT2026

On $G L_{3}$ Fourier coefficients over values of mixed powers

Qingfeng Sun, Yanxue Yu

Let $A_{I ¨} (n, 1)$ be the $(n, 1)$ -th Fourier coefficient of the Hecke-Maass cusp form $I ¨ $ for $S L_{3} (Z)$ and $I ¨ (x)$ be a smooth compactly supported function. In this pa…

cs.IR2023

Self-Supervised Multi-Modal Sequential Recommendation

Kunzhe Song, Qingfeng Sun, Can Xu +2

With the increasing development of e-commerce and online services, personalized recommendation systems have become crucial for enhancing user satisfaction and driving business reve…

Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models

Qiyuan Zhang, Yufei Wang, Tianhe Wu +5

Recent advancements in Generative Reward Models (GRMs) have demonstrated that scaling the length of Chain-of-Thought (CoT) reasoning considerably enhances the reliability of evalua…

cs.LG2026

STARE: Surprisal-Guided Token-Level Advantage Reweighting for Policy Entropy Stability

Haipeng Luo, Qingfeng Sun, Songli Wu +4

Reinforcement Learning with Verifiable Rewards algorithms like GRPO have emerged as the dominant post-training paradigm for complex reasoning in LLMs, yet commonly suffer from poli…

PromDA: Prompt-based Data Augmentation for Low-Resource NLU Tasks

Yufei Wang, Can Xu, Qingfeng Sun +4

This paper focuses on the Data Augmentation for low-resource Natural Language Understanding (NLU) tasks. We propose Prompt-based D}ata Augmentation model (PromDA) which only trains…

math.NT2018

Bounds for $G L_{3}$ $L$ -functions in depth aspect

Qingfeng Sun, Rui Zhao

Let $f$ be a Hecke-Maass cusp form for $S L_{3} (Z)$ and $I ¨ $ a primitive Dirichlet character of prime power conductor $q = p^{I^} º$ with $p$ prime and $I^º \geq 10$ . We p…

math.NT2025

On an unconditional $G L_{3}$ analog of Selberg's result

Qingfeng Sun, Hui Wang

Let $F$ be a Hecke--Maass cusp form for $SL_{3} (Z)$ with the Langlands parameter $I^¼_{F} = (I^¼_{F, 1}, I^¼_{F, 2}, I^¼_{F, 3})$ and the associated $L$ -function $L(s,…

WizardCoder: Empowering Code Large Language Models with Evol-Instruct

Ziyang Luo, Can Xu, Pu Zhao +7

Code Large Language Models (Code LLMs), such as StarCoder, have demonstrated exceptional performance in code-related tasks. However, most existing models are solely pre-trained on…

Hunyuan-TurboS: Advancing Large Language Models through Mamba-Transformer Synergy and Adaptive Chain-of-Thought

Tencent Hunyuan Team, Ao Liu, Botong Zhou +248

As Large Language Models (LLMs) rapidly advance, we introduce Hunyuan-TurboS, a novel large hybrid Transformer-Mamba Mixture of Experts (MoE) model. It synergistically combines Mam…

cond-mat2002

Parametric quantum spin pump

Wei Zheng, Junling Wu, Baigeng Wang +3

We investigate a non-adiabatic parametric quantum pump consists of a nonmagnetic scattering region connected by two ferromagnetic leads. The presence of ferromagnetic leads allows…

cs.IR2024

Towards a Unified Paradigm: Integrating Recommendation Systems as a New Language in Large Models

Kai Zheng, Qingfeng Sun, Can Xu +2

This paper explores the use of Large Language Models (LLMs) for sequential recommendation, which predicts users' future interactions based on their past behavior. We introduce a ne…

math.NT2021

Analytic twists of $G L_{2} \times G L_{2}$ automorphic forms

Bingrong Huang, Qingfeng Sun, Huimin Zhang

Let $f$ and $g$ be holomorphic or Maass cusp forms for $S L_{2} (Z)$ with normalized Fourier coefficients $I^»_{f} (n)$ and $I^»_{g} (n)$ , respectively. In this paper, we prove no…

math.NT2017

Shifted convolution sums involving theta series

Qingfeng Sun

Let $f$ be a cuspidal newform (holomorphic or Maass) of arbitrary level and nebentypus and denote by $I^»_{f} (n)$ its $n$ -th Hecke eigenvalue. Let $$ r(n)=\#\left\{(n_1,n_2)\in \mathb…

cond-mat.mes-hall2012

Shot noise of spin current and spin transfer torque

Yunjin Yu, Hongxin Zhan, Langhui Wan +4

We report the theoretical investigation of noise spectrum of spin current and spin transfer torque for non-colinear spin polarized transport in a spin-valve device which consists o…

AgentGen: Enhancing Planning Abilities for Large Language Model based Agent via Environment and Task Generation

Mengkang Hu, Pu Zhao, Can Xu +5

Large Language Model-based agents have garnered significant attention and are becoming increasingly popular. Furthermore, planning ability is a crucial component of an LLM-based ag…

Multimodal Dialogue Response Generation

Qingfeng Sun, Yujing Wang, Can Xu +7

Responsing with image has been recognized as an important capability for an intelligent conversational agent. Yet existing works only focus on exploring the multimodal dialogue mod…

math.NT2025

Equidistribution of holomorphic cusp forms on thin sets

Qingfeng Sun, Qizhi Zhang

We find some equidistribution results connected to restriction quantum unique ergodicity problem in this paper. We shows that \begin{align*} \frac{1}{|\mathcal{B}_k|}\sum_{f\in \ma…